1 2017-09-06 Wish Wu <wishwu007@gmail.com>
2 Jakub Jelinek <jakub@redhat.com>
4 * asan.c (initialize_sanitizer_builtins): Add
5 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
6 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
7 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8 BT_FN_VOID_UINT64_PTR variables.
9 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
10 (BT_FN_VOID_UINT16_UINT16): Likewise.
11 (BT_FN_VOID_UINT32_UINT32): Likewise.
12 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
13 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
14 (BT_FN_VOID_UINT64_PTR): Likewise.
15 * common.opt (flag_sanitize_coverage): New variable.
16 (fsanitize-coverage=trace-pc): Remove.
17 (fsanitize-coverage=): Add.
18 * flag-types.h (enum sanitize_coverage_code): New enum.
19 * fold-const.c (fold_range_test): Disable non-short-circuit
20 optimization if flag_sanitize_coverage.
21 (fold_truth_andor): Likewise.
22 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23 * opts.c (COVERAGE_SANITIZER_OPT): Define.
24 (coverage_sanitizer_opts): New array.
25 (get_closest_sanitizer_option): Add OPTS argument, handle also
26 OPT_fsanitize_coverage_.
27 (parse_sanitizer_options): Adjusted to also handle
28 OPT_fsanitize_coverage_.
29 (common_handle_option): Add OPT_fsanitize_coverage_.
30 * sancov.c (instrument_comparison, instrument_switch): New function.
31 (sancov_pass): Add trace-cmp support.
32 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
33 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
34 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
35 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
36 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
37 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
38 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
39 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
40 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
41 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
43 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
45 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
46 error. Only quit immediately if parsing is complete.
47 (BEGIN): Initialize fatal_err and parse_done.
48 (begin fpu, end fpu): Check number of arguments.
49 (begin arch, end arch): Likewise.
50 (begin cpu, end cpu): Likewise.
51 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
54 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
56 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
57 * config/arm/arm-isa.h: Delete. Move definitions to ...
58 * arm-cpus.in: ... here. Use new feature and fgroup values.
59 * config/arm/arm.c (arm_option_override): Use lower case for feature
61 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
62 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
63 * config/arm/parsecpu.awk (END): Add new command 'isa'.
65 (print_isa_bits_for): New function.
66 (gen_isa): New function.
67 (gen_comm_data): Use print_isa_bits_for.
68 (define feature): New keyword.
69 (define fgroup): New keyword.
70 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
71 (arm-isa.h): Add rule to generate file.
72 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
73 case for feature bit names.
75 2017-09-06 Richard Biener <rguenther@suse.de>
77 * tree-ssa-pre.c (NECESSARY): Remove.
78 (create_expression_by_pieces): Do not touch pass-local flags.
79 (insert_into_preds_of_block): Likewise.
80 (do_pre_regular_insertion): Likewise.
81 (eliminate_insert): Likewise.
82 (eliminate_dom_walker::before_dom_children): Likewise.
83 (fini_eliminate): Do not look at inserted_exprs.
84 (mark_operand_necessary): Remove.
85 (remove_dead_inserted_code): Replace with simple work-list
86 algorithm based on inserted_exprs and SSA uses.
87 (pass_pre::execute): Re-order fini_eliminate and
88 remove_dead_inserted_code.
90 2017-09-06 Olivier Hainque <hainque@adacore.com>
92 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
93 for VxWorks 7. Adjust surrounding comments.
95 2017-09-06 Richard Biener <rguenther@suse.de>
97 * gimple-ssa-strength-reduction.c
98 (find_candidates_dom_walker::before_dom_children): Also allow
101 2017-09-06 Richard Biener <rguenther@suse.de>
103 PR tree-optimization/82108
104 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
105 for gap in the non-permutation SLP case.
107 2017-09-06 Martin Jambor <mjambor@suse.cz>
109 PR tree-optimization/82078
110 * tree-sra.c (sort_and_splice_var_accesses): Move call to
111 add_access_to_work_queue...
112 (build_accesses_from_assign): ...here.
113 (propagate_all_subaccesses): Make sure racc is the group
114 representative, if there is one.
116 2017-09-06 Jakub Jelinek <jakub@redhat.com>
119 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
122 2017-09-06 Richard Biener <rguenther@suse.de>
124 * gimple-ssa-strength-reduction.c
125 (find_candidates_dom_walker::before_doom_children): Use a
126 type and not a mode check.
128 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
131 * config/arm/predicates.md (arm_general_adddi_operand): Create new
133 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
135 2017-09-05 Jeff Law <law@redhat.com>
137 PR tree-optimization/64910
138 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
139 cases where we have 3 or more operands.
141 2017-09-05 Jakub Jelinek <jakub@redhat.com>
144 * omp-low.c (lower_omp_for): Recompute tree invariant if
145 gimple_omp_for_initial/final is ADDR_EXPR.
148 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
149 into gimple val before gimplification fo the COND_EXPR.
151 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
153 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
154 REGION_COPY argument.
155 (thread_through_all_blocks): Remove unused argument to
156 duplicate_thread_path.
158 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
159 Alan Hayward <alan.hayward@arm.com>
160 David Sherwood <david.sherwood@arm.com>
162 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
163 Take a scalar_mode rather than a machine_mode.
164 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
165 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
166 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
167 (aarch64_gen_adjusted_ldpstp): Likewise.
168 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
170 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
171 Alan Hayward <alan.hayward@arm.com>
172 David Sherwood <david.sherwood@arm.com>
174 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
175 Take a scalar_int_mode instead of a machine_mode.
176 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
177 (aarch64_output_scalar_simd_mov_immediate): Likewise.
178 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
179 (aarch64_simd_attr_length_rglist): Delete.
180 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
181 a scalar_int_mode instead of a machine_mode.
182 (aarch64_add_offset): Likewise.
183 (aarch64_internal_mov_immediate): Likewise
184 (aarch64_add_constant_internal): Likewise.
185 (aarch64_add_constant): Likewise.
186 (aarch64_movw_imm): Likewise.
187 (aarch64_rtx_arith_op_extract_p): Likewise.
188 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
189 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
190 Remove assert that the mode isn't a vector.
191 (aarch64_output_scalar_simd_mov_immediate): Likewise.
192 (aarch64_expand_mov_immediate): Update calls after above changes.
193 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
194 (aarch64_and_bitmask_imm): Check for scalar integer modes.
195 (aarch64_move_imm): Likewise.
196 (aarch64_can_const_movi_rtx_p): Likewise.
197 (aarch64_strip_extend): Likewise.
198 (aarch64_extr_rtx_p): Likewise.
199 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
200 a CONST_INT when the mode parameter is VOIDmode.
201 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
203 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
205 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
206 * stor-layout.c (bitwise_mode_for_mode): Likewise.
207 (bitwise_type_for_mode): Update accordingly.
209 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
211 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
212 * stor-layout.c (mode_for_size_tree): Likewise.
213 (mode_for_array): Update accordingly.
214 (layout_decl): Likewise.
215 (compute_record_mode): Likewise. Only set the mode once.
217 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
219 * target.def (get_mask_mode): Change return type to opt_mode.
221 * doc/tm.texi: Regenerate.
222 * targhooks.h (default_get_mask_mode): Return an opt_mode.
223 * targhooks.c (default_get_mask_mode): Likewise.
224 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
225 * optabs-query.c (can_vec_mask_load_store_p): Update use of
226 targetm.get_mask_mode.
227 * tree.c (build_truth_vector_type): Likewise.
229 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
231 * machmode.h (mode_for_vector): Return an opt_mode.
232 * stor-layout.c (mode_for_vector): Likewise.
233 (mode_for_int_vector): Update accordingly.
234 (layout_type): Likewise.
235 * config/i386/i386.c (emit_memmov): Likewise.
236 (ix86_expand_set_or_movmem): Likewise.
237 (ix86_expand_vector_init): Likewise.
238 (ix86_get_mask_mode): Likewise.
239 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
241 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
242 * expmed.c (extract_bit_field_1): Likewise.
243 * expr.c (expand_expr_real_2): Likewise.
244 * optabs-query.c (can_vec_perm_p): Likewise.
245 (can_vec_mask_load_store_p): Likewise.
246 * optabs.c (expand_vec_perm): Likewise.
247 * targhooks.c (default_get_mask_mode): Likewise.
248 * tree-vect-stmts.c (vectorizable_store): Likewise.
249 (vectorizable_load): Likewise.
250 (get_vectype_for_scalar_type_and_size): Likewise.
252 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
254 * machmode.h (mode_for_int_vector): New function.
255 * stor-layout.c (mode_for_int_vector): Likewise.
256 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
257 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
258 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
259 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
260 (s390_expand_vcond): Likewise.
262 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
264 * machmode.h (opt_machine_mode): New type.
265 (opt_mode<T>): Allow construction from anything that can be
267 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
268 (mode_for_size): Return an opt_machine_mode.
269 * stor-layout.c (mode_for_size): Likewise.
270 (mode_for_size_tree): Update call accordingly.
271 (bitwise_mode_for_mode): Likewise.
272 (make_fract_type): Likewise.
273 (make_accum_type): Likewise.
274 * caller-save.c (replace_reg_with_saved_mem): Update call
276 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
277 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
278 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
279 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
280 * expmed.c (extract_bit_field_1): Likewise.
281 * reload.c (get_secondary_mem): Likewise.
282 * varasm.c (assemble_integer): Likewise.
283 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
286 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
288 * machmode.h (decimal_float_mode_for_size): New function.
289 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
290 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
291 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
292 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
293 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
294 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
296 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
298 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
299 (get_builtin_sync_mode): Likewise.
300 (expand_ifn_atomic_compare_exchange): Likewise.
301 (expand_builtin_atomic_clear): Likewise.
302 (expand_builtin_atomic_test_and_set): Likewise.
303 (fold_builtin_atomic_always_lock_free): Likewise.
304 * calls.c (compute_argument_addresses): Likewise.
305 (emit_library_call_value_1): Likewise.
306 (store_one_arg): Likewise.
307 * combine.c (combine_instructions): Likewise.
308 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
309 * config/arm/arm.c (arm_function_value): Likewise.
310 (aapcs_allocate_return_reg): Likewise.
311 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
312 * config/i386/i386.c (construct_container): Likewise.
313 (ix86_gimplify_va_arg): Likewise.
314 (ix86_expand_sse_cmp): Likewise.
315 (emit_memmov): Likewise.
316 (emit_memset): Likewise.
317 (expand_small_movmem_or_setmem): Likewise.
318 (ix86_expand_pextr): Likewise.
319 (ix86_expand_pinsr): Likewise.
320 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
321 * config/microblaze/microblaze.c (microblaze_block_move_straight):
323 * config/mips/mips.c (mips_function_value_1) Likewise.
324 (mips_block_move_straight): Likewise.
325 (mips_expand_ins_as_unaligned_store): Likewise.
326 * config/powerpcspe/powerpcspe.c
327 (rs6000_darwin64_record_arg_advance_flush): Likewise.
328 (rs6000_darwin64_record_arg_flush): Likewise.
329 * config/rs6000/rs6000.c
330 (rs6000_darwin64_record_arg_advance_flush): Likewise.
331 (rs6000_darwin64_record_arg_flush): Likewise.
332 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
333 (sparc_function_value_1): Likewise.
334 * config/spu/spu.c (adjust_operand): Likewise.
335 (spu_emit_branch_or_set): Likewise.
336 (arith_immediate_p): Likewise.
337 * emit-rtl.c (gen_lowpart_common): Likewise.
338 * expr.c (expand_expr_real_1): Likewise.
339 * function.c (assign_parm_setup_block): Likewise.
340 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
341 * reload1.c (alter_reg): Likewise.
342 * stor-layout.c (mode_for_vector): Likewise.
343 (layout_type): Likewise.
345 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
347 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
348 (spu_convert_move): Likewise.
349 * lower-subreg.c (resolve_simple_move): Likewise.
351 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
354 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
355 define_insn to a define_expand.
356 (altivec_vsum2sws_direct): New define_insn.
357 (altivec_vsumsws): Convert from a define_insn to a define_expand.
359 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
361 * config/arm/arm.c (arm_option_params_internal): Improve setting of
364 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
369 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
370 realign stack if stack alignment needed is less than incoming
373 2017-09-05 Marek Polacek <polacek@redhat.com>
376 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
379 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
381 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
383 2017-09-05 Richard Biener <rguenther@suse.de>
385 PR tree-optimization/82084
386 * fold-const.c (can_native_encode_string_p): Handle wide characters.
388 2017-09-05 Richard Biener <rguenther@suse.de>
390 PR tree-optimization/82102
391 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
393 2017-09-05 Martin Liska <mliska@suse.cz>
395 PR tree-optimization/82032
396 * tree-cfg.c (generate_range_test): New function.
397 * tree-cfg.h (generate_range_test): Declared here.
398 * tree-cfgcleanup.c (convert_single_case_switch): New function.
399 (cleanup_control_expr_graph): Use it.
400 * tree-switch-conversion.c (try_switch_expansion): Remove
402 (emit_case_nodes): Use generate_range_test.
404 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
407 * config/i386/i386.md (*<btsc><mode>_mask): Add
408 TARGET_USE_BT to insn constraint.
409 (*btr<mode>_mask): Ditto.
411 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
413 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
414 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
416 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
419 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
420 TARGET_NEON and TARGET_IWMMXT.
421 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
422 TARGET_NEON and TARGET_IWMMXT.
423 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
425 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
427 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
428 (ix86_rewrite_tls_address): Ditto.
429 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
430 (ix86_rewrite_tls_address_1): Ditto.
431 (ix86_rewrite_tls_address): Ditto.
432 * config/i386/predicates.md (tls_address_pattern): New predicate.
433 * config/i386/i386.md (TLS address splitter): New splitter.
435 2017-09-04 Richard Biener <rguenther@suse.de>
437 PR tree-optimization/82084
438 * fold-const.h (can_native_encode_string_p): Declare.
439 * fold-const.c (can_native_encode_string_p): Factor out from ...
440 (native_encode_string): ... here.
441 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
442 vectorizing stores from constants we later cannot handle.
444 2017-09-04 Marek Polacek <polacek@redhat.com>
447 * doc/invoke.texi: Update -Wtautological-compare documentation.
449 2017-09-04 Jeff Law <law@redhat.com>
451 PR tree-optimization/64910
452 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
453 swap the first and last operand if the last is a constant.
455 2017-09-04 Marek Polacek <polacek@redhat.com>
458 * convert.c (do_narrow): When sanitizing signed integer overflows,
459 bail out for signed types.
460 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
462 2017-09-04 Richard Biener <rguenther@suse.de>
464 PR tree-optimization/82060
465 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
466 Move devirtualization after stmt folding and before EH/AB/noreturn
467 cleanup to get the stmt refs canonicalized. Use a bool instead
468 of gimple_modified_p since that doesn't work for NOPs. Schedule
469 NOPs generated by folding for removal.
471 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
472 Alan Hayward <alan.hayward@arm.com>
473 David Sherwood <david.sherwood@arm.com>
475 * coretypes.h (pad_direction): New enum.
476 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
477 (FUNCTION_ARG_PADDING): Likewise.
478 * target.def (function_arg_padding): New hook.
479 * targhooks.h (default_function_arg_padding): Declare.
480 * targhooks.c (default_function_arg_padding): New function.
481 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
482 (TARGET_FUNCTION_ARG_PADDING): ...this.
483 * doc/tm.texi: Regenerate.
484 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
485 instead of direction.
486 (compute_argument_addresses): Likewise.
487 (load_register_parameters): Likewise.
488 (emit_library_call_value_1): Likewise.
489 (store_one_arg): Use targetm.calls.function_arg_padding instead
490 of FUNCTION_ARG_PADDING.
491 (must_pass_in_stack_var_size_or_pad): Likewise.
492 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
493 (emit_group_store): Likewise.
494 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
495 instead of FUNCTION_ARG_PADDING.
496 (emit_push_insn): Likewise, and propagate enum change throughout
498 * function.h (direction): Delete.
499 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
501 * function.c (assign_parm_find_stack_rtl): Likewise.
502 (assign_parm_setup_block_p): Likewise.
503 (assign_parm_setup_block): Likewise.
504 (gimplify_parameters): Likewise.
505 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
506 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
508 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
509 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
510 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
511 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
512 (aarch64_function_arg_padding): ...this new function.
513 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
514 (TARGET_FUNCTION_ARG_PADDING): Redefine.
515 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
516 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
517 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
518 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
519 (arm_pad_arg_upward): Replace with...
520 (arm_function_arg_padding): ...this new function.
521 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
523 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
524 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
525 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
526 (ia64_hpux_function_arg_padding): Replace with...
527 (ia64_function_arg_padding): ...this new function. Use pad_direction
528 instead of direction. Check for TARGET_HPUX.
529 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
530 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
531 (iq2000_function_arg_padding): New function.
532 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
533 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
534 (mips_function_arg_padding): ...this new function.
535 (mips_pad_reg_upward): Update accordingly.
536 (TARGET_FUNCTION_ARG_PADDING): Redefine.
537 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
538 targetm.calls.function_arg_padding.
539 (FUNCTION_ARG_PADDING): Delete.
540 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
541 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
542 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
543 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
544 (nios2_block_reg_padding): Return pad_direction instead of direction.
545 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
546 instead of direction.
547 (nios2_function_arg_padding): Likewise. Make static.
548 (TARGET_FUNCTION_ARG_PADDING): Redefine.
549 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
550 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
551 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
552 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
553 (pa_function_arg_padding): Make static. Return pad_direction instead
555 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
556 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
557 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
558 instead of direction. Use targetm.calls.function_arg_padding.
559 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
560 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
561 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
562 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
563 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
565 (function_arg_padding): Rename to...
566 (rs6000_function_arg_padding): ...this. Make static. Return
567 pad_direction instead of direction.
568 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
569 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
570 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
571 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
572 instead of direction. Use targetm.calls.function_arg_padding.
573 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
574 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
575 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
576 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
577 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
578 (function_arg_padding): Rename to...
579 (rs6000_function_arg_padding): ...this. Make static. Return
580 pad_direction instead of direction.
581 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
582 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
583 * config/s390/s390.c (s390_function_arg_padding): New function.
584 (TARGET_FUNCTION_ARG_PADDING): Redefine.
585 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
586 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
587 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
588 (function_arg_padding): Rename to...
589 (sparc_function_arg_padding): ...this. Make static. Return
590 pad_direction instead of direction.
591 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
592 * config/spu/spu.c (spu_function_arg_padding): New function.
593 (TARGET_FUNCTION_ARG_PADDING): Redefine.
594 * system.h (FUNCTION_ARG_PADDING): Poison.
596 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
597 Alan Hayward <alan.hayward@arm.com>
598 David Sherwood <david.sherwood@arm.com>
600 * target.def (modes_tieable_p): New hook.
601 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
602 (TARGET_MODES_TIEABLE_P): ...this.
603 * doc/tm.texi.in: Regenerate.
604 * hooks.h (hook_bool_mode_mode_true): Declare.
605 * hooks.c (hook_bool_mode_mode_true): New function.
606 * combine.c (subst): Use targetm.modes_tieable_p instead of
608 * dse.c (find_shift_sequence): Likewise.
609 * expmed.c (extract_low_bits): Likewise.
610 * lower-subreg.c: Include target.h.
611 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
613 * rtlanal.c (rtx_cost): Likewise.
614 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
615 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
616 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
617 (TARGET_MODES_TIEABLE_P): Redefine.
618 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
619 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
620 (TARGET_MODES_TIEABLE_P): Redefine.
621 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
622 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
623 (arc_modes_tieable_p): New function.
624 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
625 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
626 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
627 (arm_modes_tieable_p): Make static.
628 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
629 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
630 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
631 (TARGET_MODES_TIEABLE_P): Redefine.
632 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
633 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
634 (TARGET_MODES_TIEABLE_P): Redefine.
635 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
636 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
637 (cr16_modes_tieable_p): New function.
638 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
639 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
640 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
641 (TRULY_NOOP_TRUNCATION): Update comment.
642 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
643 (TRULY_NOOP_TRUNCATION): Update comment.
644 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
645 (frv_modes_tieable_p): New function.
646 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
647 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
648 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
649 (TARGET_MODES_TIEABLE_P): Redefine.
650 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
651 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
652 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
653 (TARGET_MODES_TIEABLE_P): Redefine.
654 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
655 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
656 (ia64_modes_tieable_p): New function.
657 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
658 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
659 (iq2000_modes_tieable_p): New function.
660 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
661 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
662 (lm32_modes_tieable_p): New function.
663 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
664 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
665 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
666 (TARGET_MODES_TIEABLE_P): Redefine.
667 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
668 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
669 (m32r_modes_tieable_p): New function.
670 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
671 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
672 (m68k_modes_tieable_p): New function.
673 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
674 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
675 (mcore_modes_tieable_p): New function.
676 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
677 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
679 (TARGET_MODES_TIEABLE_P): Redefine.
680 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
681 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
682 * config/mips/mips.c (mips_modes_tieable_p): Make static.
683 (TARGET_MODES_TIEABLE_P): Redefine.
684 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
685 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
686 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
687 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
688 (mn10300_modes_tieable_p): ...this and make static.
689 (TARGET_MODES_TIEABLE_P): Redefine.
690 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
691 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
692 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
693 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
694 (msp430_modes_tieable_p): Make static.
695 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
696 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
697 (TARGET_MODES_TIEABLE_P): Redefine.
698 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
699 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
700 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
701 (TARGET_MODES_TIEABLE_P): Redefine.
702 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
703 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
704 * config/pa/pa.c (pa_modes_tieable_p): Make static.
705 (TARGET_MODES_TIEABLE_P): Redefine.
706 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
707 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
708 (pdp11_modes_tieable_p): New function.
709 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
710 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
711 (rs6000_modes_tieable_p): New function.
712 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
713 * config/powerpcspe/powerpcspe.md: Update comment.
714 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
715 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
716 (TARGET_MODES_TIEABLE_P): Redefine.
717 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
718 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
719 (rl78_modes_tieable_p): New function.
720 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
721 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
722 (rs6000_modes_tieable_p): New function.
723 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
724 * config/rs6000/rs6000.md: Update comment.
725 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
726 * config/rx/rx.c (rx_modes_tieable_p): New function.
727 (TARGET_MODES_TIEABLE_P): Redefine.
728 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
729 * config/s390/s390.c (s390_modes_tieable_p): New function.
730 (TARGET_MODES_TIEABLE_P): Redefine.
731 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
732 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
733 (sh_modes_tieable_p): New function.
734 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
735 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
736 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
737 (sparc_modes_tieable_p): Make static.
738 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
739 * config/spu/spu.c (spu_modes_tieable_p): New function.
740 (TARGET_MODES_TIEABLE_P): Redefine.
741 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
742 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
743 (TARGET_MODES_TIEABLE_P): Redefine.
744 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
745 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
746 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
747 * config/v850/v850.c (v850_modes_tieable_p): New function.
748 (TARGET_MODES_TIEABLE_P): Redefine.
749 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
750 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
751 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
752 (visium_modes_tieable_p): New function.
753 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
754 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
755 (xtensa_modes_tieable_p): New function.
756 * system.h (MODES_TIEABLE_P): Poison.
758 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
759 Alan Hayward <alan.hayward@arm.com>
760 David Sherwood <david.sherwood@arm.com>
762 * target.def (hard_regno_mode_ok): New hook.
763 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
764 (TARGET_HARD_REGNO_MODE_OK): ...this.
765 * doc/tm.texi.in: Regenerate.
766 * hooks.h (hook_bool_uint_mode_true): Declare.
767 * hooks.c (hook_bool_uint_mode_true): New function.
768 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
770 * genpreds.c (write_insn_preds_c): Add an include of target.h.
771 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
772 instead of HARD_REGNO_MODE_OK.
773 * caller-save.c: Include target.h.
774 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
776 * combine.c (can_combine_p): Likewise.
777 (combinable_i3pat): Likewise.
778 (can_change_dest_mode): Likewise.
779 * expr.c (init_expr_target): Likewise.
780 (convert_move): Likewise.
781 (convert_modes): Likewise.
782 * ira.c (setup_prohibited_class_mode_regs): Likewise.
783 (setup_prohibited_mode_move_regs): Likewise.
784 * ira.h (target_ira): Likewise.
785 * lra-assigns.c (find_hard_regno_for_1): Likewise.
786 * lra-constraints.c (process_alt_operands): Likewise.
787 (split_reg): Likewise.
788 * recog.c (peep2_find_free_register): Likewise.
789 * ree.c (combine_reaching_defs): Likewise.
790 * regcprop.c (maybe_mode_change): Likewise.
791 * reginfo.c (init_reg_sets_1): Likewise.
792 (choose_hard_reg_mode): Likewise.
793 (simplifiable_subregs): Likewise.
794 * regrename.c (check_new_reg_p): Likewise.
795 * reload.c (find_valid_class): Likewise.
796 (find_valid_class_1): Likewise.
797 (reload_inner_reg_of_subreg): Likewise.
798 (push_reload): Likewise.
799 (combine_reloads): Likewise.
800 (find_dummy_reload): Likewise.
801 (find_reloads): Likewise.
802 * reload1.c (find_reg): Likewise.
803 (set_reload_reg): Likewise.
804 (allocate_reload_reg): Likewise.
805 (choose_reload_regs): Likewise.
806 (reload_adjust_reg_for_temp): Likewise.
807 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
808 (simplify_subreg_regno): Likewise.
809 * sel-sched.c (init_regs_for_mode): Likewise.
810 * varasm.c (make_decl_rtl): Likewise.
811 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
812 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
814 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
815 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
816 (TARGET_HARD_REGNO_MODE_OK): Redefine.
817 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
818 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
819 (TARGET_HARD_REGNO_MODE_OK): Redefine.
820 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
821 (arc_mode_class): Delete.
822 (HARD_REGNO_MODE_OK): Delete.
823 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
824 (arc_hard_regno_mode_ok): Rename old array to...
825 (arc_hard_regno_mode_ok_modes): ...this.
826 (arc_conditional_register_usage): Update accordingly.
827 (arc_mode_class): Make static.
828 (arc_hard_regno_mode_ok): New function.
829 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
830 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
831 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
832 (arm_hard_regno_mode_ok): Make static.
833 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
835 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
836 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
837 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
839 (TARGET_HARD_REGNO_MODE_OK): Redefine.
840 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
841 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
842 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
843 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
844 (TARGET_HARD_REGNO_MODE_OK): Redefine.
845 * config/bfin/predicates.md (valid_reg_operand): Use
846 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
847 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
848 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
849 (TARGET_HARD_REGNO_MODE_OK): Redefine.
850 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
851 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
852 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
853 (cr16_hard_regno_mode_ok): Make static and return a bool.
854 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
855 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
856 (cris_hard_regno_mode_ok): New function.
857 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
858 (epiphany_mode_class): Delete.
859 (HARD_REGNO_MODE_OK): Delete.
860 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
861 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
862 (hard_regno_mode_ok): Rename to...
863 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
864 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
865 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
867 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
868 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
869 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
870 (frv_hard_regno_mode_ok): Make static and return a bool.
871 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
873 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
874 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
875 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
876 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
878 (TARGET_HARD_REGNO_MODE_OK): Redefine.
879 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
880 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
881 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
883 (TARGET_HARD_REGNO_MODE_OK): Redefine.
884 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
885 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
886 (ia64_hard_regno_mode_ok): New function.
887 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
888 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
889 (iq2000_hard_regno_mode_ok): New function.
890 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
891 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
892 (lm32_hard_regno_mode_ok): New function.
893 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
894 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
895 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
896 instead of HARD_REGNO_MODE_OK.
897 (m32c_hard_regno_ok): Rename to...
898 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
899 (m32c_cannot_change_mode_class): Update accordingly.
900 (TARGET_HARD_REGNO_MODE_OK): Redefine.
901 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
902 (m32r_mode_class): Delete.
903 (HARD_REGNO_MODE_OK): Delete.
904 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
905 (m32r_hard_regno_mode_ok): Rename to...
906 (m32r_hard_regno_modes): ...this.
907 (m32r_mode_class): Make static.
908 (m32r_hard_regno_mode_ok): New function.
909 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
910 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
911 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
912 (m68k_hard_regno_mode_ok): Make static.
913 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
914 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
915 (mcore_hard_regno_mode_ok): New function.
916 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
917 (HARD_REGNO_MODE_OK): Delete.
918 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
920 (microblaze_hard_regno_mode_ok_p): ...this and make static.
921 (microblaze_hard_regno_mode_ok): New function.
922 (TARGET_HARD_REGNO_MODE_OK): Redefine.
923 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
924 (mips_hard_regno_mode_ok): Delete.
925 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
926 (mips_hard_regno_mode_ok_p): ...this and make static.
927 (mips_hard_regno_mode_ok_p): Rename to...
928 (mips_hard_regno_mode_ok_uncached): ...this.
929 (mips_hard_regno_mode_ok): New function.
930 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
931 of HARD_REGNO_MODE_OK.
932 (mips_option_override): Update after above name changes.
933 (TARGET_HARD_REGNO_MODE_OK): Redefine.
934 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
935 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
936 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
937 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
938 (TARGET_HARD_REGNO_MODE_OK): Redefine.
939 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
940 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
941 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
942 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
943 (msp430_hard_regno_mode_ok): Make static and return a bool.
944 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
945 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
946 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
948 (TARGET_HARD_REGNO_MODE_OK): Redefine.
949 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
950 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
951 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
952 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
953 (PA_HARD_REGNO_MODE_OK): ...this
954 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
955 (PA_HARD_REGNO_MODE_OK): ...this.
956 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
957 (pa_hard_regno_mode_ok): New function.
958 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
959 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
960 (pdp11_hard_regno_mode_ok): New function.
961 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
962 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
964 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
966 (TARGET_HARD_REGNO_MODE_OK): Redefine.
967 (rs6000_hard_regno_mode_ok): Rename to...
968 (rs6000_hard_regno_mode_ok_uncached): ...this.
969 (rs6000_init_hard_regno_mode_ok): Update accordingly.
970 (rs6000_hard_regno_mode_ok): New function.
971 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
972 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
973 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
974 (riscv_hard_regno_mode_ok): ...this and make static.
975 (TARGET_HARD_REGNO_MODE_OK): Redefine.
976 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
977 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
978 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
979 (rl78_hard_regno_mode_ok): Make static and return bool.
980 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
981 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
983 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
984 (TARGET_HARD_REGNO_MODE_OK): Redefine.
985 (rs6000_hard_regno_mode_ok): Rename to...
986 (rs6000_hard_regno_mode_ok_uncached): ...this.
987 (rs6000_init_hard_regno_mode_ok): Update accordingly.
988 (rs6000_hard_regno_mode_ok): New function.
989 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
990 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
991 (TARGET_HARD_REGNO_MODE_OK): Redefine.
992 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
993 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
994 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
995 (TARGET_HARD_REGNO_MODE_OK): Redefine.
996 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
997 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
998 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
999 (sh_hard_regno_mode_ok): Make static.
1000 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
1001 instead of HARD_REGNO_MODE_OK.
1002 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
1003 (sparc_mode_class): Delete.
1004 (HARD_REGNO_MODE_OK): Delete.
1005 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1006 (hard_regno_mode_classes): Make static.
1007 (sparc_mode_class): Likewise.
1008 (sparc_hard_regno_mode_ok): New function.
1009 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
1010 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
1011 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
1013 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1014 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
1015 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
1016 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
1017 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
1018 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1019 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
1020 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
1021 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1022 (visium_hard_regno_mode_ok): New function.
1023 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
1024 instead of HARD_REGNO_MODE_OK.
1025 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
1026 (HARD_REGNO_MODE_OK): Delete.
1027 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
1028 (xtensa_hard_regno_mode_ok_p): ...this and make static.
1029 (xtensa_option_override): Update accordingly.
1030 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1031 (xtensa_hard_regno_mode_ok): New function.
1032 * system.h (HARD_REGNO_MODE_OK): Poison.
1034 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1035 Alan Hayward <alan.hayward@arm.com>
1036 David Sherwood <david.sherwood@arm.com>
1038 * target.def (hard_regno_call_part_clobbered): New hook.
1039 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
1040 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
1041 * doc/tm.texi: Regenerate.
1042 * hooks.h (hook_bool_uint_mode_false): Declare.
1043 * hooks.c (hook_bool_uint_mode_false): New function.
1044 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1045 * cselib.c (cselib_process_insn): Use
1046 targetm.hard_regno_call_part_clobbered instead of
1047 HARD_REGNO_CALL_PART_CLOBBERED.
1048 * ira-conflicts.c (ira_build_conflicts): Likewise.
1049 * ira-costs.c (ira_tune_allocno_costs): Likewise.
1050 * lra-constraints.c (need_for_call_save_p): Likewise.
1051 * lra-lives.c: Include target.h.
1052 (check_pseudos_live_through_calls): Use
1053 targetm.hard_regno_call_part_clobbered instead of
1054 HARD_REGNO_CALL_PART_CLOBBERED.
1055 * regcprop.c: Include target.h.
1056 (copyprop_hardreg_forward_1): Use
1057 targetm.hard_regno_call_part_clobbered instead of
1058 HARD_REGNO_CALL_PART_CLOBBERED.
1059 * reginfo.c (choose_hard_reg_mode): Likewise.
1060 * regrename.c (check_new_reg_p): Likewise.
1061 * reload.c (find_equiv_reg): Likewise.
1062 * reload1.c (emit_reload_insns): Likewise.
1063 * sched-deps.c (deps_analyze_insn): Likewise.
1064 * sel-sched.c (init_regs_for_mode): Likewise.
1065 (mark_unavailable_hard_regs): Likewise.
1066 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
1067 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1068 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
1070 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1071 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1072 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
1074 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
1076 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1077 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1078 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
1080 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1081 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1082 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
1084 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1085 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
1087 * config/powerpcspe/powerpcspe.c
1088 (rs6000_hard_regno_call_part_clobbered): New function.
1089 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1090 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1091 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
1093 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1094 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1095 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
1097 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1098 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1099 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
1101 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1102 Alan Hayward <alan.hayward@arm.com>
1103 David Sherwood <david.sherwood@arm.com>
1105 * rtl.h (subreg_memory_offset): Declare.
1106 * emit-rtl.c (subreg_memory_offset): New function.
1107 * expmed.c (store_bit_field_1): Use it.
1108 * expr.c (undefined_operand_subword_p): Likewise.
1109 * simplify-rtx.c (simplify_subreg): Likewise.
1111 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
1113 PR rtl-optimization/57448
1116 * optabs.c (expand_atomic_load): Place compiler memory barriers if
1117 using atomic_load pattern.
1118 (expand_atomic_store): Likewise.
1120 2017-09-04 Jakub Jelinek <jakub@redhat.com>
1123 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
1124 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
1125 handling. Use replace_call_with_value with NULL instead of
1126 gsi_replace, unlink_stmt_vdef and release_defs.
1128 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
1131 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
1133 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1136 * rtl.h (emit_library_call_value_1): Declare.
1137 (emit_library_call): Replace declaration with a series of overloads.
1138 Remove the parameter count argument.
1139 (emit_library_call_value): Likewise.
1140 * calls.c (emit_library_call_value_1): Make global. Replace varargs
1141 with an "rtx_mode_t *".
1142 (emit_library_call_value): Delete.
1143 (emit_library_call): Likewise.
1144 * asan.c (asan_emit_stack_protection): Update calls accordingly.
1145 (asan_emit_allocas_unpoison): Likewise.
1146 * builtins.c (expand_builtin_powi): Likewise.
1147 (expand_asan_emit_allocas_unpoison): Likewise.
1148 * cfgexpand.c (expand_main_function): Likewise.
1149 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
1150 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
1151 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
1152 * config/arm/arm.c (arm_trampoline_init): Likewise.
1153 (arm_call_tls_get_addr): Likewise.
1154 (arm_expand_divmod_libfunc): Likewise.
1155 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
1156 (smulsi3_highpart): Likewise.
1157 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
1158 (c6x_expand_compare): Likewise.
1159 (c6x_expand_movmem): Likewise.
1160 * config/frv/frv.c (frv_trampoline_init): Likewise.
1161 * config/i386/i386.c (ix86_trampoline_init): Likewise.
1162 (ix86_expand_divmod_libfunc): Likewise.
1163 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
1164 (ia64_expand_compare): Likewise.
1165 (ia64_profile_hook): Likewise.
1166 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
1167 (nonlocal_goto): Likewise.
1168 (restore_stack_nonlocal): Likewise.
1169 * config/m32r/m32r.c (block_move_call): Likewise.
1170 (m32r_trampoline_init): Likewise.
1171 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
1172 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
1173 (m68k_call_m68k_read_tp): Likewise.
1174 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
1175 (microblaze_expand_divide): Likewise.
1176 * config/mips/mips.h (mips_args): Likewise.
1177 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
1178 (MIPS_ICACHE_SYNC): Likewise.
1179 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
1180 (nios2_trampoline_init): Likewise.
1181 * config/pa/pa.c (hppa_tls_call): Likewise.
1182 (pa_trampoline_init): Likewise.
1183 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
1184 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
1185 (expand_strn_compare): Likewise.
1186 (rs6000_generate_compare): Likewise.
1187 (rs6000_expand_float128_convert): Likewise.
1188 (output_profile_hook): Likewise.
1189 (rs6000_trampoline_init): Likewise.
1190 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
1191 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
1192 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
1193 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
1194 (rs6000_generate_compare): Likewise.
1195 (rs6000_expand_float128_convert): Likewise.
1196 (output_profile_hook): Likewise.
1197 (rs6000_trampoline_init): Likewise.
1198 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
1199 * config/sh/sh.c (sh_trampoline_init): Likewise.
1200 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
1201 (sparc_emit_float_lib_cmp): Likewise.
1202 (sparc32_initialize_trampoline): Likewise.
1203 (sparc64_initialize_trampoline): Likewise.
1204 (sparc_profile_hook): Likewise.
1205 * config/spu/spu.c (ea_load_store): Likewise.
1206 * config/spu/spu.md (floatunssidf2): Likewise.
1207 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
1208 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
1209 * config/visium/visium.c (expand_block_move_4): Likewise.
1210 (expand_block_move_2): Likewise.
1211 (expand_block_move_1): Likewise.
1212 (expand_block_set_4): Likewise.
1213 (expand_block_set_2): Likewise.
1214 (expand_block_set_1): Likewise.
1215 (visium_trampoline_init): Likewise.
1216 (visium_profile_hook): Likewise.
1217 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
1218 (xtensa_setup_frame_addresses): Likewise.
1219 (xtensa_trampoline_init): Likewise.
1220 * except.c (sjlj_emit_function_enter): Likewise.
1221 (sjlj_emit_function_exit): Likewise.
1222 * explow.c (allocate_dynamic_stack_space): Likewise.
1223 (probe_stack_range): Likewise.
1224 * expr.c (convert_mode_scalar): Likewise.
1225 * optabs.c (expand_binop): Likewise.
1226 (expand_twoval_binop_libfunc): Likewise.
1227 (expand_unop): Likewise.
1228 (prepare_cmp_insn): Likewise.
1229 (prepare_float_lib_cmp): Likewise.
1230 (expand_float): Likewise.
1231 (expand_fix): Likewise.
1232 (expand_fixed_convert): Likewise.
1233 (maybe_emit_sync_lock_test_and_set): Likewise.
1234 (expand_atomic_compare_and_swap): Likewise.
1235 (expand_mem_thread_fence): Likewise.
1236 (expand_atomic_fetch_op): Likewise.
1238 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
1240 * doc/generic.texi (OpenACC): Adjust URL.
1241 * doc/invoke.texi (C Dialect Options): Ditto.
1243 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
1245 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
1246 predicate for operand 1. Add (m,<S>) constraint.
1247 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
1248 Prevent memory operand 1 with register operand 2.
1250 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
1252 PR rtl-optimization/82024
1253 * combine.c (try_combine): If the combination result is a PARALLEL,
1254 and we only need to retain the SET in there that would be placed
1255 at I2, check that we can place that at I3 instead, before doing so.
1257 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1260 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
1262 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
1263 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
1266 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
1267 Jeff Law <law@redhat.com>
1269 * varasm.c (bss_initializer_p): Do not put constants into .bss
1270 (categorize_decl_for_section): Handle bss_initializer_p returning
1271 false when DECL_INITIAL is NULL.
1273 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1276 * config/s390/s390.c (s390_can_inline_p): New function.
1278 2017-09-01 Jeff Law <law@redhat.com>
1280 PR tree-optimization/82052
1281 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
1282 Always initialize the returned slot after a hash table miss
1283 when INSERT is true.
1285 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
1287 * config/s390/s390.md (mem_signal_fence): Remove.
1288 * doc/md.texi (mem_signal_fence): Remove.
1289 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
1291 * target-insns.def (mem_signal_fence): Remove.
1293 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1296 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
1299 * asan.c (create_odr_indicator): Strip name encoding from assembler
1300 name before appending it after __odr_asan_.
1302 2017-09-01 Martin Liska <mliska@suse.cz>
1304 PR tree-optimization/82059
1305 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
1306 frequency only when an edge is redirected.
1308 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1310 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
1311 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
1312 (arc_conditional_register_usage): Remove ARC600 lp_count
1314 (arc_file_start): Emit Tag_ARC_CPU_variation.
1315 (arc_can_use_doloop_p): New conditions to use ZOLs.
1316 (hwloop_fail): New function.
1317 (hwloop_optimize): Likewise.
1318 (hwloop_pattern_reg): Likewise.
1319 (arc_doloop_hooks): New struct, to be used with reorg_loops.
1320 (arc_reorg_loops): New function, calls reorg_loops.
1321 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
1322 (arc600_corereg_hazard): Remove ZOL checking, case handled by
1324 (arc_loop_hazard): Remove function, functionality moved into
1326 (arc_hazard): Remove arc_loop_hazard call.
1327 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
1328 into hwloop_optimize.
1329 (arc_label_align): Remove ZOL handling.
1330 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
1331 * config/arc/arc.md (doloop_begin): Remove pattern.
1332 (doloop_begin_i): Likewise.
1333 (doloop_end_i): Likewise.
1334 (doloop_fallback): Likewise.
1335 (doloop_fallback_m): Likewise.
1336 (doloop_end): Reimplement expand.
1337 (arc_lp): New pattern for LP instruction.
1338 (loop_end): New pattern.
1339 (loop_fail): Likewise.
1340 (decrement_and_branch_until_zero): Likewise.
1341 * config/arc/arc.opt (mlpc-width): New option.
1342 * doc/invoke.texi (mlpc-width): Document option.
1344 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1346 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
1347 (arc_ccfsm_advance): Fix checking for delay slots.
1348 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
1350 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1352 * config/arc/arc.md (movqi_insn): Add stores to save constant long
1354 (movhi_insn): Update store instruction constraint which are saving
1355 6-bit short immediates.
1356 (movsi_insn): Consider also short scaled load operations.
1357 (zero_extendhisi2_i): Use Usd constraint instead of T.
1358 (extendhisi2_i): Add q constraint.
1359 (arc_clzsi2): Add type and length attributes.
1360 (arc_ctzsi2): Likewise.
1361 * config/arc/constraints.md (Usc): Update constraint, the
1362 assembler can parse two relocations for a single instruction.
1364 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1366 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
1367 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
1369 2017-08-31 Olivier Hainque <hainque@adacore.com>
1371 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
1372 match as powerpc-wrs-vxworks*.
1374 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
1376 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
1377 register constraint for by-element operand.
1378 (aarch64_mls_elt_merge<mode>): Likewise.
1380 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1382 * config/arc/arc.c (arc_can_follow_jump): Check for short
1385 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1387 * config.gcc: Use g.opt for arc.
1388 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
1389 functionality moved to ...
1390 (legitimate_scaled_address_p): New function, ...here.
1391 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
1392 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
1393 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
1395 (arc_override_options): Handle G option.
1396 (arc_output_pic_addr_const): Correct function definition.
1397 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
1398 (arc_decl_anon_ns_mem_p): Delete.
1399 (arc_in_small_data_p): Overhaul this function to take into
1400 consideration the value given via G option.
1401 (arc_rewrite_small_data_1): Renamed and corrected old
1402 arc_rewrite_small_data function.
1403 (arc_rewrite_small_data): New function.
1404 (small_data_pattern): Don't use pic_offset_table_rtx.
1405 * config/arc/arc.h (CC1_SPEC): Recognize G option.
1406 * config/arc/simdext.md (movmisalignv2hi): Use
1407 prepare_move_operands function.
1409 (movmisalign*): Likewise.
1410 * doc/invoke.texi (ARC options): Document -G option.
1412 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1414 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
1416 * config/arc/arc.c (arc_print_operand): Output scalled address for
1417 sdata whenever is possible.
1418 (arc_in_small_data_p): Allow sdata for 64bit datum when double
1419 load/stores are available.
1420 (compact_sda_memory_operand): Check for the alignment required by
1421 code density instructions.
1422 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
1424 * config/arc/constraints.md (Usd): Update constraint.
1425 (Us0): New constraint.
1426 (Usc): Update constraint.
1428 2017-08-31 Richard Biener <rguenther@suse.de>
1431 * dwarf2out.c (dwarf2out_early_global_decl): Process each
1434 2017-08-31 Tamar Christina <tamar.christina@arm.com>
1436 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
1437 Resize type_signature.
1439 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1440 Alan Hayward <alan.hayward@arm.com>
1441 David Sherwood <david.sherwood@arm.com>
1443 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
1444 subregs whose inner modes can be stored in GPRs.
1445 (aarch64_classify_index): Likewise.
1447 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1448 Alan Hayward <alan.hayward@arm.com>
1449 David Sherwood <david.sherwood@arm.com>
1451 * config/aarch64/iterators.md (V_cmp_result): Rename to...
1452 (V_INT_EQUIV): ...this.
1453 (v_cmp_result): Rename to...
1454 (v_int_equiv): ...this.
1455 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
1456 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
1457 (copysign<mode>3): Likewise.
1458 (aarch64_simd_bsl<mode>_internal): Likewise.
1459 (aarch64_simd_bsl<mode>): Likewise.
1460 (vec_cmp<mode><mode>): Likewise.
1461 (vcond<mode><mode>): Likewise.
1462 (vcond<v_cmp_mixed><mode>): Likewise.
1463 (vcondu<mode><v_cmp_mixed>): Likewise.
1464 (aarch64_cm<optab><mode>): Likewise.
1465 (aarch64_cmtst<mode>): Likewise.
1466 (aarch64_fac<optab><mode>): Likewise.
1467 (vec_perm_const<mode>): Likewise.
1468 (vcond_mask_<mode><v_cmp_result>): Rename to...
1469 (vcond_mask_<mode><v_int_equiv>): ...this.
1470 (vec_cmp<mode><v_cmp_result>): Rename to...
1471 (vec_cmp<mode><v_int_equiv>): ...this.
1473 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1474 Alan Hayward <alan.hayward@arm.com>
1475 David Sherwood <david.sherwood@arm.com>
1477 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
1479 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
1480 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
1481 (UNSPEC_LD4_DREG): New unspecs.
1482 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
1483 (aarch64_ld2<mode>_dreg_be): Replace with...
1484 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
1486 (aarch64_ld3<mode>_dreg_le)
1487 (aarch64_ld3<mode>_dreg_be): Replace with...
1488 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
1490 (aarch64_ld4<mode>_dreg_le)
1491 (aarch64_ld4<mode>_dreg_be): Replace with...
1492 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
1495 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1497 PR tree-optimization/81987
1498 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
1499 insert an initializer in a location not dominated by the stride
1502 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1504 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
1505 on the entire header of the finally block in the fallthru case.
1507 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1509 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
1511 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
1513 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
1514 rs6000_emit_move_from_cr and call renamed function.
1515 (rs6000_emit_prologue): Call renamed functions.
1516 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
1517 movesi_from_cr, remove volatile CRs.
1519 2017-08-30 Jon Beniston <jon@beniston.com>
1520 Richard Biener <rguenther@suse.de>
1522 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
1523 of VECTOR_MODE_P check.
1524 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
1525 element vector types.
1527 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1529 * df.h (df_read_modify_subreg_p): Remove in favor of...
1530 * rtl.h (read_modify_subreg_p): ...this new function. Take a
1531 const_rtx instead of an rtx.
1532 * cprop.c (local_cprop_find_used_regs): Update accordingly.
1533 * df-problems.c (df_word_lr_mark_ref): Likewise.
1534 * ira-lives.c (mark_pseudo_reg_live): Likewise.
1535 (mark_pseudo_reg_dead): Likewise.
1536 (mark_ref_dead): Likewise.
1537 * reginfo.c (init_subregs_of_mode): Likewise.
1538 * sched-deps.c (sched_analyze_1): Likewise.
1539 * df-scan.c (df_def_record_1): Likewise.
1540 (df_uses_record): Likewise.
1541 (df_read_modify_subreg_p): Remove in favor of...
1542 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
1543 const_rtx instead of an rtx.
1545 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1546 Alan Hayward <alan.hayward@arm.com>
1547 David Sherwood <david.sherwood@arm.com>
1549 * rtl.h (partial_subreg_p): New function.
1550 * caller-save.c (save_call_clobbered_regs): Use it.
1551 * calls.c (expand_call): Likewise.
1552 * combine.c (combinable_i3pat): Likewise.
1553 (simplify_set): Likewise.
1554 (make_extraction): Likewise.
1555 (make_compound_operation_int): Likewise.
1556 (gen_lowpart_or_truncate): Likewise.
1557 (force_to_mode): Likewise.
1558 (make_field_assignment): Likewise.
1559 (reg_truncated_to_mode): Likewise.
1560 (record_truncated_value): Likewise.
1561 (move_deaths): Likewise.
1562 * cse.c (record_jump_cond): Likewise.
1563 (cse_insn): Likewise.
1564 * cselib.c (cselib_lookup_1): Likewise.
1565 * expmed.c (extract_bit_field_using_extv): Likewise.
1566 * function.c (assign_parm_setup_reg): Likewise.
1567 * ifcvt.c (noce_convert_multiple_sets): Likewise.
1568 * ira-build.c (create_insn_allocnos): Likewise.
1569 * lra-coalesce.c (merge_pseudos): Likewise.
1570 * lra-constraints.c (match_reload): Likewise.
1571 (simplify_operand_subreg): Likewise.
1572 (curr_insn_transform): Likewise.
1573 * lra-lives.c (process_bb_lives): Likewise.
1574 * lra.c (new_insn_reg): Likewise.
1575 (lra_substitute_pseudo): Likewise.
1576 * regcprop.c (mode_change_ok): Likewise.
1577 (maybe_mode_change): Likewise.
1578 (copyprop_hardreg_forward_1): Likewise.
1579 * reload.c (push_reload): Likewise.
1580 (find_reloads): Likewise.
1581 (find_reloads_subreg_address): Likewise.
1582 * reload1.c (alter_reg): Likewise.
1583 (eliminate_regs_1): Likewise.
1584 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1586 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
1588 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
1589 back to if statements, including unpack.
1591 2017-08-30 Martin Liska <mliska@suse.cz>
1594 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
1596 (func_checker::compare_asm_inputs_outputs): ... this function.
1597 (func_checker::compare_gimple_asm): Use the function to compare
1598 also ASM constrains.
1599 * ipa-icf-gimple.h: Rename the function.
1601 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1602 Alan Hayward <alan.hayward@arm.com>
1603 David Sherwood <david.sherwood@arm.com>
1605 * coretypes.h (complex_mode): New type.
1606 * gdbhooks.py (build_pretty_printer): Handle it.
1607 * machmode.h (complex_mode): New class.
1608 (complex_mode::includes_p): New function.
1609 (is_complex_int_mode): Likewise.
1610 (is_complex_float_mode): Likewise.
1611 * genmodes.c (get_mode_class): Handle complex mode classes.
1612 * function.c (expand_function_end): Use is_complex_int_mode.
1614 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1615 Alan Hayward <alan.hayward@arm.com>
1616 David Sherwood <david.sherwood@arm.com>
1618 * coretypes.h (scalar_mode_pod): New typedef.
1619 * gdbhooks.py (build_pretty_printer): Handle it.
1620 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
1621 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
1622 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
1623 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
1626 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1627 Alan Hayward <alan.hayward@arm.com>
1628 David Sherwood <david.sherwood@arm.com>
1630 * machmode.h (mode_for_vector): Take a scalar_mode instead
1632 * stor-layout.c (mode_for_vector): Likewise.
1633 * explow.c (promote_mode): Use as_a <scalar_mode>.
1634 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
1636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1637 Alan Hayward <alan.hayward@arm.com>
1638 David Sherwood <david.sherwood@arm.com>
1640 * target.def (preferred_simd_mode): Take a scalar_mode
1641 instead of a machine_mode.
1642 * targhooks.h (default_preferred_simd_mode): Likewise.
1643 * targhooks.c (default_preferred_simd_mode): Likewise.
1644 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
1645 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
1646 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
1647 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
1648 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
1649 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
1650 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
1651 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
1653 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
1654 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
1655 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
1656 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
1657 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
1658 * doc/tm.texi: Regenerate.
1659 * optabs-query.c (can_vec_mask_load_store_p): Return false for
1662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1663 Alan Hayward <alan.hayward@arm.com>
1664 David Sherwood <david.sherwood@arm.com>
1666 * target.def (scalar_mode_supported_p): Take a scalar_mode
1667 instead of a machine_mode.
1668 * targhooks.h (default_scalar_mode_supported_p): Likewise.
1669 * targhooks.c (default_scalar_mode_supported_p): Likewise.
1670 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
1671 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1672 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
1673 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
1674 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
1675 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
1676 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
1677 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
1678 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
1679 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
1680 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
1681 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
1683 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
1684 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
1685 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1686 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1687 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
1689 * doc/tm.texi: Regenerate.
1691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1692 Alan Hayward <alan.hayward@arm.com>
1693 David Sherwood <david.sherwood@arm.com>
1695 * coretypes.h (opt_scalar_mode): New typedef.
1696 * gdbhooks.py (build_pretty_printers): Handle it.
1697 * machmode.h (mode_iterator::get_2xwider): Add overload for
1699 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
1701 * expr.c (convert_mode_scalar): Likewise.
1702 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1703 * optabs.c (expand_float): Likewise.
1704 (expand_fix): Likewise.
1705 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1707 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1708 Alan Hayward <alan.hayward@arm.com>
1709 David Sherwood <david.sherwood@arm.com>
1711 * optabs.c (expand_float): Explicitly check for scalars before
1712 using a branching expansion.
1713 (expand_fix): Likewise.
1715 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1716 Alan Hayward <alan.hayward@arm.com>
1717 David Sherwood <david.sherwood@arm.com>
1719 * expr.c (convert_mode): Split scalar handling out into...
1720 (convert_mode_scalar): ...this new function. Treat the modes
1723 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1724 Alan Hayward <alan.hayward@arm.com>
1725 David Sherwood <david.sherwood@arm.com>
1727 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
1729 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
1731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1732 Alan Hayward <alan.hayward@arm.com>
1733 David Sherwood <david.sherwood@arm.com>
1735 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
1736 rather than a machine_mode.
1737 (fixed_from_string): Likewise.
1738 (fixed_convert): Likewise.
1739 (fixed_convert_from_int): Likewise.
1740 (fixed_convert_from_real): Likewise.
1741 (real_convert_from_fixed): Likewise.
1742 * fixed-value.c (fixed_from_double_int): Likewise.
1743 (fixed_from_string): Likewise.
1744 (fixed_convert): Likewise.
1745 (fixed_convert_from_int): Likewise.
1746 (fixed_convert_from_real): Likewise.
1747 (real_convert_from_fixed): Likewise.
1748 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
1750 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1751 Alan Hayward <alan.hayward@arm.com>
1752 David Sherwood <david.sherwood@arm.com>
1754 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
1755 of separate mode class checks. Do not allow vector modes here.
1756 (immed_wide_int_const): Use as_a <scalar_mode>.
1757 * explow.c (trunc_int_for_mode): Likewise.
1758 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
1759 (wi::shwi): Likewise.
1760 (wi::min_value): Likewise.
1761 (wi::max_value): Likewise.
1762 * dwarf2out.c (loc_descriptor): Likewise.
1763 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
1766 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1767 Alan Hayward <alan.hayward@arm.com>
1768 David Sherwood <david.sherwood@arm.com>
1770 * tree.h (SCALAR_TYPE_MODE): New macro.
1771 * expr.c (expand_expr_addr_expr_1): Use it.
1772 (expand_expr_real_2): Likewise.
1773 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
1774 (fold_convert_const_fixed_from_int): Likewise.
1775 (fold_convert_const_fixed_from_real): Likewise.
1776 (native_encode_fixed): Likewise
1777 (native_encode_complex): Likewise
1778 (native_encode_vector): Likewise.
1779 (native_interpret_fixed): Likewise.
1780 (native_interpret_real): Likewise.
1781 (native_interpret_complex): Likewise.
1782 (native_interpret_vector): Likewise.
1783 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
1784 (simd_clone_adjust_argument_types): Likewise.
1785 (simd_clone_init_simd_arrays): Likewise.
1786 (simd_clone_adjust): Likewise.
1787 * stor-layout.c (layout_type): Likewise.
1788 * tree.c (build_minus_one_cst): Likewise.
1789 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
1790 * tree-inline.c (estimate_move_cost): Likewise.
1791 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
1792 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
1793 (vectorizable_reduction): Likewise.
1794 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
1795 (vect_recog_mixed_size_cond_pattern): Likewise.
1796 (check_bool_pattern): Likewise.
1797 (adjust_bool_pattern): Likewise.
1798 (search_type_for_mask_1): Likewise.
1799 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
1800 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1801 (vectorizable_load): Likewise.
1802 (vectorizable_store): Likewise.
1803 * ubsan.c (ubsan_encode_value): Likewise.
1804 * varasm.c (output_constant): Likewise.
1806 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1807 Alan Hayward <alan.hayward@arm.com>
1808 David Sherwood <david.sherwood@arm.com>
1810 * coretypes.h (scalar_mode): New class.
1811 * machmode.h (scalar_mode): Likewise.
1812 (scalar_mode::includes_p): New function.
1813 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
1814 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
1815 * genmodes.c (get_mode_class): Handle remaining scalar modes.
1816 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
1817 * expmed.c (store_bit_field_1): Likewise.
1818 (extract_bit_field_1): Likewise.
1819 * expr.c (write_complex_part): Likewise.
1820 (read_complex_part): Likewise.
1821 (emit_move_complex_push): Likewise.
1822 (expand_expr_real_2): Likewise.
1823 * function.c (assign_parm_setup_reg): Likewise.
1824 (assign_parms_unsplit_complex): Likewise.
1825 * optabs.c (expand_binop): Likewise.
1826 * rtlanal.c (subreg_get_info): Likewise.
1827 * simplify-rtx.c (simplify_immed_subreg): Likewise.
1828 * varasm.c (output_constant_pool_2): Likewise.
1830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1831 Alan Hayward <alan.hayward@arm.com>
1832 David Sherwood <david.sherwood@arm.com>
1834 * expmed.c (extract_high_half): Use scalar_int_mode and remove
1836 (expmed_mult_highpart_optab): Likewise.
1837 (expmed_mult_highpart): Likewise.
1839 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1840 Alan Hayward <alan.hayward@arm.com>
1841 David Sherwood <david.sherwood@arm.com>
1843 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
1844 instead of a machine_mode.
1845 (builtin_memset_read_str): Likewise.
1846 * builtins.c (c_readstr): Likewise.
1847 (builtin_memcpy_read_str): Likewise.
1848 (builtin_strncpy_read_str): Likewise.
1849 (builtin_memset_read_str): Likewise.
1850 (builtin_memset_gen_str): Likewise.
1851 (expand_builtin_signbit): Use scalar_int_mode for local variables.
1852 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
1853 instead of a machine_mode.
1854 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
1856 (make_extraction): Likewise.
1857 (try_widen_shift_mode): Take and return scalar_int_modes instead
1859 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
1860 a scalar_int_mode instead of a machine_mode.
1861 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
1862 (avr_addr_space_pointer_mode): Likewise.
1863 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
1864 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
1865 (msp430_unwind_word_mode): Likewise.
1866 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
1867 (spu_addr_space_pointer_mode): Likewise.
1868 (spu_addr_space_address_mode): Likewise.
1869 (spu_libgcc_cmp_return_mode): Likewise.
1870 (spu_libgcc_shift_count_mode): Likewise.
1871 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
1872 (rl78_addr_space_pointer_mode): Likewise.
1873 (fl78_unwind_word_mode): Likewise.
1874 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
1876 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
1877 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
1878 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
1879 (mips_valid_pointer_mode): Likewise.
1880 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
1881 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
1882 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
1884 (ft32_addr_space_address_mode): Likewise.
1885 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
1886 scalar_int_mode instead of a machine_mode.
1887 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
1889 (m32c_addr_space_address_mode): Likewise.
1890 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
1891 (rs6000_eh_return_filter_mode): Likewise.
1892 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
1893 (rs6000_eh_return_filter_mode): Likewise.
1894 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
1895 (s390_libgcc_shift_count_mode): Likewise.
1896 (s390_unwind_word_mode): Likewise.
1897 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
1899 * target.def (mode_rep_extended): Likewise.
1900 (valid_pointer_mode): Likewise.
1901 (addr_space.valid_pointer_mode): Likewise.
1902 (eh_return_filter_mode): Return a scalar_int_mode rather than
1904 (libgcc_cmp_return_mode): Likewise.
1905 (libgcc_shift_count_mode): Likewise.
1906 (unwind_word_mode): Likewise.
1907 (addr_space.pointer_mode): Likewise.
1908 (addr_space.address_mode): Likewise.
1909 * doc/tm.texi: Regenerate.
1910 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
1912 (do_jump): Use scalar_int_mode for local variables.
1913 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
1914 rather than a machine_mode.
1915 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
1916 (scompare_loc_descriptor_wide): Likewise.
1917 (scompare_loc_descriptor_narrow): Likewise.
1918 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
1920 * except.c (sjlj_emit_dispatch_table): Likewise.
1921 (expand_builtin_eh_copy_values): Likewise.
1922 * explow.c (convert_memory_address_addr_space_1): Likewise.
1923 Take a scalar_int_mode rather than a machine_mode.
1924 (convert_memory_address_addr_space): Take a scalar_int_mode rather
1925 than a machine_mode.
1926 (memory_address_addr_space): Use scalar_int_mode for local variables.
1927 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
1928 rather than a machine_mode.
1929 * expmed.c (mask_rtx): Likewise.
1930 (init_expmed_one_conv): Likewise.
1931 (expand_mult_highpart_adjust): Likewise.
1932 (extract_high_half): Likewise.
1933 (expmed_mult_highpart_optab): Likewise.
1934 (expmed_mult_highpart): Likewise.
1935 (expand_smod_pow2): Likewise.
1936 (expand_sdiv_pow2): Likewise.
1937 (emit_store_flag_int): Likewise.
1938 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
1940 (extract_low_bits): Likewise.
1941 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
1943 * expr.c (pieces_addr::adjust): Likewise.
1944 (can_store_by_pieces): Likewise.
1945 (store_by_pieces): Likewise.
1946 (clear_by_pieces_1): Likewise.
1947 (expand_expr_addr_expr_1): Likewise.
1948 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
1949 (expand_expr_real_1): Likewise.
1950 (try_casesi): Likewise.
1951 * final.c (shorten_branches): Likewise.
1952 * fold-const.c (fold_convert_const_int_from_fixed): Change the
1953 type of "mode" to machine_mode.
1954 * internal-fn.c (expand_arith_overflow_result_store): Take a
1955 scalar_int_mode rather than a machine_mode.
1956 (expand_mul_overflow): Use scalar_int_mode for local variables.
1957 * loop-doloop.c (doloop_modify): Likewise.
1958 (doloop_optimize): Likewise.
1959 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
1960 than a machine_mode.
1961 (expand_doubleword_shift_condmove): Likewise.
1962 (expand_doubleword_shift): Likewise.
1963 (expand_doubleword_clz): Likewise.
1964 (expand_doubleword_popcount): Likewise.
1965 (expand_doubleword_parity): Likewise.
1966 (expand_absneg_bit): Use scalar_int_mode for local variables.
1967 (prepare_float_lib_cmp): Likewise.
1968 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
1969 rather than a machine_mode.
1970 (convert_memory_address_addr_space): Likewise.
1971 (get_mode_bounds): Likewise.
1972 (get_address_mode): Return a scalar_int_mode rather than a
1974 * rtlanal.c (get_address_mode): Likewise.
1975 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
1976 than a machine_mode.
1977 * targhooks.c (default_mode_rep_extended): Likewise.
1978 (default_valid_pointer_mode): Likewise.
1979 (default_addr_space_valid_pointer_mode): Likewise.
1980 (default_eh_return_filter_mode): Return a scalar_int_mode rather
1981 than a machine_mode.
1982 (default_libgcc_cmp_return_mode): Likewise.
1983 (default_libgcc_shift_count_mode): Likewise.
1984 (default_unwind_word_mode): Likewise.
1985 (default_addr_space_pointer_mode): Likewise.
1986 (default_addr_space_address_mode): Likewise.
1987 * targhooks.h (default_eh_return_filter_mode): Likewise.
1988 (default_libgcc_cmp_return_mode): Likewise.
1989 (default_libgcc_shift_count_mode): Likewise.
1990 (default_unwind_word_mode): Likewise.
1991 (default_addr_space_pointer_mode): Likewise.
1992 (default_addr_space_address_mode): Likewise.
1993 (default_mode_rep_extended): Take a scalar_int_mode rather than
1995 (default_valid_pointer_mode): Likewise.
1996 (default_addr_space_valid_pointer_mode): Likewise.
1997 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
1999 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
2000 rather than a machine_mode.
2001 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
2002 for local variables.
2003 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2004 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
2005 than a machine_mode.
2007 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2008 Alan Hayward <alan.hayward@arm.com>
2009 David Sherwood <david.sherwood@arm.com>
2011 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
2012 the mode argument to scalar_int_mode.
2013 (do_jump_by_parts_zero_rtx): Likewise.
2014 (do_jump_by_parts_equality_rtx): Likewise.
2015 (do_jump_by_parts_greater): Take a mode argument.
2016 (do_jump_by_parts_equality): Likewise.
2017 (do_jump_1): Update calls accordingly.
2019 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2020 Alan Hayward <alan.hayward@arm.com>
2021 David Sherwood <david.sherwood@arm.com>
2023 * is-a.h (safe_dyn_cast): New function.
2024 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
2025 (jump_table_for_label): Likewise.
2026 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
2027 instead of an rtx_insn *.
2028 (shorten_branches): Use dyn_cast instead of LABEL_P and
2029 JUMP_TABLE_DATA_P. Use jump_table_for_label and
2030 rtx_jump_table_data::get_data_mode.
2031 (final_scan_insn): Likewise.
2033 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2034 Alan Hayward <alan.hayward@arm.com>
2035 David Sherwood <david.sherwood@arm.com>
2037 * combine.c (try_combine): Use is_a <scalar_int_mode> when
2038 trying to combine a full-register integer set with a subreg
2041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2042 Alan Hayward <alan.hayward@arm.com>
2043 David Sherwood <david.sherwood@arm.com>
2045 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
2046 that is always either address_mode or pointer_mode.
2048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2049 Alan Hayward <alan.hayward@arm.com>
2050 David Sherwood <david.sherwood@arm.com>
2052 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
2053 when the two are known to be equal.
2055 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2056 Alan Hayward <alan.hayward@arm.com>
2057 David Sherwood <david.sherwood@arm.com>
2059 * simplify-rtx.c (simplify_const_unary_operation): Use
2060 is_a <scalar_int_mode> instead of checking for a nonzero
2061 precision. Forcibly convert op_mode to a scalar_int_mode
2062 in that case. More clearly differentiate the operand and
2063 result modes and use the former when deciding what the value
2064 of a count-bits operation should be. Use is_int_mode instead
2065 of checking for a MODE_INT. Remove redundant check for whether
2066 this mode has a zero precision.
2068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2069 Alan Hayward <alan.hayward@arm.com>
2070 David Sherwood <david.sherwood@arm.com>
2072 * optabs.c (widen_leading): Change the type of the mode argument
2073 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
2074 (widen_bswap): Likewise.
2075 (expand_parity): Likewise.
2076 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
2077 (expand_ffs): Likewise.
2078 (epand_unop): Check for scalar integer modes before calling the
2081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2082 Alan Hayward <alan.hayward@arm.com>
2083 David Sherwood <david.sherwood@arm.com>
2085 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
2087 (expand_expr_real_1): Update call accordingly.
2089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2090 Alan Hayward <alan.hayward@arm.com>
2091 David Sherwood <david.sherwood@arm.com>
2093 * expmed.c (store_bit_field_using_insv): Add op0_mode and
2094 value_mode arguments. Use scalar_int_mode internally.
2095 (store_bit_field_1): Rename the new integer mode from imode
2096 to op0_mode and use it instead of GET_MODE (op0). Update calls
2097 to store_split_bit_field, store_bit_field_using_insv and
2098 store_fixed_bit_field.
2099 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
2100 Use scalar_int_mode internally. Use a bit count rather than a mode
2101 when calculating the largest bit size for get_best_mode.
2102 Update calls to store_split_bit_field and store_fixed_bit_field_1.
2103 (store_fixed_bit_field_1): Add mode and value_mode arguments.
2104 Remove assertion that OP0 has a scalar integer mode.
2105 (store_split_bit_field): Add op0_mode and value_mode arguments.
2106 Update calls to extract_fixed_bit_field.
2107 (extract_bit_field_using_extv): Add an op0_mode argument.
2108 Use scalar_int_mode internally.
2109 (extract_bit_field_1): Rename the new integer mode from imode to
2110 op0_mode and use it instead of GET_MODE (op0). Update calls to
2111 extract_split_bit_field, extract_bit_field_using_extv and
2112 extract_fixed_bit_field.
2113 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
2114 to extract_split_bit_field and extract_fixed_bit_field_1.
2115 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
2116 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
2118 (extract_split_bit_field): Add an op0_mode argument. Update call
2119 to extract_fixed_bit_field.
2121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2122 Alan Hayward <alan.hayward@arm.com>
2123 David Sherwood <david.sherwood@arm.com>
2125 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
2126 * explow.c (hard_function_value): Likewise.
2127 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
2128 convert_to_mode call outside the loop.
2129 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
2130 for the mode iterator. Require the mode specified by max_pieces
2132 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
2134 (copy_blkmode_to_reg): Likewise.
2135 (set_storage_via_setmem): Likewise.
2136 * optabs.c (prepare_cmp_insn): Likewise.
2137 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2138 * stor-layout.c (finish_bitfield_representative): Likewise.
2140 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2141 Alan Hayward <alan.hayward@arm.com>
2142 David Sherwood <david.sherwood@arm.com>
2144 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
2145 * expr.c (convert_move): Use them.
2146 (convert_modes): Likewise.
2147 (store_expr_with_bounds): Likewise.
2149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2150 Alan Hayward <alan.hayward@arm.com>
2151 David Sherwood <david.sherwood@arm.com>
2153 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
2154 parameter for the mode of "x". Remove the "known_x", "known_mode"
2155 and "known_ret" arguments. Change the type of the mode argument
2157 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
2158 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
2159 (reg_num_sign_bit_copies_for_combine): Likewise.
2160 * rtlanal.c (nonzero_bits1): Likewise.
2161 (num_sign_bit_copies1): Likewise.
2162 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
2163 (reg_num_sign_bit_copies_general): Likewise.
2164 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
2165 (reg_nonzero_bits_general): Likewise.
2167 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2168 Alan Hayward <alan.hayward@arm.com>
2169 David Sherwood <david.sherwood@arm.com>
2171 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
2172 than in subroutines. Return 1 for non-integer modes.
2173 (cached_num_sign_bit_copies): Change the type of the mode parameter
2175 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
2176 classes. Handle CONST_INT_P first and then check whether X also
2177 has a scalar integer mode. Check the same thing for inner registers
2178 of a SUBREG and for values that are being extended or truncated.
2180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2181 Alan Hayward <alan.hayward@arm.com>
2182 David Sherwood <david.sherwood@arm.com>
2184 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
2185 in subroutines. Return the mode mask for non-integer modes.
2186 (cached_nonzero_bits): Change the type of the mode parameter
2188 (nonzero_bits1): Likewise. Remove early exit for other mode
2189 classes. Handle CONST_INT_P first and then check whether X
2190 also has a scalar integer mode.
2192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2193 Alan Hayward <alan.hayward@arm.com>
2194 David Sherwood <david.sherwood@arm.com>
2196 * expr.c (widest_int_mode_for_size): Make the comment match the code.
2197 Return a scalar_int_mode and assert that the size is greater than
2199 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
2200 (op_by_pieces_d::op_by_pieces_d): Likewise.
2201 (op_by_pieces_d::run): Likewise.
2202 (can_store_by_pieces): Likewise.
2204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2205 Alan Hayward <alan.hayward@arm.com>
2206 David Sherwood <david.sherwood@arm.com>
2208 * combine.c (extract_left_shift): Add a mode argument and update
2210 (make_compound_operation_int): Change the type of the mode parameter
2211 to scalar_int_mode and update the call to extract_left_shift.
2213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2214 Alan Hayward <alan.hayward@arm.com>
2215 David Sherwood <david.sherwood@arm.com>
2217 * combine.c (simplify_and_const_int): Change the type of the mode
2218 parameter to scalar_int_mode.
2219 (simplify_and_const_int_1): Likewise. Update recursive call.
2221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2222 Alan Hayward <alan.hayward@arm.com>
2223 David Sherwood <david.sherwood@arm.com>
2225 * combine.c (simplify_compare_const): Check that the mode is a
2226 scalar_int_mode (rather than VOIDmode) before testing its
2228 (simplify_comparison): Move COMPARISON_P handling out of the
2229 loop and restrict the latter part of the loop to scalar_int_modes.
2230 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
2231 and when considering SUBREG_REGs. Use is_int_mode instead of
2232 checking GET_MODE_CLASS against MODE_INT.
2234 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2235 Alan Hayward <alan.hayward@arm.com>
2236 David Sherwood <david.sherwood@arm.com>
2238 * combine.c (try_widen_shift_mode): Move check for equal modes to...
2239 (simplify_shift_const_1): ...here. Use scalar_int_mode for
2240 shift_unit_mode and for modes involved in scalar shifts.
2242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2243 Alan Hayward <alan.hayward@arm.com>
2244 David Sherwood <david.sherwood@arm.com>
2246 * combine.c (force_int_to_mode): New function, split out from...
2247 (force_to_mode): ...here. Keep xmode up-to-date and use it
2248 instead of GET_MODE (x).
2250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2251 Alan Hayward <alan.hayward@arm.com>
2252 David Sherwood <david.sherwood@arm.com>
2254 * optabs-query.h (extraction_insn::struct_mode): Change type to
2255 opt_scalar_int_mode and update comment.
2256 (extraction_insn::field_mode): Change type to scalar_int_mode.
2257 (extraction_insn::pos_mode): Likewise.
2258 * combine.c (make_extraction): Update accordingly.
2259 * optabs-query.c (get_traditional_extraction_insn): Likewise.
2260 (get_optab_extraction_insn): Likewise.
2261 * recog.c (simplify_while_replacing): Likewise.
2262 * expmed.c (narrow_bit_field_mem): Change the type of the mode
2263 parameter to opt_scalar_int_mode.
2265 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2266 Alan Hayward <alan.hayward@arm.com>
2267 David Sherwood <david.sherwood@arm.com>
2269 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
2270 to a scalar_int_mode instead of a machine_mode.
2271 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
2272 (get_best_mode): Return a boolean and use a pointer argument to store
2273 the selected mode. Replace the limit mode parameter with a bit limit.
2274 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
2275 for the values returned by bit_field_mode_iterator::next_mode.
2276 (store_bit_field): Update call to get_best_mode.
2277 (store_fixed_bit_field): Likewise.
2278 (extract_fixed_bit_field): Likewise.
2279 * expr.c (optimize_bitfield_assignment_op): Likewise.
2280 * fold-const.c (optimize_bit_field_compare): Likewise.
2281 (fold_truth_andor_1): Likewise.
2282 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
2283 Update for new type of m_mode.
2284 (get_best_mode): As above.
2286 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2287 Alan Hayward <alan.hayward@arm.com>
2288 David Sherwood <david.sherwood@arm.com>
2290 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
2291 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
2292 (store_bit_field): Check is_a <scalar_int_mode> before calling
2293 strict_volatile_bitfield_p.
2294 (extract_bit_field): Likewise.
2296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2297 Alan Hayward <alan.hayward@arm.com>
2298 David Sherwood <david.sherwood@arm.com>
2300 * target.def (cstore_mode): Return a scalar_int_mode.
2301 * doc/tm.texi: Regenerate.
2302 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
2303 * targhooks.h (default_cstore_mode): Likewise.
2304 * targhooks.c (default_cstore_mode): Likewise, using a forced
2306 * expmed.c (emit_cstore): Expect the target of the cstore to be
2309 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2310 Alan Hayward <alan.hayward@arm.com>
2311 David Sherwood <david.sherwood@arm.com>
2313 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
2315 (niter_desc): Likewise mode.
2316 (iv_analyze): Add a mode parameter.
2318 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
2319 and change its type to scalar_int_mode.
2320 * loop-iv.c: Update commentary at head of file.
2321 (iv_constant): Pass the mode paraeter before the rtx it describes
2322 and change its type to scalar_int_mode. Remove VOIDmode handling.
2323 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
2324 (iv_extend): Likewise.
2325 (shorten_into_mode): Likewise.
2326 (iv_add): Use scalar_int_mode.
2327 (iv_mult): Likewise.
2328 (iv_shift): Likewise.
2329 (canonicalize_iv_subregs): Likewise.
2330 (get_biv_step_1): Pass the outer_mode parameter before the rtx
2331 it describes and change its mode to scalar_int_mode. Also change
2332 the type of the returned inner_mode to scalar_int_mode.
2333 (get_biv_step): Likewise, turning outer_mode from a pointer
2334 into a direct parameter. Update call to get_biv_step_1.
2335 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
2336 iv_constant and get_biv_step.
2337 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
2338 and change its type to scalar_int_mode. Don't initialise iv->mode
2339 to VOIDmode and remove later checks for its still being VOIDmode.
2340 Update calls to iv_analyze_op and iv_analyze_expr. Check
2341 is_a <scalar_int_mode> when changing the mode under consideration.
2342 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
2343 Update call to iv_analyze_expr.
2344 (iv_analyze_op): Add a mode parameter. Reject subregs whose
2345 inner register is not also a scalar_int_mode. Update call to
2347 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
2348 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
2349 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
2350 separate mode class checks. Update calls to iv_analyze. Remove
2351 fix-up of VOIDmodes after iv_analyze_biv.
2352 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
2353 don't have a scalar_int_mode. Update call to biv_p.
2355 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2356 Alan Hayward <alan.hayward@arm.com>
2357 David Sherwood <david.sherwood@arm.com>
2359 * cfgexpand.c (convert_debug_memory_address): Use
2360 as_a <scalar_int_mode>.
2361 * combine.c (expand_compound_operation): Likewise.
2362 (make_extraction): Likewise.
2363 (change_zero_ext): Likewise.
2364 (simplify_comparison): Likewise.
2365 * cse.c (cse_insn): Likewise.
2366 * dwarf2out.c (minmax_loc_descriptor): Likewise.
2367 (mem_loc_descriptor): Likewise.
2368 (loc_descriptor): Likewise.
2369 * expmed.c (init_expmed_one_mode): Likewise.
2370 (synth_mult): Likewise.
2371 (emit_store_flag_1): Likewise.
2372 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
2373 of a comparison with size.
2374 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
2375 (reduce_to_bit_field_precision): Likewise.
2376 * function.c (expand_function_end): Likewise.
2377 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
2378 * loop-doloop.c (doloop_modify): Likewise.
2379 * optabs.c (expand_binop): Likewise.
2380 (expand_unop): Likewise.
2381 (expand_copysign_absneg): Likewise.
2382 (prepare_cmp_insn): Likewise.
2383 (maybe_legitimize_operand): Likewise.
2384 * recog.c (const_scalar_int_operand): Likewise.
2385 * rtlanal.c (get_address_mode): Likewise.
2386 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2387 (simplify_cond_clz_ctz): Likewise.
2388 * tree-nested.c (get_nl_goto_field): Likewise.
2389 * tree.c (build_vector_type_for_mode): Likewise.
2390 * var-tracking.c (use_narrower_mode): Likewise.
2392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2393 Alan Hayward <alan.hayward@arm.com>
2394 David Sherwood <david.sherwood@arm.com>
2396 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
2397 * builtins.c (expand_builtin_signbit): Use it.
2398 * cfgexpand.c (expand_debug_expr): Likewise.
2399 * dojump.c (do_jump): Likewise.
2400 (do_compare_and_jump): Likewise.
2401 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
2402 * expmed.c (make_tree): Likewise.
2403 * expr.c (expand_expr_real_2): Likewise.
2404 (expand_expr_real_1): Likewise.
2405 (try_casesi): Likewise.
2406 * fold-const-call.c (fold_const_call_ss): Likewise.
2407 * fold-const.c (unextend): Likewise.
2408 (extract_muldiv_1): Likewise.
2409 (fold_single_bit_test): Likewise.
2410 (native_encode_int): Likewise.
2411 (native_encode_string): Likewise.
2412 (native_interpret_int): Likewise.
2413 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
2414 * internal-fn.c (expand_addsub_overflow): Likewise.
2415 (expand_neg_overflow): Likewise.
2416 (expand_mul_overflow): Likewise.
2417 (expand_arith_overflow): Likewise.
2418 * match.pd: Likewise.
2419 * stor-layout.c (layout_type): Likewise.
2420 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2421 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
2422 * tree-ssanames.c (get_range_info): Likewise.
2423 * tree-switch-conversion.c (array_value_type) Likewise.
2424 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
2425 (vect_recog_divmod_pattern): Likewise.
2426 (vect_recog_mixed_size_cond_pattern): Likewise.
2427 * tree-vrp.c (extract_range_basic): Likewise.
2428 (simplify_float_conversion_using_ranges): Likewise.
2429 * tree.c (int_fits_type_p): Likewise.
2430 * ubsan.c (instrument_bool_enum_load): Likewise.
2431 * varasm.c (mergeable_string_section): Likewise.
2432 (narrowing_initializer_constant_valid_p): Likewise.
2433 (output_constant): Likewise.
2435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2436 Alan Hayward <alan.hayward@arm.com>
2437 David Sherwood <david.sherwood@arm.com>
2439 * machmode.h (NARROWEST_INT_MODE): New macro.
2440 * expr.c (alignment_for_piecewise_move): Use it instead of
2441 GET_CLASS_NARROWEST_MODE (MODE_INT).
2442 (push_block): Likewise.
2443 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2445 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2448 Alan Hayward <alan.hayward@arm.com>
2449 David Sherwood <david.sherwood@arm.com>
2451 * postreload.c (move2add_valid_value_p): Change the type of the
2452 mode parameter to scalar_int_mode.
2453 (move2add_use_add2_insn): Add a mode parameter and use it instead
2455 (move2add_use_add3_insn): Likewise.
2456 (reload_cse_move2add): Update accordingly.
2458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2459 Alan Hayward <alan.hayward@arm.com>
2460 David Sherwood <david.sherwood@arm.com>
2462 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
2464 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
2465 * optabs.c (expand_unop): Likewise.
2467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2468 Alan Hayward <alan.hayward@arm.com>
2469 David Sherwood <david.sherwood@arm.com>
2471 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
2472 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
2473 (popcount_loc_descriptor): Likewise.
2474 (bswap_loc_descriptor): Likewise.
2475 (rotate_loc_descriptor): Likewise.
2476 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
2477 calling the functions above.
2479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2480 Alan Hayward <alan.hayward@arm.com>
2481 David Sherwood <david.sherwood@arm.com>
2483 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
2485 (try_combine): Likewise.
2486 (simplify_if_then_else): Likewise.
2487 * cse.c (cse_insn): Likewise.
2488 * dwarf2out.c (mem_loc_descriptor): Likewise.
2489 * emit-rtl.c (gen_lowpart_common): Likewise.
2490 * simplify-rtx.c (simplify_truncation): Likewise.
2491 (simplify_binary_operation_1): Likewise.
2492 (simplify_const_relational_operation): Likewise.
2493 (simplify_ternary_operation): Likewise.
2494 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
2496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2497 Alan Hayward <alan.hayward@arm.com>
2498 David Sherwood <david.sherwood@arm.com>
2500 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
2501 * reload.c (push_reload): Likewise.
2502 (find_reloads): Likewise.
2504 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2505 Alan Hayward <alan.hayward@arm.com>
2506 David Sherwood <david.sherwood@arm.com>
2508 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
2509 (make_compound_operation_int): Likewise.
2510 (change_zero_ext): Likewise.
2511 * expr.c (convert_move): Likewise.
2512 (convert_modes): Likewise.
2513 * fwprop.c (forward_propagate_subreg): Likewise.
2514 * loop-iv.c (get_biv_step_1): Likewise.
2515 * optabs.c (widen_operand): Likewise.
2516 * postreload.c (move2add_valid_value_p): Likewise.
2517 * recog.c (simplify_while_replacing): Likewise.
2518 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2519 (simplify_binary_operation_1): Likewise. Remove redundant
2520 mode equality check.
2522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2523 Alan Hayward <alan.hayward@arm.com>
2524 David Sherwood <david.sherwood@arm.com>
2526 * combine.c (combine_simplify_rtx): Add checks for
2527 is_a <scalar_int_mode>.
2528 (simplify_if_then_else): Likewise.
2529 (make_field_assignment): Likewise.
2530 (simplify_comparison): Likewise.
2531 * ifcvt.c (noce_try_bitop): Likewise.
2532 * loop-invariant.c (canonicalize_address_mult): Likewise.
2533 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2536 Alan Hayward <alan.hayward@arm.com>
2537 David Sherwood <david.sherwood@arm.com>
2539 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
2540 is_a <scalar_int_mode> instead of != BLKmode.
2542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2543 Alan Hayward <alan.hayward@arm.com>
2544 David Sherwood <david.sherwood@arm.com>
2546 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
2547 instead of != VOIDmode.
2548 * combine.c (if_then_else_cond): Likewise.
2549 (change_zero_ext): Likewise.
2550 * dwarf2out.c (mem_loc_descriptor): Likewise.
2551 (loc_descriptor): Likewise.
2552 * rtlanal.c (canonicalize_condition): Likewise.
2553 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
2555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2556 Alan Hayward <alan.hayward@arm.com>
2557 David Sherwood <david.sherwood@arm.com>
2559 * simplify-rtx.c (simplify_binary_operation_1): Use
2560 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
2562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2563 Alan Hayward <alan.hayward@arm.com>
2564 David Sherwood <david.sherwood@arm.com>
2566 * wide-int.h (int_traits<unsigned char>) New class.
2567 (int_traits<unsigned short>) Likewise.
2568 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
2569 Use GET_MODE_UNIT_PRECISION and remove redundant test for
2571 * combine.c (set_nonzero_bits_and_sign_copies): Use
2572 is_a <scalar_int_mode>.
2573 (find_split_point): Likewise.
2574 (combine_simplify_rtx): Likewise.
2575 (simplify_logical): Likewise.
2576 (expand_compound_operation): Likewise.
2577 (expand_field_assignment): Likewise.
2578 (make_compound_operation): Likewise.
2579 (extended_count): Likewise.
2580 (change_zero_ext): Likewise.
2581 (simplify_comparison): Likewise.
2582 * dwarf2out.c (scompare_loc_descriptor): Likewise.
2583 (ucompare_loc_descriptor): Likewise.
2584 (minmax_loc_descriptor): Likewise.
2585 (mem_loc_descriptor): Likewise.
2586 (loc_descriptor): Likewise.
2587 * expmed.c (init_expmed_one_mode): Likewise.
2588 * lra-constraints.c (lra_constraint_offset): Likewise.
2589 * optabs.c (prepare_libcall_arg): Likewise.
2590 * postreload.c (move2add_note_store): Likewise.
2591 * reload.c (operands_match_p): Likewise.
2592 * rtl.h (load_extend_op): Likewise.
2593 * rtlhooks.c (gen_lowpart_general): Likewise.
2594 * simplify-rtx.c (simplify_truncation): Likewise.
2595 (simplify_unary_operation_1): Likewise.
2596 (simplify_binary_operation_1): Likewise.
2597 (simplify_const_binary_operation): Likewise.
2598 (simplify_const_relational_operation): Likewise.
2599 (simplify_subreg): Likewise.
2600 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2601 * var-tracking.c (adjust_mems): Likewise.
2602 (prepare_call_arguments): Likewise.
2604 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2605 Alan Hayward <alan.hayward@arm.com>
2606 David Sherwood <david.sherwood@arm.com>
2608 * machmode.h (is_int_mode): New fuction.
2609 * combine.c (find_split_point): Use it.
2610 (combine_simplify_rtx): Likewise.
2611 (simplify_if_then_else): Likewise.
2612 (simplify_set): Likewise.
2613 (simplify_shift_const_1): Likewise.
2614 (simplify_comparison): Likewise.
2615 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
2616 * cse.c (notreg_cost): Likewise.
2617 (cse_insn): Likewise.
2618 * cselib.c (cselib_lookup_1): Likewise.
2619 * dojump.c (do_jump_1): Likewise.
2620 (do_compare_rtx_and_jump): Likewise.
2621 * dse.c (get_call_args): Likewise.
2622 * dwarf2out.c (rtl_for_decl_init): Likewise.
2623 (native_encode_initializer): Likewise.
2624 * expmed.c (emit_store_flag_1): Likewise.
2625 (emit_store_flag): Likewise.
2626 * expr.c (convert_modes): Likewise.
2627 (store_field): Likewise.
2628 (expand_expr_real_1): Likewise.
2629 * fold-const.c (fold_read_from_constant_string): Likewise.
2630 * gimple-ssa-sprintf.c (get_format_string): Likewise.
2631 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
2632 * optabs.c (expand_binop): Likewise.
2633 (expand_unop): Likewise.
2634 (expand_abs_nojump): Likewise.
2635 (expand_one_cmpl_abs_nojump): Likewise.
2636 * simplify-rtx.c (mode_signbit_p): Likewise.
2637 (val_signbit_p): Likewise.
2638 (val_signbit_known_set_p): Likewise.
2639 (val_signbit_known_clear_p): Likewise.
2640 (simplify_relational_operation_1): Likewise.
2641 * tree.c (vector_type_mode): Likewise.
2643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2644 Alan Hayward <alan.hayward@arm.com>
2645 David Sherwood <david.sherwood@arm.com>
2647 * machmode.h (smallest_mode_for_size): Fix formatting.
2648 (smallest_int_mode_for_size): New function.
2649 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
2650 instead of smallest_mode_for_size.
2651 * combine.c (make_extraction): Likewise.
2652 * config/arc/arc.c (arc_expand_movmem): Likewise.
2653 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
2654 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2655 * config/s390/s390.c (s390_expand_insv): Likewise.
2656 * config/sparc/sparc.c (assign_int_registers): Likewise.
2657 * config/spu/spu.c (spu_function_value): Likewise.
2658 (spu_function_arg): Likewise.
2659 * coverage.c (get_gcov_type): Likewise.
2660 (get_gcov_unsigned_t): Likewise.
2661 * dse.c (find_shift_sequence): Likewise.
2662 * expmed.c (store_bit_field_1): Likewise.
2663 * expr.c (convert_move): Likewise.
2664 (store_field): Likewise.
2665 * internal-fn.c (expand_arith_overflow): Likewise.
2666 * optabs-query.c (get_best_extraction_insn): Likewise.
2667 * optabs.c (expand_twoval_binop_libfunc): Likewise.
2668 * stor-layout.c (layout_type): Likewise.
2669 (initialize_sizetypes): Likewise.
2670 * targhooks.c (default_get_mask_mode): Likewise.
2671 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
2673 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2674 Alan Hayward <alan.hayward@arm.com>
2675 David Sherwood <david.sherwood@arm.com>
2677 * machmode.h (opt_mode::else_blk): New function.
2678 (int_mode_for_mode): Declare.
2679 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
2680 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
2682 * cfgexpand.c (expand_debug_expr): Likewise.
2683 * combine.c (gen_lowpart_or_truncate): Likewise.
2684 (gen_lowpart_for_combine): Likewise.
2685 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
2686 * config/avr/avr.c (avr_to_int_mode): Likewise.
2687 (avr_out_plus_1): Likewise.
2688 (avr_out_plus): Likewise.
2689 (avr_out_round): Likewise.
2690 * config/i386/i386.c (ix86_split_to_parts): Likewise.
2691 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2692 (s390_expand_vcond): Likewise.
2693 * config/spu/spu.c (spu_split_immediate): Likewise.
2694 (spu_expand_mov): Likewise.
2695 * dse.c (get_stored_val): Likewise.
2696 * expmed.c (store_bit_field_1): Likewise.
2697 (convert_extracted_bit_field): Use int_mode_for_mode instead of
2699 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
2700 (extract_low_bits): Likewise.
2701 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
2702 handling rather than repeating the check.
2703 (emit_group_store): Likewise.
2704 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
2705 * optabs.c (expand_absneg_bit): Likewise.
2706 (expand_copysign_absneg): Likewise.
2707 (expand_copysign_bit): Likewise.
2708 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
2709 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
2710 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
2711 * var-tracking.c (prepare_call_arguments): Likewise.
2712 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
2713 int_mode_for_mode instead of mode_for_size.
2714 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2717 Alan Hayward <alan.hayward@arm.com>
2718 David Sherwood <david.sherwood@arm.com>
2720 * machmode.h (int_mode_for_size): New function.
2721 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
2722 instead of mode_for_size.
2723 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
2725 * combine.c (expand_field_assignment): Use int_mode_for_size
2726 instead of mode_for_size.
2727 (make_extraction): Likewise.
2728 (simplify_shift_const_1): Likewise.
2729 (simplify_comparison): Likewise.
2730 * dojump.c (do_jump): Likewise.
2731 * dwarf2out.c (mem_loc_descriptor): Likewise.
2732 * emit-rtl.c (init_derived_machine_modes): Likewise.
2733 * expmed.c (flip_storage_order): Likewise.
2734 (convert_extracted_bit_field): Likewise.
2735 * expr.c (copy_blkmode_from_reg): Likewise.
2736 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
2737 * internal-fn.c (expand_mul_overflow): Likewise.
2738 * lower-subreg.c (simple_move): Likewise.
2739 * optabs-libfuncs.c (init_optabs): Likewise.
2740 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2741 * tree.c (vector_type_mode): Likewise.
2742 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
2743 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
2744 * tree-vect-generic.c (expand_vector_parallel): Likewise.
2745 * tree-vect-stmts.c (vectorizable_load): Likewise.
2746 (vectorizable_store): Likewise.
2748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2749 Alan Hayward <alan.hayward@arm.com>
2750 David Sherwood <david.sherwood@arm.com>
2752 * coretypes.h (pod_mode): New type.
2753 (scalar_int_mode_pod): New typedef.
2754 * machmode.h (pod_mode): New class.
2755 (int_n_data_t::m): Change type to scalar_int_mode_pod.
2756 * genmodes.c (emit_mode_int_n): Update accordingly.
2757 * lower-subreg.h (target_lower_subreg): Change type to
2758 scalar_int_mode_pod.
2759 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
2760 scalar_int_mode_pod.
2762 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2763 Alan Hayward <alan.hayward@arm.com>
2764 David Sherwood <david.sherwood@arm.com>
2766 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
2767 machine_mode to scalar_int_mode.
2768 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
2769 (rs6000_option_override_internal): Remove cast to int.
2770 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
2771 machine_mode to scalar_int_mode.
2772 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
2773 (rs6000_option_override_internal): Remove cast to int.
2774 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
2775 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
2777 * config/s390/s390.c (s390_expand_builtin): Likewise.
2778 * coretypes.h (scalar_int_mode): New type.
2779 (opt_scalar_int_mode): New typedef.
2780 * machmode.h (scalar_int_mode): New class.
2781 (scalar_int_mode::includes_p): New function.
2782 (byte_mode): Change type to scalar_int_mode.
2783 (word_mode): Likewise.
2784 (ptr_mode): Likewise.
2785 * emit-rtl.c (byte_mode): Likewise.
2786 (word_mode): Likewise.
2787 (ptr_mode): Likewise.
2788 (init_derived_machine_modes): Update accordingly.
2789 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
2790 and MODE_PARTIAL_INT.
2791 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
2792 opt_scalar_int_mode.
2794 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2795 Alan Hayward <alan.hayward@arm.com>
2796 David Sherwood <david.sherwood@arm.com>
2798 * target.def (libgcc_floating_mode_supported_p): Take a
2800 * doc/tm.texi: Regenerate.
2801 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
2803 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
2804 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
2807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2808 Alan Hayward <alan.hayward@arm.com>
2809 David Sherwood <david.sherwood@arm.com>
2811 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
2812 * doc/tm.texi: Regenerate.
2813 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
2814 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
2815 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
2816 * targhooks.h (default_floatn_mode): Likewise.
2817 * targhooks.c (default_floatn_mode): Likewise.
2818 * tree.c (build_common_tree_nodes): Update accordingly.
2820 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2821 Alan Hayward <alan.hayward@arm.com>
2822 David Sherwood <david.sherwood@arm.com>
2824 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
2825 (mode_iterator::iterate_p): Likewise.
2826 (mode_iterator::get_wider): Likewise.
2827 * expr.c (init_expr_target): Use opt_scalar_float_mode.
2829 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2830 Alan Hayward <alan.hayward@arm.com>
2831 David Sherwood <david.sherwood@arm.com>
2833 * coretypes.h (opt_scalar_float_mode): New typedef.
2834 * machmode.h (float_mode_for_size): New function.
2835 * emit-rtl.c (double_mode): Delete.
2836 (init_emit_once): Use float_mode_for_size.
2837 * stor-layout.c (layout_type): Likewise.
2838 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
2840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2841 Alan Hayward <alan.hayward@arm.com>
2842 David Sherwood <david.sherwood@arm.com>
2844 * output.h (assemble_real): Take a scalar_float_mode.
2845 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
2846 * config/arm/arm.md (consttable_4): Likewise.
2847 (consttable_8): Likewise.
2848 (consttable_16): Likewise.
2849 * config/mips/mips.md (consttable_float): Likewise.
2850 * config/s390/s390.c (s390_output_pool_entry): Likewise.
2851 * varasm.c (assemble_real): Take a scalar_float_mode.
2852 (output_constant_pool_2): Update accordingly.
2853 (output_constant): Likewise.
2855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2856 Alan Hayward <alan.hayward@arm.com>
2857 David Sherwood <david.sherwood@arm.com>
2859 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
2860 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
2861 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
2862 (native_encode_real): Likewise.
2863 (native_interpret_real): Likewise.
2864 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
2865 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2867 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2868 Alan Hayward <alan.hayward@arm.com>
2869 David Sherwood <david.sherwood@arm.com>
2871 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
2872 <scalar_float_mode>. Simplify.
2873 (gen_extend_conv_libfunc): Likewise.
2875 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2876 Alan Hayward <alan.hayward@arm.com>
2877 David Sherwood <david.sherwood@arm.com>
2879 * coretypes.h (scalar_float_mode): New type.
2880 * machmode.h (mode_traits::from_int): Use machine_mode if
2881 USE_ENUM_MODES is defined.
2882 (is_a): New function.
2884 (dyn_cast): Likewise.
2885 (scalar_float_mode): New class.
2886 (scalar_float_mode::includes_p): New function.
2887 (is_float_mode): Likewise.
2888 * gdbhooks.py (MachineModePrinter): New class.
2889 (build_pretty_printer): Use it for scalar_float_mode.
2890 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
2891 (format_helper::format_helper): Turn into a template.
2892 * genmodes.c (get_mode_class): New function.
2893 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
2894 or machine_mode if none.
2895 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
2896 as_a <scalar_float_mode>.
2897 * dwarf2out.c (mem_loc_descriptor): Likewise.
2898 (insert_float): Likewise.
2899 (add_const_value_attribute): Likewise.
2900 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2901 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
2902 (expand_unop): Update accordingly.
2903 (expand_abs_nojump): Likewise.
2904 (expand_copysign_absneg): Take a scalar_float_mode.
2905 (expand_copysign_bit): Likewise.
2906 (expand_copysign): Update accordingly.
2908 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2909 Alan Hayward <alan.hayward@arm.com>
2910 David Sherwood <david.sherwood@arm.com>
2912 * coretypes.h (opt_mode): New class.
2913 * machmode.h (opt_mode): Likewise.
2914 (opt_mode::else_void): New function.
2915 (opt_mode::require): Likewise.
2916 (opt_mode::exists): Likewise.
2917 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
2918 (GET_MODE_2XWIDER_MODE): Likewise.
2919 (mode_iterator::get_wider): Update accordingly.
2920 (mode_iterator::get_2xwider): Likewise.
2921 (mode_iterator::get_known_wider): Likewise, turning into a template.
2922 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
2923 forcing a wider mode to exist.
2924 * config/cr16/cr16.h (LONG_REG_P): Likewise.
2925 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2926 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
2927 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
2928 * lower-subreg.c (init_lower_subreg): Likewise.
2929 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
2930 on the final iteration.
2931 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
2932 a wider mode exists before asking for a move pattern.
2933 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
2934 forcing a wider mode to exist.
2935 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
2936 returning false if no such mode exists.
2937 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
2938 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
2939 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
2940 Avoid checking for a MODE_INT if we already know the mode is not a
2942 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
2943 forcing a wider mode to exist.
2944 (expmed_mult_highpart_optab): Likewise.
2945 (expmed_mult_highpart): Likewise.
2946 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
2948 * lto-streamer-in.c (lto_input_mode_table): Likewise.
2949 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
2950 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
2951 * internal-fn.c (expand_mul_overflow): Update use of
2952 GET_MODE_2XWIDER_MODE.
2953 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2954 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
2955 GET_MODE_WIDER_MODE.
2956 (convert_plusminus_to_widen): Likewise.
2957 * tree-switch-conversion.c (array_value_type): Likewise.
2958 * var-tracking.c (emit_note_insn_var_location): Likewise.
2959 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2960 Return false inside rather than outside the loop if no wider mode
2962 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
2963 and GET_MODE_2XWIDER_MODE
2964 (can_compare_p): Use else_void.
2965 * gdbhooks.py (OptMachineModePrinter): New class.
2966 (build_pretty_printer): Use it for opt_mode.
2968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2969 Alan Hayward <alan.hayward@arm.com>
2970 David Sherwood <david.sherwood@arm.com>
2972 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
2973 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
2975 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2976 Alan Hayward <alan.hayward@arm.com>
2977 David Sherwood <david.sherwood@arm.com>
2979 * machmode.h (mode_traits): New structure.
2980 (get_narrowest_mode): New function.
2981 (mode_iterator::start): Likewise.
2982 (mode_iterator::iterate_p): Likewise.
2983 (mode_iterator::get_wider): Likewise.
2984 (mode_iterator::get_known_wider): Likewise.
2985 (mode_iterator::get_2xwider): Likewise.
2986 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
2987 (FOR_EACH_MODE): Likewise.
2988 (FOR_EACH_MODE_FROM): Likewise.
2989 (FOR_EACH_MODE_UNTIL): Likewise.
2990 (FOR_EACH_WIDER_MODE): Likewise.
2991 (FOR_EACH_2XWIDER_MODE): Likewise.
2992 * builtins.c (expand_builtin_strlen): Use new mode iterators.
2993 * combine.c (simplify_comparison): Likewise
2994 * config/i386/i386.c (type_natural_mode): Likewise.
2995 * cse.c (cse_insn): Likewise.
2996 * dse.c (find_shift_sequence): Likewise.
2997 * emit-rtl.c (init_derived_machine_modes): Likewise.
2998 (init_emit_once): Likewise.
2999 * explow.c (hard_function_value): Likewise.
3000 * expmed.c (extract_fixed_bit_field_1): Likewise.
3001 (extract_bit_field_1): Likewise.
3002 (expand_divmod): Likewise.
3003 (emit_store_flag_1): Likewise.
3004 * expr.c (init_expr_target): Likewise.
3005 (convert_move): Likewise.
3006 (alignment_for_piecewise_move): Likewise.
3007 (widest_int_mode_for_size): Likewise.
3008 (emit_block_move_via_movmem): Likewise.
3009 (copy_blkmode_to_reg): Likewise.
3010 (set_storage_via_setmem): Likewise.
3011 (compress_float_constant): Likewise.
3012 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3013 * optabs-query.c (get_best_extraction_insn): Likewise.
3014 * optabs.c (expand_binop): Likewise.
3015 (expand_twoval_unop): Likewise.
3016 (expand_twoval_binop): Likewise.
3017 (widen_leading): Likewise.
3018 (widen_bswap): Likewise.
3019 (expand_parity): Likewise.
3020 (expand_unop): Likewise.
3021 (prepare_cmp_insn): Likewise.
3022 (prepare_float_lib_cmp): Likewise.
3023 (expand_float): Likewise.
3024 (expand_fix): Likewise.
3025 (expand_sfix_optab): Likewise.
3026 * postreload.c (move2add_use_add2_insn): Likewise.
3027 * reg-stack.c (reg_to_stack): Likewise.
3028 * reginfo.c (choose_hard_reg_mode): Likewise.
3029 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3030 * stor-layout.c (mode_for_size): Likewise.
3031 (smallest_mode_for_size): Likewise.
3032 (mode_for_vector): Likewise.
3033 (finish_bitfield_representative): Likewise.
3034 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
3035 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
3036 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3037 * var-tracking.c (prepare_call_arguments): Likewise.
3039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3040 Alan Hayward <alan.hayward@arm.com>
3041 David Sherwood <david.sherwood@arm.com>
3043 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
3044 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
3045 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
3046 * machmode.h (mode_size): Move earlier in file.
3047 (mode_precision): Likewise.
3048 (mode_inner): Likewise.
3049 (mode_nunits): Likewise.
3050 (mode_unit_size): Likewise.
3051 (unit_unit_precision): Likewise.
3052 (mode_wider): Likewise.
3053 (mode_2xwider): Likewise.
3054 (machine_mode): New class.
3055 (mode_to_bytes): New function.
3056 (mode_to_bits): Likewise.
3057 (mode_to_precision): Likewise.
3058 (mode_to_inner): Likewise.
3059 (mode_to_unit_size): Likewise.
3060 (mode_to_unit_precision): Likewise.
3061 (mode_to_nunits): Likewise.
3062 (GET_MODE_SIZE): Use mode_to_bytes.
3063 (GET_MODE_BITSIZE): Use mode_to_bits.
3064 (GET_MODE_PRECISION): Use mode_to_precision.
3065 (GET_MODE_INNER): Use mode_to_inner.
3066 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
3067 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
3068 (GET_MODE_NUNITS): Use mode_to_nunits.
3069 * system.h (ALWAYS_INLINE): New macro.
3070 * config/powerpcspe/powerpcspe-c.c
3071 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
3072 int for arg1_mode and arg2_mode.
3074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3075 Alan Hayward <alan.hayward@arm.com>
3076 David Sherwood <david.sherwood@arm.com>
3078 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
3079 Prefix mode names with E_ in case statements.
3080 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3081 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
3082 (aarch64_split_simd_move): Likewise.
3083 (aarch64_gen_storewb_pair): Likewise.
3084 (aarch64_gen_loadwb_pair): Likewise.
3085 (aarch64_gen_store_pair): Likewise.
3086 (aarch64_gen_load_pair): Likewise.
3087 (aarch64_get_condition_code_1): Likewise.
3088 (aarch64_constant_pool_reload_icode): Likewise.
3089 (get_rsqrte_type): Likewise.
3090 (get_rsqrts_type): Likewise.
3091 (get_recpe_type): Likewise.
3092 (get_recps_type): Likewise.
3093 (aarch64_gimplify_va_arg_expr): Likewise.
3094 (aarch64_simd_container_mode): Likewise.
3095 (aarch64_emit_load_exclusive): Likewise.
3096 (aarch64_emit_store_exclusive): Likewise.
3097 (aarch64_expand_compare_and_swap): Likewise.
3098 (aarch64_gen_atomic_cas): Likewise.
3099 (aarch64_emit_bic): Likewise.
3100 (aarch64_emit_atomic_swap): Likewise.
3101 (aarch64_emit_atomic_load_op): Likewise.
3102 (aarch64_evpc_trn): Likewise.
3103 (aarch64_evpc_uzp): Likewise.
3104 (aarch64_evpc_zip): Likewise.
3105 (aarch64_evpc_ext): Likewise.
3106 (aarch64_evpc_rev): Likewise.
3107 (aarch64_evpc_dup): Likewise.
3108 (aarch64_gen_ccmp_first): Likewise.
3109 (aarch64_gen_ccmp_next): Likewise.
3110 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3111 (alpha_emit_xfloating_libcall): Likewise.
3112 (emit_insxl): Likewise.
3113 (alpha_arg_type): Likewise.
3114 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
3115 (arc_preferred_simd_mode): Likewise.
3116 (arc_secondary_reload): Likewise.
3117 (get_arc_condition_code): Likewise.
3118 (arc_print_operand): Likewise.
3119 (arc_legitimate_constant_p): Likewise.
3120 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3121 * config/arc/arc.md (casesi_load): Likewise.
3122 (casesi_compact_jump): Likewise.
3123 * config/arc/predicates.md (proper_comparison_operator): Likewise.
3124 (cc_use_register): Likewise.
3125 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3126 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
3127 (arm_init_iwmmxt_builtins): Likewise.
3128 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
3129 (neon_expand_vector_init): Likewise.
3130 (arm_attr_length_move_neon): Likewise.
3131 (maybe_get_arm_condition_code): Likewise.
3132 (arm_emit_vector_const): Likewise.
3133 (arm_preferred_simd_mode): Likewise.
3134 (arm_output_iwmmxt_tinsr): Likewise.
3135 (thumb1_output_casesi): Likewise.
3136 (thumb2_output_casesi): Likewise.
3137 (arm_emit_load_exclusive): Likewise.
3138 (arm_emit_store_exclusive): Likewise.
3139 (arm_expand_compare_and_swap): Likewise.
3140 (arm_evpc_neon_vuzp): Likewise.
3141 (arm_evpc_neon_vzip): Likewise.
3142 (arm_evpc_neon_vrev): Likewise.
3143 (arm_evpc_neon_vtrn): Likewise.
3144 (arm_evpc_neon_vext): Likewise.
3145 (arm_validize_comparison): Likewise.
3146 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
3147 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
3148 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
3149 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
3150 (c6x_preferred_simd_mode): Likewise.
3151 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
3152 (epiphany_rtx_costs): Likewise.
3153 * config/epiphany/predicates.md (proper_comparison_operator):
3155 * config/frv/frv.c (condexec_memory_operand): Likewise.
3156 (frv_emit_move): Likewise.
3157 (output_move_single): Likewise.
3158 (output_condmove_single): Likewise.
3159 (frv_hard_regno_mode_ok): Likewise.
3160 (frv_matching_accg_mode): Likewise.
3161 * config/h8300/h8300.c (split_adds_subs): Likewise.
3162 (h8300_rtx_costs): Likewise.
3163 (h8300_print_operand): Likewise.
3164 (compute_mov_length): Likewise.
3165 (output_logical_op): Likewise.
3166 (compute_logical_op_length): Likewise.
3167 (compute_logical_op_cc): Likewise.
3168 (h8300_shift_needs_scratch_p): Likewise.
3169 (output_a_shift): Likewise.
3170 (compute_a_shift_length): Likewise.
3171 (compute_a_shift_cc): Likewise.
3172 (expand_a_rotate): Likewise.
3173 (output_a_rotate): Likewise.
3174 * config/i386/i386.c (classify_argument): Likewise.
3175 (function_arg_advance_32): Likewise.
3176 (function_arg_32): Likewise.
3177 (function_arg_64): Likewise.
3178 (function_value_64): Likewise.
3179 (ix86_gimplify_va_arg): Likewise.
3180 (ix86_legitimate_constant_p): Likewise.
3181 (put_condition_code): Likewise.
3182 (split_double_mode): Likewise.
3183 (ix86_avx256_split_vector_move_misalign): Likewise.
3184 (ix86_expand_vector_logical_operator): Likewise.
3185 (ix86_split_idivmod): Likewise.
3186 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
3187 (ix86_build_const_vector): Likewise.
3188 (ix86_build_signbit_mask): Likewise.
3189 (ix86_match_ccmode): Likewise.
3190 (ix86_cc_modes_compatible): Likewise.
3191 (ix86_expand_branch): Likewise.
3192 (ix86_expand_sse_cmp): Likewise.
3193 (ix86_expand_sse_movcc): Likewise.
3194 (ix86_expand_int_sse_cmp): Likewise.
3195 (ix86_expand_vec_perm_vpermi2): Likewise.
3196 (ix86_expand_vec_perm): Likewise.
3197 (ix86_expand_sse_unpack): Likewise.
3198 (ix86_expand_int_addcc): Likewise.
3199 (ix86_split_to_parts): Likewise.
3200 (ix86_vectorize_builtin_gather): Likewise.
3201 (ix86_vectorize_builtin_scatter): Likewise.
3202 (avx_vpermilp_parallel): Likewise.
3203 (inline_memory_move_cost): Likewise.
3204 (ix86_tieable_integer_mode_p): Likewise.
3205 (x86_maybe_negate_const_int): Likewise.
3206 (ix86_expand_vector_init_duplicate): Likewise.
3207 (ix86_expand_vector_init_one_nonzero): Likewise.
3208 (ix86_expand_vector_init_one_var): Likewise.
3209 (ix86_expand_vector_init_concat): Likewise.
3210 (ix86_expand_vector_init_interleave): Likewise.
3211 (ix86_expand_vector_init_general): Likewise.
3212 (ix86_expand_vector_set): Likewise.
3213 (ix86_expand_vector_extract): Likewise.
3214 (emit_reduc_half): Likewise.
3215 (ix86_emit_i387_round): Likewise.
3216 (ix86_mangle_type): Likewise.
3217 (ix86_expand_round_sse4): Likewise.
3218 (expand_vec_perm_blend): Likewise.
3219 (canonicalize_vector_int_perm): Likewise.
3220 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
3221 (expand_vec_perm_1): Likewise.
3222 (expand_vec_perm_interleave3): Likewise.
3223 (expand_vec_perm_even_odd_pack): Likewise.
3224 (expand_vec_perm_even_odd_1): Likewise.
3225 (expand_vec_perm_broadcast_1): Likewise.
3226 (ix86_vectorize_vec_perm_const_ok): Likewise.
3227 (ix86_expand_vecop_qihi): Likewise.
3228 (ix86_expand_mul_widen_hilo): Likewise.
3229 (ix86_expand_sse2_abs): Likewise.
3230 (ix86_expand_pextr): Likewise.
3231 (ix86_expand_pinsr): Likewise.
3232 (ix86_preferred_simd_mode): Likewise.
3233 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
3234 * config/i386/sse.md (*andnot<mode>3): Likewise.
3235 (<mask_codefor><code><mode>3<mask_name>): Likewise.
3236 (*<code><mode>3): Likewise.
3237 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
3238 (ia64_expand_atomic_op): Likewise.
3239 (ia64_arg_type): Likewise.
3240 (ia64_mode_to_int): Likewise.
3241 (ia64_scalar_mode_supported_p): Likewise.
3242 (ia64_vector_mode_supported_p): Likewise.
3243 (expand_vec_perm_broadcast): Likewise.
3244 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
3245 (iq2000_function_arg_advance): Likewise.
3246 (iq2000_function_arg): Likewise.
3247 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
3248 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
3249 (m68k_libcall_value): Likewise.
3250 (m68k_function_value): Likewise.
3251 (sched_attr_op_type): Likewise.
3252 * config/mcore/mcore.c (mcore_output_move): Likewise.
3253 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
3255 (microblaze_function_arg): Likewise.
3256 * config/mips/mips.c (mips16_build_call_stub): Likewise.
3257 (mips_print_operand): Likewise.
3258 (mips_mode_ok_for_mov_fmt_p): Likewise.
3259 (mips_vector_mode_supported_p): Likewise.
3260 (mips_preferred_simd_mode): Likewise.
3261 (mips_expand_vpc_loongson_even_odd): Likewise.
3262 (mips_expand_vec_unpack): Likewise.
3263 (mips_expand_vi_broadcast): Likewise.
3264 (mips_expand_vector_init): Likewise.
3265 (mips_expand_vec_reduc): Likewise.
3266 (mips_expand_msa_cmp): Likewise.
3267 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
3268 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
3269 (cc_flags_for_mode): Likewise.
3270 * config/msp430/msp430.c (msp430_print_operand): Likewise.
3271 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
3272 (nds32_output_casesi_pc_relative): Likewise.
3273 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3274 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
3275 (nvptx_gen_unpack): Likewise.
3276 (nvptx_gen_pack): Likewise.
3277 (nvptx_gen_shuffle): Likewise.
3278 (nvptx_gen_wcast): Likewise.
3279 (nvptx_preferred_simd_mode): Likewise.
3280 * config/pa/pa.c (pa_secondary_reload): Likewise.
3281 * config/pa/predicates.md (base14_operand): Likewise.
3282 * config/powerpcspe/powerpcspe-c.c
3283 (altivec_resolve_overloaded_builtin): Likewise.
3284 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
3286 (rs6000_preferred_simd_mode): Likewise.
3287 (output_vec_const_move): Likewise.
3288 (rs6000_expand_vector_extract): Likewise.
3289 (rs6000_split_vec_extract_var): Likewise.
3290 (reg_offset_addressing_ok_p): Likewise.
3291 (rs6000_legitimate_offset_address_p): Likewise.
3292 (rs6000_legitimize_address): Likewise.
3293 (rs6000_emit_set_const): Likewise.
3294 (rs6000_const_vec): Likewise.
3295 (rs6000_emit_move): Likewise.
3296 (spe_build_register_parallel): Likewise.
3297 (rs6000_darwin64_record_arg_recurse): Likewise.
3298 (swap_selector_for_mode): Likewise.
3299 (spe_init_builtins): Likewise.
3300 (paired_init_builtins): Likewise.
3301 (altivec_init_builtins): Likewise.
3302 (do_load_for_compare): Likewise.
3303 (rs6000_generate_compare): Likewise.
3304 (rs6000_expand_float128_convert): Likewise.
3305 (emit_load_locked): Likewise.
3306 (emit_store_conditional): Likewise.
3307 (rs6000_output_function_epilogue): Likewise.
3308 (rs6000_handle_altivec_attribute): Likewise.
3309 (rs6000_function_value): Likewise.
3310 (emit_fusion_gpr_load): Likewise.
3311 (emit_fusion_p9_load): Likewise.
3312 (emit_fusion_p9_store): Likewise.
3313 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
3314 (fusion_gpr_mem_load): Likewise.
3315 (fusion_addis_mem_combo_load): Likewise.
3316 (fusion_addis_mem_combo_store): Likewise.
3317 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
3318 (fusion_gpr_mem_load): Likewise.
3319 (fusion_addis_mem_combo_load): Likewise.
3320 (fusion_addis_mem_combo_store): Likewise.
3321 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3323 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
3324 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
3325 (rs6000_preferred_simd_mode): Likewise.
3326 (output_vec_const_move): Likewise.
3327 (rs6000_expand_vector_extract): Likewise.
3328 (rs6000_split_vec_extract_var): Likewise.
3329 (reg_offset_addressing_ok_p): Likewise.
3330 (rs6000_legitimate_offset_address_p): Likewise.
3331 (rs6000_legitimize_address): Likewise.
3332 (rs6000_emit_set_const): Likewise.
3333 (rs6000_const_vec): Likewise.
3334 (rs6000_emit_move): Likewise.
3335 (rs6000_darwin64_record_arg_recurse): Likewise.
3336 (swap_selector_for_mode): Likewise.
3337 (paired_init_builtins): Likewise.
3338 (altivec_init_builtins): Likewise.
3339 (rs6000_expand_float128_convert): Likewise.
3340 (emit_load_locked): Likewise.
3341 (emit_store_conditional): Likewise.
3342 (rs6000_output_function_epilogue): Likewise.
3343 (rs6000_handle_altivec_attribute): Likewise.
3344 (rs6000_function_value): Likewise.
3345 (emit_fusion_gpr_load): Likewise.
3346 (emit_fusion_p9_load): Likewise.
3347 (emit_fusion_p9_store): Likewise.
3348 * config/rx/rx.c (rx_gen_move_template): Likewise.
3349 (flags_from_mode): Likewise.
3350 * config/s390/predicates.md (s390_alc_comparison): Likewise.
3351 (s390_slb_comparison): Likewise.
3352 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
3353 (s390_vector_mode_supported_p): Likewise.
3354 (s390_cc_modes_compatible): Likewise.
3355 (s390_match_ccmode_set): Likewise.
3356 (s390_canonicalize_comparison): Likewise.
3357 (s390_emit_compare_and_swap): Likewise.
3358 (s390_branch_condition_mask): Likewise.
3359 (s390_rtx_costs): Likewise.
3360 (s390_secondary_reload): Likewise.
3361 (__SECONDARY_RELOAD_CASE): Likewise.
3362 (s390_expand_cs): Likewise.
3363 (s390_preferred_simd_mode): Likewise.
3364 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
3365 * config/sh/sh.c (sh_print_operand): Likewise.
3366 (dump_table): Likewise.
3367 (sh_secondary_reload): Likewise.
3368 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3369 * config/sh/sh.md (casesi_worker_1): Likewise.
3370 (casesi_worker_2): Likewise.
3371 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
3372 (fcc_comparison_operator): Likewise.
3373 * config/sparc/sparc.c (sparc_expand_move): Likewise.
3374 (emit_soft_tfmode_cvt): Likewise.
3375 (sparc_preferred_simd_mode): Likewise.
3376 (output_cbranch): Likewise.
3377 (sparc_print_operand): Likewise.
3378 (sparc_expand_vec_perm_bmask): Likewise.
3379 (vector_init_bshuffle): Likewise.
3380 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3381 (spu_vector_mode_supported_p): Likewise.
3382 (spu_expand_insv): Likewise.
3383 (spu_emit_branch_or_set): Likewise.
3384 (spu_handle_vector_attribute): Likewise.
3385 (spu_builtin_splats): Likewise.
3386 (spu_builtin_extract): Likewise.
3387 (spu_builtin_promote): Likewise.
3388 (spu_expand_sign_extend): Likewise.
3389 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3390 (tilegx_simd_int): Likewise.
3391 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
3392 (tilepro_simd_int): Likewise.
3393 * config/v850/v850.c (const_double_split): Likewise.
3394 (v850_print_operand): Likewise.
3395 (ep_memory_offset): Likewise.
3396 * config/vax/vax.c (vax_rtx_costs): Likewise.
3397 (vax_output_int_move): Likewise.
3398 (vax_output_int_add): Likewise.
3399 (vax_output_int_subtract): Likewise.
3400 * config/visium/predicates.md (visium_branch_operator): Likewise.
3401 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
3402 (visium_print_operand_address): Likewise.
3403 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3404 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
3405 (xtensa_expand_conditional_branch): Likewise.
3406 (xtensa_copy_incoming_a7): Likewise.
3407 (xtensa_output_literal): Likewise.
3408 * dfp.c (decimal_real_maxval): Likewise.
3409 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3411 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3412 Alan Hayward <alan.hayward@arm.com>
3413 David Sherwood <david.sherwood@arm.com>
3415 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
3416 (mode_nunits_inline): Likewise.
3417 (mode_inner_inline): Likewise.
3418 (mode_unit_size_inline): Likewise.
3419 (mode_unit_precision_inline): Likewise.
3420 (emit_insn_modes_h): Likewise. Also emit a #define of the
3422 (emit_mode_wider): Add an E_ prefix to mode names.
3423 (emit_mode_complex): Likewise.
3424 (emit_mode_inner): Likewise.
3425 (emit_mode_adjustments): Likewise.
3426 (emit_mode_int_n): Likewise.
3427 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
3428 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
3429 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
3430 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
3431 (CRC32_BUILTIN, ENTRY): Likewise.
3432 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
3433 (aarch64_pop_regs): Likewise.
3434 (aarch64_process_components): Likewise.
3435 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
3436 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
3437 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
3438 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
3439 * config/arm/arm.c (arm_init_libfuncs): Likewise.
3440 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
3442 * config/i386/i386-builtin.def (pcmpestr): Likewise.
3443 (pcmpistr): Likewise.
3444 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
3445 * config/mmix/mmix.c (mmix_output_condition): Likewise.
3446 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
3448 * config/rl78/rl78.c (mduc_regs): Likewise.
3449 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3450 (htm_expand_builtin): Likewise.
3451 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
3452 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
3453 * config/xtensa/xtensa.c (print_operand): Likewise.
3454 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
3455 (NUM_MODE_VECTOR_INT): Likewise.
3456 * genoutput.c (null_operand): Likewise.
3457 (output_operand_data): Likewise.
3458 * genrecog.c (print_parameter_value): Likewise.
3459 * lra.c (debug_operand_data): Likewise.
3461 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3463 * dwarf2out.c (scompare_loc_descriptor_wide)
3464 (scompare_loc_descriptor_narrow): New functions, split out from...
3465 (scompare_loc_descriptor): ...here.
3466 * expmed.c (emit_store_flag_int): New function, split out from...
3467 (emit_store_flag): ...here.
3469 2017-08-30 Richard Biener <rguenther@suse.de>
3471 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
3472 (dwarf2out_early_finish): Move setting of AT_pubnames from
3473 early debug output to early finish.
3475 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3477 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
3478 and -mdata-region to the assembler.
3480 2017-08-30 Richard Biener <rguenther@suse.de>
3482 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
3484 (gen_subprogram_die): Add DW_AT_object_pointer only early.
3485 (dwarf2out_early_global_decl): Only generate a DIE for the
3486 abstract origin if it doesn't already exist or is a declaration DIE.
3487 (resolve_addr): Do not add the linkage name twice when
3488 generating a stub DIE for the DW_TAG_GNU_call_site target.
3490 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3492 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3493 Use machine_mode rather than int for arg1_mode.
3495 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
3498 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
3499 that the second argument of the built-in functions to unpack
3500 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
3501 switch statement instead a lot of if statements.
3502 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
3503 Allow 64-bit values to be in Altivec registers as well as
3504 traditional floating point registers.
3505 (pack<mode>, FMOVE128_VSX iterator): Likewise.
3507 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
3509 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
3510 MAX_REGS_PER_ADDRESS == 1.
3512 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
3514 * config/i386/i386.opt (flag_fentry): Do not init to -1.
3515 * config/i386/i386.c (ix86_option_override_internal): Simplify
3516 setting of opts->x_flag_entry.
3518 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3519 Jakub Jelinek <jakub@redhat.com>
3520 Richard Biener <rguenther@suse.de>
3522 PR tree-optimization/81503
3523 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
3524 folded constant fits in the target type; reorder tests for clarity.
3526 2017-08-29 Martin Liska <mliska@suse.cz>
3528 * passes.def: Include pass_lower_switch.
3529 * stmt.c (dump_case_nodes): Remove and move to
3530 tree-switch-conversion.
3531 (case_values_threshold): Likewise.
3532 (expand_switch_as_decision_tree_p): Likewise.
3533 (emit_case_decision_tree): Likewise.
3534 (expand_case): Likewise.
3535 (balance_case_nodes): Likewise.
3536 (node_has_low_bound): Likewise.
3537 (node_has_high_bound): Likewise.
3538 (node_is_bounded): Likewise.
3539 (emit_case_nodes): Likewise.
3540 (struct simple_case_node): New struct.
3541 (add_case_node): Remove.
3542 (emit_case_dispatch_table): Use vector instead of case_list.
3543 (reset_out_edges_aux): Remove.
3544 (compute_cases_per_edge): Likewise.
3545 (expand_case): Build list of simple_case_node.
3546 (expand_sjlj_dispatch_table): Use it.
3547 * tree-switch-conversion.c (struct case_node): Moved from
3548 stmt.c and adjusted.
3549 (emit_case_nodes): Likewise.
3550 (node_has_low_bound): Likewise.
3551 (node_has_high_bound): Likewise.
3552 (node_is_bounded): Likewise.
3553 (case_values_threshold): Likewise.
3554 (reset_out_edges_aux): Likewise.
3555 (compute_cases_per_edge): Likewise.
3556 (add_case_node): Likewise.
3557 (dump_case_nodes): Likewise.
3558 (balance_case_nodes): Likewise.
3559 (expand_switch_as_decision_tree_p): Likewise.
3560 (emit_jump): Likewise.
3561 (emit_case_decision_tree): Likewise.
3562 (try_switch_expansion): Likewise.
3563 (do_jump_if_equal): Likewise.
3564 (emit_cmp_and_jump_insns): Likewise.
3565 (fix_phi_operands_for_edge): New function.
3566 (record_phi_operand_mapping): Likewise.
3567 (class pass_lower_switch): New pass.
3568 (pass_lower_switch::execute): New function.
3569 (make_pass_lower_switch): Likewise.
3570 (conditional_probability):
3571 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
3572 * tree-pass.h: Add make_pass_lower_switch.
3574 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3577 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
3580 2017-08-29 Richard Biener <rguenther@suse.de>
3582 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
3583 we do not add a DW_AT_inline attribute twice.
3584 (gen_subprogram_die): Remove code setting DW_AT_inline on
3585 DECL_ABSTRACT_P nodes.
3587 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
3589 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
3590 calls to internal functions.
3591 (gimplify_modify_expr): Likewise.
3592 * tree-call-cdce.c (use_internal_fn): Likewise.
3593 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
3594 (convert_to_divmod): Set the nothrow flag.
3595 * tree-if-conv.c (predicate_mem_writes): Likewise.
3596 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3597 (vectorizable_call): Likewise.
3598 (vectorizable_store): Likewise.
3599 (vectorizable_load): Likewise.
3600 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
3601 (vect_recog_mask_conversion_pattern): Likewise.
3603 2017-08-29 Martin Liska <mliska@suse.cz>
3606 * gcc.c (driver_handle_option): Add new argument.
3607 * opts-common.c (handle_option): Pass
3608 target_option_override_hook.
3609 * opts-global.c (lang_handle_option): Add new option.
3610 (set_default_handlers): Add new argument.
3611 (decode_options): Likewise.
3612 * opts.c (target_handle_option): Likewise.
3613 (common_handle_option): Call target_option_override_hook.
3614 * opts.h (struct cl_option_handler_func): Add hook for
3615 target option override.
3616 (struct cl_option_handlers): Likewise.
3617 (set_default_handlers): Add new argument.
3618 (decode_options): Likewise.
3619 (common_handle_option): Likewise.
3620 (target_handle_option): Likewise.
3621 * toplev.c (toplev::main): Pass targetm.target_option.override
3624 2017-08-29 Richard Biener <rguenther@suse.de>
3625 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3627 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
3628 life to the active subtree.
3630 2017-08-28 Jeff Law <law@redhat.com>
3632 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
3633 derive_equivalences.
3634 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
3636 (edge_info::derive_equivalences): New private member function
3638 * tree-ssa-dom.c (class edge_info): Changed from a struct
3639 to a class. Add ctor/dtor, methods and data members.
3640 (edge_info::edge_info): Renamed from allocate_edge_info.
3641 Initialize additional members.
3642 (edge_info::~edge_info): New.
3643 (free_dom_edge_info): Delete the edge info.
3644 (record_edge_info): Use new class & associated member functions.
3645 Tighten forms for testing for edge equivalences.
3646 (record_temporary_equivalences): Iterate over the simple
3647 equivalences rather than assuming there's only one per edge.
3648 (cprop_into_successor_phis): Iterate over the simple
3649 equivalences rather than assuming there's only one per edge.
3650 (optimize_stmt): Use operand_equal_p rather than pointer
3651 equality for mini-DSE code.
3653 2017-08-28 Nathan Sidwell <nathan@acm.org>
3655 * gcc.c (execute): Fold SIGPIPE handling into switch
3656 statement. Adjust internal error message.
3658 2017-08-28 Richard Biener <rguenther@suse.de>
3661 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
3662 Do nothing for removed DIEs.
3664 2017-08-28 Richard Biener <rguenther@suse.de>
3666 PR tree-optimization/81977
3667 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
3670 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
3673 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
3674 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
3675 using targetm.gen_mem_thread_fence.
3677 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
3680 * config/i386/i386.md (*<btsc><mode>): Change operand 2
3681 predicate to register_operand. Reorder operands.
3682 (*btr<mode>): Ditto.
3683 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
3684 (*btr<mode>_mask): Ditto.
3686 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
3688 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
3689 * config/rs6000/xmmintrin.h: New file.
3690 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
3692 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3695 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
3696 parameter and_insn and return it.
3697 (recombine_lvx_pattern): Insert a copy to ensure availability of
3698 the base register of the copied masking operation at the point of
3699 the instruction replacement.
3700 (recombine_stvx_pattern): Likewise.
3702 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3704 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
3705 undocumented switches.
3706 (-mpower9-dform-vector): Likewise.
3707 (-mpower9-dform): Likewise.
3708 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
3709 comments to delete references to -mpower9-dform* switches.
3710 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
3711 Delete reference to -mpower9-dform* switches, test for
3712 -mpower9-vector instead.
3713 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
3714 (OTHER_P9_VECTOR_MASKS): Likewise.
3715 (POWERPC_MASKS): Likewise.
3716 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
3717 tests against -mpower9-dform* to -mpower9-vector. Delete code
3718 that checked for -mpower9-dform* consistancy with other options.
3719 Add test for -mpower9-misc to enable other power9 switches.
3720 (rs6000_init_hard_regno_mode_ok): Likewise.
3721 (rs6000_option_override_internal): Likewise.
3722 (rs6000_emit_prologue): Likewise.
3723 (rs6000_emit_epilogue): Likewise.
3724 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
3725 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
3726 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
3728 (emit_fusion_p9_store): Likewise.
3729 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
3730 resetting these macros if the assembler does not support ISA 3.0
3732 (TARGET_P9_DFORM_VECTOR): Likewise.
3733 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
3734 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
3736 2017-08-25 Alan Modra <amodra@gmail.com>
3738 PR rtl-optimization/81747
3739 * cse.c (cse_extended_basic_block): Don't attempt to record
3740 equivalences for degenerate conditional branches that branch
3741 to their fall-through.
3743 2017-08-24 Martin Sebor <msebor@redhat.com>
3746 * gimple-fold.c (size_must_be_zero_p): New function.
3747 (gimple_fold_builtin_memory_op): Call it.
3749 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
3751 * config/rs6000/mm_malloc.h: New file.
3753 2017-08-24 Bin Cheng <bin.cheng@arm.com>
3755 PR tree-optimization/81913
3756 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
3757 analysis when either IVs in condition can wrap.
3759 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
3761 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
3762 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
3764 2017-08-24 Richard Biener <rguenther@suse.de>
3767 * targhooks.c (default_target_can_inline_p): Properly
3768 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
3769 is present and always compare.
3770 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
3771 infer -mfpmath=sse from TARGET_SSE_P.
3772 (ix86_can_inline_p): Properly use target_option_default_node when
3773 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
3775 2017-08-24 Richard Biener <rguenther@suse.de>
3778 * dwarf2out.c (output_die): Handle flag_generate_offload like
3780 (output_comp_unit): Likewise.
3781 (gen_array_type_die): Likewise.
3782 (dwarf2out_early_finish): Likewise.
3783 (note_variable_value_in_expr): Likewise.
3784 (dwarf2out_finish): Likewise. Adjust assert.
3785 * cgraphunit.c (symbol_table::compile): Move setting of
3786 flag_generate_offload earlier ...
3787 (symbol_table::finalize_compilation_unit): ... here, before
3788 early debug finalization.
3790 2017-08-24 Richard Biener <rguenther@suse.de>
3792 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
3793 and ipa-fnsummary.h.
3794 (ix86_can_inline_p): When ix86_fpmath flags do not match
3795 check whether the callee uses FP math at all.
3797 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
3800 * tree-ssanames.c (get_nonzero_bits): Use element_precision
3801 instead of TYPE_PRECISION.
3803 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
3804 Alan Hayward <alan.hayward@arm.com>
3805 David Sherwood <david.sherwood@arm.com>
3807 * combine.c (make_extraction): Use subreg_offset_from_lsb.
3809 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
3811 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
3813 (ix86_frame::stack_realign_allocate): New field.
3814 (struct machine_frame_state): Modify comments.
3815 (machine_frame_state::sp_realigned_fp_end): New field.
3816 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
3818 (sp_valid_at): Add assertion to assure no attempt to access invalid
3819 offset of a realigned stack.
3820 (fp_valid_at): Likewise.
3821 (choose_baseaddr): Modify comments.
3822 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
3823 ix86_expand_prologue.
3824 (ix86_expand_prologue): Modify stack realignment and allocation.
3825 (ix86_expand_epilogue): Modify comments.
3826 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
3827 avx2_runtime, avx512f, and avx512f_runtime.
3829 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
3831 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
3832 (mstackrealign): Do not init to -1.
3833 * config/i386/i386.c (ix86_option_override_internal):
3834 Check opts_set, not opts when setting default value of
3835 opts->x_ix86_force_align_arg_pointer.
3837 2017-08-23 Richard Biener <rguenther@suse.de>
3839 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
3840 lang_hooks.decl_printable_name.
3841 * print-rtl-function.c (print_rtx_function): Likewise.
3842 * tree-pretty-print.c (dump_function_header): Likewise.
3844 2017-08-23 Richard Biener <rguenther@suse.de>
3847 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
3848 -g0 at compile-time.
3850 2017-08-23 Tamar Christina <tamar.christina@arm.com>
3853 * doc/sourcebuild.texi (Other hardware attributes):
3856 2017-08-23 Tamar Christina <tamar.christina@arm.com>
3859 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
3860 Fix single-use check.
3862 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
3864 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
3866 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
3868 * doc/install.texi: Modify to add more details on running selected
3871 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
3873 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
3874 is combined with -mabi=ms.
3875 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
3878 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3880 PR tree-optimization/81488
3881 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
3882 and cached_basis fields.
3883 (MAX_SPREAD): New constant.
3884 (alloc_cand_and_find_basis): Initialize new fields.
3885 (clear_visited): New function.
3886 (create_phi_basis_1): Rename from create_phi_basis, set visited
3887 and cached_basis fields.
3888 (create_phi_basis): New wrapper function.
3889 (phi_add_costs_1): Rename from phi_add_costs, add spread
3890 parameter, set visited field, short-circuit when limits reached.
3891 (phi_add_costs): New wrapper function.
3892 (record_phi_increments_1): Rename from record_phi_increments, set
3894 (record_phi_increments): New wrapper function.
3895 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
3896 (phi_incr_cost): New wrapper function.
3897 (all_phi_incrs_profitable_1): Rename from
3898 all_phi_incrs_profitable, set visited field.
3899 (all_phi_incrs_profitable): New wrapper function.
3901 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
3902 Alan Hayward <alan.hayward@arm.com>
3903 David Sherwood <david.sherwood@arm.com>
3905 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
3906 that takes the outer and inner modes.
3907 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
3908 comparison as the canonical test for a paradoxical subreg.
3909 * combine.c (simplify_set): Use paradoxical_subreg_p.
3910 (make_extraction): Likewise.
3911 (force_to_mode): Likewise.
3912 (rtx_equal_for_field_assignment_p): Likewise.
3913 (gen_lowpart_for_combine): Likewise.
3914 (simplify_comparison): Likewise.
3915 * cse.c (equiv_constant): Likewise.
3916 * expmed.c (store_bit_field_1): Likewise.
3917 * final.c (alter_subreg): Likewise.
3918 * fwprop.c (propagate_rtx): Likewise.
3919 (forward_propagate_subreg): Likewise.
3920 * ira-conflicts.c (ira_build_conflicts): Likewise.
3921 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
3922 * lra-constraints.c (curr_insn_transform): Likewise.
3923 (split_reg): Likewise.
3924 * lra-eliminations.c (move_plus_up): Likewise.
3925 (lra_eliminate_regs_1): Likewise.
3926 * recog.c (general_operand): Likewise.
3927 * ree.c (combine_reaching_defs): Likewise.
3928 * reload.c (push_reload): Likewise.
3929 (find_reloads): Likewise.
3930 * reload1.c (elimination_effects): Likewise.
3931 (compute_reload_subreg_offset): Likewise.
3932 (choose_reload_regs): Likewise.
3933 * rtlanal.c (subreg_lsb_1): Likewise.
3934 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3935 (simplify_subreg): Likewise.
3936 * var-tracking.c (track_loc_p): Likewise.
3937 * emit-rtl.c (byte_lowpart_offset): Likewise.
3938 (paradoxical_subreg_p): Delete out-of-line definition.
3940 2017-08-22 Jeff Law <law@redhat.com>
3942 PR tree-optimization/81741
3943 PR tree-optimization/71947
3944 * tree-ssa-dom.c: Include tree-inline.h.
3945 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
3946 equivalences if one is more expensive to compute than the other.
3947 * tree-ssa-scopedtables.h (class const_or_copies): Make
3948 record_const_or_copy_raw method private.
3949 (class avail_exprs_stack): New method simplify_binary_operation.
3950 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
3951 avail_exprs_stack::simplify_binary_operation as needed.
3952 (avail_exprs_stack::simplify_binary_operation): New function.
3954 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
3956 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
3957 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
3958 (DOT_SYMBOLS): Likewise.
3959 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
3960 (RELOCATABLE_NEEDS_FIXUP): Likewise.
3961 (RS6000_ABI_NAME): Likewise.
3962 (TARGET_CMODEL): Likewise.
3963 (TOC_SECTION_ASM_OP): Likewise.
3964 (SET_CMODEL): New macro.
3965 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
3967 2017-08-22 Richard Biener <rguenther@suse.de>
3969 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
3970 to allow for free-lang-data replacements similar to verify_type_variant.
3972 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
3973 Alan Hayward <alan.hayward@arm.com>
3974 David Sherwood <david.sherwood@arm.com>
3976 * config/aarch64/aarch64.md (casesi): Use DImode rather than
3977 VOIDmode for the LABEL_REF.
3979 2017-08-22 Richard Biener <rguenther@suse.de>
3981 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
3983 2017-08-22 Richard Biener <rguenther@suse.de>
3985 * common.opt (feliminate-dwarf2-dups): Ignore.
3986 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
3987 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
3988 same_die_p_wrap, compute_section_prefix,
3989 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
3990 (comdat_symbol_id, comdat_symbol_number): Likewise.
3991 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
3993 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
3994 (output_die): Mark unreachable path unreachable.
3995 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
3996 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
3997 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
3998 (dwarf2out_early_finish): Likewise.
4000 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
4002 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
4004 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
4007 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
4008 not VAR_P. Filter attribute warnings with OPT_Wattributes.
4009 (avr_attribute_table) <io, io_low, address>: Initialize
4010 .decl_required with true.
4012 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
4014 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
4015 undocumented debugging options.
4016 (-mvsx-scalar-double): Likewise.
4017 (-mallow-df-permute): Likewise.
4018 (-mvectorize-builtins): Likewise.
4019 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4020 (rs6000_builtin_vectorized_function): Likewise.
4021 (rs6000_builtin_md_vectorized_function): Likewise.
4022 (rs6000_opt_vars): Likewise.
4024 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
4027 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
4028 (*btrq_imm): Rename from *btrq.
4029 (*btcq_imm): Rename from *btcq.
4030 (btsc): New code attribute.
4031 (*<btsc><mode>): New insn pattern.
4032 (*btr<mode>): Ditto.
4033 (*<btsc><mode>_mask): New insn_and_split pattern.
4034 (*btr<mode>_mask): Ditto.
4036 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4037 Alan Hayward <alan.hayward@arm.com>
4038 David Sherwood <david.sherwood@arm.com>
4040 * function.c (pad_below): Simplify padding calculation.
4042 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4043 Alan Hayward <alan.hayward@arm.com>
4044 David Sherwood <david.sherwood@arm.com>
4046 * target.def (function_prologue): Remove frame size argument.
4047 (function_epilogue): Likewise.
4048 * doc/tm.texi: Regenerate.
4049 * final.c (final_start_function): Update call to function_prologue.
4050 (final_end_function): Update call to function_epilogue.
4051 (default_function_pro_epilogue): Remove frame size argument.
4052 * output.h (default_function_pro_epilogue): Likewise.
4053 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
4054 (arm_output_function_prologue): Likewise.
4055 * config/frv/frv.c (frv_function_prologue): Likewise.
4056 (frv_function_epilogue): Likewise.
4057 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
4058 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
4059 (ia64_output_function_epilogue): Likewise.
4060 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
4061 (m32r_output_function_epilogue): Likewise.
4062 * config/microblaze/microblaze.c (microblaze_function_prologue)
4063 (microblaze_function_epilogue): Likewise.
4064 * config/mips/mips.c (mips_output_function_prologue): Likewise.
4065 (mips_output_function_epilogue): Likewise.
4066 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
4067 (mmix_target_asm_function_epilogue): Likewise.
4068 * config/msp430/msp430.c (msp430_start_function): Likewise.
4069 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
4070 (nds32_asm_function_epilogue): Likewise.
4071 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
4072 * config/pa/pa.c (pa_output_function_prologue): Likewise.
4073 (pa_output_function_epilogue): Likewise.
4074 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
4075 (rs6000_output_function_epilogue): Likewise.
4076 * config/rl78/rl78.c (rl78_start_function): Likewise.
4077 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
4078 (rs6000_output_function_epilogue): Likewise.
4079 * config/rx/rx.c (rx_output_function_prologue): Likewise.
4080 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
4081 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
4082 (sparc_asm_function_epilogue): Likewise.
4084 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4086 * tree.h (type_has_mode_precision_p): New function.
4087 * convert.c (convert_to_integer_1): Use it.
4088 * expr.c (expand_expr_real_2): Likewise.
4089 (expand_expr_real_1): Likewise.
4090 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
4091 * match.pd: Likewise.
4092 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
4093 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
4094 * tree-tailcall.c (process_assignment): Likewise.
4095 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4096 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
4097 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
4098 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4099 (vectorizable_assignment): Likewise.
4100 (vectorizable_shift): Likewise.
4101 (vectorizable_operation): Likewise.
4102 * tree-vrp.c (register_edge_assert_for_2): Likewise.
4104 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
4106 * match.pd: Add pow (C, x) simplification.
4108 2017-08-21 Richard Biener <rguenther@suse.de>
4110 PR tree-optimization/81900
4111 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
4112 for blocks with abnormal predecessors.
4113 (compute_antic): Do not set visited flag prematurely.
4115 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
4118 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
4120 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4122 * stor-layout.h (vector_type_mode): Move to...
4123 * tree.h (vector_type_mode): ...here.
4124 * stor-layout.c (vector_type_mode): Move to...
4125 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
4127 2017-08-21 Richard Biener <rguenther@suse.de>
4129 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
4130 register_external_die hooks.
4131 (debug_false_tree_charstarstar_uhwistar): Declare.
4132 (debug_nothing_tree_charstar_uhwi): Likewise.
4133 * debug.c (do_nothing_debug_hooks): Adjust.
4134 (debug_false_tree_charstarstar_uhwistar): New do nothing.
4135 (debug_nothing_tree_charstar_uhwi): Likewise.
4136 * dbxout.c (dbx_debug_hooks): Adjust.
4137 (xcoff_debug_hooks): Likewise.
4138 * sdbout.c (sdb_debug_hooks): Likewise.
4139 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4140 * dwarf2out.c (macinfo_label_base): New global.
4141 (dwarf2out_register_external_die): New function for the
4142 register_external_die hook.
4143 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
4144 (dwarf2_debug_hooks): Use them.
4145 (dwarf2_lineno_debug_hooks): Adjust.
4146 (struct die_struct): Add with_offset flag.
4147 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
4148 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
4149 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
4150 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
4151 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
4152 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
4153 defining section names for the early LTO debug variants.
4154 (reset_indirect_string): New helper.
4155 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
4156 (print_dw_val): Add support for offsetted symbol references.
4157 (get_ultimate_context): Split out from is_cxx.
4158 (is_cxx): Use get_ultimate_context.
4159 (is_fortran): Add decl overload.
4160 (compute_comp_unit_symbol): Split out worker from
4161 compute_section_prefix.
4162 (compute_section_prefix): Call compute_comp_unit_symbol and
4163 set comdat_type_p here.
4164 (output_die): Skip DIE symbol output for the LTO added one.
4165 Handle DIE symbol references with offset.
4166 (output_comp_unit): Guard section name mangling properly.
4167 For LTO debug sections emit a symbol at the section beginning
4168 which we use to refer to its DIEs.
4169 (add_abstract_origin_attribute): For DIEs registered via
4170 dwarf2out_register_external_die directly refer to the early
4171 DIE rather than indirectly through the shadow one we created.
4172 Remove obsolete call to dwarf2out_abstract_function for
4173 non-function/block origins.
4174 (gen_array_type_die): When generating early LTO debug do
4175 not emit DW_AT_string_length.
4176 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
4177 late when in LTO. As suggested place a gcc_unreachable for
4178 the DECL_ABSTRACT_P case.
4179 (gen_subprogram_die): Avoid another specification DIE
4180 for early built declarations/definitions for the late LTO case.
4181 (gen_variable_die): Add type references for late duplicated VLA dies
4183 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
4184 we have the abstract instance already.
4185 (process_scope_var): Adjust decl DIE contexts in LTO which
4186 first puts them in limbo.
4187 (gen_decl_die): Do not generate type DIEs late apart from
4188 types for VLAs or for decls we do not yet have a DIE. Do not
4189 call dwarf2out_abstract_function late.
4190 (dwarf2out_early_global_decl): Make sure to create DIEs
4191 for abstract instances of a decl first.
4192 (dwarf2out_late_global_decl): Adjust comment.
4193 (output_macinfo_op): With multiple macro sections use
4194 macinfo_label_base to distinguish labels.
4195 (output_macinfo): Likewise. Update macinfo_label_base.
4196 Pass in the line info label.
4197 (note_variable_value_in_expr): When generating LTO resolve
4198 all variable values here by generating DIEs as needed.
4199 (init_sections_and_labels): Add early LTO debug flag parameter
4200 and generate different sections and names if set. Add generation
4201 counter for the labels so we can have multiple of them.
4202 (reset_dies): Helper to allow DIEs to be output multiple times.
4203 (dwarf2out_finish): When outputting DIEs to the fat part of an
4204 LTO object first reset DIEs.
4205 (dwarf2out_early_finish): Output early DIEs when generating LTO.
4206 (modified_type_die): Check for decl_ultimate_origin being self
4208 (gen_type_die_with_usage): Likewise.
4209 (gen_typedef_die): Allow decl_ultimate_origin being self.
4210 (set_decl_abstract_flags): Remove.
4211 (set_block_abstract_flags): Likewise.
4212 (dwarf2out_abstract_function): Treat the early generated DIEs
4213 as the abstract copy and only add DW_AT_inline and
4214 DW_AT_artificial here and call set_decl_origin_self.
4215 If the DIE has an abstract origin don't do anything.
4216 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
4217 if we have none yet (Go fails to build one, PR78628).
4218 (variably_modified_type_p): Prevent endless recursion for Ada
4219 cyclic pointer types.
4220 * lto-streamer-in.c: Include debug.h.
4221 (dref_queue): New global.
4222 (lto_read_tree_1): Stream in DIE references.
4223 (lto_input_tree): Register DIE references.
4224 (input_function): Stream DECL_DEBUG_ARGS.
4225 * lto-streamer-out.c: Include debug.h.
4226 (lto_write_tree_1): Output DIE references.
4227 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
4228 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
4229 (output_function): Stream DECL_DEBUG_ARGS.
4230 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
4231 Stream DECL_ABSTRACT_ORIGIN.
4232 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
4233 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
4234 DECL_CONTEXT for file-scope decls.
4235 * lto-streamer.h (struct dref_entry): Declare.
4236 (dref_queue): Likewise.
4237 * cfgexpand.c (pass_expand::execute): Do not call the
4238 outlining_inline_function hook here.
4239 * lto-wrapper.c (debug_obj): New global.
4240 (tool_cleanup): Unlink it if required.
4241 (debug_objcopy): New function.
4242 (run_gcc): Handle early debug sections in the IL files by
4243 extracting them to separate files, partially linkin them and
4244 feeding the result back as result to the linker.
4245 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
4246 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
4247 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
4248 sections into a separate segment.
4249 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
4251 (darwin_asm_dwarf_section): Likewise.
4252 (darwin_asm_output_dwarf_offset): Likewise.
4253 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
4255 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4256 Alan Hayward <alan.hayward@arm.com>
4257 David Sherwood <david.sherwood@arm.com>
4259 * read-md.h (md_reader::record_potential_iterator_use): Replace
4260 pointer argument with an rtx and an index.
4261 * read-rtl.c (iterator_group::apply_iterator): Likewise.
4262 (apply_mode_iterator): Likewise.
4263 (apply_code_iterator): Likewise.
4264 (apply_int_iterator): Likewise.
4265 (apply_subst_iterator): Likewise.
4266 (record_iterator_use): Likewise.
4267 (record_attribute_use): Likewise.
4268 (md_reader::record_potential_iterator_use): Likewise. Update calls
4269 to record_iterator_use and apply_iterator.
4270 (iterator_use): Replace ptr with x and index.
4271 (attribute_use): Likewise.
4272 (apply_attribute_uses): Update calls to apply_iterator.
4273 (apply_iterators): Likewise. Update initialization of iterator_use.
4274 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
4275 and record_potential_iterator_use.
4276 (rtx_reader::read_rtx_operand): Likewise.
4278 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4279 Alan Hayward <alan.hayward@arm.com>
4280 David Sherwood <david.sherwood@arm.com>
4282 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
4285 2017-08-21 Richard Biener <rguenther@suse.de>
4288 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
4289 at struct end conservatively when comparing common bases.
4291 2017-08-21 Richard Biener <rguenther@suse.de>
4293 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
4294 (mem_ref_in_stmt): Remove.
4295 (determine_max_movement): Use ref index to get at the reference.
4296 (invariantness_dom_walker::before_dom_children): Deal with
4297 lim data already initialized.
4298 (gather_mem_refs_stmt): Initialize lim data and record ref index.
4300 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4302 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
4303 (TARGET_ISA_ROUND): Ditto.
4304 (TARGET_ROUND): Ditto.
4305 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
4306 * config/i386/i386.md: Ditto.
4307 * config/i386/sse.md: Ditto.
4308 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
4309 with OPTION_MASK_ISA_SSE4_1.
4311 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4314 * doc/extend.texi (x86 Built-in Functions): Correct the name of
4315 __builtin_ia32_lzcnt_u16.
4317 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4320 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
4321 (rs6000_set_current_function): Rewrite function to use it.
4323 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
4326 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
4327 and TYPE_WARN_IF_NOT_ALIGN.
4328 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
4329 (handle_warn_if_not_align): New.
4330 (place_union_field): Call handle_warn_if_not_align.
4331 (place_field): Call handle_warn_if_not_align.
4332 Copy TYPE_WARN_IF_NOT_ALIGN.
4333 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
4334 (layout_type): Likewise.
4335 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
4337 (tree_decl_common): Add warn_if_not_align.
4338 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
4339 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
4340 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
4341 (DECL_WARN_IF_NOT_ALIGN): Likewise.
4342 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
4343 * doc/extend.texi: Document warn_if_not_aligned attribute.
4344 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
4346 2017-08-17 Martin Liska <mliska@suse.cz>
4349 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
4350 (get_data_dependence): Use it as pointer type.
4351 (distribute_loop): Likewise.
4353 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4355 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
4356 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
4357 (p8_vmrgow_<mode>_direct): New define_insn.
4358 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
4359 handle endianness for vmrgew and vmrgow permute patterns.
4361 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4363 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
4364 * config/rs6000/rs6000-cpus.def: Remove comment.
4365 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
4366 (POWERPC_MASKS): Likewise.
4367 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
4368 use of TARGET_VSX_TIMODE.
4369 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
4370 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
4371 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
4372 (rs6000_option_override_internal): Remove dead code.
4373 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
4374 (rs6000_legitimize_reload_address): Likewise.
4375 (rs6000_legitimate_address_p): Likewise.
4376 (rs6000_opt_masks): Delete "vsx-timode".
4377 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
4378 from function comment.
4379 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
4380 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
4381 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
4383 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
4384 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
4385 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
4388 (define_peephole2): Likewise.
4390 2017-08-17 Martin Sebor <msebor@redhat.com>
4393 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
4394 past the end of an array.
4395 (test_pp_format): Add test cases.
4397 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
4399 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
4400 missing ECF_NOTHROW flags.
4402 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4405 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
4406 operands residing in integer registers.
4407 (*vsx_le_perm_load_<mode>): Likewise.
4408 (*vsx_le_perm_store_<mode>): Likewise.
4409 (define_peephole2): Add peepholes to optimize the above.
4411 2017-08-17 Marek Polacek <polacek@redhat.com>
4414 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
4415 to mimic what shorten_compare did. Change the return type to bool.
4416 (fold_cond_expr_with_comparison): Update call to
4417 operand_equal_for_comparison_p.
4418 (fold_ternary_loc): Likewise.
4420 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
4422 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
4424 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
4426 2017-08-17 Richard Biener <rguenther@suse.de>
4428 * tree-ssa-structalias.c (solve_graph): When propagating
4429 to successors update the graphs succ edges and avoid duplicate work.
4431 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
4434 * config/i386/i386.c (ix86_option_override_internal): Save target
4435 specific options after ix86_stack_protector_guard_reg was changed.
4437 2017-08-17 Richard Biener <rguenther@suse.de>
4439 PR tree-optimization/81827
4440 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
4441 (new_var_info): Initialize it conservatively.
4442 (get_call_vi): Mark register vars.
4443 (new_scalar_tmp_constraint_exp): Likewise.
4444 (handle_rhs_call): Likewise.
4445 (handle_const_call): Likewise.
4446 (create_function_info_for): Likewise.
4447 (solve_constraints): Sort varinfos to separate register from
4448 non-register vars to pack points-to solution bitmaps during
4451 2017-08-17 Marek Polacek <polacek@redhat.com>
4453 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
4455 2017-08-17 Richard Biener <rguenther@suse.de>
4457 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
4458 to true when overflow is undefined and we saturated the result.
4460 2017-08-17 Alan Modra <amodra@gmail.com>
4463 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
4464 Don't use store multiple if only one reg needs saving.
4465 (interesting_frame_related_regno): New function.
4466 (rs6000_frame_related): Don't emit frame info for regs that
4468 (rs6000_emit_epilogue): Likewise.
4470 2017-08-16 Nathan Sidwell <nathan@acm.org>
4472 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
4473 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
4474 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
4475 * tree.c (free_lang_data_in_type): Use else-if chain. Always
4476 clear TYPE_LANG_1. Remove obsolete member-function stripping.
4477 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
4478 (verify_type): Adjust for TYPE_BINFO move.
4479 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
4480 process TYPE_BINFO directly.
4481 (hash_tree): Likewise.
4482 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
4484 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
4487 2017-08-16 David Malcolm <dmalcolm@redhat.com>
4489 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
4491 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
4494 * config/i386/i386.md (*anddi_1_btr): Change predicates of
4495 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
4496 Add ix86_binary_operator_ok to insn constraint.
4497 (*iordi_1_bts): Ditto.
4498 (*xordi_1_btc): Ditto.
4499 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
4500 Update corresponding peephole2 pattern.
4504 2017-08-16 Bin Cheng <bin.cheng@arm.com>
4506 PR tree-optimization/81832
4507 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
4508 copy loop header which has IFN_LOOP_DIST_ALIAS call.
4510 2017-08-16 Marek Polacek <polacek@redhat.com>
4513 * fold-const.c (fold_indirect_ref_1): Restore original behavior
4514 regarding size_zero_node.
4516 2017-08-16 Martin Liska <mliska@suse.cz>
4519 * config.gcc: Respect previously set extra_objs in case
4522 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
4524 PR tree-optimization/81835
4525 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
4526 the phi SSA_NAME. Check that the condition in a COND_EXPR does
4527 not depend on the phi.
4529 2017-08-16 Alan Modra <amodra@gmail.com>
4531 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
4534 2017-08-16 Alan Modra <amodra@gmail.com>
4536 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
4537 (save_reg_p): ..into this. Update all callers.
4538 (first_reg_to_save): Simplify.
4540 2017-08-16 Alan Modra <amodra@gmail.com>
4542 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
4545 2017-08-15 Joseph Myers <joseph@codesourcery.com>
4549 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
4550 constant count if that count is less than 32.
4552 2017-08-15 Nathan Sidwell <nathan@acm.org>
4554 * gcc.c (execute): Emit friendlier message if inferior is killed
4555 by an external cause.
4557 2017-08-15 Richard Biener <rguenther@suse.de>
4559 PR tree-optimization/81790
4560 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
4561 CONSTRUCTORs from simplifying and VN.
4563 2017-08-14 Martin Sebor <msebor@redhat.com>
4565 * builtin-attrs.def: Add comments.
4567 2017-08-14 Martin Sebor <msebor@redhat.com>
4570 * doc/extend.texi (attribute nonstring): Document new attribute.
4572 2017-08-14 Martin Sebor <msebor@redhat.com>
4575 * tree-diagnostic.c (default_tree_printer): Handle %G.
4576 * gimple-pretty-print.h (percent_G_format): Declare new function.
4577 * gimple-pretty-print.c (percent_G_format): Define.
4578 * tree-pretty-print.c (percent_K_format): Add argument.
4580 2017-08-14 Martin Sebor <msebor@redhat.com>
4582 PR translation/79998
4583 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
4584 Remove a stray space.
4586 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
4589 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
4590 (*iordi_1_bts): Ditto.
4591 (*xordi_1_btc): Ditto.
4593 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4596 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
4597 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4599 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
4600 quoted strings, and make more translator-friendly.
4601 (darwin_rs6000_override_options): Likewise.
4602 (rs6000_option_override_internal): Likewise.
4603 (rs6000_return_in_memory): Fix overlong line.
4604 (init_cmulative_args): Use quoted strings, and make more
4605 translator-friendly.
4606 (rs6000_pass_by_reference): Fix overlong line.
4607 (def_builtin): Use quoted strings.
4608 (altivec_expand_predicate_builtin): Use quoted strings, and make
4609 more translator-friendly.
4610 (htm_expand_builtin): Use quoted strings.
4611 (cpu_expand_builtin): Use quoted strings, and make more
4612 translator-friendly.
4613 (altivec_expand_builtin): Likewise.
4614 (paired_expand_predicate_builtin): Likewise.
4615 (rs6000_invalid_builtin): Likewise.
4616 (builtin_function_type): Use quoted strings.
4617 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
4618 more translator-friendly.
4619 (rs6000_trampoline_init): Likewise.
4620 (rs6000_handle_altivec_attribute): Likewise.
4621 (rs6000_inner_target_options): Use quoted strings.
4622 (rs6000_disable_incompatible_switches): Likewise.
4623 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
4624 strings, and make more translator-friendly.
4625 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
4627 2017-08-14 Bin Cheng <bin.cheng@arm.com>
4629 PR tree-optimization/81799
4630 * tree-loop-distribution.c (version_loop_by_alias_check): Force
4631 cond_expr to simple gimple operand.
4633 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
4636 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
4638 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
4642 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
4643 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
4644 TARGET_GASISR_PROLOGUES.
4645 * config/avr/avr.c (avr_option_override): Same.
4646 (avr_pass_pre_proep::execute): Same.
4648 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
4651 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
4652 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
4654 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
4656 * config/i386/i386.md (*load_tp_<mode>): Redefine as
4657 define_insn_and_split. Split to a memory load from 0 in
4658 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
4659 using PTR mode iterator.
4660 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
4661 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
4662 (*add_tp_<mode>): Redefine as define_insn_and_split.
4663 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
4664 address space. Merge with *add_tp_x32 using PTR mode iterator.
4665 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
4666 Split to an add with a memory load from 0 in
4667 DEFAULT_TLS_SEG_REG address space.
4669 2017-08-12 Andrew Pinski <apinski@cavium.com>
4671 * config/aarch64/aarch64-option-extensions.def (rdma):
4672 Fix feature string to what Linux prints out in /proc/cpuinfo.
4674 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
4677 * dwarf2out.c (modified_type_die): For C typedef types that have
4678 an ultimate origin, process the ultimate origin instead of the
4680 (gen_typedef_die): Assert that input DECLs have no ultimate
4682 (gen_type_die_with_usage): For typedef variants that have an
4683 ultimate origin, just call gen_decl_die on the original DECL.
4684 (process_scope_var): Avoid creating DIEs for local typedefs and
4685 concrete static variables.
4687 2017-08-12 Alan Modra <amodra@gmail.com>
4691 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
4692 match gnu-user.h startfile.
4693 (ENDFILE_LINUX_SPEC): Similarly.
4695 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
4698 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
4700 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
4702 2017-08-11 Tamar Christina <tamar.christina@arm.com>
4703 * config/aarch64/aarch64.md (mov<mode>): Change.
4704 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4705 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
4706 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
4708 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
4710 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
4711 for storage order barriers.
4713 2017-08-11 Martin Liska <mliska@suse.cz>
4716 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
4717 variables of void type.
4719 2017-08-11 Martin Liska <mliska@suse.cz>
4721 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
4722 TARGET_SUPPORTS_ALIASES.
4723 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
4724 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
4725 (optimize_weakref): Likewise.
4726 * symtab.c (symtab_node::noninterposable_alias): Likewise.
4727 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
4728 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
4730 2017-08-11 Martin Liska <mliska@suse.cz>
4733 * config/i386/i386.c (make_resolver_func): Do complete
4734 refactoring of the function.
4736 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4739 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
4740 * config/i386/i386.c (ix86_stack_protect_guard): Use
4741 ix86_stack_protect_guard_symbol_str to generate varible declaration.
4742 * doc/invoke.texi (x86 Options): Document
4743 -mstack-protector-guard-symbol= option.
4745 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4747 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
4748 * config/i386/i386.c (ix86_split_stack_guard): New function.
4749 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
4750 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
4751 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
4752 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
4753 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
4754 (split_stack_space_check): Call ix86_split_stack_guard.
4756 2017-08-10 Martin Sebor <msebor@redhat.com>
4758 * print-tree.c (print_node): Print location using the established
4760 Replace spaces with colons.
4761 (debug_raw, debug): Ditto.
4763 2017-08-10 Martin Sebor <msebor@redhat.com>
4766 * pretty-print.c (pp_format): Correct the handling of %s precision.
4768 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
4771 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
4773 (ix86_finalize_stack_frame_flags): This. Also clear
4774 frame_pointer_needed if -fno-omit-frame-pointer is used without
4776 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
4777 with ix86_finalize_stack_frame_flags.
4778 (ix86_expand_epilogue): Likewise.
4779 (ix86_expand_split_stack_prologue): Likewise.
4780 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
4782 2017-08-10 Martin Liska <mliska@suse.cz>
4785 * c-attribs.c (handle_target_attribute):
4786 Report warning for an empty string argument of target attribute.
4788 2017-08-09 Jakub Jelinek <jakub@redhat.com>
4791 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
4793 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
4794 or DECL_NONLOCAL labels.
4795 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
4796 or DECL_NONLOCAL labels here.
4798 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
4800 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
4801 to indicate when early gimple folding has been disabled.
4802 (rs6000_gimple_fold_builtin): Add debug content.
4803 (rs6000_invalid_builtin): Fix whitespace.
4804 (rs6000_expand_builtin): Fix whitespace.
4805 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
4807 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
4810 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
4811 SAVE_MULTIPLE if not all the registers that saves, should be saved.
4813 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
4815 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
4816 (qdf24xx): Likewise.
4817 * config/aarch64/aarch64.md: Include falkor.md.
4818 * config/aarch64/falkor.md: New.
4820 2017-08-09 Marek Polacek <polacek@redhat.com>
4823 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
4824 * diagnostic.c (emit_diagnostic): Add a comment.
4825 (emit_diagnostic_valist): New function.
4827 2017-08-09 Marek Polacek <polacek@redhat.com>
4830 * input.c (make_location): New overload.
4831 * input.h (make_location): Declare.
4833 2017-08-08 Alan Modra <amodra@gmail.com>
4834 H.J. Lu <hongjiu.lu@intel.com>
4837 * gcc.c (NO_PIE_SPEC): Delete.
4838 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
4840 (LINK_PIE_SPEC): ..to here.
4841 (LINK_COMMAND_SPEC): Support -no-pie.
4842 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
4843 chain of crtbegin*.o selection, update for PIE_SPEC changes and
4845 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
4846 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
4847 (ENDFILE_CRTEND_SPEC): Similarly.
4849 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
4852 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
4853 (mstack-protector-guard-offset=): Ditto.
4854 * config/i386/i386.c (ix86_option_override): Handle
4855 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
4857 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
4858 ix86_stack_protect_guard_offset variables.
4859 (TARGET_STACK_PROTECT_GUARD): Always define.
4860 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
4861 and -mstack-protector-guard-offset= options.
4863 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4865 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
4866 boundary case for the last candidate.
4868 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4870 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
4871 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
4873 2017-08-08 Tamar Christina <tamar.christina@arm.com>
4876 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
4877 * config/aarch64/aarch64-builtins.c
4878 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
4879 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
4880 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
4882 2017-08-08 Tamar Christina <tamar.christina@arm.com>
4883 Andrew Pinski <pinskia@gmail.com>
4886 * internal-fn.def (XORSIGN): New.
4887 * optabs.def (xorsign_optab): New.
4888 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
4889 (convert_expand_mult_copysign): New.
4890 (pass_optimize_widening_mul::execute): Call
4891 convert_expand_mult_copysign.
4893 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4895 PR tree-optimization/81354
4896 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
4897 Insert on edges rather than explicitly creating landing pads.
4898 (analyze_candidates_and_replace): Commit edge inserts.
4900 2017-08-08 Richard Biener <rguenther@suse.de>
4903 * tree-ssa-loop-niter.c: Include tree-dfa.h.
4904 (expand_simple_operations): Also look through ADDR_EXPRs with
4905 MEM_REF bases treating them as POINTER_PLUS_EXPR.
4907 2017-08-08 Richard Biener <rguenther@suse.de>
4909 PR tree-optimization/81723
4910 * tree-vect-slp.c (struct bst_traits): New hash traits.
4911 (bst_fail): New global.
4912 (vect_build_slp_tree_2): New worker, split out from ...
4913 (vect_build_slp_tree): ... this now wrapping it with using
4914 bst_fail set to cache SLP tree build fails. Properly handle
4916 (vect_analyze_slp_instance): Allocate and free bst_fail.
4918 2017-08-08 Martin Liska <mliska@suse.cz>
4921 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
4922 LABEL_DECLs that can be from a different function.
4924 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4926 PR tree-optimization/81744
4927 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
4928 loop's number of iterations.
4930 2017-08-08 Martin Liska <mliska@suse.cz>
4932 * asan.c: Include header files.
4933 * attribs.c (build_decl_attribute_variant): New function moved
4935 (build_type_attribute_qual_variant): Likewise.
4936 (cmp_attrib_identifiers): Likewise.
4937 (simple_cst_list_equal): Likewise.
4938 (omp_declare_simd_clauses_equal): Likewise.
4939 (attribute_value_equal): Likewise.
4940 (comp_type_attributes): Likewise.
4941 (build_type_attribute_variant): Likewise.
4942 (lookup_ident_attribute): Likewise.
4943 (remove_attribute): Likewise.
4944 (merge_attributes): Likewise.
4945 (merge_type_attributes): Likewise.
4946 (merge_decl_attributes): Likewise.
4947 (merge_dllimport_decl_attributes): Likewise.
4948 (handle_dll_attribute): Likewise.
4949 (attribute_list_equal): Likewise.
4950 (attribute_list_contained): Likewise.
4951 * attribs.h (lookup_attribute): New function moved from tree.[ch].
4952 (lookup_attribute_by_prefix): Likewise.
4953 * bb-reorder.c: Include header files.
4954 * builtins.c: Likewise.
4955 * calls.c: Likewise.
4956 * cfgexpand.c: Likewise.
4957 * cgraph.c: Likewise.
4958 * cgraphunit.c: Likewise.
4959 * convert.c: Likewise.
4960 * dwarf2out.c: Likewise.
4961 * final.c: Likewise.
4962 * fold-const.c: Likewise.
4963 * function.c: Likewise.
4964 * gimple-expr.c: Likewise.
4965 * gimple-fold.c: Likewise.
4966 * gimple-pretty-print.c: Likewise.
4967 * gimple.c: Likewise.
4968 * gimplify.c: Likewise.
4969 * hsa-common.c: Likewise.
4970 * hsa-gen.c: Likewise.
4971 * internal-fn.c: Likewise.
4972 * ipa-chkp.c: Likewise.
4973 * ipa-cp.c: Likewise.
4974 * ipa-devirt.c: Likewise.
4975 * ipa-fnsummary.c: Likewise.
4976 * ipa-inline.c: Likewise.
4977 * ipa-visibility.c: Likewise.
4979 * lto-cgraph.c: Likewise.
4980 * omp-expand.c: Likewise.
4981 * omp-general.c: Likewise.
4982 * omp-low.c: Likewise.
4983 * omp-offload.c: Likewise.
4984 * omp-simd-clone.c: Likewise.
4985 * opts-global.c: Likewise.
4986 * passes.c: Likewise.
4987 * predict.c: Likewise.
4988 * sancov.c: Likewise.
4989 * sanopt.c: Likewise.
4990 * symtab.c: Likewise.
4991 * toplev.c: Likewise.
4992 * trans-mem.c: Likewise.
4993 * tree-chkp.c: Likewise.
4994 * tree-eh.c: Likewise.
4995 * tree-into-ssa.c: Likewise.
4996 * tree-object-size.c: Likewise.
4997 * tree-parloops.c: Likewise.
4998 * tree-profile.c: Likewise.
4999 * tree-ssa-ccp.c: Likewise.
5000 * tree-ssa-live.c: Likewise.
5001 * tree-ssa-loop.c: Likewise.
5002 * tree-ssa-sccvn.c: Likewise.
5003 * tree-ssa-structalias.c: Likewise.
5004 * tree-ssa.c: Likewise.
5005 * tree-streamer-in.c: Likewise.
5006 * tree-vectorizer.c: Likewise.
5007 * tree-vrp.c: Likewise.
5009 * ubsan.c: Likewise.
5010 * varasm.c: Likewise.
5011 * varpool.c: Likewise.
5012 * tree.c: Remove functions moved to attribs.[ch].
5014 * config/aarch64/aarch64.c: Add attrs.h header file.
5015 * config/alpha/alpha.c: Likewise.
5016 * config/arc/arc.c: Likewise.
5017 * config/arm/arm.c: Likewise.
5018 * config/avr/avr.c: Likewise.
5019 * config/bfin/bfin.c: Likewise.
5020 * config/c6x/c6x.c: Likewise.
5021 * config/cr16/cr16.c: Likewise.
5022 * config/cris/cris.c: Likewise.
5023 * config/darwin.c: Likewise.
5024 * config/epiphany/epiphany.c: Likewise.
5025 * config/fr30/fr30.c: Likewise.
5026 * config/frv/frv.c: Likewise.
5027 * config/ft32/ft32.c: Likewise.
5028 * config/h8300/h8300.c: Likewise.
5029 * config/i386/winnt.c: Likewise.
5030 * config/ia64/ia64.c: Likewise.
5031 * config/iq2000/iq2000.c: Likewise.
5032 * config/lm32/lm32.c: Likewise.
5033 * config/m32c/m32c.c: Likewise.
5034 * config/m32r/m32r.c: Likewise.
5035 * config/m68k/m68k.c: Likewise.
5036 * config/mcore/mcore.c: Likewise.
5037 * config/microblaze/microblaze.c: Likewise.
5038 * config/mips/mips.c: Likewise.
5039 * config/mmix/mmix.c: Likewise.
5040 * config/mn10300/mn10300.c: Likewise.
5041 * config/moxie/moxie.c: Likewise.
5042 * config/msp430/msp430.c: Likewise.
5043 * config/nds32/nds32-isr.c: Likewise.
5044 * config/nds32/nds32.c: Likewise.
5045 * config/nios2/nios2.c: Likewise.
5046 * config/nvptx/nvptx.c: Likewise.
5047 * config/pa/pa.c: Likewise.
5048 * config/pdp11/pdp11.c: Likewise.
5049 * config/powerpcspe/powerpcspe.c: Likewise.
5050 * config/riscv/riscv.c: Likewise.
5051 * config/rl78/rl78.c: Likewise.
5052 * config/rx/rx.c: Likewise.
5053 * config/s390/s390.c: Likewise.
5054 * config/sh/sh.c: Likewise.
5055 * config/sol2.c: Likewise.
5056 * config/sparc/sparc.c: Likewise.
5057 * config/spu/spu.c: Likewise.
5058 * config/stormy16/stormy16.c: Likewise.
5059 * config/tilegx/tilegx.c: Likewise.
5060 * config/tilepro/tilepro.c: Likewise.
5061 * config/v850/v850.c: Likewise.
5062 * config/vax/vax.c: Likewise.
5063 * config/visium/visium.c: Likewise.
5064 * config/xtensa/xtensa.c: Likewise.
5066 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5069 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
5070 constraints since the -mupper-regs-* switches have been
5072 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
5073 into a vector from a double word element that was extracted from
5074 another vector, and eliminate extra XXPERMDI instructions.
5075 (vsx_concat_<mode>_2): Likewise.
5076 (vsx_concat_<mode>_3): Likewise.
5077 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
5078 concat to allow optimizing inserts from previous extracts.
5080 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
5082 * config/i386/i386.c (ix86_stack_protect_guard): Generate
5083 memory reference to a SSP offset in TLS address space.
5084 (ix86_print_operand) <case '@'>: Remove.
5085 (ix86_print_operand_punct_valid_p): Remove '@' code.
5086 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
5088 (stack_tls_protect_set_<mode>): Remove.
5089 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
5090 (stack_tls_protect_test_<mode>): Remove.
5091 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
5093 2017-08-07 Olivier Hainque <hainque@adacore.com>
5096 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
5098 2017-08-07 Douglas Rupp <rupp@adacore.com>
5100 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
5101 variable was referenced as multidir in command.
5103 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5106 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
5109 2017-08-07 Martin Liska <mliska@suse.cz>
5111 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
5112 * config/rl78/rl78.c: Add include of attribs.h.
5113 * config/sh/sh.c: Likewise.
5114 * config/v850/v850.c: Likewise.
5116 2017-08-07 Tom de Vries <tom@codesourcery.com>
5119 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
5121 2017-08-07 Martin Liska <mliska@suse.cz>
5123 * config/mips/mips.c: Include attribs.h.
5125 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
5128 * doc/invoke.texi: Document change in behvaior for -Ofast for
5131 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
5133 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
5135 (aarch64_pop_regs): Likewise.
5136 (aarch64_gen_load_pair): Likewise.
5137 (aarch64_save_callee_saves): Likewise.
5138 (aarch64_restore_callee_saves): Likewise.
5140 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5142 * config/i386/i386.c: Revert the last change.
5144 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5147 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5149 (ix86_finalize_stack_frame_flags): This. Also clear
5150 frame_pointer_needed if -fno-omit-frame-pointer is used without
5152 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5153 with ix86_finalize_stack_frame_flags.
5154 (ix86_expand_epilogue): Likewise.
5155 (ix86_expand_split_stack_prologue): Likewise.
5157 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5160 * config/i386/i386.c (get_builtin_code_for_version): Set priority
5161 to P_AES for Westmere.
5163 2017-08-07 Jonathan Yong <10walls@gmail.com>
5165 * config/i386/mingw.opt (fset-stack-executable): Removed.
5166 * config/i386/cygming.opt (fset-stack-executable): Moved
5168 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
5170 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
5172 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
5174 2017-08-07 Marek Polacek <polacek@redhat.com>
5177 * fold-const.c (fold_indirect_ref_1): Check type_domain.
5179 2017-08-07 Martin Liska <mliska@suse.cz>
5181 * attribs.h (canonicalize_attr_name): New function.
5182 (cmp_attribs): Move from c-format.c and adjusted.
5183 (is_attribute_p): Moved from tree.h.
5184 * tree-inline.c: Add new includes.
5185 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
5186 (private_is_attribute_p): Remove.
5187 (private_lookup_attribute): Likewise.
5188 (private_lookup_attribute_by_prefix): Simplify.
5189 (remove_attribute): Use is_attribute_p.
5190 * tree.h: Remove removed declarations.
5192 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5195 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
5196 instead of computing it in the function. Formatting fix.
5197 (expand_case): Don't rely on default_edge being the first edge,
5198 clear it if removing it, pass default_edge to
5199 emit_case_dispatch_table.
5200 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
5203 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
5205 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
5206 insn in the function, emit NOP after the insn.
5208 2017-08-06 Tom de Vries <tom@codesourcery.com>
5210 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
5213 2017-08-06 Tom de Vries <tom@codesourcery.com>
5215 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
5218 2017-08-04 Yury Gribov <tetra2005@gmail.com>
5220 PR tree-optimization/57371
5221 * match.pd: New pattern.
5223 2017-08-04 Marek Polacek <polacek@redhat.com>
5226 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
5227 perform the computation in offset_int.
5229 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5231 PR tree-optimization/81136
5232 * tree-vectorizer.h: Include tree-hash-traits.h.
5233 (vec_base_alignments): New typedef.
5234 (vec_info): Add a base_alignments field.
5235 (vect_record_base_alignments): Declare.
5236 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
5238 (DR_IS_CONDITIONAL_IN_STMT): New macro.
5239 (create_data_ref): Add an is_conditional_in_stmt argument.
5240 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
5241 the is_conditional_in_stmt field.
5242 (data_ref_loc): Add an is_conditional_in_stmt field.
5243 (get_references_in_stmt): Set the is_conditional_in_stmt field.
5244 (find_data_references_in_stmt): Update call to create_data_ref.
5245 (graphite_find_data_references_in_stmt): Likewise.
5246 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
5247 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
5248 (vect_record_base_alignment): New function.
5249 (vect_record_base_alignments): Likewise.
5250 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
5251 for nested statements even if we fail to compute a misalignment.
5252 Use pooled base alignments for unconditional references.
5253 (vect_find_same_alignment_drs): Compare base addresses instead
5255 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
5256 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
5258 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5260 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
5261 Add an explicit name for the enum. Use auto_vec for slp_instances
5263 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
5265 (_bb_vec_info): Add a constructor and destructor.
5266 (vinfo_for_stmt): Return NULL for uids of -1 as well.
5267 (destroy_loop_vec_info): Delete.
5268 (vect_destroy_datarefs): Likewise.
5269 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
5270 (vec_info::vec_info): New function.
5271 (vec_info::~vec_info): Likewise.
5272 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
5273 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
5274 destroy_loop_vec_info.
5275 * tree-vect-loop.c (new_loop_vec_info): Replace with...
5276 (_loop_vec_info::_loop_vec_info): ...this.
5277 (destroy_loop_vec_info): Replace with...
5278 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
5279 the stmt_vec_infos. Leave handling of vec_info information to its
5280 destructor. Remove explicit vector releases.
5281 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
5282 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
5283 * tree-vect-slp.c (new_bb_vec_info): Replace with...
5284 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
5285 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
5286 (destroy_bb_vec_info): Replace with...
5287 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
5288 information to its destructor.
5289 (vect_slp_analyze_bb_1): Use new and delete instead of
5290 new_bb_vec_info and destroy_bb_vec_info.
5291 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
5294 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5296 * tree-data-ref.h (subscript): Add access_fn field.
5297 (data_dependence_relation): Add could_be_independent_p.
5298 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
5299 (same_access_functions): Move to tree-data-ref.c.
5300 * tree-data-ref.c (ref_contains_union_access_p): New function.
5301 (access_fn_component_p): Likewise.
5302 (access_fn_components_comparable_p): Likewise.
5303 (dr_analyze_indices): Add a reference to access_fn_component_p.
5304 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
5306 (constant_access_functions): Likewise.
5307 (add_other_self_distances): Likewise.
5308 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
5309 (initialize_data_dependence_relation): Use XCNEW and remove
5310 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
5311 of access functions that have the same type. Allow the
5312 subsequence to end with different bases in some circumstances.
5313 Record the chosen access functions in SUB_ACCESS_FN.
5314 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
5315 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
5316 (subscript_dependence_tester_1): Likewise dra and drb.
5317 (build_classic_dist_vector): Update calls accordingly.
5318 (subscript_dependence_tester): Likewise.
5319 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
5320 DDR_COULD_BE_INDEPENDENT_P.
5321 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
5322 comp_alias_ddrs instead of may_alias_ddrs.
5323 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
5325 (vect_analyze_data_ref_dependence): Use it if
5326 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
5327 distance vectors if that fails.
5328 (dependence_distance_ge_vf): New function.
5329 (vect_prune_runtime_alias_test_list): Use it. Don't clear
5330 LOOP_VINFO_MAY_ALIAS_DDRS.
5332 2017-08-04 Richard Biener <rguenther@suse.de>
5335 * fold-const.c (fold_binary_loc): Properly restrict
5336 minus_var0 && minus_var1 case when associating undefined overflow
5339 2017-08-04 Tom de Vries <tom@codesourcery.com>
5341 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
5343 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5345 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5346 Don't start diagnostic messages with a capital letter.
5347 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5349 (rs6000_invalid_builtin): Likewise.
5350 (rs6000_trampoline_init): Likewise.
5352 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5355 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
5356 after setting changeable df flags.
5358 2017-08-03 Richard Biener <rguenther@suse.de>
5360 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
5361 up if the use is in USE - X.
5363 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
5365 * toplev.c (dumpfile.h): New include.
5366 (internal_error_reentered): New static function. Use it...
5367 (internal_error_function): ...here to handle reentered internal_error.
5369 2017-08-03 Richard Biener <rguenther@suse.de>
5372 * fold-const.c (split_tree): Add minus_var and minus_con
5373 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
5374 here but always use minus_*.
5375 (associate_trees): Assert we never associate with MINUS_EXPR
5376 and NULL first operand. Do not recurse for PLUS_EXPR operands
5377 when associating as MINUS_EXPR either.
5378 (fold_binary_loc): Track minus_var and minus_con.
5380 2017-08-03 Tom de Vries <tom@codesourcery.com>
5383 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
5384 ACCEL_COMPILER, apply finish_options on
5385 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
5387 2017-08-03 Tom de Vries <tom@codesourcery.com>
5390 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
5391 function_entry_patch_area_size > 0.
5393 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5396 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
5397 instead of 10??LU, perform unit multiplication in wide_int,
5398 don't change alloc_object_size_limit if the limit is larger
5401 PR tree-optimization/81655
5402 PR tree-optimization/81588
5403 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
5404 the case when ranges[i].low and high are 1 for unsigned type with
5408 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
5409 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
5411 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5413 * tree-vrp.h: Add include guard.
5415 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
5418 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
5419 (ud2): New insn pattern.
5420 * config/i386/i386.c (ix86_expand_epilogue):
5421 For naked functions, generate ud2 instead of trap insn.
5423 2017-08-02 Marek Polacek <polacek@redhat.com>
5426 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
5428 2017-08-02 Tom de Vries <tom@codesourcery.com>
5429 Cesar Philippidis <cesar@codesourcery.com>
5431 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
5432 Add missing edge probabilities.
5434 2017-08-02 Tamar Christina <tamar.christina@arm.com>
5436 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
5439 2017-08-02 Jakub Jelinek <jakub@redhat.com>
5442 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
5443 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
5444 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
5445 prologue_seq sequences - if any.
5447 2017-08-02 Richard Biener <rguenther@suse.de>
5449 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
5450 via vectors if supported, integer extracts via punning if supported
5451 or otherwise vector extracts.
5453 2017-08-02 Richard Biener <rguenther@suse.de>
5455 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
5457 (bitmap_insert_into_set): ... this.
5459 2017-08-02 Richard Biener <rguenther@suse.de>
5461 PR tree-optimization/81633
5463 2015-08-17 Alan Hayward <alan.hayward@arm.com>
5465 PR tree-optimization/71752
5466 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
5468 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
5470 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
5471 (machine_function::call_ms2sysv_pad_out): Remove field.
5472 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
5473 (ix86_compute_frame_layout): Likewise.
5475 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
5478 * config/i386/i386.c (ix86_set_func_type): Disallow naked
5479 attribute with interrupt attribute.
5481 2017-08-01 Andrew Pinski <apinski@cavium.com>
5483 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
5484 BIT_INSERT_EXPR's operand 1
5485 to see if the types precision matches.
5487 2017-08-01 Martin Liska <mliska@suse.cz>
5490 * builtins.c (expand_builtin_memcpy_args): Remove.
5491 (expand_builtin_memcpy): Call newly added function
5492 expand_builtin_memory_copy_args.
5493 (expand_builtin_memcpy_with_bounds): Likewise.
5494 (expand_builtin_mempcpy): Remove last argument.
5495 (expand_builtin_mempcpy_with_bounds): Likewise.
5496 (expand_builtin_memory_copy_args): New function created from
5497 expand_builtin_mempcpy_args with small modifications.
5498 (expand_builtin_mempcpy_args): Remove.
5499 (expand_builtin_stpcpy): Remove unused argument.
5500 (expand_builtin): Likewise.
5501 (expand_builtin_with_bounds): Likewise.
5503 2017-08-01 Martin Liska <mliska@suse.cz>
5506 Make mempcpy more optimal (PR middle-end/70140).
5508 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5511 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5512 __builtin_vec_cmpne verify both arguments are compatible vectors
5513 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
5514 verify arg1_type is a pointer or array type. For __builtin_vec_st,
5515 move computation of aligned to after checking the argument types.
5519 * config/rs6000/vsx.md (vextract_fp_from_shorth,
5520 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
5523 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
5524 Doug Rupp <rupp@adacore.com>
5525 Olivier Hainque <hainque@adacore.com>
5527 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
5528 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
5530 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
5531 for TARGET_OS_CPP_BUILTIN.
5532 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
5533 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
5535 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
5536 passing required abi options to the assembler for EABI configurations.
5537 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
5538 of .text.hot and .text.unlikely sections for kernel modules when
5539 using ARM style exceptions.
5540 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
5541 options. Add EXTRA_CC1_SPEC.
5542 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
5544 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
5546 (ARM_TARGET2_DWARF_FORMAT): Define.
5547 * config/arm/t-vxworks: Adjust multilib control to removal of the
5548 Diab command line options.
5550 2017-08-01 Martin Liska <mliska@suse.cz>
5552 PR gcov-profile/81561
5553 * gcov.c (unblock): Make unblocking safe as we need to preserve
5554 index correspondence of blocks and block_lists.
5556 2017-08-01 Richard Biener <rguenther@suse.de>
5558 PR tree-optimization/81181
5559 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5560 (compute_antic): ... end of iteration here.
5562 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
5564 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
5565 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
5566 (ftree-slp-vectorize): Likewise.
5567 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
5568 can no longer be set independent of flag_tree_loop_vectorize.
5569 * omp-general.c (emp_max_vf): Likewise.
5570 * opts.c (enable_fdo_optimizations): Remove references to
5571 flag_tree_vectorize, these are now implicit.
5572 (common_handle_option): Remove handling for OPT_ftree_vectorize,
5573 and leave it for the options machinery.
5575 2017-08-01 Martin Liska <mliska@suse.cz>
5578 * builtins.c (expand_builtin_memcpy_args): Remove.
5579 (expand_builtin_memcpy): Call newly added function
5580 expand_builtin_memory_copy_args.
5581 (expand_builtin_memcpy_with_bounds): Likewise.
5582 (expand_builtin_mempcpy): Remove last argument.
5583 (expand_builtin_mempcpy_with_bounds): Likewise.
5584 (expand_builtin_memory_copy_args): New function created from
5585 expand_builtin_mempcpy_args with small modifications.
5586 (expand_builtin_mempcpy_args): Remove.
5587 (expand_builtin_stpcpy): Remove unused argument.
5588 (expand_builtin): Likewise.
5589 (expand_builtin_with_bounds): Likewise.
5591 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5594 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
5595 print "ds:" only for immediates in generic address space.
5597 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5600 * config/i386/i386.c (ix86_funciton_naked): New prototype.
5601 (ix86_function_ok_for_sibcall): Return false for naked functions.
5603 2017-08-01 Richard Biener <rguenther@suse.de>
5605 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
5606 (compute_antic): Seed worklist with exit block predecessors.
5607 * cfganal.c (dfs_find_deadend): For a cycle return the source
5608 of the edge closing it.
5610 2017-08-01 Tamar Christina <tamar.christina@arm.com>
5612 * config/aarch64/aarch64.c
5613 (aarch64_can_const_movi_rtx_p): Move 0 check.
5615 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5617 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
5618 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
5620 * match.pd: Ditto in address comparison pattern.
5622 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5624 PR tree-optimization/81627
5625 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
5626 closed ssa form for store-store chain.
5628 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5630 PR tree-optimization/81620
5631 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
5632 for store-store chain.
5634 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5636 PR tree-optimization/81588
5637 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
5638 ranges[i].in_p, invert comparison code ccode. For >/>=,
5639 swap rhs1 and rhs2 and comparison code unconditionally,
5640 for </<= don't do that. Don't swap rhs1/rhs2 again if
5641 ranges[i].in_p, instead invert comparison code ccode if
5642 opcode or oe->rank is BIT_IOR_EXPR.
5645 * optabs.def (vec_extract_optab, vec_init_optab): Change from
5646 a direct optab to conversion optab.
5647 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
5648 with GET_MODE_INNER as last argument instead of optab_handler.
5649 * expmed.c (extract_bit_field_1): Likewise. Use vector from
5650 vector extraction if possible and optab is available.
5651 * expr.c (store_constructor): Use convert_optab_handler instead
5652 of optab_handler. Use vector initialization from smaller
5653 vectors if possible and optab is available.
5654 * tree-vect-stmts.c (vectorizable_load): Likewise.
5655 * doc/md.texi (vec_extract, vec_init): Document that the optabs
5657 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
5658 of vec_init from half-sized vectors with the same element mode.
5659 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
5660 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
5661 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
5662 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
5663 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
5664 after mode in gen_vec_extract* calls.
5665 (vec_extract<mode>): Renamed to ...
5666 (vec_extract<mode><ssescalarmodelower>): ... this.
5667 (vec_extract<mode><ssehalfvecmodelower>): New expander.
5668 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
5669 element mode after mode in gen_vec_init* calls.
5670 (VEC_INIT_HALF_MODE): New mode iterator.
5671 (vec_init<mode>): Renamed to ...
5672 (vec_init<mode><ssescalarmodelower>): ... this.
5673 (vec_init<mode><ssehalfvecmodelower>): New expander.
5674 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
5675 (vec_extractv2sfsf): ... this.
5676 (vec_initv2sf): Renamed to ...
5677 (vec_initv2sfsf): ... this.
5678 (vec_extractv2si): Renamed to ...
5679 (vec_extractv2sisi): ... this.
5680 (vec_initv2si): Renamed to ...
5681 (vec_initv2sisi): ... this.
5682 (vec_extractv4hi): Renamed to ...
5683 (vec_extractv4hihi): ... this.
5684 (vec_initv4hi): Renamed to ...
5685 (vec_initv4hihi): ... this.
5686 (vec_extractv8qi): Renamed to ...
5687 (vec_extractv8qiqi): ... this.
5688 (vec_initv8qi): Renamed to ...
5689 (vec_initv8qiqi): ... this.
5690 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
5691 (vec_init<mode>): Renamed to ...
5692 (vec_init<mode><VEC_base_l>): ... this.
5693 (vec_extract<mode>): Renamed to ...
5694 (vec_extract<mode><VEC_base_l>): ... this.
5695 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
5696 (vec_initv2sfsf): ... this.
5697 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
5698 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5699 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
5700 element mode after mode in gen_vec_init* calls.
5701 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
5702 (vec_init<mode><Vel>): ... this.
5703 (vec_extract<mode>): Renamed to ...
5704 (vec_extract<mode><Vel>): ... this.
5705 * config/aarch64/iterators.md (Vel): New mode attribute.
5706 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
5707 Add element mode after mode in gen_vec_extract* calls.
5708 * config/s390/vector.md (non_vec_l): New mode attribute.
5709 (vec_extract<mode>): Renamed to ...
5710 (vec_extract<mode><non_vec_l>): ... this.
5711 (vec_init<mode>): Renamed to ...
5712 (vec_init<mode><non_vec_l>): ... this.
5713 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
5714 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
5716 * config/arm/iterators.md (V_elem_l): New mode attribute.
5717 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
5718 (vec_extract<mode><V_elem_l>): ... this.
5719 (vec_extractv2di): Renamed to ...
5720 (vec_extractv2didi): ... this.
5721 (vec_init<mode>): Renamed to ...
5722 (vec_init<mode><V_elem_l>): ... this.
5723 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
5724 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
5725 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
5726 Add element mode after gen_vec_extract* calls.
5727 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
5728 (vec_init<mode><unitmode>): ... this.
5729 (vec_extract<mode>): Renamed to ...
5730 (vec_extract<mode><unitmode>): ... this.
5731 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
5732 (vec_init<mode><unitmode>): ... this.
5733 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
5734 (vec_initv2sfsf): ... this.
5735 (vec_extractv2sf): Renamed to ...
5736 (vec_extractv2sfsf): ... this.
5737 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
5738 Add element mode after gen_vec_extract* calls.
5739 * config/mips/mips.md (unitmode): New mode iterator.
5740 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
5741 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
5742 * config/spu/spu.md (inner_l): New mode attribute.
5743 (vec_init<mode>): Renamed to ...
5744 (vec_init<mode><inner_l>): ... this.
5745 (vec_extract<mode>): Renamed to ...
5746 (vec_extract<mode><inner_l>): ... this.
5747 * config/sparc/sparc.md (veltmode): New mode iterator.
5748 (vec_init<VMALL:mode>): Renamed to ...
5749 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
5750 * config/ia64/vect.md (vec_initv2si): Renamed to ...
5751 (vec_initv2sisi): ... this.
5752 (vec_initv2sf): Renamed to ...
5753 (vec_initv2sfsf): ... this.
5754 (vec_extractv2sf): Renamed to ...
5755 (vec_extractv2sfsf): ... this.
5756 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
5757 (vec_init<mode>): Renamed to ...
5758 (vec_init<mode><VEC_base_l>): ... this.
5759 (vec_extract<mode>): Renamed to ...
5760 (vec_extract<mode><VEC_base_l>): ... this.
5761 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
5762 (vec_initv2sfsf): ... this.
5763 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
5764 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5765 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
5766 gen_vec_init* calls.
5768 2017-08-01 Richard Biener <rguenther@suse.de>
5770 PR tree-optimization/81297
5771 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
5772 TREE_OVERFLOW from INTEGER_CSTs.
5774 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
5776 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
5778 2017-07-31 Carl Love <cel@us.ibm.com>
5780 * config/rs6000/rs6000-c: Add support for built-in functions
5781 vector signed char vec_xl_be (signed long long, signed char *);
5782 vector unsigned char vec_xl_be (signed long long, unsigned char *);
5783 vector signed int vec_xl_be (signed long long, signed int *);
5784 vector unsigned int vec_xl_be (signed long long, unsigned int *);
5785 vector signed long long vec_xl_be (signed long long, signed long long *);
5786 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
5787 vector signed short vec_xl_be (signed long long, signed short *);
5788 vector unsigned short vec_xl_be (signed long long, unsigned short *);
5789 vector double vec_xl_be (signed long long, double *);
5790 vector float vec_xl_be (signed long long, float *);
5791 * config/rs6000/altivec.h (vec_xl_be): Add #define.
5792 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
5793 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
5795 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
5796 (altivec_expand_builtin): Add switch statement to call
5797 altivec_expand_xl_be for each builtin.
5798 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
5799 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
5800 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
5801 __builtin_vsx_le_be_v16qi.
5802 * doc/extend.texi: Update the built-in documentation file for the
5803 new built-in functions.
5805 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5808 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
5810 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5812 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5814 * config.gcc: Add z14.
5815 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
5816 CPU model numbers for z13s and z14.
5817 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
5819 * config/s390/s390-opts.h (enum processor_type): Rename
5820 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5821 * config/s390/s390.c (processor_table): Add field for CPU name to
5822 be passed to Binutils.
5823 (s390_asm_output_machine_for_arch): Use the new field in
5824 processor_table for Binutils.
5825 (s390_expand_builtin): Replace arch12 with z14.
5826 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5827 (s390_get_sched_attrmask): Likewise.
5828 (s390_get_unit_mask): Likewise.
5829 * config/s390/s390.opt: Add z14 to processor_type enum.
5831 2017-07-31 Martin Jambor <mjambor@suse.cz>
5834 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
5835 regardless of optimization level.
5837 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
5838 Martin Liska <mliska@suse.cz>
5840 * predict.def: Remove old comment and adjust probability.
5841 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
5844 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5847 * config/i386/i386.c (ix86_function_naked): New function.
5848 (ix86_can_use_return_insn_p): Return false for naked functions.
5849 (ix86_expand_prologue): Skip prologue for naked functions.
5850 (ix86_expand_epilogue): Skip epilogue for naked functions
5851 and emit trap instruction.
5852 (ix86_warn_func_return): New function.
5853 (ix86_attribute_table): Add "naked" attribute specification.
5854 (TARGET_WARN_FUNC_RETURN): Define.
5855 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
5857 2017-07-31 Martin Liska <mliska@suse.cz>
5859 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
5860 (dump_gimple_bb_header): Always dump BB info.
5861 (pp_cfg_jump): Do not append info about BB when dumping a jump.
5863 2017-07-31 Martin Liska <mliska@suse.cz>
5866 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
5867 also with current_function_decl non-null equality.
5869 2017-07-31 Jakub Jelinek <jakub@redhat.com>
5872 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
5873 change type to the element type, instead add eltype variable and
5874 use it where we are interested in the element type.
5876 PR tree-optimization/81603
5877 * ipa-polymorphic-call.c
5878 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
5879 offset arithmetic in offset_int, bail out if the resulting bit offset
5880 doesn't fit into shwi.
5882 2017-07-31 Martin Liska <mliska@suse.cz>
5884 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
5885 (gimplify_save_expr): Fix comment.
5887 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
5890 * config/i386/i386.c (ix86_function_arg): Update arguments for
5892 (ix86_compute_frame_layout): Set the initial stack offset to
5893 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
5894 INCOMING_FRAME_SP_OFFSET.
5895 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
5896 stack before exception handler returns.
5897 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
5898 the 'ERROR_CODE' for exception handler.
5900 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
5902 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
5903 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
5904 (ASM_OUTPUT_REG_POP): Ditto.
5905 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
5906 instead of asm_fprintf to output pure string.
5908 2017-07-29 Jakub Jelinek <jakub@redhat.com>
5910 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
5911 to imported_module_or_decl hook.
5912 (debug_nothing_tree_tree_tree_bool): Remove.
5913 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
5914 * debug.c (do_nothing_debug_hooks): Use
5915 debug_nothing_tree_tree_tree_bool_bool instead of
5916 debug_nothing_tree_tree_tree_bool.
5917 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5918 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
5919 * sdbout.c (sdb_debug_hooks): Likewise.
5920 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
5921 (gen_namespace_die): Add DW_AT_export_symbols attribute if
5923 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
5924 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
5925 attribute, don't add anything.
5927 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5929 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
5930 (fold_build2_stat_loc): Likewise.
5931 (fold_build3_stat_loc): Likewise.
5932 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
5933 (fold_build1_loc): Remove macro.
5934 (fold_build2_loc): Likewise.
5935 (fold_build3_loc): Likewise.
5937 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5939 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
5940 (gimple_build_debug_bind_source_stat): Likewise.
5941 * gimple.h (gimple_build_debug_bind): Remove macro.
5942 (gimple_build_debug_bind_source): Likewise.
5944 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5946 * bitmap.c (bitmap_alloc): Adjust.
5947 (bitmap_gc_alloc): Likewise.
5948 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
5950 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5952 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
5953 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
5954 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
5955 (bitmap_gc_alloc_stat): Likewise.
5956 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
5958 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5960 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
5961 * rtl.h (shallow_copy_rtx): Remove macro.
5963 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5965 * emit-rtl.c (gen_raw_REG): Adjust.
5966 * gengenrtl.c (gendef): Likewise.
5967 * rtl.c (rtx_alloc_stat): Remove _stat from name.
5968 * rtl.h (rtx_alloc): Remove macro.
5970 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5972 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
5973 (build_tree_list_stat): Likewise.
5974 * tree.h (build_tree_list): Remove macro.
5975 (build_tree_list_vec): Likewise.
5977 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5979 * tree.c (make_vector_stat): Remove _stat from name.
5980 (build_vector_stat): Likewise.
5981 * tree.h (make_vector_stat): Remove macro.
5982 (build_vector_stat): Likewise.
5984 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5986 * tree.h (build_var_debug_value): Remove prototype.
5988 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5990 * tree.c (tree_cons_stat): Remove _stat from name.
5991 * tree.h (tree_cons): Remove macro.
5993 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5995 * tree.c (build_vl_exp_stat): Remove _stat from name.
5996 * tree.h (build_vl_exp): Remove macro.
5998 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6000 * tree.c (build_decl_stat): Remove _stat from name.
6001 * tree.h (build_decl): Remove macro.
6003 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6005 * gimple.c (gimple_build_with_ops_stat): Adjust.
6006 (gimple_alloc_stat): Remove _stat from name.
6007 * gimple.h (gimple_alloc): Remove macro.
6009 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6011 * tree.c (make_tree_vec_stat): Remove _stat from name.
6012 (grow_tree_vec_stat): Likewise.
6013 * tree.h (make_tree_vec_stat): Adjust prototype.
6014 (grow_tree_vec_stat): Likewise.
6015 (make_tree_vec): Remove macro.
6016 (grow_tree_vec): Likewise.
6018 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6020 * fold-const.c (fold_build1_stat_loc): Adjust.
6021 (fold_build2_stat_loc): Likewise.
6022 (fold_build3_stat_loc): Likewise.
6023 * tree.c (build0_stat): Remove _stat from name.
6024 (build1_stat): Likewise.
6025 (build2_stat): Likewise.
6026 (build3_stat): Likewise.
6027 (build4_stat): Likewise.
6028 (build5_stat): Likewise.
6029 * tree.h (build1_loc): Remove macro, and rename _stat function
6031 (build2_loc): Likewise.
6032 (build3_loc): Likewise.
6033 (build4_loc): Likewise.
6034 (build5_loc): Likewise.
6036 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6038 * tree.c (make_int_cst_stat): Remove _stat from name.
6039 * tree.h (make_int_cst_stat): Adjust prototype.
6040 (make_int_cst): Remove macro.
6042 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6044 * tree.c (make_tre_binfo_stat): Remove _stat from name.
6045 * tree.h (make_tree_binfo_stat): Adjust prototype.
6046 (make_tree_binfo): Remove.
6048 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6050 * tree.c (copy_node_stat): Rename to copy_node.
6051 (build_distinct_type_copy): Adjust.
6052 * tree.h (copy_node_stat): Adjust prototype.
6053 (copy_node): Remove macro.
6055 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6057 * tree.c (make_node_stat): rename to make_node.
6058 (build_tree_list_stat): Adjust.
6059 (build0_stat): Likewise.
6060 (build2_stat): Likewise.
6061 (build3_stat): Likewise.
6062 (build4_stat): Likewise.
6063 (build5_stat): Likewise.
6064 (build_decl_stat): Likewise.
6065 * tree.h (make_node_stat): Adjust prototype.
6066 (make_node): remove macro.
6068 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
6070 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
6071 (PPC_FEATURE2_SCV): Likewise.
6072 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
6074 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6076 * config/aarch64/aarch64.c
6077 (aarch64_internal_mov_immediate): Add new special pattern.
6078 * config/aarch64/aarch64.md (*movdi_aarch64):
6079 Add reg/32bit const mov case.
6081 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6082 Richard Sandiford <richard.sandiford@linaro.org>
6084 * config/aarch64/aarch64.md (mov<mode>): Generalize.
6085 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6086 Add integer and movi cases.
6087 (movi-split-hf-df-sf split, fp16): New.
6088 (enabled): Added TARGET_FP_F16INST.
6089 * config/aarch64/iterators.md (GPF_HF): New.
6090 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
6092 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6094 * config/aarch64/aarch64.c
6095 (aarch64_simd_container_mode): Add prototype.
6096 (aarch64_expand_mov_immediate): Add HI support.
6097 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
6098 (aarch64_can_const_movi_rtx_p): New.
6099 (aarch64_preferred_reload_class):
6100 Remove restrictions of using FP registers for certain SIMD operations.
6101 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
6102 (aarch64_valid_floating_const): Add integer move validation.
6103 (aarch64_simd_imm_scalar_p): Remove.
6104 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
6105 (aarch64_legitimate_constant_p): Expand list of supported cases.
6106 * config/aarch64/aarch64-protos.h
6107 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
6108 (aarch64_reinterpret_float_as_int): New.
6109 (aarch64_simd_imm_scalar_p): Remove.
6110 * config/aarch64/constraints.md (Uvi): New.
6111 (Dd): Split into Ds and new Dd.
6112 * config/aarch64/aarch64.md (*movsi_aarch64):
6114 (*movdi_aarch64): Add SIMD mov case.
6116 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6118 * tree-predcom.c: (struct chain): Handle store-store chain in which
6119 stores for elimination only store loop invariant values.
6120 (execute_pred_commoning_chain): Ditto.
6121 (prepare_initializers_chain_store_elim): Ditto.
6122 (prepare_finalizers): Ditto.
6123 (is_inv_store_elimination_chain): New function.
6124 (initialize_root_vars_store_elim_1): New function.
6126 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6128 * tree-predcom.c: Revise general description of the pass.
6129 (enum chain_type): New enum type for store elimination.
6130 (struct chain): New field supporting store elimination.
6131 (struct component): Ditto.
6132 (dump_chain): Dump store-stores chain.
6133 (release_chain): Release resources.
6134 (split_data_refs_to_components): Compute and create component
6135 contains only stores for elimination.
6136 (get_chain_last_ref_at): New function.
6137 (make_invariant_chain): Initialization.
6138 (make_rooted_chain): Specify chain type in parameter and record it.
6139 (add_looparound_copies): Skip for store-stores chain.
6140 (determine_roots_comp): Compute type of chain and pass it to
6142 (initialize_root_vars_store_elim_2): New function.
6143 (finalize_eliminated_stores): New function.
6144 (remove_stmt): Handle store for elimination.
6145 (execute_pred_commoning_chain): Execute predictive commoning on
6147 (determine_unroll_factor): Skip unroll for store-stores chain.
6148 (prepare_initializers_chain_store_elim): New function.
6149 (prepare_initializers_chain): Hanlde store-store chain.
6150 (prepare_finalizers_chain, prepare_finalizers): New function.
6151 (tree_predictive_commoning_loop): Return integer value indicating
6152 if loop is unrolled or lcssa form is corrupted.
6153 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
6155 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6157 * tree-predcom.c (initialize_root): Delete.
6158 (execute_pred_commoning_chain): Initialize root vars and replace
6159 reference of non-combined chain directly, rather than call above
6162 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6164 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
6165 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
6167 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6169 * tree-predcom.c (struct chain): New field init_seq.
6170 (release_chain): Release init_seq.
6171 (prepare_initializers_chain): Record intialization stmts in above
6173 (insert_init_seqs): New function.
6174 (tree_predictive_commoning_loop): Call insert_init_seqs.
6176 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6178 * tree-predcom.c (determine_roots_comp): Skip trivial components.
6180 2017-07-28 Richard Biener <rguenther@suse.de>
6182 * match.pd: Remove superfluous :c.
6183 * genmatch.c (simplify::id): Add member.
6184 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
6186 (current_id): New global.
6187 (dt_node::parent): Move from ...
6188 (dt_operand::parent): ... here. Add for_id member.
6189 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
6190 (decision_tree::find_node): Relax order requirement when
6191 merging DT_TRUE nodes to ones inbetween the current simplify
6192 and the one we try to merge with. Add diagnostic whenever
6193 we need to enforce pattern order by not merging.
6194 (decision_tree::insert): Set current_id.
6195 (decision_tree::print_node): Dump parent node and for_id.
6196 (parser::last_id): Add member.
6197 (parser::push_simplify): Assign unique id.
6198 (parser::parser): Initialize last_id.
6200 2017-07-28 Martin Liska <mliska@suse.cz>
6203 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
6204 gimple_build_debug_bind.
6206 2017-07-28 Richard Biener <rguenther@suse.de>
6208 PR tree-optimization/81502
6209 * match.pd: Add pattern combining BIT_INSERT_EXPR with
6211 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
6213 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
6214 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
6215 for BIT_FIELD_REF args.
6216 * fold-const.c (make_bit_field_ref): Likewise.
6217 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6219 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6222 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
6223 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
6224 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
6225 Or it into SANITIZER_UNDEFINED.
6226 * ubsan.c: Include gimple-fold.h and varasm.h.
6227 (ubsan_expand_ptr_ifn): New function.
6228 (instrument_pointer_overflow): New function.
6229 (maybe_instrument_pointer_overflow): New function.
6230 (instrument_object_size): Formatting fix.
6231 (pass_ubsan::execute): Call instrument_pointer_overflow
6232 and maybe_instrument_pointer_overflow.
6233 * internal-fn.c (expand_UBSAN_PTR): New function.
6234 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
6235 * sanitizer.def (__ubsan_handle_pointer_overflow,
6236 __ubsan_handle_pointer_overflow_abort): New builtins.
6237 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
6238 * internal-fn.def (UBSAN_PTR): New internal function.
6239 * opts.c (sanitizer_opts): Add pointer-overflow.
6240 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
6241 * fold-const.c (build_range_check): Compute pointer range check in
6242 integral type if pointer arithmetics would be needed. Formatting
6245 2017-07-28 Martin Liska <mliska@suse.cz>
6248 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
6249 parameters that are of a variable-length.
6251 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
6253 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
6255 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
6256 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
6257 (CRT_CALL_STATIC_FUNCTION): Likewise.
6258 (ASM_DEFAULT_SPEC): New define.
6259 (ASM_SPEC32): Likewise.
6260 (ASM_SPEC64): Likewise.
6261 (ASM_SPEC_COMMON): Likewise.
6262 (ASM_SPEC): Likewise.
6263 (INVALID_64BIT): Likewise.
6264 (LINK_OS_DEFAULT_SPEC): Likewise.
6265 (LINK_OS_SPEC32): Likewise.
6266 (LINK_OS_SPEC64): Likewise.
6267 (POWERPC_LINUX): Likewise.
6268 (PTRDIFF_TYPE): Likewise.
6269 (RESTORE_FP_PREFIX): Likewise.
6270 (RESTORE_FP_SUFFIX): Likewise.
6271 (SAVE_FP_PREFIX): Likewise.
6272 (SAVE_FP_SUFFIX): Likewise.
6273 (SIZE_TYPE): Likewise.
6274 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
6275 (TARGET_64BIT): Likewise.
6276 (TARGET_64BIT): Likewise.
6277 (TARGET_AIX): Likewise.
6278 (WCHAR_TYPE_SIZE): Likewise.
6279 (WCHAR_TYPE): Undefine.
6280 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
6281 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
6282 (CPP_OS_RTEMS_SPEC): Delete.
6283 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
6284 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
6286 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
6288 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6290 PR tree-optimization/81578
6291 * tree-parloops.c (build_new_reduction): Bail out if
6292 reduction_code isn't one of the standard OpenMP reductions.
6293 Move the details printing after that decision.
6295 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6297 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
6298 related to reload_in_progress.
6299 (splat_input_operand): Likewise.
6300 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
6302 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
6304 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
6305 (TARGET_INSTANTIATE_DECLS): Likewise.
6306 (legitimate_indexed_address_p): Delete reload_in_progress code.
6307 (rs6000_debug_legitimate_address_p): Likewise.
6308 (rs6000_eliminate_indexed_memrefs): Likewise.
6309 (rs6000_emit_le_vsx_store): Likewise.
6310 (rs6000_emit_move_si_sf_subreg): Likewise.
6311 (rs6000_emit_move): Likewise.
6312 (register_to_reg_type): Likewise.
6313 (rs6000_pre_atomic_barrier): Likewise.
6314 (rs6000_machopic_legitimize_pic_address): Likewise.
6315 (rs6000_allocate_stack_temp): Likewise.
6316 (rs6000_address_for_fpconvert): Likewise.
6317 (rs6000_address_for_altivec): Likewise.
6318 (rs6000_secondary_memory_needed_rtx): Delete function.
6319 (rs6000_check_sdmode): Likewise.
6320 (rs6000_alloc_sdmode_stack_slot): Likewise.
6321 (rs6000_instantiate_decls): Likewise.
6322 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
6323 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
6324 Delete reload_in_progress.
6325 (*vec_reload_and_plus_<mptrsize>): Likewise.
6326 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
6327 (vsx_div_v2di): Likewise.
6328 (vsx_udiv_v2di): Likewise.
6330 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6332 * config/rs6000/rs6000.opt (mlra): Replace with stub.
6333 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
6334 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
6335 (rs6000_debug_reg_global): Delete print of LRA status.
6336 (rs6000_option_override_internal): Delete dead LRA related code.
6337 (rs6000_lra_p): Delete function.
6338 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
6340 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
6342 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
6343 * config/riscv/rtems.h: New file.
6345 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6346 Sudakshina Das <sudi.das@arm.com>
6348 * config/aarch64/aarch64.md
6349 (define_split for and<mode>3nr_compare): Move
6350 non aarch64_logical_operand to a register.
6351 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
6352 register immediate operand to a register.
6353 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
6355 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6358 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
6360 2017-07-27 Richard Biener <rguenther@suse.de>
6362 PR tree-optimization/81573
6363 PR tree-optimization/81494
6364 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
6365 multi defuse cycle case.
6367 2017-07-27 Richard Biener <rguenther@suse.de>
6369 PR tree-optimization/81571
6370 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
6373 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
6375 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
6376 earlier and only if MASK_FPU is set. Adjust formatting.
6378 2017-07-27 Martin Liska <mliska@suse.cz>
6380 * opt-functions.awk: Add validation of value of Init.
6381 * optc-gen.awk: Pass new argument.
6383 2017-07-27 Martin Liska <mliska@suse.cz>
6385 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
6386 Fix wrong condition.
6388 2017-07-27 Martin Liska <mliska@suse.cz>
6390 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
6391 BBs and edges seen by autoFDO.
6393 2017-07-27 Richard Biener <rguenther@suse.de>
6395 PR tree-optimization/81502
6396 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
6397 with incompatible but same sized type.
6398 (execute_update_addresses_taken): Likewise.
6400 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
6402 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
6403 flag_tree_loop_vectorize rather than flag_tree_vectorize.
6405 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6408 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
6409 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
6410 Change s_operand to memory_operand.
6412 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
6414 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
6415 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
6416 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
6417 Emit instructions rather than returning an expression. Handle TFmode
6418 and KFmode by casting to TImode.
6419 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
6420 (rs6000_emit_le_vsx_store): Likewise.
6421 * config/rs6000/vsx.md (VSX_TI): New iterator.
6422 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
6423 (*vsx_le_undo_permute_<mode>): Likewise.
6424 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
6425 emit the split sequence.
6426 (*vsx_le_perm_store_<mode>): Likewise.
6428 2017-07-27 Jakub Jelinek <jakub@redhat.com>
6430 PR tree-optimization/81555
6431 PR tree-optimization/81556
6432 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
6433 if true, force CHANGED for the recursive invocation.
6434 (reassociate_bb): Remember original length of ops array, pass
6435 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
6437 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
6438 attributes for noipa attribute. For naked attribute use
6439 lookup_attribute first before lookup_attribute_spec.
6440 * final.c (rest_of_handle_final): Disable IPA RA for functions with
6442 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
6443 for functions with noipa attribute.
6444 (cgraph_externally_visible_p): Return true for functions with noipa
6446 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
6447 for functions with noipa attribute.
6448 * doc/extend.texi: Document noipa function attribute.
6449 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
6450 also for functions with noipa attribute.
6451 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
6453 2017-07-26 Andrew Pinski <apinski@cavium.com>
6455 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
6456 vec_unalign_load_cost and vec_unalign_store_cost.
6458 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6460 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
6461 -mvsx-small-integer option.
6462 (ISA_3_0_MASKS_IEEE): Likewise.
6463 (OTHER_VSX_VECTOR_MASKS): Likewise.
6464 (POWERPC_MASKS): Likewise.
6465 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
6466 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
6467 code, only testing for DImode being allowed in non-VSX floating
6469 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
6470 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
6472 (rs6000_option_override_internal): Delete -mvsx-small-integer.
6473 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
6474 TARGET_P8_VECTOR test.
6475 (rs6000_secondary_reload_simple_move): Likewise.
6476 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
6477 since TARGET_P9_VECTOR was already tested.
6478 (rs6000_opt_masks): Remove -mvsx-small-integer.
6479 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
6480 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6482 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
6483 test for TARGET_VEXTRACTUB was used, and that uses
6485 (p9 extract splitter): Likewise.
6486 (vsx_extract_<mode>_di_p9): Likewise.
6487 (vsx_extract_<mode>_store_p9): Likewise.
6488 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
6489 for TARGET_P9_VECTOR was used. Delete code that is now dead with
6490 the elimination of TARGET_VSX_SMALL_INTEGER.
6491 (vsx_extract_<mode>_p8): Likewise.
6492 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
6493 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
6494 (vsx_set_<mode>_p9): Likewise.
6495 (vsx_set_v4sf_p9): Likewise.
6496 (vsx_set_v4sf_p9_zero): Likewise.
6497 (vsx_insert_extract_v4sf_p9): Likewise.
6498 (vsx_insert_extract_v4sf_p9_2): Likewise.
6499 * config/rs6000/rs6000.md (sign extend splitter): Change
6500 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
6501 (floatsi<mode>2_lfiwax_mem): Likewise.
6502 (floatunssi<mode>2_lfiwzx_mem): Likewise.
6503 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
6504 since a test for TARGET_P9_VECTOR was used.
6505 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6506 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
6507 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6508 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6509 TARGET_P8_VECTOR test.
6510 (fix_trunc<mode>si2_stfiwx): Likewise.
6511 (fix_trunc<mode>si2_internal): Likewise.
6512 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
6513 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6515 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6516 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6517 TARGET_P8_VECTOR test.
6518 (fixuns_trunc<mode>si2_stfiwx): Likewise.
6519 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
6520 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6522 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6523 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
6524 since a test for TARGET_P9_VECTOR was used.
6525 (splitter for loading small constants): Likewise.
6527 2017-07-26 Andrew Pinski <apinski@cavium.com>
6529 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
6532 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
6535 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
6536 (fp_valid_at): Likewise.
6538 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6540 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
6541 (qdf24xx_addrcost_table): Likewise.
6542 (cortexa57_tunings): Update to use generic_branch_cost.
6543 (cortexa72_tunings): Likewise.
6544 (cortexa73_tunings): Likewise.
6545 (qdf24xx_tunings): Likewise.
6547 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6549 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
6550 (thunderx2t99_branch_cost): Likewise.
6551 (cortexa35_tunings): Update to use generic_branch_cost.
6552 (cortexa53_tunings): Likewise.
6553 (cortexa57_tunings): Likewise.
6554 (cortexa72_tunings): Likewise.
6555 (cortexa73_tunings): Likewise.
6556 (thunderx2t99_tunings): Likewise.
6558 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6560 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
6561 (sparc_option_override): Honour MASK_FSMULD.
6562 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
6563 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
6564 * config/sparc/sparc.opt (mfsmuld): New option.
6565 * doc/invoke.texi (mfsmuld): Document option.
6567 2017-07-26 Marek Polacek <polacek@redhat.com>
6570 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
6572 2017-07-26 Richard Biener <rguenther@suse.de>
6574 * gimple-match-head.c (do_valueize): Return OP if valueize
6576 (get_def): New helper to get at the def stmt of a SSA name
6578 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
6579 do_valueize to get at the def stmt.
6580 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
6582 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
6585 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
6587 2017-07-26 Martin Liska <mliska@suse.cz>
6590 * function.c (expand_function_start): Make expansion of
6591 nonlocal_goto_save_area after parm_birth_insn.
6593 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6595 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
6596 from all CPU target flags enable members.
6598 2017-07-26 Richard Biener <rguenther@suse.de>
6600 * genmatch.c (dt_simplify::gen): Make iterator vars const.
6601 (decision_tree::gen): Make 'type' const.
6602 (write_predicate): Likewise.
6604 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6606 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
6607 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
6608 (rs6000_option_override_internal): Likewise.
6609 (rs6000_expand_vector_set): Likewise.
6610 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
6611 (TARGET_UPPER_REGS_SF): Likewise.
6612 (TARGET_UPPER_REGS_DI): Likewise.
6613 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
6614 (TARGET_DIRECT_MOVE_64BIT): Likewise.
6615 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
6616 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6617 (Splitters for DI constants in Altivec registers): Likewise.
6618 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
6619 (vsx_set_v4sf_p9): Likewise.
6620 (vsx_set_v4sf_p9_zero): Likewise.
6621 (vsx_insert_extract_v4sf_p9): Likewise.
6622 (vsx_insert_extract_v4sf_p9_2): Likewise.
6624 2017-07-25 Carl Love <cel@us.ibm.com>
6626 * doc/extend.texi: Update the built-in documentation file for the
6627 existing built-in functions
6628 vector signed char vec_cnttz (vector signed char);
6629 vector unsigned char vec_cnttz (vector unsigned char);
6630 vector signed short vec_cnttz (vector signed short);
6631 vector unsigned short vec_cnttz (vector unsigned short);
6632 vector signed int vec_cnttz (vector signed int);
6633 vector unsigned int vec_cnttz (vector unsigned int);
6634 vector signed long long vec_cnttz (vector signed long long);
6635 vector unsigned long long vec_cnttz (vector unsigned long long);
6637 2017-07-25 Andrew Pinski <apinski@cavium.com>
6639 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
6640 accesses where the use is for the first operand of a BIT_INSERT.
6642 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6645 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6646 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6648 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6650 * config/i386/gstabs.h: Delete.
6651 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
6653 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
6655 * config/i386/i386.c (ix86_decompose_address): Do not check for
6656 register RTX when looking at index_reg or base_reg.
6657 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
6659 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
6661 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
6662 to update EH info here.
6664 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6666 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
6668 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6670 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
6672 2017-07-25 Torsten Duwe <duwe@suse.de>
6674 * common.opt: Introduce -fpatchable-function-entry
6675 command line option, and its variables function_entry_patch_area_size
6676 and function_entry_patch_area_start.
6677 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
6678 including a two-value parser.
6679 * target.def (print_patchable_function_entry): New target hook.
6680 * targhooks.h (default_print_patchable_function_entry): New function.
6681 * targhooks.c (default_print_patchable_function_entry): Likewise.
6682 * toplev.c (process_options): Switch off IPA-RA if
6683 patchable function entries are being generated.
6684 * varasm.c (assemble_start_function): Look at the
6685 patchable-function-entry command line switch and current
6686 function attributes and maybe generate NOP instructions by
6687 calling the print_patchable_function_entry hook.
6688 * doc/extend.texi: Document patchable_function_entry attribute.
6689 * doc/invoke.texi: Document -fpatchable_function_entry
6690 command line option.
6691 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
6693 * doc/tm.texi: Re-generate.
6695 2017-07-25 Jakub Jelinek <jakub@redhat.com>
6698 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
6699 TARGET_AVX512DQ rather than TARGET_AVX512BW.
6701 2017-07-25 Tamar Christina <tamar.christina@arm.com>
6703 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
6705 2017-07-25 Richard Biener <rguenther@suse.de>
6707 PR tree-optimization/81455
6708 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
6709 not walk in cycles when looking for guards.
6711 2017-07-25 Richard Biener <rguenther@suse.de>
6713 PR tree-optimization/81529
6714 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
6715 when optimizing backedge uses.
6717 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
6719 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
6721 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
6722 to dl_section_ref. On AIX, append an expression to subtract
6723 the size of the section length to dl_section_ref.
6725 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
6727 * configure.ac: If any of the config.* scripts fail, exit 1.
6728 * configure: Regenerate.
6730 2017-07-25 Richard Biener <rguenther@suse.de>
6733 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
6734 of immediate uses, be more verbose on errors.
6736 2017-07-25 Richard Biener <rguenther@suse.de>
6738 PR tree-optimization/81510
6739 * tree-vect-loop.c (vect_is_simple_reduction): When the
6740 reduction stmt is not inside the loop bail out.
6742 2017-07-25 Richard Biener <rguenther@suse.de>
6744 PR tree-optimization/81303
6745 * tree-vect-loop-manip.c (vect_loop_versioning): Build
6746 profitability check against LOOP_VINFO_NITERSM1.
6748 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6750 * domwalk.c (cmp_bb_postorder): Simplify.
6751 (sort_bbs_postorder): New function. Use it...
6752 (dom_walker::walk): ...here to optimize common cases.
6754 2017-07-25 Martin Liska <mliska@suse.cz>
6757 * ipa-visibility.c (function_and_variable_visibility): Make the
6758 redirection just on target that supports aliasing.
6759 Fix GNU coding style.
6761 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
6764 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
6766 * config/arm/rtems.h: Likewise.
6767 * config/bfin/rtems.h: Likewise.
6768 * config/i386/rtemself.h: Likewise.
6769 * config/lm32/rtems.h: Likewise.
6770 * config/m32c/rtems.h: Likewise.
6771 * config/m68k/rtemself.h: Likewise.
6772 * config/microblaze/rtems.h: Likewise.
6773 * config/mips/rtems.h: Likewise.
6774 * config/moxie/rtems.h: Likewise.
6775 * config/nios2/rtems.h: Likewise.
6776 * config/powerpcspe/rtems.h: Likewise.
6777 * config/rs6000/rtems.h: Likewise.
6778 * config/rtems.h: Likewise.
6779 * config/sh/rtems.h: Likewise.
6780 * config/sh/rtemself.h: Likewise.
6781 * config/sparc/rtemself.h: Likewise.
6783 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6786 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
6787 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
6788 * tree-ssa-structalias.c (alias_get_name): Same.
6790 2017-07-25 Bin Cheng <bin.cheng@arm.com>
6793 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
6794 instructions if no du chain is found.
6796 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6798 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
6800 2017-07-25 Richard Biener <rguenther@suse.de>
6803 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
6806 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6808 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
6810 (ISA_2_7_MASKS_SERVER): Likewise.
6811 (ISA_3_0_MASKS_IEEE): Likewise.
6812 (OTHER_P8_VECTOR_MASKS): Likewise.
6813 (OTHER_VSX_VECTOR_MASKS): Likewise.
6814 (POWERPC_MASKS): Likewise.
6815 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
6816 duplicate list of options.
6817 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
6818 explicit -mupper-regs options.
6819 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
6820 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
6821 alias for -mupper-regs-df.
6822 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6823 (rs6000_init_hard_regno_mode_ok): Likewise.
6824 (rs6000_option_override_internal): Likewise.
6825 (rs6000_opt_masks): Likewise.
6826 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
6827 options in terms of whether -mvsx or -mpower8-vector was used.
6828 (TARGET_UPPER_REGS_DI): Likewise.
6829 (TARGET_UPPER_REGS_SF): Likewise.
6830 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
6831 -mupper-regs-* options.
6833 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6835 * passes.c (emergency_dump_function): Print some empty lines and a
6836 header before the RTL dump.
6838 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6840 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
6842 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
6845 * config/aarch64/aarch64.c (aarch64_classify_symbol):
6846 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
6848 2017-07-24 Carl Love <cel@us.ibm.com>
6850 * config/rs6000/rs6000-c.c: Add support for built-in functions
6851 vector float vec_extract_fp32_from_shorth (vector unsigned short);
6852 vector float vec_extract_fp32_from_shortl (vector unsigned short);
6853 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
6854 vec_extract_fp_from_shortl): Add defines for the two builtins.
6855 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
6856 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
6858 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
6859 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
6860 * doc/extend.texi: Update the built-in documentation file for the
6861 new built-in function.
6863 2017-07-24 Jakub Jelinek <jakub@redhat.com>
6866 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
6868 * doc/generic.texi: Likewise.
6869 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6870 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6872 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
6874 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
6875 (aarch64_mls_elt_merge<mode>): Likewise.
6877 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
6879 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
6880 having __cxa_atexit.
6882 2017-07-23 Michael Collison <michael.collison@arm.com>
6884 * config/arm/arm.c (arm_option_override): Deprecate
6885 use of -mstructure-size-boundary.
6886 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
6887 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
6889 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6892 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
6893 Reduce cost estimate for direct moves.
6895 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
6898 * config/i386/i386.c (ix86_option_override_internal): Disable
6899 BMI, BMI2 and TBM instructions for -m16.
6901 2017-07-21 Carl Love <cel@us.ibm.com>
6903 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6904 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6905 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
6906 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
6907 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
6908 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
6909 VMULOSW): New enum "unspec" values.
6910 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
6911 altivec_vmulosw): New patterns.
6912 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
6913 VMULOSW): Add definitions.
6915 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
6917 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
6918 (qdf24xx): Likewise.
6919 * config/aarch64/aarch64-options-extensions.def (rdma); New.
6920 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
6921 (AARCH64_FL_V8_1): Renumber.
6922 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
6923 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
6924 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
6925 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
6926 rdma to feature modifiers list.
6928 2017-07-21 Yury Gribov <tetra2005@gmail.com>
6931 * ipa-visibility (function_and_variable_visibility): Convert
6932 recursive PLT call to direct call if appropriate.
6934 2017-07-21 Andrew Pinski <apinski@cavium.com>
6936 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
6937 operand 1 to see if the types precision matches.
6938 * fold-const.c (operand_equal_p): Likewise.
6940 2017-07-21 Richard Biener <rguenther@suse.de>
6942 PR tree-optimization/81303
6943 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
6944 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
6945 (vect_peeling_hash_get_lowest_cost): Adjust.
6946 (vect_enhance_data_refs_alignment): Likewise. Use
6947 vect_get_peeling_costs_all_drs to compute the penalty for no
6948 peeling to match up costs.
6950 2017-07-21 Richard Biener <rguenther@suse.de>
6952 PR tree-optimization/81500
6953 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
6954 we didn't identify a reduction path.
6956 2017-07-21 Tom de Vries <tom@codesourcery.com>
6957 Cesar Philippidis <cesar@codesourcery.com>
6959 PR gcov-profile/81442
6960 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
6963 2017-07-21 Tom de Vries <tom@codesourcery.com>
6966 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
6968 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
6969 nvptx_override_options_after_change.
6971 2017-07-21 Ulrich Drepper <drepper@redhat.com>
6973 * dwarf2out.c (output_file_names): Avoid double testing for
6976 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
6978 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
6980 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
6982 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
6984 (try_crossjump_to_edge): Do not punt on partitioned functions.
6986 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
6988 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
6989 Put all BBs reachable only via paths crossing cold region to cold
6991 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
6993 2016-07-21 Richard Biener <rguenther@suse.de>
6995 PR tree-optimization/81303
6996 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
6997 into account prologue and epilogue iterations when raising
6998 min_profitable_iters to sth at least covering one vector iteration.
7000 2017-07-21 Tamar Christina <tamar.christina@arm.com>
7002 * config/arm/arm.c (arm_test_cpu_arch_dat):
7005 2017-07-20 Nathan Sidwell <nathan@acm.org>
7007 Remove TYPE_METHODS.
7008 * tree.h (TYPE_METHODS): Delete.
7009 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
7010 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
7011 (dbxout_type_methods): Scan TYPE_FIELDS.
7012 (dbxout_type): Don't check TYPE_METHODS here.
7013 * function.c (use_register_for_decl): Always ignore register for
7014 class types when not optimizing.
7015 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
7016 * tree.c (free_lang_data_in_type): Stitch out member functions and
7017 templates from TYPE_FIELDS.
7018 (build_distinct_type_copy, verify_type_variant,
7019 verify_type): Member fns are on TYPE_FIELDS.
7020 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
7021 * tree-pretty-print.c (dump_generic_node): Likewise.
7023 2017-07-20 Jakub Jelinek <jakub@redhat.com>
7026 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
7027 V2TImode and V4TImode.
7028 (ix86_expand_vector_extract): Likewise.
7029 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
7030 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
7031 (ssescalarmode): Handle V4TImode and V2TImode.
7032 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
7033 (*vec_extractv2ti, *vec_extractv4ti): New insns.
7034 (VEXTRACTI128_MODE): New mode iterator.
7035 (splitter for *vec_extractv?ti first element): New.
7036 (VEC_INIT_MODE): New mode iterator.
7037 (vec_init<mode>): Consolidate 3 expanders into one using
7038 VEC_INIT_MODE mode iterator.
7040 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7042 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
7043 non_spilled_static_chain_regno_p.
7045 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7047 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
7049 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
7051 * bb-reorder.c (connect_traces): Allow copying of blocks within
7054 2017-07-20 Richard Biener <rguenther@suse.de>
7056 * gimple.h (gimple_phi_result): Add gphi * overload.
7057 (gimple_phi_result_ptr): Likewise.
7058 (gimple_phi_arg): Likewise. Adjust index assert to only
7059 allow actual argument accesses rather than all slots available
7061 (gimple_phi_arg_def): Add gphi * overload.
7062 * tree-phinodes.c (make_phi_node): Initialize only actual
7064 (resize_phi_node): Clear memory not covered by old node,
7065 do not initialize excess argument slots.
7066 (reserve_phi_args_for_new_edge): Initialize new argument slot
7069 2017-07-20 Bin Cheng <bin.cheng@arm.com>
7071 PR tree-optimization/81388
7073 2016-07-21 Bin Cheng <bin.cheng@arm.com>
7075 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
7076 by removing computation of may_be_zero.
7078 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7079 Tom de Vries <tom@codesourcery.com>
7082 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
7083 when gimple level profile disagrees with what RTL expander did.
7085 2017-07-20 Richard Biener <rguenther@suse.de>
7087 PR tree-optimization/61171
7088 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
7089 (vect_analyze_stmt): Add slp instance parameter.
7090 (vectorizable_reduction): Likewise.
7091 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
7092 (vect_is_simple_reduction): Deal with chains not detected
7093 as SLP reduction chain, specifically not properly associated
7094 chains containing a mix of plus/minus.
7095 (get_reduction_op): Remove.
7096 (get_initial_defs_for_reduction): Simplify, pass in whether
7097 this is a reduction chain, pass in the SLP node for the PHIs.
7098 (vect_create_epilog_for_reduction): Get the SLP instance as
7100 (vectorizable_reduction): Get the SLP instance as arg.
7101 During analysis remember the SLP node with the PHIs in the
7102 instance. Simplify getting at the vectorized reduction PHIs.
7103 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
7104 through SLP instance.
7105 (vect_slp_analyze_operations): Likewise.
7106 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
7107 (vect_transform_stmt): Likewise.
7109 2017-07-20 Tom de Vries <tom@codesourcery.com>
7111 PR tree-optimization/81489
7112 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
7113 read of phi arg location to before loop that modifies phi.
7115 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7117 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
7120 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7123 * except.c (execute): Fix ordering issue.
7125 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7127 PR rtl-optimization/81423
7128 * combine.c (make_compound_operation_int): Don't try to optimize
7129 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
7131 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7133 PR rtl-optimization/81423
7134 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
7135 with a constant that is -1 in the truncated to mode.
7137 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7139 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
7140 (determine_unlikely_bbs): ... here.
7141 * predict.h (propagate_unlikely_bbs_forward): Declare.
7142 * cfgexpand.c (pass_expand::execute): Use it.
7143 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
7145 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
7146 propagate_unlikely_bbs_forward.
7148 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7151 * except.c (maybe_add_nop_after_section_switch): New function.
7154 2017-07-19 Tom de Vries <tom@codesourcery.com>
7156 * gimple.h (gimple_phi_set_arg): Make assert more strict.
7158 2017-07-19 Tom de Vries <tom@codesourcery.com>
7160 * gimple.h (gimple_phi_arg): Make assert more strict.
7162 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
7164 * config.gcc (powerpc*-*-*): Add mmintrin.h.
7165 * config/rs6000/mmintrin.h: New file.
7166 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
7168 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7170 PR tree-optimization/81346
7171 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
7173 2017-07-19 Tom de Vries <tom@codesourcery.com>
7175 * config/nvptx/nvptx.md (VECIM): Add V2DI.
7177 2017-07-19 Tom de Vries <tom@codesourcery.com>
7179 * config/nvptx/nvptx-modes.def: Add V2DImode.
7180 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
7181 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
7182 (nvptx_output_mov_insn): Handle lack of mov.b128.
7183 (nvptx_print_operand): Handle 'H' and 'L' codes.
7184 (nvptx_vector_mode_supported): Allow V2DImode.
7185 (nvptx_preferred_simd_mode): New function.
7186 (nvptx_data_alignment): New function.
7187 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
7188 nvptx_preferred_simd_mode.
7189 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
7191 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
7193 2017-07-19 Tom de Vries <tom@codesourcery.com>
7195 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
7196 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
7197 (nvptx_vector_mode_supported): New function. Allow V2SImode.
7198 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
7199 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
7200 (mov<VECIM>_insn): New define_insn.
7201 (define_expand "mov<VECIM>): New define_expand.
7203 2017-07-19 Tom de Vries <tom@codesourcery.com>
7205 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
7207 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7209 PR tree-optimization/81346
7210 * fold-const.h (fold_div_compare, range_check_type): Declare.
7211 * fold-const.c (range_check_type): New function.
7212 (build_range_check): Use range_check_type.
7213 (fold_div_compare): No longer static, rewritten into
7214 a match.pd helper function.
7215 (fold_comparison): Don't call fold_div_compare here.
7216 * match.pd (X / C1 op C2): New optimization using fold_div_compare
7219 2017-07-19 Nathan Sidwell <nathan@acm.org>
7221 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
7222 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
7223 * tree.c (find_decls_types_r, verify_type): Use
7224 TYPE_{MIN,MAX}_VALUE_RAW.
7225 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
7226 (hash_tree): Likewise.
7227 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7229 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7232 2017-07-18 Tom de Vries <tom@codesourcery.com>
7235 * omp-expand.c (expand_omp_for_static_chunk): Handle
7236 equal-argument loop exit phi.
7238 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
7241 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
7242 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
7243 operand 2 predicate.
7244 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
7245 operand 2 predicate.
7246 (ror,rol -> rorx splitters): Use const_int_operand as
7247 operand 2 predicate.
7249 2017-06-18 Richard Biener <rguenther@suse.de>
7251 PR tree-optimization/81410
7252 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
7253 the gap in the ! slp_perm SLP case after each group.
7255 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7258 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
7261 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7264 * predict.c (set_even_probabilities): Cleanup; do not affect
7265 probabilities that are already known.
7266 (combine_predictions_for_bb): Call even when count is set.
7268 2017-07-18 Nathan Sidwell <nathan@acm.org>
7270 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
7273 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7276 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
7277 optimization for loop niter analysis.
7279 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
7282 * config/avr/avr.c (avr_optimize_casesi): Don't use
7283 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
7285 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7287 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
7288 body_cost_vec from _vect_peel_extended_info.
7289 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
7290 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
7293 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7295 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
7297 2017-07-18 Richard Biener <rguenther@suse.de>
7299 PR tree-optimization/80620
7300 PR tree-optimization/81403
7301 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
7302 info when re-using a VN table entry.
7304 2017-07-18 Richard Biener <rguenther@suse.de>
7306 PR tree-optimization/81418
7307 * tree-vect-loop.c (vectorizable_reduction): Properly compute
7308 vectype_in. Verify that with lane-reducing reduction operations
7309 we have a single def-use cycle.
7311 2017-07-17 Carl Love <cel@us.ibm.com>
7313 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
7315 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7316 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7317 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7318 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7319 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7320 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7321 VMULOSW): New enum "unspec" values.
7322 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
7323 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
7324 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7325 altivec_vmulosw): New patterns.
7326 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7327 VMULOSW): Add definitions.
7328 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
7330 * config/alpha/alpha.c: Include predict.h.
7332 2017-07-17 Yury Gribov <tetra2005@gmail.com>
7334 * tree-vrp.c (compare_assert_loc): Fix comparison function
7335 to return predictable results.
7337 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7339 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
7342 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
7343 * doc/invoke.texi (mexpand-adddi): Update text.
7345 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7347 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
7348 that also clobbers the CC register. The old expand code is moved
7350 (*arc_clzsi2): ... here.
7351 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
7352 the CC register. The old expand code is moved to ...
7353 (arc_ctzsi2): ... here.
7355 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7357 * config/arc/arc.opt (mindexed-loads): Use initial value
7358 TARGET_INDEXED_LOADS_DEFAULT.
7359 (mauto-modify-reg): Use initial value
7360 TARGET_AUTO_MODIFY_REG_DEFAULT.
7361 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7362 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7363 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7364 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7366 2017-07-17 Martin Liska <mliska@suse.cz>
7369 * opts.c (finish_options): Do not allow -fgnu-tm
7370 w/ -fsanitize={kernel-,}address. Say sorry.
7372 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7375 * tree-loop-distribution.c (classify_partition): Only assert on
7376 numer of iterations.
7377 (merge_dep_scc_partitions): Delete prameter. Update function call.
7378 (distribute_loop): Remove code handling loop with unknown niters.
7379 (pass_loop_distribution::execute): Skip loop with unknown niters.
7381 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7384 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
7385 function sort_partitions_by_post_order.
7387 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7389 PR tree-optimization/81374
7390 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
7391 the max index of basic blocks, rather than number of basic blocks.
7393 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7395 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
7397 (arc_legitimate_pic_operand_p): Likewise.
7398 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
7400 (arc_needs_pcl_p): Likewise.
7401 (arc_legitimate_pc_offset_p): Likewise.
7402 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
7403 function is also used in constrains.md.
7404 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
7405 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
7406 PLUS. Only return true/false in known cases, otherwise assert.
7407 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
7408 is already called in arc_legitimate_constant_p.
7409 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
7411 (LEGITIMATE_PIC_OPERAND_P): Use
7412 arc_raw_symbolic_reference_mentioned_p function.
7413 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
7418 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7419 Andrew Burgess <andrew.burgess@embecosm.com>
7421 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
7422 (arc_return_address_register): New function.
7423 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
7424 (arc_handle_fndecl_attribute): Add naked attribute.
7425 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7426 (TARGET_WARN_FUNC_RETURN): Likewise.
7427 (arc_allocate_stack_slots_for_args): New function.
7428 (arc_warn_func_return): Likewise.
7429 (machine_function): Change type fn_type.
7430 (arc_compute_function_type): Consider new naked function type,
7431 change function return type.
7432 (arc_must_save_register): Adapt to handle new
7433 arc_compute_function_type's return type.
7434 (arc_expand_prologue): Likewise.
7435 (arc_expand_epilogue): Likewise.
7436 (arc_return_address_regs): Delete.
7437 (arc_return_address_register): New function.
7438 (arc_epilogue_uses): Use above function.
7439 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
7440 (arc_function_type): Change encoding, add naked type.
7441 (ARC_INTERRUPT_P): Change to handle the new encoding.
7442 (ARC_FAST_INTERRUPT_P): Likewise.
7443 (ARC_NORMAL_P): Define.
7444 (ARC_NAKED_P): Likewise.
7445 (arc_compute_function_type): Delete prototype.
7446 * config/arc/arc.md (in_ret_delay_slot): Use
7447 arc_return_address_register function.
7448 (simple_return): Likewise.
7449 (p_return_i): Likewise.
7451 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7453 PR tree-optimization/81428
7454 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
7455 can't be built for those types.
7457 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7459 Remove stuff dead since r239246.
7461 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
7462 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
7463 (avr_inform_devices): Remove dead stuff.
7465 2017-07-17 Tamar Christina <tamar.christina@arm.com>
7467 * config/arm/arm_neon.h: Fix softp typo.
7469 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7471 PR tree-optimization/81365
7472 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
7473 aggregate moves onto bb predecessor edges, make sure there are no
7474 loads that could alias the lhs in between the start of bb and the
7477 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7480 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
7481 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
7482 [LSHIFTRT, outer_code = TRUNCATE]: Same.
7484 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7486 PR tree-optimization/81396
7487 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
7488 (init_symbolic_number): Initialize it to 1.
7489 (perform_symbolic_merge): Add n_ops from both operands into the new
7491 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
7492 without base_addr as useless if they need more than one operation.
7493 (bswap_replace): Handle !bswap case for NULL base_addr.
7495 2017-07-17 Tom de Vries <tom@codesourcery.com>
7498 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
7501 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
7503 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
7504 conditional builtin define __FIX_LEON3FT_B2BST.
7506 2017-07-17 Daniel Cederman <cederman@gaisler.com>
7508 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
7509 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
7512 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7514 PR rtl-optimization/81424
7515 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
7516 to remove potential trapping from operands if -fnon-call-exceptions.
7518 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7520 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
7521 profile_proability for scalling.
7522 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
7524 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7526 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
7528 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7530 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
7531 fixpoint arithmetics.
7533 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7535 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
7536 fixpoint arithmetics.
7538 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7540 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
7541 fixpoint arithmetics.
7543 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7545 * profile-count.h (profile_probability::from_reg_br_prob_note,
7546 profile_probability::to_reg_br_prob_note): New functions.
7547 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
7548 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
7549 * predict.c (probability_reliable_p): Update.
7550 (edge_probability_reliable_p): Update.
7551 (br_prob_note_reliable_p): Update.
7552 (invert_br_probabilities): Update.
7553 (add_reg_br_prob_note): New function.
7554 (combine_predictions_for_insn): Update.
7555 * asan.c (asan_clear_shadow): Update.
7556 * cfgbuild.c (compute_outgoing_frequencies): Update.
7557 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
7558 (update_br_prob_note): Update.
7559 (rtl_verify_edges): Update.
7560 (purge_dead_edges): Update.
7561 (fixup_reorder_chain): Update.
7562 * emit-rtl.c (try_split): Update.
7563 * ifcvt.c (cond_exec_process_insns): Update.
7564 (cond_exec_process_if_block): Update.
7565 (dead_or_predicable): Update.
7566 * internal-fn.c (expand_addsub_overflow): Update.
7567 (expand_neg_overflow): Update.
7568 (expand_mul_overflow): Update.
7569 * loop-doloop.c (doloop_modify): Update.
7570 * loop-unroll.c (compare_and_jump_seq): Update.
7571 * optabs.c (emit_cmp_and_jump_insn_1): Update.
7572 * predict.h: Update.
7573 * reorg.c (mostly_true_jump): Update.
7575 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
7576 * config/alpha/alpha.c (emit_unlikely_jump): Update.
7577 * config/arc/arc.c: (emit_unlikely_jump): Update.
7578 * config/arm/arm.c: (emit_unlikely_jump): Update.
7579 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
7580 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
7581 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
7582 (ix86_print_operand): Update.
7583 (ix86_split_fp_branch): Update.
7584 (predict_jump): Update.
7585 * config/ia64/ia64.c (ia64_print_operand): Update.
7586 * config/mmix/mmix.c (mmix_print_operand): Update.
7587 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
7588 (rs6000_expand_split_stack_prologue): Update.
7589 * config/rs6000/rs6000.c: Update.
7590 * config/s390/s390.c (s390_expand_vec_strlen): Update.
7591 (s390_expand_vec_movstr): Update.
7592 (s390_expand_cs_tdsi): Update.
7593 (s390_expand_split_stack_prologue): Update.
7594 * config/sh/sh.c (sh_print_operand): Update.
7595 (expand_cbranchsi4): Update.
7596 (expand_cbranchdi4): Update.
7597 * config/sparc/sparc.c (output_v9branch): Update.
7598 * config/spu/spu.c (get_branch_target): Update.
7599 (ea_load_store_inline): Update.
7600 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
7601 * config/tilepro/tilepro.c: Update.
7603 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7605 * gimplify.c (mostly_copy_tree_r): Revert latest change.
7606 (gimplify_save_expr): Likewise.
7608 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7610 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
7612 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7614 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
7616 * timevar.def (TV_IPA_FNSUMMARY): Define.
7618 2017-07-16 Daniel Cederman <cederman@gaisler.com>
7620 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
7621 to back store errata sensitive sequence from being generated.
7622 (sqrtdf2_fix): Likewise.
7624 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7626 * tree-ssa-threadupdate.c (compute_path_counts,
7627 update_joiner_offpath_counts): Use profile_probability.
7629 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
7632 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7634 * config/arm/arm-c.c (arm_cpu_builtins): Define
7635 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7637 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
7639 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7640 array entries to represent __ieee128 versions of the
7641 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
7642 scalar_extract_sig, and scalar_insert_exp built-in functions.
7643 (altivec_resolve_overloaded_builtin): Add special case handling
7644 for the __builtin_scalar_insert_exp function, as represented by
7645 the P9V_BUILTIN_VEC_VSIEDP constant.
7646 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
7647 exponent support for __ieee128 argument.
7648 (VSESQP): Add scalar extract signature support for __ieee128
7650 (VSTDCNQP): Add scalar test negative support for __ieee128
7652 (VSIEQP): Add scalar insert exponent support for __int128 argument
7653 with __ieee128 result.
7654 (VSIEQPF): Add scalar insert exponent support for __ieee128
7655 argument with __ieee128 result.
7656 (VSTDCQP): Add scalar test data class support for __ieee128
7658 (VSTDCNQP): Add overload support for scalar test negative with
7660 (VSTDCQP): Add overload support for scalar test data class
7662 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
7664 (UNSPEC_VSX_SIEXPQP): New constant.
7665 (xsxexpqp): New insn for VSX scalar extract exponent quad
7667 (xsxsigqp): New insn for VSX scalar extract significand quad
7669 (xsiexpqpf): New insn for VSX scalar insert exponent quad
7670 precision with floating point argument.
7671 (xststdcqp): New expand for VSX scalar test data class quad
7673 (xststdcnegqp): New expand for VSX scalar test negative quad
7675 (xststdcqp): New insn to match expansions for VSX scalar test data
7676 class quad precision and VSX scalar test negative quad precision.
7677 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7678 special case operand checking to enforce that second operand of
7679 VSX scalar test data class with quad precision argument is a 7-bit
7681 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
7682 prototypes and descriptions of __ieee128 versions of
7683 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
7684 scalar_test_data_class, and scalar_test_neg built-in functions.
7686 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7688 PR tree-optimization/81162
7689 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
7690 replace a negate with an add.
7692 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
7694 * doc/invoke.texi (arm/-mcpu): Document +crypto.
7696 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7698 * config/arm/arm-c.c (arm_cpu_builtins): Define
7699 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7701 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7703 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
7704 (armv8-r): Set ARM Cortex-R52 as default CPU.
7705 * config/arm/arm-tables.opt: Regenerate.
7706 * config/arm/arm-tune.md: Regenerate.
7707 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7709 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
7710 extension for -mcpu=cortex-r52.
7712 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7714 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
7715 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
7716 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
7717 (fp-armv8): Define it as FP_ARMv8 only.
7718 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
7719 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
7721 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
7722 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
7723 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
7724 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
7725 than TARGET_FPU_ARMV8.
7726 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
7727 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
7728 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
7730 * config/arm/neon.md (neon_vrint): Likewise.
7731 (neon_vcvt): Likewise.
7732 (neon_<fmaxmin_op><mode>): Likewise.
7733 (<fmaxmin><mode>3): Likewise.
7734 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
7735 * config/arm/predicates.md (arm_cond_move_operator): Check against
7736 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
7738 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
7740 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
7743 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7745 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
7746 loop in comment with memset.
7748 2017-07-14 Martin Liska <mliska@suse.cz>
7750 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
7751 * dwarf2out.c (is_java): Remove the function.
7752 (output_pubname): Remove usage of the function.
7753 (lower_bound_default): Remove usage of DW_LANG_Java.
7754 (gen_compile_unit_die): Likewise.
7755 * gcc.c: Remove compiler defaults for .java and .zip files.
7756 * gimple-expr.c (remove_suffix): Change as there's no longer
7757 extension than 4-letter one.
7758 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
7759 (gimplify_save_expr): Likewise.
7760 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
7761 as it's possible even for other languages than Java.
7762 * langhooks.h (struct lang_hooks): Remove Java from a comment.
7763 * lto-opts.c (lto_write_options): Remove reference to Java.
7764 * opts.c (strip_off_ending): Update file extension handling.
7765 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
7766 * tree-eh.c (lower_resx): Likewise.
7767 * tree.c (free_lang_data_in_type): Remove dead code.
7768 (find_decls_types_r): Likewise.
7769 (build_common_builtin_nodes): Remove Java from a comment.
7770 (verify_type): Remove dead code.
7771 * varasm.c (assemble_external): Remove Java from a comment.
7773 2017-07-14 Martin Liska <mliska@suse.cz>
7775 * opts.c (finish_options): Add quotes.
7776 (common_handle_option): Likewise.
7778 2017-07-14 Martin Liska <mliska@suse.cz>
7780 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
7781 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
7783 * dwarf2out.c (lower_bound_default): Do not handle
7785 (gen_compile_unit_die): Likewise.
7786 * gcc.c: Remove default extension binding for GNU Pascal.
7787 * stmt.c: Remove Pascal language from a comment.
7788 * xcoffout.c: Likewise.
7790 2017-07-13 David Malcolm <dmalcolm@redhat.com>
7793 * diagnostic-show-locus.c (fixit_cmp): New function.
7794 (layout::layout): Sort m_fixit_hints.
7795 (column_range::column_range): Assert that the values are valid.
7796 (struct char_span): New struct.
7797 (correction::overwrite): New method.
7798 (struct source_line): New struct.
7799 (line_corrections::add_hint): Add assertions. Reimplement memcpy
7800 calls in terms of classes source_line and char_span, and
7801 correction::overwrite.
7802 (selftest::test_overlapped_fixit_printing_2): New function.
7803 (selftest::diagnostic_show_locus_c_tests): Call it.
7805 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7807 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
7808 early if there is no lhs.
7810 2017-07-13 Martin Liska <mliska@suse.cz>
7812 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
7813 (gen_reference_type_die): Likewise.
7814 * stor-layout.c: Remove Pascal-related comment.
7816 2017-07-13 Martin Liska <mliska@suse.cz>
7818 * opts.c (finish_options): Add quotes to error messages.
7819 (parse_sanitizer_options): Likewise.
7821 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
7823 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
7825 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
7827 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
7829 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
7831 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
7833 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
7835 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7838 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
7839 provides the hardware capability bits, define the macro
7840 __BUILTIN_CPU_SUPPORTS__.
7841 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
7842 if GLIBC does not provide the hardware capability bits. Add a
7843 gcc_unreachable call if the built-in cpu function is neither
7844 __builtin_cpu_is nor __builtin_cpu_supports.
7845 (rs6000_get_function_versions_dispatcher): Change the warning
7846 that an old GLIBC is used which does not export the capability
7847 bits to be an error.
7848 * doc/extend.texi (target_clones attribute): Document the
7849 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
7850 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
7851 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
7852 the macros defined by GCC if the newer GLIBC is available.
7854 2017-07-12 Jeff Law <law@redhat.com>
7856 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
7857 remaining includes slightly.
7858 * config/riscv/riscv-builtins.c: Include profile-count.h.
7860 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
7863 * config/avr/avr.c (avr_set_current_function): In diagnostic
7864 messages: Quote keywords and (parts of) identifiers.
7865 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
7868 2017-07-12 Carl Love <cel@us.ibm.com>
7870 * config/rs6000/rs6000-c.c: Add support for built-in functions
7871 vector bool char vec_revb (vector bool char);
7872 vector bool short vec_revb (vector short char);
7873 vector bool int vec_revb (vector bool int);
7874 vector bool long long vec_revb (vector bool long long);
7875 * doc/extend.texi: Update the built-in documentation file for the
7876 new built-in functions.
7878 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7880 * config/s390/s390.md: Remove movcc splitter.
7882 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7884 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
7885 load/store on condition.
7887 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
7890 * config/avr/avr.c (avr_encode_section_info)
7891 [progmem && !TREE_READONLY]: Error if progmem object needs
7894 2017-07-11 Michael Collison <michael.collison@arm.com>
7896 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
7899 2017-07-11 Carl Love <cel@us.ibm.com>
7901 * config/rs6000/rs6000-c.c: Add support for builtins
7902 vector unsigned int vec_parity_lsbb (vector signed int);
7903 vector unsigned int vec_parity_lsbb (vector unsigned int);
7904 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
7905 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
7906 vector unsigned long long vec_parity_lsbb (vector signed long long);
7907 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
7908 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
7909 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
7910 * doc/extend.texi: Update the built-in documentation file for the
7911 new built-in functions.
7913 2017-07-11 David Malcolm <dmalcolm@redhat.com>
7915 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
7916 (layout::m_primary_loc): New field.
7917 (layout::layout): Initialize new field. Move location filtering
7918 logic from here to...
7919 (layout::maybe_add_location_range): ...this new method. Add
7920 support for filtering to just the lines already specified by other
7922 (layout::will_show_line_p): New method.
7923 (gcc_rich_location::add_location_if_nearby): New method.
7924 (selftest::test_add_location_if_nearby): New test function.
7925 (selftest::diagnostic_show_locus_c_tests): Call it.
7926 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
7929 2017-07-11 Tom de Vries <tom@codesourcery.com>
7931 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
7932 (bb_first_real_insn): New function.
7933 (nvptx_single): Add extra initialization of broadcasted condition
7936 2017-07-11 Nathan Sidwell <nathan@acm.org>
7938 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
7940 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
7942 * doc/extend.texi (AVR Function Attributes): Remove weblink to
7943 Binutils doc as TEXI will mess them up.
7944 * doc/invoke.texi (AVR Options): Same here.
7946 2017-07-11 Daniel Cederman <cederman@gaisler.com>
7948 * config/sparc/sparc.opt (mfix-ut700): New option.
7949 (mfix-gr712rc): Likewise.
7950 (sparc_fix_b2bst): New variable.
7951 * doc/invoke.texi (SPARC options): Document them.
7952 (ARM options): Fix warnings.
7953 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
7954 instructions to prevent sequences that can trigger the store-store
7955 errata for certain LEON3FT processors.
7956 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
7957 (sparc_option_override): Set sparc_fix_b2bst appropriately.
7958 * config/sparc/sparc.md (fix_b2bst): New attribute.
7959 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
7961 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
7964 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
7966 (*rsqrtsf2_sse): Ditto.
7968 (div<mode>3): Macroize insn from divdf3 and divsf3
7969 using MODEF mode iterator.
7971 2017-07-10 Martin Sebor <msebor@redhat.com>
7973 PR tree-optimization/80397
7974 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
7975 instead of testing for equality to INTEGER_TYPE.
7977 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
7979 * config.gcc: Remove uclibc from arc target spec.
7981 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
7983 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
7985 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7988 * lto-wrapper.c (remove_option): New function.
7989 (merge_and_complain): Merge PIC/PIE options more realistically.
7991 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
7993 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
7997 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
7998 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
7999 * config.in: Regenerate.
8000 * configure: Regenerate.
8001 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
8002 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
8003 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
8004 (TARGET_GASISR_PROLOGUES): ...target mask.
8005 * common/config/avr/avr-common.c
8006 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
8007 Set -mgas-isr-prologues.
8008 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
8009 INSERT_PASS_BEFORE for it.
8010 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
8011 * config/avr/avr.c (avr_option_override)
8012 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
8013 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
8014 (avr_attribute_table) <no_gccisr>: Add new function attribute.
8015 (avr_set_current_function) <is_no_gccisr>: Init machine field.
8016 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
8018 (make_avr_pass_pre_proep): New function.
8019 (emit_push_sfr) <treg>: Add argument to function and use it
8021 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
8022 and set machine->gasisr.yes.
8023 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
8024 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
8025 __gcc_isr.n_pushed to .L__stack_usage.
8026 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
8027 (avr_asm_final_postscan_insn): ...this new static function.
8028 * config/avr/avr.h (machine_function)
8029 <is_no_gccisr, use_L__stack_usage>: New fields.
8030 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
8031 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
8032 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
8033 (gasisr, *gasisr): New expander and insn.
8034 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
8035 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
8036 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
8038 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
8040 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
8043 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8045 Move jump-tables out of .text again.
8048 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
8049 (ASM_OUTPUT_ADDR_VEC): New function.
8050 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
8051 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
8052 INSN_ADDRESSes as asm comment.
8053 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
8054 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
8055 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
8056 * config/avr/avr.md (*tablejump): Adjust comment.
8057 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
8058 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
8060 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
8061 (avr_output_addr_vec): New proto.
8062 (avr_log_t) <insn_addresses>: New field.
8064 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
8067 * config/i386/i386.c (ix86_function_arg_advance): Set
8068 outgoing_args_on_stack to true if there are outgoing arguments
8070 (ix86_function_arg): Likewise.
8071 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
8072 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
8073 * config/i386/i386.h (machine_function): Add
8074 outgoing_args_on_stack.
8076 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
8078 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8080 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
8082 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
8084 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
8085 (REAL_H): Remove $(MACHMODE_H).
8086 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
8088 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
8089 $(MACHMODE_H) and double-int.h.
8090 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
8092 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
8095 2017-07-07 Andrew Pinski <apinski@cavium.com>
8097 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
8098 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
8100 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8102 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
8103 Add warning if GCC was not configured to link against a GLIBC that
8104 exports the hardware capability bits.
8105 (make_resolver_func): Make resolver function private and not a
8106 COMDAT function. Create the name with clone_function_name instead
8107 of make_unique_name.
8110 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
8111 correct operand in doing the split.
8113 2017-07-07 Carl Love <cel@us.ibm.com>
8115 * config/rs6000/rs6000-c: Add support for built-in function
8116 vector unsigned short vec_pack_to_short_fp32 (vector float,
8118 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
8119 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
8120 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
8121 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
8122 (convert_4f32_8i16): Add define_expand.
8123 * doc/extend.texi: Update the built-in documentation file for the
8124 new built-in function.
8126 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8128 * config/sparc/m8.md: New file.
8129 * config/sparc/sparc.md: Include m8.md.
8131 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8133 * config/sparc/sparc.opt: New option -mvis4b.
8134 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
8135 (sparc_option_override): Handle VIS4B.
8136 (enum sparc_builtins): Define
8137 SPARC_BUILTIN_DICTUNPACK{8,16,32},
8138 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
8139 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
8140 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
8141 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
8142 (check_constant_argument): New function.
8143 (sparc_vis_init_builtins): Define builtins
8144 __builtin_vis_dictunpack{8,16,32},
8145 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
8146 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
8147 __builtin_vis_fpcmpde{8,16,32}shl and
8148 __builtin_vis_fpcmpur{8,16,32}shl.
8149 (sparc_expand_builtin): Check that the constant operands to
8150 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
8151 constant and in range.
8152 * config/sparc/sparc-c.c (sparc_target_macros): Handle
8154 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
8155 (SPARC_IMM5_P): Likewise.
8156 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
8157 (enabled): Handle vis4b.
8158 (UNSPEC_DICTUNPACK): New unspec.
8159 (UNSPEC_FPCMPSHL): Likewise.
8160 (UNSPEC_FPUCMPSHL): Likewise.
8161 (UNSPEC_FPCMPDESHL): Likewise.
8162 (UNSPEC_FPCMPURSHL): Likewise.
8163 (cpu_feature): New CPU feature `vis4b'.
8164 (dictunpack{8,16,32}): New insns.
8165 (FPCSMODE): New mode iterator.
8166 (fpcscond): New code iterator.
8167 (fpcsucond): Likewise.
8168 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
8169 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
8170 (fpcmpde{8,16,32}{si,di}shl): Likewise.
8171 (fpcmpur{8,16,32}{si,di}shl): Likewise.
8172 * config/sparc/constraints.md: Define constraints `q' for unsigned
8173 2-bit integer constants and `t' for unsigned 5-bit integer
8175 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
8177 (imm5_operand_dictunpack16): Likewise.
8178 (imm5_operand_dictunpack32): Likewise.
8179 (imm2_operand): Likewise.
8180 * doc/invoke.texi (SPARC Options): Document -mvis4b.
8181 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
8182 ditunpack* and fpcmp*shl builtins.
8184 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8186 * config.gcc: Handle m8 in --with-{cpu,tune} options.
8187 * config.in: Add HAVE_AS_SPARC6 define.
8188 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
8190 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
8192 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
8193 (CPP_CPU_SPEC): Handle m8.
8194 (ASM_CPU_SPEC): Likewise.
8195 * config/sparc/sparc-opts.h (enum processor_type): Add
8197 * config/sparc/sparc.c (m8_costs): New struct.
8198 (sparc_option_override): Handle TARGET_CPU_m8.
8199 (sparc32_initialize_trampoline): Likewise.
8200 (sparc64_initialize_trampoline): Likewise.
8201 (sparc_issue_rate): Likewise.
8202 (sparc_register_move_cost): Likewise.
8203 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
8204 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
8205 (ASM_CPU64_DEFAULT_SPEC): Likewise.
8206 (CPP_CPU_SPEC): Handle M8.
8207 (ASM_CPU_SPEC): Likewise.
8208 (AS_M8_FLAG): Define.
8209 * config/sparc/sparc.md: Add m8 to the cpu attribute.
8210 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
8211 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
8213 * configure: Regenerate.
8214 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
8217 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8219 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
8221 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
8222 ("*movdi_insn_sp32"): Do not set v3pipe.
8223 ("*movsi_insn"): Likewise.
8224 ("*movdi_insn_sp64"): Likewise.
8225 ("*movsf_insn"): Likewise.
8226 ("*movdf_insn_sp32"): Likewise.
8227 ("*movdf_insn_sp64"): Likewise.
8228 ("*zero_extendsidi2_insn_sp64"): Likewise.
8229 ("*sign_extendsidi2_insn"): Likewise.
8230 ("*mov<VM32:mode>_insn"): Likewise.
8231 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8232 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8233 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8234 ("<vlop:code><VL:mode>3"): Likewise.
8235 ("*not_<vlop:code><VL:mode>3"): Likewise.
8236 ("*nand<VL:mode>_vis"): Likewise.
8237 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
8238 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
8239 ("one_cmpl<VL:mode>2"): Likewise.
8240 ("faligndata<VM64:mode>_vis"): Likewise.
8241 ("alignaddrsi_vis"): Likewise.
8242 ("alignaddrdi_vis"): Likweise.
8243 ("alignaddrlsi_vis"): Likewise.
8244 ("alignaddrldi_vis"): Likewise.
8245 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8246 ("bmaskdi_vis"): Likewise.
8247 ("bmasksi_vis"): Likewise.
8248 ("bshuffle<VM64:mode>_vis"): Likewise.
8249 ("cmask8<P:mode>_vis"): Likewise.
8250 ("cmask16<P:mode>_vis"): Likewise.
8251 ("cmask32<P:mode>_vis"): Likewise.
8252 ("pdistn<P:mode>_vis"): Likewise.
8253 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8255 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8257 * config/sparc/sparc.md ("subtype"): New insn attribute.
8258 ("*wrgsr_sp64"): Set insn subtype.
8259 ("*rdgsr_sp64"): Likewise.
8260 ("alignaddrsi_vis"): Likewise.
8261 ("alignaddrdi_vis"): Likewise.
8262 ("alignaddrlsi_vis"): Likewise.
8263 ("alignaddrldi_vis"): Likewise.
8264 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8265 ("fexpand_vis"): Likewise.
8266 ("fpmerge_vis"): Likewise.
8267 ("faligndata<VM64:mode>_vis"): Likewise.
8268 ("bshuffle<VM64:mode>_vis"): Likewise.
8269 ("cmask8<P:mode>_vis"): Likewise.
8270 ("cmask16<P:mode>_vis"): Likewise.
8271 ("cmask32<P:mode>_vis"): Likewise.
8272 ("fchksm16_vis"): Likewise.
8273 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
8274 ("fmean16_vis"): Likewise.
8275 ("fp<plusminus_insn>64_vis"): Likewise.
8276 ("<plusminus_insn>v8qi3"): Likewise.
8277 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8278 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
8279 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
8280 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
8281 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
8282 ("*movqi_insn"): Likewise.
8283 ("*movhi_insn"): Likewise.
8284 ("*movsi_insn"): Likewise.
8285 ("movsi_pic_gotdata_op"): Likewise.
8286 ("*movdi_insn_sp32"): Likewise.
8287 ("*movdi_insn_sp64"): Likewise.
8288 ("movdi_pic_gotdata_op"): Likewise.
8289 ("*movsf_insn"): Likewise.
8290 ("*movdf_insn_sp32"): Likewise.
8291 ("*movdf_insn_sp64"): Likewise.
8292 ("*zero_extendhisi2_insn"): Likewise.
8293 ("*zero_extendqihi2_insn"): Likewise.
8294 ("*zero_extendqisi2_insn"): Likewise.
8295 ("*zero_extendqidi2_insn"): Likewise.
8296 ("*zero_extendhidi2_insn"): Likewise.
8297 ("*zero_extendsidi2_insn_sp64"): Likewise.
8298 ("ldfsr"): Likewise.
8299 ("prefetch_64"): Likewise.
8300 ("prefetch_32"): Likewise.
8301 ("tie_ld32"): Likewise.
8302 ("tie_ld64"): Likewise.
8303 ("*tldo_ldub_sp32"): Likewise.
8304 ("*tldo_ldub1_sp32"): Likewise.
8305 ("*tldo_ldub2_sp32"): Likewise.
8306 ("*tldo_ldub_sp64"): Likewise.
8307 ("*tldo_ldub1_sp64"): Likewise.
8308 ("*tldo_ldub2_sp64"): Likewise.
8309 ("*tldo_ldub3_sp64"): Likewise.
8310 ("*tldo_lduh_sp32"): Likewise.
8311 ("*tldo_lduh1_sp32"): Likewise.
8312 ("*tldo_lduh_sp64"): Likewise.
8313 ("*tldo_lduh1_sp64"): Likewise.
8314 ("*tldo_lduh2_sp64"): Likewise.
8315 ("*tldo_lduw_sp32"): Likewise.
8316 ("*tldo_lduw_sp64"): Likewise.
8317 ("*tldo_lduw1_sp64"): Likewise.
8318 ("*tldo_ldx_sp64"): Likewise.
8319 ("*mov<VM32:mode>_insn"): Likewise.
8320 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8321 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8323 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8325 * config/sparc/sparc.md ("type"): New insn type viscmp.
8326 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
8328 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
8329 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
8330 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8331 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
8333 ("n7_vis_logical_11cycle"): Likewise.
8334 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
8335 * config/sparc/niagara2.md ("niag3_vis": Likewise.
8336 * config/sparc/niagara.md ("niag_vis"): Likewise.
8337 * config/sparc/ultra3.md ("us3_fga"): Likewise.
8338 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
8340 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8342 * config/sparc/sparc.md: New instruction type `bmask'.
8343 (bmaskdi_vis): Use the `bmask' type.
8344 (bmasksi_vis): Likewise.
8345 * config/sparc/ultra3.md (us3_array): Likewise.
8346 * config/sparc/niagara7.md (n7_array): Likewise.
8347 * config/sparc/niagara4.md (n4_array): Likewise.
8348 * config/sparc/niagara2.md (niag2_vis): Likewise.
8349 (niag3_vis): Likewise.
8350 * config/sparc/niagara.md (niag_vis): Likewise.
8352 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8354 * ipa-comdats.c: Remove optimize check from gate.
8355 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
8356 for functions not optimized.
8357 (ipa_fn_summary_read): Skip optimize check.
8358 (ipa_fn_summary_write): Likewise.
8359 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
8361 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
8363 (can_inline_edge_p): Check flag_pcc_struct_return for match.
8364 (check_callers): Give up on caller which is not optimized.
8365 (inline_small_functions): Likewise.
8366 (ipa_inline): Do not give up when not optimizing.
8367 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
8368 away unoptimizes cdtors.
8369 (whole_program_function_and_variable_visibility): Do
8370 ipa_discover_readonly_nonaddressable_vars in LTO mode.
8371 * ipa.c (process_references): Do not check optimize.
8372 (symbol_table::remove_unreachable_nodes): Update optimize check.
8373 (set_writeonly_bit): Update optimize check.
8374 (pass_ipa_cdtor_merge::gate): Do not check optimize.
8375 (pass_ipa_single_use::gate): Remove.
8377 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8379 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
8380 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
8381 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
8382 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
8383 permute_load, permute_store, adjust_extract, adjust_splat,
8384 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
8385 replace_swap_with_copy, dump_swap_insn_table,
8386 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
8387 recombine_lvx_pattern, recombine_stvx_pattern,
8388 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
8389 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
8390 to file rs6000-p8swap.c.
8391 * config/rs6000/rs6000-p8swap.c: New file.
8392 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
8393 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
8394 and rs6000*-*-* targets.
8396 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8398 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
8400 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8402 * lto-wrapper.c (merge_and_complain): Do not merge
8403 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
8404 fsigned_zeros, ftrapping_math, fwrapv.
8405 (append_compiler_options): Do not track these options.
8406 (append_linker_options): Likewie
8408 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8410 * cgraphunit.c (cgraph_node::finalize_function): When
8411 !flag_toplevel_reorde set no_reorder flag.
8412 (varpool_node::finalize_decl): Likewise.
8413 (symbol_table::compile): Drop no toplevel reorder path.
8415 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8417 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
8418 edges; zero probability is not better than uninitialized.
8420 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8422 * asan.h (asan_sanitize_allocas_p): Declare.
8423 * asan.c (asan_sanitize_allocas_p): New function.
8424 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
8425 (handle_builtin_alloca): Likewise.
8426 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
8427 if !asan_sanitize_allocas_p.
8428 * params.def (asan-instrument-allocas): Add new option.
8429 * params.h (ASAN_PROTECT_ALLOCAS): Define.
8430 * opts.c (common_handle_option): Disable allocas sanitization for
8433 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8435 * asan.c: Include gimple-fold.h.
8436 (get_last_alloca_addr): New function.
8437 (handle_builtin_stackrestore): Likewise.
8438 (handle_builtin_alloca): Likewise.
8439 (asan_emit_allocas_unpoison): Likewise.
8440 (get_mem_refs_of_builtin_call): Add new parameter, remove const
8441 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
8442 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
8443 (instrument_builtin_call): Pass gimple iterator to
8444 get_mem_refs_of_builtin_call.
8445 (last_alloca_addr): New global.
8446 * asan.h (asan_emit_allocas_unpoison): Declare.
8447 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
8448 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
8449 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
8450 if function calls alloca.
8451 * gimple-fold.c (replace_call_with_value): Remove static keyword.
8452 * gimple-fold.h (replace_call_with_value): Declare.
8453 * internal-fn.c: Include asan.h.
8454 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
8455 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
8457 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8459 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
8460 (C_SELFTEST_FLAGS): New.
8461 (CPP_SELFTEST_FLAGS): New.
8462 (SELFTEST_DEPS): New, from deps of s-selftest.
8463 (C_SELFTEST_DEPS): New, from deps of s-selftest.
8464 (CPP_SELFTEST_DEPS): New.
8465 (selftest): Add dependency on s-selftest-c++.
8466 (s-selftest): Rename to...
8467 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
8468 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
8469 than SELFTEST_FLAGS.
8470 (selftest-gdb): Rename to...
8471 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
8473 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
8474 (selftest-valgrind): Rename to...
8475 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
8477 (selftest-valgrind): Reintroduce as an alias for
8478 selftest-c-valgrind.
8479 (s-selftest-c++): New.
8480 (selftest-c++-gdb): New.
8481 (selftest-c++-valgrind): New.
8483 2017-07-06 Olivier Hainque <hainque@adacore.com>
8485 * gcc.c (process_command): When deciding if undefined variables
8486 should be ignored when processing specs, accept "gcc -v" as well.
8488 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8490 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
8491 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
8493 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
8495 * config/arm/arm-cpus.in (armv8-r): Add new entry.
8496 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
8497 * config/arm/arm-tables.opt: Regenerate.
8498 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
8500 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
8502 2017-07-06 Carl Love <cel@us.ibm.com>
8504 * ChangeLog: Clean up from mid air collision
8506 2017-07-06 Carl Love <cel@us.ibm.com>
8508 * config/rs6000/rs6000-c.c: Add support for built-in functions
8509 vector signed int vec_subc (vector signed int, vector signed int);
8510 vector signed __int128 vec_subc (vector signed __int128,
8511 vector signed __int128);
8512 vector unsigned __int128 vec_subc (vector unsigned __int128,
8513 vector unsigned __int128);
8514 vector signed int vec_sube (vector signed int, vector signed int,
8516 vector unsigned int vec_sube (vector unsigned int,
8517 vector unsigned int,
8518 vector unsigned int);
8519 vector signed __int128 vec_sube (vector signed __int128,
8520 vector signed __int128,
8521 vector signed__int128);
8522 vector unsigned __int128 vec_sube (vector unsigned __int128,
8523 vector unsigned __int128,
8524 vector unsigned __int128);
8525 vector signed int vec_subec (vector signed int, vector signed int,
8527 vector unsigned int vec_subec (vector unsigned int,
8528 vector unsigned int,
8529 vector unsigned int);
8530 vector signed __int128 vec_subec (vector signed __int128,
8531 vector signed __int128,
8532 vector signed__int128);
8533 vector unsigned __int128 vec_subec (vector unsigned __int128,
8534 vector unsigned __int128,
8535 vector unsigned __int128);
8536 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
8537 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
8538 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
8539 BU_ALTIVEC_OVERLOAD_X definitions.
8540 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
8541 * doc/extend.texi: Update the built-in documentation file for the new
8544 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8547 * diagnostic-show-locus.c (layout::layout): Use start and finish
8548 spelling location for the start and finish of each range.
8549 * genmatch.c (linemap_client_expand_location_to_spelling_point):
8550 Add unused aspect param.
8551 * input.c (expand_location_1): Add "aspect" param, and use it
8552 to access the correct part of the location.
8553 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
8555 (expand_location_to_spelling_point): Likewise.
8556 (linemap_client_expand_location_to_spelling_point): Add "aspect"
8557 param, and pass it to expand_location_1.
8559 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
8561 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
8562 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
8563 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
8564 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
8565 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
8566 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
8567 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
8568 _mm_maskz_getmant_ss): New intrinsics.
8569 (__builtin_ia32_getexpss128_mask): Changed to ...
8570 __builtin_ia32_getexpss128_round ... this.
8571 (__builtin_ia32_getexpsd128_mask): Changed to ...
8572 __builtin_ia32_getexpsd128_round ... this.
8573 * config/i386/i386-builtin-types.def
8574 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
8575 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
8576 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
8577 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
8578 __builtin_ia32_getmantss_mask_round): New builtins.
8579 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
8580 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
8581 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
8582 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
8583 * config/i386/sse.md
8584 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
8585 avx512f_sgetexp<mode><mask_scalar_name>
8586 <round_saeonly_scalar_name> ... this.
8587 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
8588 %0, %1, %2<round_saeonly_op3>}): Changed to ...
8589 vgetexp<ssescalarmodesuffix>
8590 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8591 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
8592 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
8593 avx512f_vgetmant<mode><mask_scalar_name>
8594 <round_saeonly_scalar_name> ... this.
8595 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
8596 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
8597 vgetmant<ssescalarmodesuffix>
8598 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
8599 %0<mask_scalar_operand4>, %1, %2
8600 <round_saeonly_scalar_mask_op4>, %3} ... this.
8601 * config/i386/subst.md (mask_scalar_operand4,
8602 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
8603 round_saeonly_scalar_nimm_predicate): New subst attributes.
8605 2017-07-06 Julia Koval <julia.koval@intel.com>
8607 * config/i386/i386.c (ix86_erase_embedded_rounding):
8608 Remove code for old rounding pattern.
8610 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
8612 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
8614 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
8616 * doc/sourcebuild.texi (Test Directives, Variants of
8617 dg-require-support): Add documentation for dg-require-stack-check.
8619 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
8621 * config/i386/subst.md (mask_scalar, round_scalar,
8622 round_saeonly_scalar): New meta-templates.
8623 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
8624 round_scalar_mask_operand3, round_scalar_mask_op3,
8625 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
8626 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
8627 round_saeonly_scalar_constraint,
8628 round_saeonly_scalar_prefix): New subst attribute.
8629 * config/i386/sse.md
8630 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
8631 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
8632 <round_scalar_name> ... this.
8633 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
8634 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
8635 <round_scalar_name> ... this.
8636 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
8637 <sse>_vm<code><mode>3<mask_scalar_name>
8638 <round_saeonly_scalar_name> ... this.
8639 (v<plusminus_mnemonic><ssescalarmodesuffix>
8640 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8641 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8642 v<plusminus_mnemonic><ssescalarmodesuffix>
8643 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8644 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8645 (v<multdiv_mnemonic><ssescalarmodesuffix>
8646 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8647 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8648 v<multdiv_mnemonic><ssescalarmodesuffix>
8649 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8650 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8651 (v<maxmin_float><ssescalarmodesuffix>
8652 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
8653 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
8654 v<maxmin_float><ssescalarmodesuffix>
8655 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8656 %0<mask_scalar_operand3>, %1, %<iptr>2
8657 <round_saeonly_scalar_mask_op3>} ... this.
8659 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
8661 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
8662 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
8664 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
8665 Alan Hayward <alan.hayward@arm.com>
8666 David Sherwood <david.sherwood@arm.com>
8668 * combine.c (simplify_if_then_else): Remove "enum" before
8670 * compare-elim.c (can_eliminate_compare): Likewise.
8671 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8673 (aarch64_lookup_simd_builtin_type): Likewise.
8674 (aarch64_simd_builtin_type): Likewise.
8675 (aarch64_init_simd_builtin_types): Likewise.
8676 (aarch64_simd_expand_args): Likewise.
8677 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
8679 (aarch64_reverse_mask): Likewise.
8680 (aarch64_simd_emit_reg_reg_move): Likewise.
8681 (aarch64_gen_adjusted_ldpstp): Likewise.
8682 (aarch64_ccmp_mode_to_code): Likewise.
8683 (aarch64_operands_ok_for_ldpstp): Likewise.
8684 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8685 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
8687 (aarch64_min_divisions_for_recip_mul): Likewise.
8688 (aarch64_reassociation_width): Likewise.
8689 (aarch64_get_condition_code_1): Likewise.
8690 (aarch64_simd_emit_reg_reg_move): Likewise.
8691 (aarch64_simd_attr_length_rglist): Likewise.
8692 (aarch64_reverse_mask): Likewise.
8693 (aarch64_operands_ok_for_ldpstp): Likewise.
8694 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8695 (aarch64_gen_adjusted_ldpstp): Likewise.
8696 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
8698 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
8699 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8700 (arm_lookup_simd_builtin_type): Likewise.
8701 (arm_simd_builtin_type): Likewise.
8702 (arm_init_simd_builtin_types): Likewise.
8703 (arm_expand_builtin_args): Likewise.
8704 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
8705 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
8706 (ft32_setup_incoming_varargs): Likewise.
8707 (ft32_function_arg): Likewise.
8708 (ft32_function_arg_advance): Likewise.
8709 (ft32_pass_by_reference): Likewise.
8710 (ft32_arg_partial_bytes): Likewise.
8711 (ft32_valid_pointer_mode): Likewise.
8712 (ft32_addr_space_pointer_mode): Likewise.
8713 (ft32_addr_space_legitimate_address_p): Likewise.
8714 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
8716 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
8717 (ix86_emit_outlined_ms2sysv_restore): Likewise.
8718 (iamcu_alignment): Likewise.
8719 (canonicalize_vector_int_perm): Likewise.
8720 (ix86_noce_conversion_profitable_p): Likewise.
8721 (ix86_mpx_bound_mode): Likewise.
8722 (ix86_operands_ok_for_move_multiple): Likewise.
8723 * config/microblaze/microblaze-protos.h
8724 (microblaze_expand_conditional_branch_reg): Likewise.
8725 * config/microblaze/microblaze.c
8726 (microblaze_expand_conditional_branch_reg): Likewise.
8727 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8729 (rs6000_reassociation_width): Likewise.
8730 (rs6000_invalid_binary_op): Likewise.
8731 (fusion_p9_p): Likewise.
8732 (emit_fusion_p9_load): Likewise.
8733 (emit_fusion_p9_store): Likewise.
8734 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
8736 (riscv_hard_regno_mode_ok_p): Likewise.
8737 (riscv_address_insns): Likewise.
8738 (riscv_split_symbol): Likewise.
8739 (riscv_legitimize_move): Likewise.
8740 (riscv_function_value): Likewise.
8741 (riscv_hard_regno_nregs): Likewise.
8742 (riscv_expand_builtin): Likewise.
8743 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
8744 (riscv_build_integer): Likewise.
8745 (riscv_split_integer): Likewise.
8746 (riscv_legitimate_constant_p): Likewise.
8747 (riscv_cannot_force_const_mem): Likewise.
8748 (riscv_regno_mode_ok_for_base_p): Likewise.
8749 (riscv_valid_base_register_p): Likewise.
8750 (riscv_valid_offset_p): Likewise.
8751 (riscv_valid_lo_sum_p): Likewise.
8752 (riscv_classify_address): Likewise.
8753 (riscv_legitimate_address_p): Likewise.
8754 (riscv_address_insns): Likewise.
8755 (riscv_load_store_insns): Likewise.
8756 (riscv_force_binary): Likewise.
8757 (riscv_split_symbol): Likewise.
8758 (riscv_force_address): Likewise.
8759 (riscv_legitimize_address): Likewise.
8760 (riscv_move_integer): Likewise.
8761 (riscv_legitimize_const_move): Likewise.
8762 (riscv_legitimize_move): Likewise.
8763 (riscv_address_cost): Likewise.
8764 (riscv_subword): Likewise.
8765 (riscv_output_move): Likewise.
8766 (riscv_canonicalize_int_order_test): Likewise.
8767 (riscv_emit_int_order_test): Likewise.
8768 (riscv_function_arg_boundary): Likewise.
8769 (riscv_pass_mode_in_fpr_p): Likewise.
8770 (riscv_pass_fpr_single): Likewise.
8771 (riscv_pass_fpr_pair): Likewise.
8772 (riscv_get_arg_info): Likewise.
8773 (riscv_function_arg): Likewise.
8774 (riscv_function_arg_advance): Likewise.
8775 (riscv_arg_partial_bytes): Likewise.
8776 (riscv_function_value): Likewise.
8777 (riscv_pass_by_reference): Likewise.
8778 (riscv_setup_incoming_varargs): Likewise.
8779 (riscv_print_operand): Likewise.
8780 (riscv_elf_select_rtx_section): Likewise.
8781 (riscv_save_restore_reg): Likewise.
8782 (riscv_for_each_saved_reg): Likewise.
8783 (riscv_register_move_cost): Likewise.
8784 (riscv_hard_regno_mode_ok_p): Likewise.
8785 (riscv_hard_regno_nregs): Likewise.
8786 (riscv_class_max_nregs): Likewise.
8787 (riscv_memory_move_cost): Likewise.
8788 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
8789 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
8790 (rl78_addr_space_address_mode): Likewise.
8791 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8793 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8794 (rs6000_reassociation_width): Likewise.
8795 (rs6000_invalid_binary_op): Likewise.
8796 (fusion_p9_p): Likewise.
8797 (emit_fusion_p9_load): Likewise.
8798 (emit_fusion_p9_store): Likewise.
8799 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
8800 (ok_for_simple_move_operands): Likewise.
8801 (ok_for_simple_move_strict_operands): Likewise.
8802 (ok_for_simple_arith_logic_operands): Likewise.
8803 (visium_legitimize_reload_address): Likewise.
8804 (visium_select_cc_mode): Likewise.
8805 (output_cbranch): Likewise.
8806 (visium_split_double_move): Likewise.
8807 (visium_expand_copysign): Likewise.
8808 (visium_expand_int_cstore): Likewise.
8809 (visium_expand_fp_cstore): Likewise.
8810 * config/visium/visium.c (visium_pass_by_reference): Likewise.
8811 (visium_function_arg): Likewise.
8812 (visium_function_arg_advance): Likewise.
8813 (visium_libcall_value): Likewise.
8814 (visium_setup_incoming_varargs): Likewise.
8815 (visium_legitimate_constant_p): Likewise.
8816 (visium_legitimate_address_p): Likewise.
8817 (visium_legitimize_address): Likewise.
8818 (visium_secondary_reload): Likewise.
8819 (visium_register_move_cost): Likewise.
8820 (visium_memory_move_cost): Likewise.
8821 (prepare_move_operands): Likewise.
8822 (ok_for_simple_move_operands): Likewise.
8823 (ok_for_simple_move_strict_operands): Likewise.
8824 (ok_for_simple_arith_logic_operands): Likewise.
8825 (visium_function_value_1): Likewise.
8826 (rtx_ok_for_offset_p): Likewise.
8827 (visium_legitimize_reload_address): Likewise.
8828 (visium_split_double_move): Likewise.
8829 (visium_expand_copysign): Likewise.
8830 (visium_expand_int_cstore): Likewise.
8831 (visium_expand_fp_cstore): Likewise.
8832 (visium_split_cstore): Likewise.
8833 (visium_select_cc_mode): Likewise.
8834 (visium_split_cbranch): Likewise.
8835 (output_cbranch): Likewise.
8836 (visium_print_operand_address): Likewise.
8837 * expmed.c (flip_storage_order): Likewise.
8838 * expmed.h (emit_cstore): Likewise.
8839 (flip_storage_order): Likewise.
8840 * genrecog.c (validate_pattern): Likewise.
8841 * hsa-gen.c (gen_hsa_addr): Likewise.
8842 * internal-fn.c (expand_arith_overflow): Likewise.
8843 * ira-color.c (allocno_copy_cost_saving): Likewise.
8844 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8845 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
8846 (process_invariant_for_inheritance): Likewise.
8847 * lra-eliminations.c (move_plus_up): Likewise.
8848 * omp-low.c (lower_oacc_reductions): Likewise.
8849 * simplify-rtx.c (simplify_subreg): Likewise.
8850 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
8851 (TARGET_CHKP_BOUND_MODE): Likewise..
8852 * targhooks.c (default_chkp_bound_mode): Likewise.
8853 (default_setup_incoming_vararg_bounds): Likewise.
8854 * targhooks.h (default_chkp_bound_mode): Likewise.
8855 (default_setup_incoming_vararg_bounds): Likewise.
8856 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
8857 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
8858 (have_whole_vector_shift): Likewise.
8859 * tree-vect-stmts.c (vectorizable_load): Likewise.
8860 * doc/tm.texi: Regenerate.
8862 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
8864 Graceful degrade if Binutils PR21472 is not available.
8867 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
8868 .rodata in flash test fails.
8869 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
8870 * confgure: Regenerate.
8871 * config.in: Regenerate.
8872 * config/avr/avr.c (avr_asm_named_section)
8873 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
8874 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
8875 (avr_asm_init_sections): Same.
8877 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8879 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
8880 (fma<VH:mode>4_intrinsic): Likewise.
8881 (*fmsub<VCVTF:mode>4): Likewise.
8882 (*fmsub<VH:mode>4_intrinsic): Likewise.
8884 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
8887 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
8888 Don't depend on "optimize > 0".
8889 (out_movhi_r_mr, out_movqi_mr_r): Same.
8890 (out_movhi_mr_r, out_movqi_r_mr): Same.
8891 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
8892 io_address_operand on "optimize > 0".
8894 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8896 * tree-loop-distribution.c: Add general explanantion on the pass.
8897 (generate_loops_for_partition): Mark distributed loop.
8898 (pg_add_dependence_edges): New parameter. Handle alias data
8899 dependence specially and record it in the parameter if asked.
8900 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
8901 (init_partition_graph_vertices, add_partition_graph_edge): New.
8902 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
8903 (free_partition_graph_vdata, build_partition_graph): New.
8904 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
8905 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
8906 (data_ref_segment_size, latch_dominated_by_data_ref): New.
8907 (compute_alias_check_pairs, version_loop_by_alias_check): New.
8908 (version_for_distribution_p, finalize_partitions): New.
8909 (distribute_loop): Handle alias data dependence specially. Factor
8910 out loop fusion code as functions and call these functions.
8912 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8914 * tree-loop-distribution.c (classify_partition): New parameter and
8915 better handle reduction statement.
8916 (rdg_build_partitions): Revise comment.
8917 (distribute_loop): Compute statements in all partitions and pass it
8918 to classify_partition.
8920 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8922 * tree-loop-distribution.c (enum partition_type): New.
8923 (struct partition): New field type.
8924 (partition_merge_into): Add parameter. Update partition type.
8925 (data_dep_in_cycle_p, update_type_for_merge): New functions.
8926 (build_rdg_partition_for_vertex): Compute partition type.
8927 (rdg_build_partitions): Dump partition type.
8928 (distribute_loop): Update calls to partition_merge_into.
8930 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8932 * tree-loop-distribution.c (struct ddr_hasher): New.
8933 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
8935 (classify_partition): Call get_data_dependence.
8936 (pg_add_dependence_edges): Ditto.
8937 (distribute_loop): Release data dependence hash table.
8939 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8941 * tree-loop-distribution.c (ref_base_address): Delete.
8942 (similar_memory_accesses): Rename ...
8943 (share_memory_accesses): ... to this. Check if partitions access
8944 the same memory reference.
8945 (distribute_loop): Call share_memory_accesses.
8947 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8949 * tree-loop-distribution.c (struct partition): New field recording
8951 (partition_alloc, partition_free): Init and release data refs.
8952 (partition_merge_into): Merge data refs.
8953 (build_rdg_partition_for_vertex): Collect data refs for partition.
8954 (pg_add_dependence_edges): Change parameters from vector to bitmap.
8956 (distribute_loop): Remve data refs from vertice data of partition
8959 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8961 * tree-loop-distribution.c (params.h): Include header file.
8962 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
8963 (datarefs_vec): New global var.
8964 (create_rdg_vertices): Use datarefs_vec directly.
8965 (free_rdg): Don't free data references.
8966 (build_rdg): Update use. Don't free data references.
8967 (distribute_loop): Compute global variable for data references.
8968 Bail out if there are too many data references.
8970 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8972 * tree-loop-distribution.c (loop_nest): New global var.
8973 (build_rdg): Use loop directly, rather than loop nest.
8974 (pg_add_dependence_edges): Remove loop nest parameter. Use global
8976 (distribute_loop): Compute global variable loop nest. Update use.
8978 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8980 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
8981 (partition_merge_into): New parameter. Dump reason for fusion.
8982 (distribute_loop): Update use of partition_merge_into.
8984 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8986 * tree-loop-distribution.c (bb_top_order_index): New.
8987 (bb_top_order_index_size, bb_top_order_cmp): New.
8988 (stmts_from_loop): Use topological order.
8989 (pass_loop_distribution::execute): Compute and release topological
8990 order for basic blocks.
8992 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8994 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
8997 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8999 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
9000 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
9001 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
9002 * internal-fn.def (LOOP_DIST_ALIAS): New.
9003 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
9004 (fold_loop_internal_call): ... this.
9005 (vect_loop_dist_alias_call): New function.
9006 (set_uid_loop_bbs): Call fold_loop_internal_call.
9007 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
9010 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9013 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
9014 Require dead FLAGS_REG at the beginning of a peephole.
9016 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9019 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
9020 arguments in the call to __builtin_ia32_sbb_u32.
9021 (_subborrow_u64): Swap _X and _Y arguments in the call to
9022 __builtin_ia32_sbb_u64.
9024 2017-07-04 Jakub Jelinek <jakub@redhat.com>
9027 * tree-vrp.c (compare_assert_loc): Turn into a function template
9028 with stable template parameter. Only test if a->e is NULL,
9029 !a->e == !b->e has been verified already. Use e == NULL or
9030 e != NULL instead of e or ! e tests. If stable is true, don't use
9031 iterative_hash_expr, on the other side allow a or b or both NULL
9032 and sort the NULLs last.
9033 (process_assert_insertions): Sort using compare_assert_loc<false>
9034 instead of compare_assert_loc, later sort using
9035 compare_assert_loc<true> before calling process_assert_insertions_for
9036 in a loop. Use break instead of continue once seen NULL pointer.
9038 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9040 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9041 Cortex-R7 and Cortex-R8 processors.
9043 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9045 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
9046 uninitialized while src is not.
9048 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
9050 * common/config/arm/arm-common.c: Adjust include path for
9052 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
9053 (arm-cpu.h): Create in build directory. Adjust dependency rules.
9054 (arm-cpu-data.h): Likewise.
9055 (arm-cpu-cdata.h): Likewise.
9056 * config/arm/arm-cpu.h: Delete.
9057 * config/arm/arm-cpu-cdata.h: Delete.
9058 * config/arm/arm-cpu-data.h: Delete.
9060 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
9062 * config/arm/arm-cpus.in (cortex-a55): New.
9063 (cortex-a75): Likewise.
9064 (cortex-a75.cortex-a55): Likewise.
9065 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
9067 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
9068 * config/arm/arm-cpu-cdata.h: Regenerate.
9069 * config/arm/arm-cpu-data.h: Regenerate.
9070 * config/arm/arm-cpu.h: Regenerate.
9071 * config/arm/arm-tables.opt: Regenerate.
9072 * config/arm/arm-tune.md: Regenerate.
9074 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9076 * haifa-sched.c (sched_create_recovery_edges): Update profile.
9078 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9080 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
9083 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
9085 PR tree-optimization/81292
9086 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
9087 full_string_p, also call adjust_related_strinfos if the adjustment
9088 is simple, otherwise invalidate related strinfos.
9090 2017-07-04 Martin Liska <mliska@suse.cz>
9093 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
9094 newly created variable as DECL_IGNORED_P.
9096 2017-07-04 Martin Liska <mliska@suse.cz>
9099 * ipa-inline.c (inline_small_functions):
9100 Use xstrdup_for_dump.
9102 2017-07-04 Tom de Vries <tom@codesourcery.com>
9104 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
9106 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9109 * config/darwin.c (darwin_function_switched_text_sections):
9112 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
9114 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
9116 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9118 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
9120 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9122 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
9123 min_profitable_iters, and th as inclusive lower bounds.
9124 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
9125 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
9126 for min_profitable_iters and min_profitable_estimate.
9127 (vect_transform_loop): Treat th as an inclusive lower bound.
9128 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
9130 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9133 * config/darwin.c (darwin_function_switched_text_sections):
9134 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
9135 in two pieces, and suppress the use of buf.
9137 2017-07-03 Nathan Sidwell <nathan@acm.org>
9139 * hash-table.h (hash_table_mod1): Fix indentation.
9141 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9144 * predict.c (force_edge_cold): Be more careful about propagation
9146 * profile-count.h (profile_probability::guessed,
9147 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
9149 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
9151 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
9153 * doc/invoke.texi (rcpc architecture extension): Document it.
9155 2017-07-03 Richard Biener <rguenther@suse.de>
9157 PR tree-optimization/60510
9158 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
9159 the scalar reduction PHI and use it.
9160 (vectorizable_reduction): Properly guard the single_defuse_cycle
9161 path for non-SLP reduction chains where we cannot use it.
9162 Rework reduc_def/index and vector type deduction. Rework
9163 vector operand gathering during reduction op code-gen.
9164 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
9165 chains dissolve the chain and leave it to non-SLP reduction
9168 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9170 * tree-data-ref.h (dr_alignment): Declare.
9171 * tree-data-ref.c (dr_alignment): New function.
9172 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
9173 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
9175 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
9177 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9179 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
9180 and base_misalignment fields.
9181 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
9182 * tree-data-ref.c: Include builtins.h.
9183 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
9184 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
9185 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
9186 * tree-vect-data-refs.c: Include tree-cfg.h.
9187 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
9188 fields instead of calculating an alignment here.
9189 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
9190 innermost_loop_behavior fields.
9192 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9194 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
9196 (DR_STEP_ALIGNMENT): New macro.
9197 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
9198 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
9199 (create_data_ref): Print it.
9200 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
9201 to tell whether the step preserves vector (mis)alignment.
9202 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9203 Move the check for an integer step and generalise to all INTEGER_CST.
9204 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
9205 Print the outer step alignment.
9207 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9209 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
9210 with offset_alignment.
9211 (DR_ALIGNED_TO): Delete.
9212 (DR_OFFSET_ALIGNMENT): New macro.
9213 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
9214 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
9215 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
9216 (create_data_ref): Likewise.
9217 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9218 (vect_analyze_data_refs): Likewise.
9219 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
9220 creating dummy innermost behavior.
9222 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9224 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
9225 with a "innermost_loop_behavior *" and refeence tree.
9226 * tree-data-ref.c (dr_analyze_innermost): Likewise.
9227 (create_data_ref): Update call accordingly.
9228 * tree-predcom.c (find_looparound_phi): Likewise.
9230 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9232 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
9233 fields with dr_wrt_vec_loop.
9234 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
9235 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
9236 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
9237 (vect_dr_behavior): New function.
9238 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9239 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
9240 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
9241 track whether the step preserves the misalignment.
9242 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9243 Use vect_dr_behavior.
9244 (vect_setup_realignment): Update call accordingly.
9245 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
9246 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
9247 call to vect_create_addr_base_for_vector_ref.
9248 (vect_create_cond_for_align_checks): Likewise.
9249 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
9250 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
9251 (vect_recog_mask_conversion_pattern): Likewise.
9252 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
9253 (new_stmt_vec_info): Remove redundant zeroing.
9255 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9257 * common/config/arm/arm-common.c (arm_be8_option): New function.
9258 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
9259 (ISA_ARMv6): Add isa_bit_be8.
9260 * config/arm/arm.h (arm_be8_option): Add prototype.
9261 (BE8_SPEC_FUNCTION): New define.
9262 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
9263 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
9264 (mlittle-endian): Similarly.
9265 (mbe8, mbe32): New options.
9266 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
9267 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
9269 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9271 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
9273 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9275 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
9276 (cleanup_tree_cfg_bb): Use it.
9277 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
9279 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
9281 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9284 * loop-doloop.c (add_test): Update profile.
9286 2017-07-03 Martin Liska <mliska@suse.cz>
9289 * sanopt.c (rewrite_usage_of_param): New function.
9290 (sanitize_rewrite_addressable_params): Likewise.
9291 (pass_sanopt::execute): Call rewrite_usage_of_param.
9293 2017-07-03 Richard Biener <rguenther@suse.de>
9295 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
9296 back to using VIEW_CONVERT_EXPR.
9298 2017-07-03 Martin Liska <mliska@suse.cz>
9301 * doc/extend.texi: Document when a resolver function is
9302 generated for target_clones.
9304 2017-07-03 Martin Liska <mliska@suse.cz>
9306 * asan.c (asan_emit_stack_protection): Unpoison just red zones
9307 and shadow memory of auto variables which are subject of
9308 use-after-scope sanitization.
9309 (asan_expand_mark_ifn): Add do set only when is_poison.
9311 2016-07-03 Richard Biener <rguenther@suse.de>
9313 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
9315 (vect_force_simple_reduction): Record reduction def -> phi mapping.
9316 (vectorizable_reduction): Perform reduction PHI creation when
9317 visiting a reduction PHI and adjust and simplify code generation
9318 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
9319 (vect_transform_loop): Visit reduction PHIs.
9320 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
9321 defs into the SLP tree.
9322 (vect_build_slp_tree): Reduction defs terminate the recursion.
9323 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
9325 (vect_get_vec_defs_for_stmt_copy): Export.
9326 (vect_get_vec_defs): Likewise.
9327 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
9329 (vect_get_vec_defs_for_stmt_copy): Declare.
9330 (vect_get_vec_defs): Likewise.
9332 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9334 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
9335 parameter with a "loop" parameter and use it instead of the
9336 loop containing DR_STMT. Don't check simple_iv when doing
9337 BB analysis. Describe the two analysis modes in the comment.
9339 2017-07-03 Tom de Vries <tom@codesourcery.com>
9341 PR tree-optimization/69468
9342 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
9343 (find_same_succ_bb): Handle ignore_edge_flags.
9345 2017-07-03 Tom de Vries <tom@codesourcery.com>
9347 PR tree-optimization/81192
9348 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
9350 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
9352 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
9354 2017-07-03 Tom de Vries <tom@codesourcery.com>
9356 PR tree-optimization/81192
9357 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
9358 BB_SAME_SUCC (bb) == NULL.
9360 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9362 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
9365 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9367 * dumpfile.c: Include profile-count.h
9368 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
9370 (insert_cond_bb): Update profile.
9371 * tree-cfg.h (insert_cond_bb): Update prototype.
9372 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
9373 * tree-dump.c: Do not include tree-cfg.
9375 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9377 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
9379 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9381 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
9384 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9386 * tree-complex.c (expand_complex_div_wide): update profile.
9388 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9389 Alan Hayward <alan.hayward@arm.com>
9390 David Sherwood <david.sherwood@arm.com>
9392 * Makefile.in (MACHMODE_H): Remove insn-modes.h
9393 (CORETYPES_H): New define.
9394 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
9395 (insn-modes-inline.h, s-modes-inline-h): New rules.
9396 (generated_files): Add insn-modes-inline.h.
9397 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
9398 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
9399 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
9400 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
9401 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
9402 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
9403 (build/gencodes.o, build/genconditions.o): Likewise.
9404 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
9405 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
9406 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
9407 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
9408 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
9409 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
9410 * coretypes.h: Include everything up to real.h for generators.
9411 Include insn-modes.h first. Include wide-int-print.h after
9412 wide-int.h. Include insn-modes-inline.h and then machmode.h.
9413 * machmode.h: Don't include insn-modes.h here.
9414 * function-tests.c: Remove includes of signop.h, machmode.h,
9415 double-int.h and wide-int.h.
9417 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
9419 * optc-save-gen.awk: Likewise.
9420 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
9421 * godump.c: Remove include of wide-int-print.h.
9422 * pretty-print.h: Likewise.
9423 * wide-int-print.cc: Likewise.
9424 * wide-int.cc: Likewise.
9425 * hash-map-tests.c: Remove include of signop.h.
9426 * hash-set-tests.c: Likewise.
9427 * rtl-tests.c: Likewise.
9428 * mkconfig.sh: Remove include of machmode.h.
9429 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
9431 (emit_insn_modes_inline_h): ...this new function. Emit the code
9432 into an insn-modes-inline.h header file, adding appropriate
9433 include guards and end comments.
9434 (emit_insn_modes_c_header): Remove include of machmode.h.
9435 (emit_min_insn_modes_c_header): Include coretypes.h rather than
9437 (main): Handle -i flag and call emit_insn_modes_inline_h when
9440 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9442 * tree-ssa-strlen.c (strinfo): Rename the length field to
9443 nonzero_chars. Add a full_string_p field.
9444 (compare_nonzero_chars, zero_length_string_p): New functions.
9445 (get_addr_stridx): Add an offset_out parameter.
9446 Use compare_nonzero_chars.
9447 (get_stridx): Update accordingly. Use compare_nonzero_chars.
9448 (new_strinfo): Update after above changes to strinfo.
9449 (set_endptr_and_length): Set full_string_p.
9450 (get_string_length): Update after above changes to strinfo.
9451 (unshare_strinfo): Update call to new_strinfo.
9452 (maybe_invalidate): Likewise.
9453 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
9454 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
9455 as a uhwi instead of an shwi. Update after above changes to
9456 strinfo and new_strinfo.
9457 (zero_length_string): Assert that chainsi contains full strings.
9458 Use zero_length_string_p. Update call to new_strinfo.
9459 (adjust_related_strinfos): Update after above changes to strinfo.
9460 Copy full_string_p from origsi.
9461 (adjust_last_stmt): Use zero_length_string_p.
9462 (handle_builtin_strlen): Update after above changes to strinfo and
9463 new_strinfo. Install the lhs as the string length if the previous
9464 entry didn't describe a full string.
9465 (handle_builtin_strchr): Update after above changes to strinfo
9467 (handle_builtin_strcpy): Likewise.
9468 (handle_builtin_strcat): Likewise.
9469 (handle_builtin_malloc): Likewise.
9470 (handle_pointer_plus): Likewise.
9471 (handle_builtin_memcpy): Likewise. Track nonzero characters
9472 that aren't necessarily followed by a nul terminator.
9473 (handle_char_store): Likewise.
9475 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9477 PR tree-optimization/80769
9478 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
9479 for malloc and calloc. Document the new invariant that all related
9480 strinfos have delayed lengths or none do.
9481 (verify_related_strinfos): Move earlier in file.
9482 (set_endptr_and_length): New function, split out from...
9483 (get_string_length): ...here. Also set the lengths of related
9485 (zero_length_string): Assert that chainsi has known (rather than
9487 (adjust_related_strinfos): Likewise.
9489 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9491 PR tree-optimization/81136
9492 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
9493 assert that two references with the same misalignment have the same
9494 compile-time misalignment if those compile-time misalignments
9497 2017-07-01 Andi Kleen <ak@linux.intel.com>
9499 * print-tree.c (print_node): Print all attributes.
9501 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9503 * cfg.c (scale_bbs_frequencies): New function.
9504 * cfg.h (scale_bbs_frequencies): Declare it.
9505 * cfgloopanal.c (single_likely_exit): Cleanup.
9506 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
9508 (scale_loop_profile): Likewise.
9509 (loop_version): Likewise.
9510 (create_empty_loop_on_edge): Update.
9511 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
9512 scale_loop_frequencies, scale_loop_profile, loopify,
9513 loop_version): Update prototypes.
9514 * modulo-sched.c (sms_schedule): Update.
9515 * predict.c (unlikely_executed_edge_p): Also check probability.
9516 (probably_never_executed_edge_p): Fix typo.
9517 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9518 * tree-parloops.c (gen_parallel_loop): Update.
9519 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
9520 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9521 * tree-ssa-loop-split.c (split_loop): Update.
9522 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9523 * tree-vect-loop-manip.c (vect_do_peeling): Update.
9524 (vect_loop_versioning): Update.
9525 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9527 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9529 * trans-mem.c (split_bb_make_tm_edge): Update profile.
9531 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9533 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
9534 to keep profile consistent.
9536 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9538 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
9539 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
9540 * profile-count.h (max_safe_multiplier): Make unsigned.
9541 (profile_count::guessed_zero): New.
9543 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9545 * bb-reorder.c (fix_up_crossing_landing_pad,
9546 fix_crossing_conditional_branches): Use make_single_succ_edge
9547 to keep profile consistent.
9549 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9551 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
9554 2017-07-01 Jakub Jelinek <jakub@redhat.com>
9557 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
9558 the right scopes, make sure cond_jump isn't preserved between multiple
9559 iterations. Search for fallthru edge whenever there are 3+ edges and
9560 use find_fallthru_edge for it.
9562 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9564 Patch by Alexander Monakov <amonakov@ispras.ru>
9565 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
9566 probabilities consistently.
9568 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9570 * pa.c (pa_expand_compare_and_swap_loop): Update call of
9571 emit_cmp_and_jump_insns.
9573 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9576 * tree-inline.c (expand_call_inline): Combine profile statuses.
9578 2017-06-30 Andrew Pinski <apinski@cavium.com>
9580 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
9581 fold_stmt returned true.
9583 2017-06-30 Nathan Sidwell <nathan@acm.org>
9585 * ggc.h (empty_string): Delete.
9586 * cfgexpand.c (expand_asm_stmt): Use plain "".
9587 * optabs.c (expand_asm_memory_barrier): Likewise.
9588 * stringpool.c (empty_string): Delete.
9589 (digit_vector, digit_string): Delete.
9590 (ggc_alloc_string): Use plain "", don't optimize single digit
9591 strings. Use ggc_alloc_atomic.
9593 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
9595 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
9596 comparison set and one other set, use the cost of the non-comparison
9599 2017-06-30 Nathan Sidwell <nathan@acm.org>
9601 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
9604 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
9606 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
9607 loops. Remove now unneeded calls to gimple_switch_set_label() that
9608 just set removed labels to NULL_TREE.
9610 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
9612 * tree-ssanames.c (set_range_info_raw): Abstract from ...
9613 (set_range_info): ...here. Only call set_range_info_raw if domain
9615 (set_nonzero_bits): Call set_range_info_raw.
9616 * tree-ssanames.h (set_range_info_raw): New.
9618 2017-06-30 Jakub Jelinek <jakub@redhat.com>
9621 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
9622 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
9623 of nonimmediate_operand and <store_mask_constraint> instead of m
9624 for the input operand. For V8FI iterator, always split if input
9625 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
9626 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
9627 <store_mask_predicate> instead of register_operand and
9628 <store_mask_constraint> instead of v for the input operand. Make
9629 sure both operands aren't MEMs for if not <mask_applied>.
9631 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
9633 * lto-wrapper.c (copy_file) Close both file descriptors before
9636 2017-06-30 Martin Liska <mliska@suse.cz>
9639 * multiple_target.c (create_dispatcher_calls): Make ifunc
9640 also for function that don't have calls or are not referenced.
9642 2017-06-30 Richard Biener <rguenther@suse.de>
9644 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
9645 analyze the first scalar stmt. Move vector type computation
9646 for the BB case here from ...
9647 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
9648 live operation processing in the SLP case properly.
9650 2017-06-30 Richard Biener <rguenther@suse.de>
9652 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
9654 2017-06-30 Martin Liska <mliska@suse.cz>
9657 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
9658 before BUILT_IN_UNWIND_RESUME when ASAN is used.
9660 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
9662 * doc/invoke.texi (AArch64): Add missing options and remove redundant
9665 2017-06-30 Richard Biener <rguenther@suse.de>
9667 PR tree-optimization/81249
9668 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
9669 condition reduction result to original scalar type.
9671 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9673 * profile-count.h (enum profile_quality): Fix typos and whitespace
9676 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9678 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
9679 type for branch probabilities.
9681 2017-06-29 Julian Brown <julian@codesourcery.com>
9682 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9684 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
9685 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
9686 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
9687 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
9689 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9691 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
9692 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
9693 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
9694 CC usage from generic code to here.
9695 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
9696 CC usage into the target macros.
9698 2017-06-29 Maya Rashish <coypu@sdf.org>
9700 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
9703 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9705 * arm/arm-builtins.c: Include profile-count.h
9706 * except.c (sjlj_emit_function_enter): Use
9707 profile_probability::unlikely.
9709 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9711 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
9712 and tocrel_offset be pointer args rather than implicitly using
9714 (legitimate_constant_pool_address_p, rs6000_emit_move,
9715 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
9716 tocrel_offset and use in toc_relative_expr_p call.
9717 (print_operand, print_operand_address): Use static tocrel_base_oac
9718 and tocrel_offset_oac.
9719 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
9722 2017-06-29 Maya Rashish <coypu@sdf.org>
9724 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
9726 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
9728 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
9729 objects, take into account only the alignment of 'op0' and 'mode1' if
9732 2017-06-29 Steve Ellcey <sellcey@cavium.com>
9734 * ccmp.c (ccmp_tree_comparison_p): New function.
9735 (ccmp_candidate_p): Update to use above function.
9736 (get_compare_parts): New function.
9737 (expand_ccmp_next): Update to use new functions.
9738 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
9740 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
9741 take mode as argument.
9742 * ccmp.h (expand_ccmp_expr): Add mode as argument.
9743 * expr.c (expand_expr_real_1): Pass mode as argument.
9745 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
9747 * combine.c (combine_instructions): Print insns to dump_file, together
9750 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9752 * asan.c (asan_emit_stack_protection): Update.
9753 (create_cond_insert_point): Update.
9754 * auto-profile.c (afdo_propagate_circuit): Update.
9755 * basic-block.h (struct edge_def): Turn probability to
9756 profile_probability.
9757 (EDGE_FREQUENCY): Update.
9758 * bb-reorder.c (find_traces_1_round): Update.
9759 (better_edge_p): Update.
9760 (sanitize_hot_paths): Update.
9761 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
9762 (make_single_succ_edge): Update.
9763 (check_bb_profile): Update.
9764 (dump_edge_info): Update.
9765 (update_bb_profile_for_threading): Update.
9766 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
9767 probabilitycount to 0.
9768 * cfgbuild.c (compute_outgoing_frequencies): Update.
9769 * cfgcleanup.c (try_forward_edges): Update.
9770 (outgoing_edges_match): Update.
9771 (try_crossjump_to_edge): Update.
9772 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
9773 (expand_gimple_tailcall): Update.
9774 (construct_init_block): Use make_single_succ_edge.
9775 (construct_exit_block): Use make_single_succ_edge.
9776 * cfghooks.c (verify_flow_info): Update.
9777 (redirect_edge_succ_nodup): Update.
9778 (split_edge): Update.
9779 (account_profile_record): Update.
9780 * cfgloopanal.c (single_likely_exit): Update.
9781 * cfgloopmanip.c (scale_loop_profile): Update.
9782 (set_zero_probability): Remove.
9783 (duplicate_loop_to_header_edge): Update.
9784 * cfgloopmanip.h (loop_version): Update prototype.
9785 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9786 (force_nonfallthru_and_redirect): Update.
9787 (update_br_prob_note): Update.
9788 (rtl_verify_edges): Update.
9789 (purge_dead_edges): Update.
9790 (rtl_lv_add_condition_to_bb): Update.
9791 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
9792 * cgraphunit.c (init_lowered_empty_function): Update.
9793 (cgraph_node::expand_thunk): Update.
9794 * cilk-common.c: Include profile-count.h
9795 * dojump.c (inv): Remove.
9796 (jumpifnot): Update.
9797 (jumpifnot_1): Update.
9798 (do_jump_1): Update.
9800 (do_jump_by_parts_greater_rtx): Update.
9801 (do_compare_rtx_and_jump): Update.
9802 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
9803 do_jump_1. do_compare_rtx_and_jump): Update prototype.
9804 * dwarf2cfi.c: Include profile-count.h
9805 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
9806 (sjlj_emit_dispatch_table): Likewise.
9807 * explow.c: Include profile-count.h
9808 * expmed.c (emit_store_flag_force): Update.
9809 (do_cmp_and_jump): Update.
9810 * expr.c (compare_by_pieces_d::generate): Update.
9811 (compare_by_pieces_d::finish_mode): Update.
9812 (emit_block_move_via_loop): Update.
9813 (store_expr_with_bounds): Update.
9814 (store_constructor): Update.
9815 (expand_expr_real_2): Update.
9816 (expand_expr_real_1): Update.
9817 * expr.h (try_casesi, try_tablejump): Update prototypes.
9818 * gimple-pretty-print.c (dump_probability): Update.
9819 (dump_profile): New.
9820 (dump_gimple_label): Update.
9821 (dump_gimple_bb_header): Update.
9822 * graph.c (draw_cfg_node_succ_edges): Update.
9823 * hsa-gen.c (convert_switch_statements): Update.
9824 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
9825 (find_if_case_1): Update.
9826 (find_if_case_2): Update.
9827 * internal-fn.c (expand_arith_overflow_result_store): Update.
9828 (expand_addsub_overflow): Update.
9829 (expand_neg_overflow): Update.
9830 (expand_mul_overflow): Update.
9831 (expand_vector_ubsan_overflow): Update.
9832 * ipa-cp.c (good_cloning_opportunity_p): Update.
9833 * ipa-split.c (split_function): Use make_single_succ_edge.
9834 * ipa-utils.c (ipa_merge_profiles): Update.
9835 * loop-doloop.c (add_test): Update.
9836 (doloop_modify): Update.
9837 * loop-unroll.c (compare_and_jump_seq): Update.
9838 (unroll_loop_runtime_iterations): Update.
9839 * lra-constraints.c (lra_inheritance): Update.
9840 * lto-streamer-in.c (input_cfg): Update.
9841 * lto-streamer-out.c (output_cfg): Update.
9842 * mcf.c (adjust_cfg_counts): Update.
9843 * modulo-sched.c (sms_schedule): Update.
9844 * omp-expand.c (expand_omp_for_init_counts): Update.
9845 (extract_omp_for_update_vars): Update.
9846 (expand_omp_ordered_sink): Update.
9847 (expand_omp_for_ordered_loops): Update.
9848 (expand_omp_for_generic): Update.
9849 (expand_omp_for_static_nochunk): Update.
9850 (expand_omp_for_static_chunk): Update.
9851 (expand_cilk_for): Update.
9852 (expand_omp_simd): Update.
9853 (expand_omp_taskloop_for_outer): Update.
9854 (expand_omp_taskloop_for_inner): Update.
9855 * omp-simd-clone.c (simd_clone_adjust): Update.
9856 * optabs.c (expand_doubleword_shift): Update.
9857 (expand_abs): Update.
9858 (emit_cmp_and_jump_insn_1): Update.
9859 (expand_compare_and_swap_loop): Update.
9860 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
9861 * predict.c (predictable_edge_p): Update.
9862 (edge_probability_reliable_p): Update.
9863 (set_even_probabilities): Update.
9864 (combine_predictions_for_insn): Update.
9865 (combine_predictions_for_bb): Update.
9866 (propagate_freq): Update.
9867 (estimate_bb_frequencies): Update.
9868 (force_edge_cold): Update.
9869 * profile-count.c (profile_count::dump): Add missing space into dump.
9870 (profile_count::debug): Add newline.
9871 (profile_count::differs_from_p): Explicitly convert to unsigned.
9872 (profile_count::stream_in): Update.
9873 (profile_probability::dump): New member function.
9874 (profile_probability::debug): New member function.
9875 (profile_probability::differs_from_p): New member function.
9876 (profile_probability::differs_lot_from_p): New member function.
9877 (profile_probability::stream_in): New member function.
9878 (profile_probability::stream_out): New member function.
9879 * profile-count.h (profile_count_quality): Rename to ...
9880 (profile_quality): ... this one.
9881 (profile_probability): New.
9882 (profile_count): Update.
9883 * profile.c (compute_branch_probabilities): Update.
9884 * recog.c (peep2_attempt): Update.
9885 * sched-ebb.c (schedule_ebbs): Update.
9886 * sched-rgn.c (find_single_block_region): Update.
9887 (compute_dom_prob_ps): Update.
9888 (schedule_region): Update.
9889 * sel-sched-ir.c (compute_succs_info): Update.
9890 * stmt.c (struct case_node): Update.
9891 (do_jump_if_equal): Update.
9892 (get_outgoing_edge_probs): Update.
9893 (conditional_probability): Update.
9894 (emit_case_dispatch_table): Update.
9895 (expand_case): Update.
9896 (expand_sjlj_dispatch_table): Update.
9897 (emit_case_nodes): Update.
9898 * targhooks.c: Update.
9899 * tracer.c (better_p): Update.
9900 (find_best_successor): Update.
9901 * trans-mem.c (expand_transaction): Update.
9902 * tree-call-cdce.c: Update.
9903 * tree-cfg.c (gimple_split_edge): Upate.
9904 (move_sese_region_to_fn): Upate.
9905 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
9906 * tree-eh.c (lower_resx): Upate.
9907 (cleanup_empty_eh_move_lp): Upate.
9908 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9909 * tree-inline.c (copy_edges_for_bb): Update.
9910 (copy_cfg_body): Update.
9911 * tree-parloops.c (gen_parallel_loop): Update.
9912 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
9913 (gimple_gen_time_profiler): Update.
9914 * tree-ssa-dce.c (remove_dead_stmt): Update.
9915 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
9916 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
9917 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
9918 (unloop_loops): Update.
9919 (try_peel_loop): Update.
9920 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9921 * tree-ssa-loop-split.c (connect_loops): Update.
9922 (split_loop): Update.
9923 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9924 (hoist_guard): Update.
9925 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
9926 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
9927 (value_replacement): Update.
9928 * tree-ssa-reassoc.c (branch_fixup): Update.
9929 * tree-ssa-tail-merge.c (replace_block_by): Update.
9930 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
9931 (create_edge_and_update_destination_phis): Update.
9932 (compute_path_counts): Update.
9933 (recompute_probabilities): Update.
9934 (update_joiner_offpath_counts): Update.
9935 (freqs_to_counts_path): Update.
9936 (duplicate_thread_path): Update.
9937 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
9938 (struct switch_conv_info): Update.
9939 (gen_inbound_check): Update.
9940 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
9941 (vect_do_peeling): Update.
9942 (vect_loop_versioning): Update.
9943 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9944 (optimize_mask_stores): Update.
9945 * ubsan.c (ubsan_expand_null_ifn): Update.
9946 * value-prof.c (gimple_divmod_fixed_value): Update.
9947 (gimple_divmod_fixed_value_transform): Update.
9948 (gimple_mod_pow2): Update.
9949 (gimple_mod_pow2_value_transform): Update.
9950 (gimple_mod_subtract): Update.
9951 (gimple_mod_subtract_transform): Update.
9952 (gimple_ic): Update.
9953 (gimple_stringop_fixed_value): Update.
9954 (gimple_stringops_transform): Update.
9955 * value-prof.h: Update.
9957 2017-06-29 Carl Love <cel@us.ibm.com>
9959 * config/rs6000/rs6000-c.c: Add support for built-in functions
9960 vector signed int vec_signed (vector float);
9961 vector signed long long vec_signed (vector double);
9962 vector signed int vec_signed2 (vector double, vector double);
9963 vector signed int vec_signede (vector double);
9964 vector signed int vec_signedo (vector double);
9965 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
9966 instruction generator.
9967 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
9968 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
9969 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
9971 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
9972 vunsignede_v2df): Add define_expands.
9973 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
9974 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
9975 VEC_UNSIGNEDO): Add definitions.
9976 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
9977 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
9978 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
9979 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
9980 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
9981 * config/rs6000/altivec.h (vec_signed, vec_signed2,
9982 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
9983 vec_unsignede, vec_unsignedo): Add builtin defines.
9984 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
9986 * doc/extend.texi: Update the built-in documentation file for the
9987 new built-in functions.
9989 2017-06-29 Richard Biener <rguenther@suse.de>
9991 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
9992 reduction chains to LOOP_VINFO_REDUCTIONS.
9993 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
9994 SLP reductions after processing reduction chains.
9996 2017-06-29 Nathan Sidwell <nathan@acm.org>
9998 * builtins.c (fold_builtin_FUNCTION): Use
9999 lang_hooks.decl_printable_name.
10001 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
10003 PR middle-end/81194
10004 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
10005 with only one label.
10006 * stmt.c (expand_case): Assert NCASES is greater than one.
10008 2017-06-29 Richard Biener <rguenther@suse.de>
10010 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
10012 (group_case_labels): Likewise.
10013 (find_taken_edge): Push sanity checking on val to workers...
10014 (find_taken_edge_cond_expr): ... here
10015 (find_taken_edge_switch_expr): ... and here, handle cases
10016 with just a default label.
10017 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
10018 (group_case_labels): Likewise.
10019 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
10020 group_case_labels does anything cleanup the CFG again.
10022 2017-06-29 Bin Cheng <bin.cheng@arm.com>
10024 PR tree-optimization/81196
10025 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
10026 exit condition comparing two IVs.
10028 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
10030 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
10031 profile to the dummy entry at the end of the list of architectures.
10032 * config/arm/arm-cpu-cdata.h: Regenerated.
10034 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10035 Michael Collison <michael.collison@arm.com>
10038 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
10040 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
10041 (*aarch64_reg_<mode>3_minus_mask): New pattern.
10042 (*aarch64_<optab>_reg_di3_mask2): New pattern.
10043 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
10044 of shift when the shift amount is masked with constant equal to
10045 the size of the mode.
10046 * config/aarch64/predicates.md (subreg_lowpart_operator): New
10049 2017-06-29 Martin Liska <mliska@suse.cz>
10051 * config/i386/i386.opt: Change range from [1,5] to [0,5].
10053 2017-06-29 Yury Gribov <tetra2005@gmail.com>
10056 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
10059 (edge_growth_cache_entry::edge_growth_cache_entry): New
10061 (reset_edge_growth_cache): Update to use constructor.
10063 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10065 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
10066 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10067 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10069 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
10071 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
10072 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
10074 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
10076 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
10077 (*-linux-uclibc*): Add t-uclibc tmake_file.
10078 * config/t-musl: New.
10079 * config/t-uclibc: New.
10081 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
10083 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
10085 (gen_comm_data): Emit architectural setting of arch_prof.
10086 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
10088 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
10089 (armv8-m.base, armv8-m.main): Likewise.
10090 * arm-protos.h (arm_build_target): Add profile field.
10091 (arch_option): Likewise.
10092 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
10094 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
10095 arm_active_target.profile.
10097 2017-06-28 Richard Biener <rguenther@suse.de>
10099 PR middle-end/81227
10100 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
10101 TYPE_OVERFLOW_WRAPS.
10102 * match.pd (negate_expr_p): Likewise.
10103 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
10104 fold_build2, not fold_binary.
10106 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10108 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
10109 Convert memory address to Pmode.
10110 (aarch64_print_operand): Assert MEM operands are always Pmode.
10112 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10115 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
10116 Remove redundant if.
10117 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
10118 * config/arm/aarch-common-protos.h
10119 (aarch_forward_to_shift_is_not_shifted_re): Remove.
10120 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
10122 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
10125 * multiple_target.c (create_dispatcher_calls): Set the default
10126 clone to be static, not public.
10128 2017-06-28 Richard Biener <rguenther@suse.de>
10130 * tree-vect-loop.c (vectorizable_reduction): Move special
10131 cond reduction IV var creation ...
10132 (vect_create_epilog_for_reduction): ... here. Remove induction_index
10133 parameter. Use STMT_VINFO_VECTYPE.
10134 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
10137 2017-06-28 Martin Liska <mliska@suse.cz>
10140 * ipa-visibility.c (non_local_p): Handle visibility.
10142 2017-06-28 Martin Liska <mliska@suse.cz>
10145 * common.opt: Add IntegerRange to various options.
10146 * opt-functions.awk (integer_range_info): New function.
10147 * optc-gen.awk: Add integer_range_info to cl_options struct.
10148 * opts-common.c (decode_cmdline_option): Handle
10149 CL_ERR_INT_RANGE_ARG.
10150 (cmdline_handle_error): Likewise.
10151 * opts.c (print_filtered_help): Show valid interval in
10152 when --help is provided.
10153 * opts.h (struct cl_option): Add range_min and range_max fields.
10154 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
10156 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
10158 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
10159 (x * C EQ/NE y * C): New transformation.
10161 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
10163 * genmultilib (combination_space): Accept '+' in option names.
10165 2017-06-28 Martin Liska <mliska@suse.cz>
10168 * asan.c (instrument_derefs): Bail out inner references
10169 that are hard register variables.
10171 2017-06-28 Jakub Jelinek <jakub@redhat.com>
10174 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
10175 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
10177 2017-06-28 Richard Biener <rguenther@suse.de>
10179 * tree-vectorizer.h (vect_get_vec_defs): Remove.
10180 (vect_get_slp_defs): Adjust.
10181 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
10183 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
10185 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10186 get_initial_defs_for_reduction instead of vect_get_vec_defs.
10187 (vectorizable_reduction): Adjust.
10188 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
10190 (vect_get_slp_defs): Likewise.
10191 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
10192 (vectorizable_bswap): Adjust.
10193 (vectorizable_call): Likewise.
10194 (vectorizable_conversion): Likewise.
10195 (vectorizable_assignment): Likewise.
10196 (vectorizable_shift): Likewise.
10197 (vectorizable_operation): Likewise.
10198 (vectorizable_store): Likewise.
10199 (vectorizable_condition): Likewise.
10200 (vectorizable_comparison): Likewise.
10202 2017-06-28 Michael Collison <michael.collison@arm.com>
10205 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
10207 (arm_gen_movmemqi): Removed unused variable 'i'.
10208 Convert 'for' loop into 'while' loop.
10209 (arm_expand_prologue): Remove last unnecessary set of insn.
10210 (thumb_pop): Remove unused variable 'pushed_words'.
10211 (thumb_exit): Remove last unnecessary set of regs_to_pop.
10213 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10215 * config/s390/predicates.md: Use s390_rel_address_ok_p.
10216 * config/s390/s390-protos.h: Add prototype of
10217 s390_rel_address_ok_p.
10218 * config/s390/s390.c (s390_got_symbol): New function.
10219 (s390_rel_address_ok_p): New function.
10220 (legitimize_pic_address): Use s390_rel_address_ok_p.
10221 (s390_load_got): Use s390_got_symbol.
10222 (s390_option_override): Issue error if
10223 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
10224 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
10226 * config/s390/s390.opt: New option mpic-data-is-text-relative.
10228 2017-06-27 Andrew Pinski <apinski@cavium.com>
10230 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
10231 (X * copysign (1.0, X)): New pattern.
10232 (X * copysign (1.0, -X)): New pattern.
10233 (copysign (-1.0, CST)): New pattern.
10235 2017-06-27 Joseph Myers <joseph@codesourcery.com>
10237 * genmultilib (combination_space): Remove variable.
10238 Validate reuse rules against regular expression for any sequence
10239 of multilib options in any order.
10241 2017-06-27 Michael Collison <michael.collison@arm.com>
10243 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
10244 call aarch64_split_simd_combine.
10245 * (aarch64_combine_internal<mode>): Delete pattern.
10246 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
10247 Allow register and subreg operands.
10249 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10251 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
10252 specific need, just fallback on defaults.
10253 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
10255 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10256 Olivier Hainque <hainque@adacore.com>
10258 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
10260 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
10261 targets. Pick a default if no particular attempt applied.
10262 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
10265 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10267 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
10268 (x86_64-wrs-vxworks7): Likewise.
10270 2017-06-27 Marek Polacek <polacek@redhat.com>
10273 * ubsan.c (instrument_null): Check get_base_address's result for null.
10275 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10277 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
10279 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10281 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
10282 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
10283 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
10284 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
10285 New function types.
10286 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
10287 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
10288 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
10289 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
10290 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
10291 BUILT_IN_FEUPDATEENV): New builtins.
10292 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
10293 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
10294 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
10295 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
10297 (builtin_structptr_types): Adjust size.
10298 * tree.c (builtin_structptr_types): Add four entries.
10300 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10301 Olivier Hainque <hainque@adacore.com>
10303 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
10304 (TLS_SYM): New local macro, forcing reference to __tls__ on
10305 link command lines for VxWorks 7 RTPs, triggering initialization
10307 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
10308 OS features TLS support, true for RTPs on VxWorks 7.
10309 * config/vxworks.c (vxworks_override_options): Setup emutls
10312 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10314 * predict.c (test_prediction_value_range): Use -1U instead of -1
10315 to avoid narrowing conversion warning.
10316 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
10317 to avoid narrowing conversion warning.
10318 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
10320 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
10322 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10324 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
10325 64bit configurations.
10326 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
10327 (SIZE_TYPE): Likewise.
10328 * config/vxworks.c (vxworks_emutls_var_fields): Use
10329 long_unsigned_type_node instead of unsigned_type_node as the offset
10330 field type, which is "pointer" mode in emutls.c.
10332 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10335 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
10337 PR middle-end/81207
10338 * gimple-fold.c (replace_call_with_call_and_fold): Handle
10339 gimple_vuse copying separately from gimple_vdef copying.
10341 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10343 * value-prof.c (free_hist): Remove call to memset and the enclosing if
10346 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10347 Olivier Hainque <hainque@adacore.com>
10349 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
10350 for all vxworks7 targets.
10351 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
10352 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
10353 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
10354 variations for VX6/VX7 and 32/64bits later on in ...
10355 (VXWORKS_LIB_SPEC): Leverage new macros.
10356 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
10357 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
10359 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10361 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
10362 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
10364 2017-06-26 Carl Love <cel@us.ibm.com>
10366 * config/rs6000/rs6000-c.c: Add support for built-in functions
10367 vector bool char vec_reve (vector bool char);
10368 vector signed char vec_reve (vector signed char);
10369 vector unsigned char vec_reve (vector unsigned char);
10370 vector bool int vec_reve (vector bool int);
10371 vector signed int vec_reve (vector signed int);
10372 vector unsigned int vec_reve (vector unsigned int);
10373 vector bool long long vec_reve (vector bool long long);
10374 vector signed long long vec_reve (vector signed long long);
10375 vector unsigned long long vec_reve (vector unsigned long long);
10376 vector bool short vec_reve (vector bool short);
10377 vector signed short vec_reve (vector signed short);
10378 vector double vec_reve (vector double);
10379 vector float vec_reve (vector float);
10380 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
10381 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
10382 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
10383 (altivec_vreve): New pattern.
10384 * config/rs6000/altivec.h (vec_reve): New define.
10385 * doc/extend.texi (vec_rev): Update the built-in documentation file
10386 for the new built-in functions.
10388 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10390 PR tree-optimization/71815
10391 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
10393 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
10395 (slsr_process_phi): Likewise.
10396 (replace_uncond_cands_and_profitable_phis): Don't replace a
10397 multiply candidate with a stride of 1 (copy or cast).
10398 (phi_incr_cost): Call uses_consumed_by_stmt rather than
10400 (lowest_cost_path): Likewise.
10401 (total_savings): Likewise.
10403 2017-06-26 Richard Biener <rguenther@suse.de>
10406 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
10407 Use def_builtin_pure for all gather builtins.
10409 2017-06-26 Richard Biener <rguenther@suse.de>
10411 PR tree-optimization/81203
10412 * tree-tailcall.c (find_tail_calls): Do not move stmts into
10413 non-dominating BBs.
10415 2017-06-26 Marek Polacek <polacek@redhat.com>
10418 * doc/invoke.texi: Document -Wmultistatement-macros.
10420 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
10422 * doc/sourcebuild.texi (ARM-specific attributes): Document new
10423 arm_neon_ok_no_float_abi effective target.
10425 2017-06-26 Richard Biener <rguenther@suse.de>
10427 PR tree-optimization/80928
10428 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
10429 (copy_bbs): Set BB_DUPLICATED flag early.
10430 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
10432 (execute_on_shrinking_pred): Likewise.
10433 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
10434 BB_DUPLICATED blocks.
10435 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
10436 iterate over all PHIs considering removal of *gsi.
10438 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
10440 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
10443 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10445 * config/rs6000/rs6000-string.c: (expand_block_clear,
10446 do_load_for_compare, select_block_compare_mode,
10447 compute_current_alignment, expand_block_compare,
10448 expand_strncmp_align_check, expand_strn_compare,
10449 expand_block_move, rs6000_output_load_multiple)
10450 Move functions related to string/block move/compare
10451 to a separate file.
10452 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
10453 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
10454 for this function which is now used in two files.
10455 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
10456 * config.gcc: Add rs6000-string.o to extra_objs for
10457 targets powerpc*-*-* and rs6000*-*-*.
10459 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
10462 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
10463 32-bit, since indexed is not valid for DImode.
10464 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
10465 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
10466 (define_peephole2 for Altivec d-form load): Add 32-bit support.
10467 (define_peephole2 for Altivec d-form store): Likewise.
10470 * multiple_target.c (create_dispatcher_calls): Only create the
10471 dispatcher call if the function is the default clone of a
10472 versioned function.
10474 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
10476 PR middle-end/80902
10477 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
10478 a call, force the call to not be a tail call.
10480 2017-06-23 Jeff Law <law@redhat.com>
10482 * doc/contrib.texi: Add entry for Steven Pemberton's work on
10485 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
10487 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
10488 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
10489 handling for early expansion of vector shifts (sl,sr,sra,rl).
10490 (builtin_function_type): Add vector shift right instructions
10491 to the unsigned argument list.
10493 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
10495 rtl-optimizatoin/79286
10496 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
10497 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
10498 trap. PIC register plus a const unspec without offset can never trap.
10500 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
10502 * tree.h (builtin_structptr_type): New type.
10503 (builtin_structptr_types): Declare new array.
10504 * tree.c (builtin_structptr_types): New array.
10505 (free_lang_data, build_common_tree_nodes): Use it.
10507 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
10510 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
10513 2017-06-22 Matt Turner <mattst88@gmail.com>
10515 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
10516 Lake models to skylake case. Assume skylake for unknown
10517 models with clflushopt.
10519 2017-06-22 Jeff Law <law@redhat.com>
10521 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
10522 frame sizes that do not satisfy aarch64_uimm12_shift.
10524 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
10526 * profile-count.h (apply_probability,
10527 apply_scale, probability_in): Fix checks for zero.
10529 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10531 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
10532 * doc/cppdiropts.texi (-I @var{dir}): Document it.
10534 2016-06-22 Richard Biener <rguenther@suse.de>
10536 * tree-vect-loop.c (vect_model_reduction_cost): Handle
10537 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
10538 REDUC_MAX_EXPR support.
10539 (vectorizable_reduction): Likewise.
10540 (vect_create_epilog_for_reduction): Likewise.
10542 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
10544 * match.pd (A / (1 << B) -> A >> B): New.
10545 * generic-match-head.c: Include optabs-tree.h.
10546 * gimple-match-head.c: Likewise.
10547 * optabs-tree.h (target_supports_op_p): New.
10548 * optabs-tree.c (target_supports_op_p): New.
10550 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10552 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
10553 $gcc_cv_ld --help output.
10554 (gcc_cv_ld_demangle): Likewise.
10555 (gcc_cv_ld_eh_frame_hdr): Likewise.
10556 (gcc_cv_ld_pie): Likewise.
10557 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
10558 (gcc_cv_ld_buildid): Likewise.
10559 (gcc_cv_ld_sysroot): Likewise.
10560 (ld_bndplt_support): Likewise.
10561 (ld_pushpopstate_support): Likewise.
10562 * configure: Regenerate.
10563 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
10565 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10568 * config/i386/sse.md (round<mode>2): Renumber match_dup and
10569 operands indexes to avoid gap between operands and match_dups.
10571 2017-06-21 Andrew Pinski <apinski@cavium.com>
10573 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
10574 Increment Arith_shift and Arith_shift_reg by 1.
10575 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
10577 * config/aarch64/aarch64.c (thunderx_tunings): Enable
10578 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
10579 (aarch64_strip_extend): Add new argument and test for it.
10580 (aarch64_cheap_mult_shift_p): New function.
10581 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
10582 add a cost if it is true.
10583 Update calls to aarch64_strip_extend.
10584 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
10586 2017-06-21 Andrew Pinski <apinski@cavium.com>
10588 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
10590 (thunderxt88): Likewise.
10591 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
10592 (thunderx_prefetch_tune): New variable.
10593 (thunderx2t99_prefetch_tune): Update for the correct values.
10594 (thunderxt88_tunings): New variable.
10595 (thunderx_tunings): Use thunderx_prefetch_tune instead of
10596 generic_prefetch_tune.
10597 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
10599 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10601 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
10602 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
10603 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
10604 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
10605 (aarch64_atomic_cas<mode>, GPI): Likewise.
10607 2017-06-21 Martin Liska <mliska@suse.cz>
10609 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
10610 statements on cold and hot labels.
10611 * predict.c (tree_estimate_probability_bb): Remove the
10612 prediction from this place.
10614 2017-06-21 Martin Liska <mliska@suse.cz>
10616 PR tree-optimization/79489
10617 * gimplify.c (maybe_add_early_return_predict_stmt): New
10619 (gimplify_return_expr): Call the function.
10620 * predict.c (tree_estimate_probability_bb): Remove handling
10622 * predict.def: Update comment about early return predictor.
10623 * gimple-predict.h (is_gimple_predict): New function.
10624 * predict.def: Change default value of early return to 66.
10625 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
10627 * passes.def: Put pass_strip_predict_hints to the beginning of
10630 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
10632 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
10633 FUNCTION_DECL declarations.
10634 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
10636 (dwaf2out_decl): Likewise.
10637 * godump.c (go_early_global_decl): Skip call to the real debug hook
10638 for FUNCTION_DECL declarations.
10639 * passes.c (rest_of_decl_compilation): Skip call to the
10640 early_global_decl debug hook for FUNCTION_DECL declarations, unless
10641 -fdump-go-spec is passed.
10643 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10645 * config/i386/i386.c (struct builtin_isa): New field pure_p.
10646 Reorder for compactness.
10647 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
10648 (def_builtin_pure, def_builtin_pure2): New functions.
10649 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
10651 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10653 * match.pd (nop_convert): New predicate.
10654 ((A +- CST1) +- CST2): Allow some NOP conversions.
10656 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10659 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
10660 with ctors/dtors if GOVD_SHARED is set.
10662 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10664 * config/aarch64/aarch64.md (movti_aarch64):
10665 Emit mov rather than orr.
10666 (movtf_aarch64): Likewise.
10667 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
10668 Emit mov rather than orr.
10670 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10672 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
10673 Swap alternatives, make integer dup more expensive.
10675 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10677 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
10678 Return true for non-tls symbols.
10680 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
10682 * config/aarch64/aarch64-cores.def (cortex-a55): New.
10683 (cortex-a75): Likewise.
10684 (cortex-a75.cortex-a55): Likewise.
10685 * config/aarch64/aarch64-tune.md: Regenerate.
10686 * doc/invoke.texi (-mtune): Document new values for -mtune.
10688 2017-06-21 Tom de Vries <tom@codesourcery.com>
10690 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
10691 stack_size feature.
10692 (Effective-Target Keywords, Other attributes): Suggest using
10693 dg-add-options stack_size feature to get stack limit in stack_size
10694 effective target documentation.
10696 2017-06-21 Julian Brown <julian@codesourcery.com>
10697 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10699 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
10700 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
10701 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
10703 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
10704 attribute type list for neon_multiply.
10705 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
10706 attribute type list for neon_multiply.
10707 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
10708 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
10709 attribute type list for neon_multiply.
10710 * config/arm/types.md (crypto_pmull): Add.
10711 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
10712 attribute type list.
10714 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
10716 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
10719 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10721 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
10722 to make sure not to dereference a NULL cost_classes_ptr pointer.
10724 2017-06-20 Carl Love <cel@us.ibm.com>
10726 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10727 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10728 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10729 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10730 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10731 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10732 VMULOSW): New enum "unspec" values.
10733 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10734 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10735 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10736 altivec_vmulosw): New patterns.
10737 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10738 VMULOSW): Add definitions.
10740 2017-06-20 Julia Koval <julia.koval@intel.com>
10742 * config/i386/i386.c: Fix rounding expand for new pattern.
10743 * config/i386/subst.md: Fix pattern (parallel -> unspec).
10745 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10747 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
10748 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
10750 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10752 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
10755 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10757 * config/aarch64/aarch64-cores.def: Rearrange to sort by
10758 architecture, then by implementer ID.
10759 * config/aarch64/aarch64-tune.md: Regenerate.
10761 2017-06-20 Richard Biener <rguenther@suse.de>
10763 PR middle-end/81097
10764 * fold-const.c (split_tree): Fold to type before negating.
10766 2017-06-20 David Malcolm <dmalcolm@redhat.com>
10768 * diagnostic-show-locus.c
10769 (selftest::test_fixit_deletion_affecting_newline): New function.
10770 (selftest::diagnostic_show_locus_c_tests): Call it.
10772 2017-06-20 Andreas Schwab <schwab@suse.de>
10775 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
10778 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10780 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
10781 __ARM_FEATURE_COPROC according to support.
10783 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10785 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10786 Rewritten to avoid overflow for > 32-bit pointers.
10789 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
10790 by removing enum keyword.
10791 (ubsan_type_descriptor): Likewise. Formatting fix.
10794 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
10795 splitter): Require TARGET_SSE2 in the condition.
10797 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
10800 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
10801 for doing vector set of SFmode on ISA 3.0.
10802 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
10803 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
10805 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
10806 SFmode value into a V4SF variable that was extracted from another
10807 V4SF variable without converting the element to double precision
10808 and back to single precision vector format.
10809 (vsx_insert_extract_v4sf_p9_2): Likewise.
10811 2017-06-19 Jakub Jelinek <jakub@redhat.com>
10813 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
10814 in UWHI to avoid undefined overflow.
10817 * ubsan.h (enum ubsan_encode_value_phase): New.
10818 (ubsan_encode_value): Change second argument to
10819 enum ubsan_encode_value_phase with default value of
10820 UBSAN_ENCODE_VALUE_GENERIC.
10821 * ubsan.c (ubsan_encode_value): Change second argument to
10822 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
10823 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
10824 create_tmp_var_raw instead of create_tmp_var and use a
10826 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
10827 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
10828 ubsan_encode_value callers.
10831 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
10832 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
10833 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
10835 2017-06-19 Richard Biener <rguenther@suse.de>
10837 PR middle-end/81118
10838 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
10839 estimates if we changed anything.
10841 2017-06-19 Richard Biener <rguenther@suse.de>
10843 PR tree-optimization/80887
10844 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
10845 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
10846 simplified lookups, then reset mprts_hook.
10847 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
10849 (try_to_simplify): Likewise.
10851 2017-06-19 Martin Liska <mliska@suse.cz>
10854 * gimplify.c (gimplify_switch_expr):
10855 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
10857 2017-06-19 Martin Liska <mliska@suse.cz>
10859 * doc/install.texi: Document that PGO runs in 4 stages.
10861 2017-06-19 Martin Liska <mliska@suse.cz>
10864 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
10865 to dispatcher function name.
10866 * multiple_target.c (replace_function_decl): New function.
10867 (create_dispatcher_calls): Redirect both edges and references.
10869 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
10871 * profile-count.c (profile_count::dump): Dump quality.
10872 (profile_count::differs_from_p): Update for unsigned val.
10873 * profile-count.h (profile_count_quality): New enum.
10874 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
10876 2017-06-19 Richard Biener <rguenther@suse.de>
10878 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
10879 struct function as arg.
10880 (estimate_numbers_of_iterations): Export overload with loop arg.
10881 (free_numbers_of_iterations_estimates_loop): Use an overload of
10882 free_numbers_of_iterations_estimates instead.
10883 * tree-cfg.c (remove_bb): Adjust.
10884 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
10885 * tree-parloops.c (gen_parallel_loop): Likewise.
10886 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
10888 (tree_unroll_loops_completely): Likewise.
10889 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
10890 Use an overload instead and export.
10891 (estimated_loop_iterations): Adjust.
10892 (max_loop_iterations): Likewise.
10893 (likely_max_loop_iterations): Likewise.
10894 (estimate_numbers_of_iterations): Take struct function as arg
10896 (loop_exits_before_overflow): Adjust.
10897 (free_numbers_of_iterations_estimates_loop): Use an overload.
10898 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
10899 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
10901 2017-06-19 Richard Biener <rguenther@suse.de>
10904 * ipa-prop.c (find_constructor_constant_at_offset): Handle
10905 RANGE_EXPR conservatively.
10907 2017-06-16 Carl Love <cel@us.ibm.com>
10909 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10910 definitions for vec_float, vec_float2, vec_floato,
10911 vec_floate built-ins.
10912 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
10913 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
10915 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
10916 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
10917 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
10918 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
10919 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
10920 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
10921 vec_floato): Add builtin defines.
10922 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
10923 Update the built-in documentation file for the new built-in
10926 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10928 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
10929 (mthumb): Mark as the negative of -marm.
10931 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10933 * doc/invoke.texi (ARM Options, -mcpu): Document supported
10935 (ARM Options, -mtune): Document that this accepts the same
10936 extension options as -mcpu.
10937 (ARM Options, -mfpu): Document addition of -mfpu=auto.
10939 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10941 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
10942 permitted extensions.
10944 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10946 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
10947 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
10948 (armv8-m.main): Add option +nodsp.
10949 * config/arm/arm-cpu-cdata.h: Regenerated.
10951 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10953 * config/arm/t-fuchsia: New file.
10954 * config.gcc (arm*-*-fuchsia*): Use it.
10956 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10958 * config/arm/t-symbian: Rewrite for new option infrastructure.
10960 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10962 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
10963 (MULTILIB_REQUIRED): Likewise.
10965 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10967 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
10968 (MULTILIB_RESUE): Likewise.
10969 (MULTILIB_MATCHES): Likewise.
10970 (MULTLIB_REQUIRED): Likewise.
10972 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10974 * config/arm/t-rtems: Rewrite for new option framework.
10976 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10978 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
10979 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
10980 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
10981 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
10982 * config/arm/t-multilib: ... here.
10983 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
10984 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
10985 armv7-a and armv8*-a when A-profile libraries have not been built.
10986 * config/arm/t-rmprofile: Rewrite.
10988 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10990 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
10991 with a backslash. Remove the backslash after substituting unescaped
10993 * doc/fragments.texi (MULTILIB_REUSE): Document it.
10995 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10997 * config.gcc: (arm*-*-*): When building a-profile libraries, force
10998 the driver to pass through the default setting of -mfloat-abi.
10999 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
11001 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
11002 (all_feat_combs): New rule.
11003 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
11005 * config/arm/t-aprofile: Rewrite.
11007 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11009 * config/arm/arm.h (FPUTYPE_AUTO): Define.
11010 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
11011 fpu is not specified by the user/command-line.
11012 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
11013 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
11014 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
11015 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
11016 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
11017 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
11019 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11021 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
11022 * config/arm/t-arm-elf: Rewritten.
11024 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11026 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
11027 have some floating-point instructions.
11028 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
11029 (TARGET_MAYBE_HARD_FLOAT): New macro.
11030 * config/arm/arm-builtins.c (arm_init_builtins): Use
11031 TARGET_MAYBE_HARD_FLOAT.
11032 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
11034 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11036 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
11037 (configargs.h): Include it.
11038 (arm_print_hint_for_fpu_option): New function.
11039 (arm_parse_fpu_option): New function.
11040 (candidate_extension): New class.
11041 (arm_canon_for_multilib): New function.
11042 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
11043 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
11044 (ARCH_CANONICAL_SPECS): New macro.
11045 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
11047 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11049 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
11050 are set after handling multilib fragments. Set target_cpu_default2
11053 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11055 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
11057 (arm*-*-*): Set target_cpu_default2 to a quoted string.
11058 * config/arm/parsecpu.awk (check_cpu): Validate any extension
11060 (check_arch): Likewise.
11061 * config/arm/arm.c (arm_configure_build_target): Handle
11062 TARGET_CPU_DEFAULT being a string constant. Scan any feature
11063 options in the default.
11065 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11067 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
11068 when an option is an alias of another.
11069 * config/arm/parsecpu.awk (optalias): New parser token.
11070 (gen_comm_data): Mark non-alias options as such. Emit entries
11071 for extension aliases.
11072 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
11073 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
11074 (armv6kz, armv6zk, armv6t2): Likewise.
11075 (armv7): Make vfpv3-d16 an alias.
11076 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
11078 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
11079 Sort in canonical order.
11080 (armv8-a): Sort in canonical order.
11081 (armv8.1-a, armv8.2-a): Likewise.
11082 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
11084 (cortex-a9): Sort in canonical order.
11085 * config/arm/arm.c (selftests.h): Include it.
11086 (arm_test_cpu_arch_data): New function.
11087 (arm_run_self_tests): New function.
11088 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
11089 (targetm): Move declaration to the end of the file.
11090 * arm-cpu-cdata.h: Regenerated.
11092 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11094 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
11095 call to target_mode_check describing the type of option passed.
11096 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
11097 (arm_target_thumb_only): Use arm_parse_arch_option_name or
11098 arm_parse_cpu_option_name to match parameters against list of
11100 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
11101 arm_arch_core_flags data structure.
11102 * config/arm/arm-cpu_cdata.h: Regenerated.
11104 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11106 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
11108 (arm_print_hint_for_cpu_option): Likewise.
11109 (arm_print_hint_for_arch_option): Likewise.
11110 (arm_parse_cpu_option_name): Likewise.
11111 (arm_parse_arch_option_name): Likewise.
11112 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
11113 of entries in the all_fpus list.
11114 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
11115 (arm_parse_cpu_option_name): Declare.
11116 (arm_parse_arch_option_name): Declare.
11117 (arm_parse_option_features): Declare.
11118 (arm_intialize_isa): Declare.
11119 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
11121 (gen_comm_data): ... here. Make definitions non-static.
11122 * config/arm/arm-cpu-data.h: Regenerated.
11123 * config/arm/arm-cpu-cdata.h: Regenerated.
11125 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11127 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
11128 (cpu_arch_extension): New structure.
11129 (cpu_arch_option, arch_option, cpu_option): New structures.
11130 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
11131 architecture types.
11132 (gen_data): Generate new format data tables.
11133 * config/arm/arm.c (cpu_tune): New structure.
11134 (cpu_option, processors): Delete.
11135 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
11136 (arm_print_hint_for_cpu_option): ... this and ...
11137 (arm_print_hint_for_arch_option): ... this.
11138 (arm_parse_arch_cpu_name): Delete. Replace with ...
11139 (arm_parse_cpu_option_name): ... this and ...
11140 (arm_parse_arch_option_name): ... this.
11141 (arm_unrecognized_feature): Change type of target parameter to
11143 (arm_parse_arch_cpu_features): Delete. Replace with ...
11144 (arm_parse_option_features): ... this.
11145 (arm_configure_build_target): Rework to use new configuration data
11147 (arm_print_tune_info): Rework for new configuration data tables.
11148 * config/arm/arm-cpu-data.h: Regenerated.
11149 * config/arm/arm-cpu.h: Regenerated.
11151 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11153 * Makefile.in (OBJS): Move sbitmap.o from here ...
11154 (OBJS-libcommon): ... to here.
11156 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11158 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
11159 (ISA_ALL_CRYPTO): New macro.
11160 (ISA_ALL_SIMD): New macro
11161 (ISA_ALL_FP): New macro.
11162 * config/arm/arm.c (fpu_bitlist): Update initializer.
11163 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
11165 (arm9e): Add fpu. Add option for nofp
11166 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
11167 (arm926ej-s, arm1026ej-s): Likewise.
11168 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
11169 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
11170 neon-fp16, neon-vfpv4, nofp and nosimd.
11171 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
11172 (cortex-a8): Add fpu. Add option for nofp.
11173 (cortex-a9): Add fpu. Add options for nosimd and nofp.
11174 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
11175 (cortex-r4f): Add fpu.
11176 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
11177 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
11179 (cortex-r8): Likewise.
11180 (cortex-m4): Add fpu. Add option for nofp.
11181 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
11182 (cortex-a17.cortex-a7): Likewise.
11183 (cortex-a32): Add fpu. Add options for crypto and nofp.
11184 (cortex-a35, cortex-a53): Likewise.
11185 (cortex-a57): Add fpu. Add option for crypto.
11186 (cortex-a72, cortex-a73): Likewise.
11187 (exynos-m1): Likewise.
11188 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
11189 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
11190 (cortex-m33): Add fpu. Add option for nofp.
11191 * config/arm/arm-cpu-cdata.h: Regenerated
11192 * config/arm/arm-cpu-data.h: Regenerated.
11194 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11196 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
11197 (armv5te, armv5tej): Likewise.
11198 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
11199 (armv7): Add options fp and vfpv3-d16.
11200 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
11201 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
11203 (armv7ve): Likewise.
11204 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
11205 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
11206 (armv8-a): Add nocrypto option.
11207 (armv8.1-a, armv8.2-a): Likewise.
11208 (armv8-m.main): add options fp, fp.dp and nofp.
11210 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11212 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
11214 (armv8-a+crc): Delete.
11215 (armv8.1-a): Add options simd, crypto and nofp.
11216 (armv8.2-a): Add options fp16, simd, crypto and nofp.
11217 (armv8.2-a+fp16): Delete.
11218 (armv8-m.main): Add option dsp.
11219 (armv8-m.main+dsp): Delete.
11220 (cortex-a8): Add fpu. Add nofp option.
11221 (cortex-a9): Add fpu. Add nofp and nosimd options.
11222 * config/arm/parsecpu.awk (gen_data): Generate option tables and
11223 link to main cpu and architecture data structures.
11224 (gen_comm_data): Only put isa attributes from the main architecture
11226 (option): New statement for architecture and CPU entries.
11227 * arm.c (struct cpu_option): New structure.
11228 (struct processors): Add entry for options.
11229 (arm_unrecognized_feature): New function.
11230 (arm_parse_arch_cpu_name): Ignore any characters after the first
11232 (arm_parse_arch_cpu_feature): New function.
11233 (arm_configure_build_target): Separate out any CPU and architecture
11234 features and parse separately. Don't error out if -mfpu=auto is
11235 used with only an architecture string.
11236 (arm_print_asm_arch_directives): New function.
11237 (arm_file_start): Call it.
11238 * config/arm/arm-cpu-cdata.h: Regenerated.
11239 * config/arm/arm-cpu-data.h: Likewise.
11240 * config/arm/arm-tables.opt: Likewise.
11242 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11244 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
11245 assembler when it is not -mfpu=auto.
11247 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11249 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
11250 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
11251 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
11252 (ASM_CPU_SPEC): Rewrite.
11253 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
11254 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
11255 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
11256 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
11257 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
11258 copied string is NUL-terminated. Also strip any characters prefixed
11260 (arm_rewrite_selected_arch): New function.
11261 (arm_rewrite_march): New function.
11263 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11265 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
11266 (x_arm_cpu_string, x_arm_tune_string): Likewise.
11267 (march, mcpu, mtune): Convert to string-based options.
11268 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
11269 (arm_parse_arch_cpu_name): New function.
11270 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
11271 identify selected architecture or CPU.
11272 (arm_option_save): New function.
11273 (TARGET_OPTION_SAVE): Redefine.
11274 (arm_option_restore): Restore string options.
11275 (arm_option_print): Print string options.
11277 2017-06-16 Martin Sebor <msebor@redhat.com>
11279 PR tree-optimization/80933
11280 PR tree-optimization/80934
11281 * builtins.c (fold_builtin_3): Do not handle bcmp here.
11282 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
11283 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
11284 (gimple_fold_builtin): Call them.
11286 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11288 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
11289 as unlikely; update profile.
11291 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11293 * predict.c (force_edge_cold): Handle declaring edges impossible
11296 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11298 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
11300 (try_unroll_loop_completely): Fix reporting.
11302 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11304 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
11306 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
11309 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
11310 if given a non-constant argument for an intrinsic which requires a
11313 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11315 * profile.c (compare_freqs): New function.
11316 (branch_prob): Sort edge list.
11317 (find_spanning_tree): Assume that the list is priority sorted.
11319 2017-06-16 Richard Biener <rguenther@suse.de>
11321 PR tree-optimization/81090
11322 * passes.def (pass_record_bounds): Remove.
11323 * tree-pass.h (make_pass_record_bounds): Likewise.
11324 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
11325 make_pass_record_bounds): Likewise.
11326 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
11327 not free niter estimates at the beginning but at the end.
11328 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
11330 2017-06-16 Richard Biener <rguenther@suse.de>
11332 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
11333 initializer to workaround ICE in host GCC 4.8.
11335 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11337 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
11339 (clone_inlined_nodes): Update.
11341 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11343 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
11344 prefetch settings, and enable prefetching by default at -O3.
11346 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11348 * config/aarch64/aarch64.c (aarch64_override_options_internal):
11349 Set flag_prefetch_loop_arrays according to tuning data.
11351 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11353 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
11354 New tune structure.
11355 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
11356 [Unrelated to main purpose of the patch] Place the pointer field last
11357 to enable type checking errors when tune structure are wrongly merged.
11358 * config/aarch64/aarch64.c (generic_prefetch_tune,)
11359 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
11360 (thunderx2t99_prefetch_tune): New tune constants.
11361 (tune_params *_tunings): Update all tunings (no functional change).
11362 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
11363 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
11364 from tunings structures.
11366 2017-06-16 Jakub Jelinek <jakub@redhat.com>
11369 * ubsan.c (instrument_null): Add T argument, use it instead
11370 of computing it based on IS_LHS.
11371 (instrument_object_size): Likewise.
11372 (pass_ubsan::execute): Adjust instrument_null and
11373 instrument_object_size callers to pass gimple_get_lhs or
11374 gimple_assign_rhs1 result to it. Use instrument_null instead of
11375 calling get_base_address and instrument_mem_ref. Handle
11376 aggregate call arguments for object-size sanitization.
11378 2017-06-16 Yury Gribov <tetra2005@gmail.com>
11380 PR tree-optimization/81089
11381 * tree-vrp.c (is_masked_range_test): Validate operands of
11384 2017-06-15 Martin Sebor <msebor@redhat.com>
11387 * dumpfile.c (dump_register): Avoid calling memset to initialize
11388 a class with a default ctor.
11389 * gcc.c (struct compiler): Remove const qualification.
11390 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
11391 * hash-table.h: Ditto.
11392 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
11394 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
11395 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
11397 * params.h (struct param_info): Make struct members non-const.
11398 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
11399 with default initialization.
11400 * vec.h (vec_copy_construct, vec_default_construct): New helper
11402 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
11403 with vec_copy_construct.
11404 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
11405 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
11406 * doc/invoke.texi (-Wclass-memaccess): Document.
11408 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11410 * emit-rtl.h (is_leaf): Update comment about local
11411 register allocator.
11413 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11416 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
11417 for a variable to have a section before checking if the section has a
11419 Set section to.persistent if persistent attribute is set.
11420 Warn if .persistent attribute is used on an automatic variable.
11422 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
11424 PR rtl-optimization/80474
11425 * reorg.c (update_block): Do not ignore instructions in a delay slot.
11427 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
11429 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
11432 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
11434 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
11435 (casesi): Emit bounds checking as RTL.
11436 (casesi_internal_mips16_<mode>): Remove bounds checking.
11438 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
11440 * config/xtensa/xtensa.c (xtensa_option_override): Append
11441 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
11442 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
11443 xtensa_doloop_hooks): Define unconditionally.
11444 (xtensa_reorg_loops): Only call reorg_loops in the presence of
11446 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
11447 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
11448 for it in xtensa_option_override.
11449 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
11450 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
11452 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
11454 * doc/cppopts.texi: Document '-' special value to -MF.
11456 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
11458 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
11459 (cortex_a53_fconst): Likewise.
11460 (cortex_a53_fpmul): Likewise.
11461 (cortex_a53_f_load_64): Likewise.
11462 (cortex_a53_f_load_many): Likewise.
11463 (cortex_a53_advsimd_alu): Likewise.
11464 (cortex_a53_advsimd_alu_q): Likewise.
11465 (cortex_a53_advsimd_mul): Likewise.
11466 (cortex_a53_advsimd_mul_q): Likewise.
11467 (fpmac bypass): Add new bypass for fpmac-fpmac case.
11468 Add missing fmul, r2f_cvt and fconst cases.
11470 2017-06-14 Richard Biener <rguenther@suse.de>
11472 PR middle-end/81088
11473 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
11475 (fold_binary_loc): When associating do not treat pre-existing
11476 TREE_OVERFLOW on literal constants as a reason to allow
11477 TREE_OVERFLOW on associated literal constants.
11479 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11481 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
11482 (MASK_FEATURES): New macro.
11483 * config/sparc/sparc.c (sparc_option_override): Remove the special
11484 handling of -mfpu and generalize it to all MASK_FEATURES switches.
11486 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11488 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
11489 a division of 0 if non-call exceptions are enabled.
11491 2017-06-14 Andrew Pinski <apinski@cavium.com>
11492 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11495 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
11496 Improve vector initialization code gen for only variable case.
11498 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11500 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
11502 2017-06-14 Richard Biener <rguenther@suse.de>
11504 PR tree-optimization/81083
11505 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
11508 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11510 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
11511 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
11512 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
11513 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
11514 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
11515 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
11517 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11519 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
11520 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
11522 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11524 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
11526 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11528 * config/rs6000/t-rtems: Don't handle SPE.
11530 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11532 * config/rs6000/t-linux: Don't handle SPE.
11534 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11536 * config/rs6000/eabispe.h: Delete file.
11538 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11540 * config/rs6000/t-spe: Delete file.
11542 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11544 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
11545 (rs6000_legitimate_offset_address_p): Return false for anything in
11546 V2SImode or V2SFmode.
11548 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11550 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
11551 except V2SF and V2SI. Rearrange the vector modes, and add comments.
11552 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
11554 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
11555 (rs6000_legitimate_offset_address_p): Ditto.
11556 (rs6000_emit_move): Ditto.
11557 (rs6000_init_builtins): Remove V4HI_type_node.
11559 2017-06-13 Martin Liska <mliska@suse.cz>
11562 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
11563 (gate_asan): Likewise.
11564 * asan.h (asan_no_sanitize_address_p): Remove the function.
11565 (sanitize_flags_p): New function.
11566 * builtins.def: Fix coding style.
11567 * common.opt: Use renamed enum value.
11568 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
11569 * doc/extend.texi: Document no_sanitize attribute.
11570 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
11571 to SANITIZE_UNDEFINED_NONDEFAULT.
11572 * gcc.c (sanitize_spec_function): Use the renamed enum value.
11573 * gimple-fold.c (optimize_atomic_compare_exchange_p):
11574 Use sanitize_flags_p.
11575 * gimplify.c (gimplify_function_tree): Likewise.
11576 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
11577 * opts.c (parse_no_sanitize_attribute): New function.
11578 (common_handle_option): Use renamed enum value.
11579 * opts.h (parse_no_sanitize_attribute): Declare.
11580 * tree.c (sanitize_flags_p): New function.
11581 * tree.h: Declared here.
11582 * tsan.c: Use sanitize_flags_p.
11583 * ubsan.c (ubsan_expand_null_ifn): Likewise.
11584 (instrument_mem_ref): Likewise.
11585 (instrument_bool_enum_load): Likewise.
11586 (do_ubsan_in_current_function): Remove the function.
11587 (pass_ubsan::execute): Use sanitize_flags_p.
11588 * ubsan.h: Remove do_ubsan_in_current_function
11589 * tree-cfg.c (print_no_sanitize_attr_value): New function.
11590 (dump_function_to_file): Use it here.
11592 2017-06-13 Martin Jambor <mjambor@suse.cz>
11594 PR tree-optimization/80803
11595 PR tree-optimization/81063
11596 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
11597 (propagate_subaccesses_across_link): Enqueue subtree whenever
11598 necessary instead of relying on the caller.
11600 2017-06-13 Martin Jambor <mjambor@suse.cz>
11602 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
11603 that have a first_link.
11604 (sort_and_splice_var_accesses): Do not check first_link before
11606 (subtree_mark_written_and_enqueue): Likewise.
11607 (propagate_all_subaccesses): Likewise and do not stop at first
11608 parent with a first_link.
11610 2017-06-13 Martin Jambor <mjambor@suse.cz>
11612 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
11615 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11617 * match.pd: New pattern.
11619 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11621 * tree-vrp.c (is_masked_range_test): New function.
11622 (register_edge_assert_for): Determine ranges for
11625 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11627 PR tree-optimization/67328
11628 * fold-const.c (maskable_range_p): New function.
11629 (build_range_check): Generate bittests if possible.
11631 2017-06-13 Martin Liska <mliska@suse.cz>
11633 * gimple-pretty-print.c (dump_probability): Add new argument.
11634 (dump_edge_probability): Dump both probability and count.
11635 (dump_gimple_label): Likewise.
11636 (dump_gimple_bb_header): Likewise.
11638 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
11641 * config/avr/avr-devices.c: Fix indentation.
11642 * config/avr/gen-avr-mmcu-specs.c: Dito.
11644 2017-06-13 Richard Biener <rguenther@suse.de>
11646 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
11647 instead get vector type from stmt_info.
11648 (vectorizable_reduction): Adjust. Remove dead code.
11650 2017-06-13 Richard Biener <rguenther@suse.de>
11652 PR middle-end/81065
11653 * fold-const.c (extract_muldiv_1): Remove bogus distribution
11654 case of C * (x * C2 + C3).
11655 (fold_addr_of_array_ref_difference): Properly fold index difference.
11657 2017-06-12 David S. Miller <davem@davemloft.net>
11660 * config/sparc/sparc.md (return expander): Emit frame blockage if
11661 function uses alloca.
11663 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
11665 * combine.c (make_field_assignment): Check len rather than the mode
11666 precision when calling force_to_mode.
11668 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
11670 Support multilibs and devices that see flash in RAM address range.
11673 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
11674 (avr_mcu_t) <flash_pm_offset>: New field.
11675 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
11676 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
11677 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
11678 (AVR_TINY_PM_OFFSET): Remove macro.
11679 * config/avr/avr.opt (-mshort-calls): New option.
11680 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11681 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
11682 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
11683 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
11684 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
11685 instead of avr_arch->have_jmp_call.
11686 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
11687 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
11688 avr_arch->flash_pm_offset to define.
11689 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
11690 new field flash_pm_offset. Add entry for avrxmega3.
11691 (avr_texinfo): Add entry for avrxmega3.
11692 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
11693 attiny212, attiny214,
11694 attiny412, attiny414, attiny416, attiny417,
11695 attiny814, attiny816, attiny817,
11696 attiny1614, attiny1616, attiny1617,
11697 attiny3214, attiny3216, attiny3217.
11698 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
11699 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
11700 (avr_print_operand_address) [AVR_TINY]: Same.
11701 (avr_asm_init_sections) <readonly_data_section>: Only patch
11702 callback if avr_arch->flash_pm_offset = 0.
11703 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
11704 for rodata if avr_arch->flash_pm_offset != 0.
11705 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
11706 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
11707 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
11708 (m_options): Append opt_rcall.
11709 (m_dirnames): Append dir_rcall.
11710 * config/avr/t-multilib: Regenerate.
11712 * configure.ac [target=avr]: Check whether avrxmega3 default
11713 linker description file works as needed.
11714 * configure: Regenerate.
11715 * doc/avr-mmcu.texi: Regenerate.
11716 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
11717 <__AVR_ARCH__>: Document avrxmega3 and 103.
11718 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
11719 <__AVR_SHORT_CALLS__>: Document it.
11720 <__AVR_PM_BASE_ADDRESS__>: Document it.
11721 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
11722 (AVR Variable Attributes) <progmem>: Document this is
11723 not needed for avrxmega3.
11724 (AVR Named Address Spaces) <__flash>: Dito.
11726 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
11728 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
11730 2017-06-12 Doug Rupp <rupp@adacore.com>
11732 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
11733 Append vxworks-stdint.h to the tm_file list.
11734 * config/vxworks-stdint.h: New file.
11736 2017-06-12 Martin Liska <mliska@suse.cz>
11738 PR tree-optimization/81041
11739 * tree-profile.c (gimple_gen_ic_func_profiler):
11740 Create an extra BB in profile-generate
11741 (gimple_gen_time_profiler): Likewise.
11743 2017-06-12 Jakub Jelinek <jakub@redhat.com>
11745 PR tree-optimization/81003
11746 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
11747 (update_range_test): Use it instead of force_gimple_operand_gsi.
11749 2017-06-12 Richard Biener <rguenther@suse.de>
11751 PR tree-optimization/81053
11752 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
11753 with backedge value not defined in loop. Simplify def stmt
11756 2017-06-11 Tom de Vries <tom@codesourcery.com>
11759 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
11761 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
11762 nvptx_cannot_force_const_mem.
11764 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11766 * opts.c (finish_options): Move test for flag_split_stack after
11767 it has been initialized.
11769 2017-06-11 Jason Merrill <jason@redhat.com>
11771 * tree.h (id_equal): New.
11772 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
11773 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
11774 instead of strcmp of IDENTIFIER_POINTER.
11776 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11778 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
11779 (mark_all_inlined_calls_cdtor): Fix formating.
11780 (inline_transform): Rescale profile before inlining.
11782 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11784 * cgraph.h (cgraph_edge::clone): Update prototype.
11785 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
11786 (cgraph_node::create_clone): Update.
11787 (cgraph_node::create_version_clone): Update.
11788 * tree-inline.c (copy_bb): Update.
11789 (expand_call_inline): Update.
11791 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11793 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
11794 factored out from ...
11795 (rs6000_emit_prologue): ... here.
11797 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11799 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
11800 factored out from ...
11801 (rs6000_emit_prologue): ... here.
11803 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11805 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
11807 (handle_missing_profiles): Fix computation of tp_first_run.
11808 (counts_to_freqs): Do not touch freqs when count is 0.
11810 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11812 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
11815 2017-06-10 Tom de Vries <tom@codesourcery.com>
11817 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
11818 attributes): Document signal effective target.
11820 2017-06-10 Tom de Vries <tom@codesourcery.com>
11822 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
11823 Document effective target stack_size.
11825 2017-06-09 David Malcolm <dmalcolm@redhat.com>
11827 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
11828 to the edit_context if they can be auto-applied.
11830 2017-06-9 Ian Lance Taylor <iant@golang.org>
11832 * opts.c (finish_options): If -fsplit-stack, disable implicit
11833 -forder-blocks-and-partition.
11834 * doc/invoke.texi (Optimize Options): Document that when using
11835 -fsplit-stack -forder-blocks-and-partition is not implicitly
11838 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11840 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
11841 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
11842 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
11843 * builtins.def (abort, trap, unreachable): Declare cold.
11844 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
11845 * tree-core.h (ECF_COLD): New.
11846 * tree.c (set_call_expr_flags): Handle ECF_COLD.
11847 (build_common_builtin_nodes): Mark unreachable and abort as cold.
11849 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11851 * predict.c (unlikely_executed_stmt_p): Cleanup.
11853 2017-06-09 Richard Biener <rguenther@suse.de>
11855 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
11856 model if the ref is always written to.
11858 2017-06-09 Tamar Christina <tamar.christina@arm.com>
11860 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
11862 2017-06-09 Tamar Christina <tamar.christina@arm.com>
11864 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
11867 2017-06-09 Tom de Vries <tom@codesourcery.com>
11870 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
11871 "target cannot support label values" when encountering LABEL_REF.
11873 2017-06-09 Martin Liska <mliska@suse.cz>
11875 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
11876 (gimple_gen_ic_func_profiler): Emit direct comparison
11877 of __gcov_indirect_call_callee with NULL.
11878 (gimple_gen_time_profiler): Change probability from
11879 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
11881 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11883 * profile.c (edge_gcov_counts): Turn to pointer.
11884 (compute_branch_probabilities, compute_branch_probabilities): Update.
11885 (branch_prob): Do not clear edge_gcov_count.
11886 * profile.h (edge_gcov_counts): Turn to pointer.
11887 (edge_gcov_count): Update.
11889 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11891 * gimple.h (gimple_check_failed): Mark cold.
11893 2017-06-09 Richard Biener <rguenther@suse.de>
11895 PR tree-optimization/66623
11896 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
11897 refactor check_reduction into two parts, properly computing
11898 whether we have to check reduction validity for outer loop
11901 2017-06-09 Richard Biener <rguenther@suse.de>
11903 PR tree-optimization/79483
11904 * graphite-scop-detection.c (order): New global.
11905 (get_order): Compute bb to order mapping that satisfies code
11906 generation constraints.
11907 (cmp_pbbs): New helper.
11908 (build_scops): Start domwalk at entry block, sort generated
11911 2017-06-09 Richard Biener <rguenther@suse.de>
11913 PR middle-end/81007
11914 * ipa-polymorphic-call.c
11915 (ipa_polymorphic_call_context::restrict_to_inner_class):
11916 Skip FIELD_DECLs with error_mark_node type.
11917 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
11920 2017-06-09 Martin Liska <mliska@suse.cz>
11922 * predict.c (struct branch_predictor): New struct.
11923 (test_prediction_value_range): New test.
11924 (predict_c_tests): New function.
11925 * selftest-run-tests.c (selftest::run_tests): Run the function.
11926 * selftest.h: Declare new tests.
11928 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
11931 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
11932 gen_add3_insn did not fail.
11933 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
11934 r0, construct that number in a temporary reg and add that reg to r0.
11935 If asked to put the result in r0 as well, fail.
11937 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
11939 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11940 for early expansion of vec_eqv.
11942 2017-06-08 Jakub Jelinek <jakub@redhat.com>
11944 PR middle-end/81005
11945 * ubsan.c (instrument_null): Avoid pointless code temporary.
11946 (pass_ubsan::execute): Instrument aggregate arguments of calls.
11948 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
11952 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
11955 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
11958 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11960 * predict.c (unlikely_executed_edge_p): Move ahead.
11961 (probably_never_executed_edge_p): Use it.
11963 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
11965 PR middle-end/79988
11966 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
11967 gimple_call_builtin_p call.
11969 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11971 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
11972 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
11973 rtl_check_failed_type2, rtl_check_failed_code1,
11974 rtl_check_failed_code2, rtl_check_failed_code_mode,
11975 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
11976 rtvec_check_failed_bounds, rtl_check_failed_flag,
11977 _fatal_insn_not_found, _fatal_insn): Likewise.
11978 * tree.h (tree_contains_struct_check_failed,
11979 tree_check_failed, tree_not_check_failed,
11980 tree_class_check_failed, tree_range_check_failed,
11981 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
11982 tree_vec_elt_check_failed, phi_node_elt_check_failed,
11983 tree_operand_check_failed, omp_clause_check_failed,
11984 omp_clause_operand_check_failed, omp_clause_range_check_failed):
11987 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11989 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
11990 flag_branch_probabilities.
11991 * ipa-inline.c (edge_badness): Likewise.
11992 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
11993 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
11994 * predict.c (maybe_hot_frequency_p): Likewise.
11995 (probably_never_executed): Likewise.
11996 * sched-ebb.c (schedule_ebbs): Likewise.
11997 * sched-rgn.c (find_single_block_region): Likewise.
11998 * tracer.c (tail_duplicate): Likewise.
12000 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12002 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
12003 longer requires x_flag_profile_use.
12005 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12007 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
12008 instead of flag_reorder_blocks_and_partition.
12009 * dbxout.c (dbxout_function_end): Likewise.
12010 * dwarf2out.c (gen_subprogram_die): Likewise.
12011 * haifa-sched.c (sched_create_recovery_edges): Likewise.
12012 * hw-doloop.c (reorg_loops): Likewise.
12013 * varasm.c (assemble_start_function,
12014 assemble_end_function): Likewise.
12015 (decide_function_section): Do not check for
12016 flag_reorder_blocks_and_partition.
12018 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12020 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12022 (chkp_get_hard_register_fake_addr_expr): Ditto.
12023 (chkp_build_addr_expr): Add check for hard reg case.
12024 (chkp_parse_array_and_component_ref): Ditto.
12025 (chkp_find_bounds_1): Ditto.
12026 (chkp_process_stmt): Don't generate bounds store for
12029 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12031 * predict.c (maybe_hot_bb_p): Do not check profile status.
12032 (maybe_hot_edge_p): Likewise.
12033 (probably_never_executed): Check for zero counts even if profile
12035 (unlikely_executed_edge_p): New function.
12036 (unlikely_executed_stmt_p): New function.
12037 (unlikely_executed_bb_p): New function.
12038 (set_even_probabilities): Use unlikely predicates.
12039 (combine_predictions_for_bb): Likewise.
12040 (predict_paths_for_bb): Likewise.
12041 (predict_paths_leading_to_edge): Likewise.
12042 (determine_unlikely_bbs): New function.
12043 (estimate_bb_frequencies): Use it.
12044 (compute_function_frequency): Use zero counts even if profile is
12046 * profile-count.h: Fix typo.
12048 2017-08-08 Julia Koval <julia.koval@intel.com>
12050 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
12051 _mm512_mask_cvtsepi16_storeu_epi8,
12052 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
12053 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
12054 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
12055 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
12056 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
12057 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
12058 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
12059 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
12060 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
12061 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
12062 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
12063 __builtin_ia32_pmovuswb256mem_mask,
12064 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
12065 __builtin_ia32_pmovwb512mem_mask): New builtins.
12067 2017-08-08 Julia Koval <julia.koval@intel.com>
12069 PR target/73350,80862
12070 * config/i386/subst.md (round): Fix round pattern.
12071 * config/i386/i386.c (ix86_erase_embedded_rounding):
12072 Fix erasing rounding for the fixed pattern.
12074 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12076 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
12078 2017-06-08 Martin Liska <mliska@suse.cz>
12080 PR gcov-profile/80911
12081 * gcov.c (block_info::block_info): New constructor.
12083 2017-06-07 Carl Love <cel@us.ibm.com>
12085 * config/rs6000/rs6000-c: The return type of the following
12086 built-in functions was implemented as int not long long. Fix sign
12087 of return value for the unsigned version of vec_mulo and vec_mule.
12088 vector unsigned long long vec_bperm (vector unsigned long long,
12089 vector unsigned char)
12090 vector signed long long vec_mule (vector signed int,
12092 vector unsigned long long vec_mule (vector unsigned int,
12093 vector unsigned int)
12094 vector signed long long vec_mulo (vector signed int,
12096 vector unsigned long long vec_mulo (vector unsigned int,
12097 vector unsigned int)
12098 * doc/extend.texi: Fix the documentation for the built-in
12101 2017-06-07 Carl Love <cel@us.ibm.com>
12104 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
12107 2017-06-07 Carl Love <cel@us.ibm.com>
12109 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
12110 support, Generate doublehv for signed int/float for BE case only.
12112 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
12114 * doc/invoke.texi (mcx16): Rewrite.
12116 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12118 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
12119 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
12120 *mov<mode>_softfloat, and an anonymous splitter): Use
12121 nonimmediate_operand instead of rs6000_nonimmediate_operand.
12123 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12125 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
12127 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12128 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
12129 (rs6000_debug_reg_global): Adjust.
12130 (rs6000_init_hard_regno_mode_ok): Adjust.
12131 (rs6000_dbx_register_number): Adjust.
12132 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
12133 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12134 Remove SPE_ACC and SPEFSCR.
12135 (REG_ALLOC_ORDER): Ditto.
12136 (FRAME_POINTER_REGNUM): Change to 111.
12137 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
12138 (REG_CLASS_NAMES): Ditto.
12139 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
12140 (REGISTER_NAMES): Ditto.
12141 (ADDITIONAL_REG_NAMES): Ditto.
12142 (rs6000_reg_names): Ditto.
12143 * config/rs6000/rs6000.md: Renumber some register number
12146 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12148 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
12150 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12151 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
12153 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
12154 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12155 Delete the SPE high registers.
12156 (REG_ALLOC_ORDER): Ditto.
12157 (enum reg_class): Remove SPE_HIGH_REGS.
12158 (REG_CLASS_NAMES): Ditto.
12159 (REG_CLASS_CONTENTS): Delete the SPE high registers.
12160 (REGISTER_NAMES): Ditto.
12161 (rs6000_reg_names): Ditto.
12162 * doc/tm.texi.in: Remove SPE as example.
12163 * doc/tm.texi: Regenerate.
12165 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12167 * config/rs6000/8540.md (ppc8540_brinc): Delete.
12168 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
12169 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
12170 * config/rs6000/rs6000.md (type): Remove "brinc".
12172 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12174 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
12175 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
12176 * config/rs6000/linuxspe.h: Delete file.
12177 * config/rs6000/rs6000.md: Don't include spe.md.
12178 * config/rs6000/spe.h: Delete file.
12179 * config/rs6000/spe.md: Delete file.
12180 * config/rs6000/t-rs6000: Remove spe.md.
12182 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12184 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
12185 (reg_or_none500mem_operand): Delete.
12186 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
12187 instead of reg_or_none500mem_operand.
12189 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12191 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
12192 handling of SPE flags.
12193 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
12195 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12197 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
12199 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
12200 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
12201 paired_divv2sf3): Similar.
12202 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
12203 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
12204 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
12206 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
12207 Rename the paired_* instruction patterns.
12208 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
12210 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
12211 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
12212 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
12213 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
12214 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
12215 PAIRED_VECTOR_MODE.
12216 (struct machine_function): Delete field spe_insn_chain_scanned_p.
12217 (spe_func_has_64bit_regs_p): Delete.
12218 (spe_expand_predicate_builtin): Delete.
12219 (spe_expand_evsel_builtin): Delete.
12220 (TARGET_DWARF_REGISTER_SPAN): Do not define.
12221 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
12222 (invalid_e500_subreg): Delete.
12223 (rs6000_legitimize_address): Always force_reg op2 as well, for
12224 paired single memory accesses.
12225 (rs6000_member_type_forces_blk): Delete.
12226 (rs6000_spe_function_arg): Delete.
12227 (rs6000_expand_unop_builtin): Delete SPE handling.
12228 (rs6000_expand_binop_builtin): Ditto.
12229 (spe_expand_stv_builtin): Delete.
12230 (bdesc_2arg_spe): Delete.
12231 (spe_expand_builtin): Delete.
12232 (spe_expand_predicate_builtin): Delete.
12233 (spe_expand_evsel_builtin): Delete.
12234 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
12235 (spe_init_builtins): Delete.
12236 (spe_func_has_64bit_regs_p): Delete.
12237 (savres_routine_name): Delete "info" parameter. Adjust callers.
12238 (rs6000_emit_stack_reset): Ditto.
12239 (rs6000_dwarf_register_span): Delete.
12240 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
12241 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
12242 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
12244 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
12246 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
12247 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
12248 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
12249 mulv2sf3, divv2sf3): Delete expanders.
12251 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12253 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
12255 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12257 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
12258 * config/rs6000/rs6000.c: Ditto.
12260 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12262 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
12263 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
12264 comparison_operator.
12266 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12268 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
12269 * config/rs6000/rs6000.opt: Ditto.
12270 * config/rs6000/t-rtems: Ditto.
12272 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12274 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
12275 TARGET_E500_SINGLE by 0, simplify.
12276 * config/rs6000/rs6000.c: Ditto.
12277 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
12278 (spe_build_register_parallel): Delete.
12279 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
12280 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
12281 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
12282 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
12283 (E500_CONVERT): Delete.
12284 * config/rs6000/spe.md: Remove many patterns and all define_constants.
12286 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12288 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
12289 * config/rs6000/dfp.md: Ditto.
12290 (negdd2, *negdd2_fpr): Merge.
12291 (absdd2, *absdd2_fpr): Merge.
12292 (negtd2, *negtd2_fpr): Merge.
12293 (abstd2, *abstd2_fpr): Merge.
12294 * config/rs6000/e500.h: Delete file.
12295 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
12296 TARGET_FPRS by 1 and simplify.
12297 * config/rs6000/rs6000-c.c: Ditto.
12298 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
12299 TARGET_DF_SPE by 0.
12300 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
12302 * config/rs6000/rs6000.md: Ditto.
12303 (floatdidf2, *floatdidf2_fpr): Merge.
12304 (move_from_CR_gt_bit): Delete.
12305 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
12306 (E500_CR_IOR_COMPARE): Delete.
12307 (All patterns that require !TARGET_FPRS): Delete.
12308 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
12310 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12312 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
12314 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12316 * graphds.c (add_edge): Intitialize edge's attached data.
12317 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
12318 pointer parameter. Call pointed function on each edge during
12319 graph traversing. Skip traversing the edge when the function
12321 (graphds_dfs, graphds_scc): Ditto.
12322 (for_each_edge): New parameter. Pass the new parameter to callback
12324 * graphds.h (skip_edge_callback): New function pointer type.
12325 (graphds_dfs, graphds_scc): New function pointer parameter.
12326 (graphds_edge_callback, for_each_edge): New parameter.
12328 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12330 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
12331 out code checking if runtime alias check is possible to below ...
12332 Call the new function.
12333 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
12334 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
12336 2017-06-07 Marek Polacek <polacek@redhat.com>
12339 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
12340 TYPE_OVERFLOW_WRAPS check.
12342 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12344 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
12345 if versioning is required.
12346 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
12347 peeling with the check for versioning.
12349 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12351 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
12352 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
12353 Set true to new parameter if new ssa variable is defined.
12354 (vect_gen_vector_loop_niters): Refactor. Set range information
12355 for the new vector loop bound variable.
12356 (vect_do_peeling): Ditto.
12358 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12360 * tree-affine.c (ssa.h): Include header file.
12361 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
12362 has wrapping overflow behavior.
12364 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12366 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
12368 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12370 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
12371 (tree_to_aff_combination): ... here.
12373 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12375 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
12376 reg_pressure model function.
12377 (ivopts_global_cost_for_size): Delete.
12378 (determine_set_costs, iv_ca_recount_cost): Call new model function
12379 ivopts_estimate_reg_pressure.
12381 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12383 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
12384 expensive than udiv. Remove floating point cases from mod.
12386 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12388 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
12389 Increase idiv cost.
12391 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12393 * config/aarch64/aarch64.md
12394 (copysignsf3): Fix mask generation.
12396 2017-06-07 Jakub Jelinek <jakub@redhat.com>
12398 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
12400 (class dump_manager): Add register_dumps method.
12401 * dumpfile.c: Include langhooks.h.
12402 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
12403 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
12404 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
12405 (dump_manager::dump_register): Start with 512 entries instead of 32.
12406 (dump_manager::register_dumps): New method.
12407 * toplev.c (general_init): Instead of invoking register_dumps
12408 langhook, invoke register_dumps method on the dump manager.
12409 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
12412 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
12414 * doc/md.texi: Clarify the restrictions on a define_insn condition.
12415 Say that # requires an associated define_split to exist, and that
12416 the define_split must be suitable for use after register allocation.
12418 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12420 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
12421 (compute_outgoing_frequencies): Also initialize zero counts.
12422 (find_many_sub_basic_blocks): Do not produce uninitialized profile
12423 around loops; preserve more of profile when nothing changes.
12425 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
12427 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
12429 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
12430 * config/arm/arm-cpu-cdata.h: Regenerate.
12431 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
12432 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
12433 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
12434 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
12435 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
12437 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
12439 * config/arm/t-rmprofile: Likewise.
12440 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
12442 2017-06-06 David S. Miller <davem@davemloft.net>
12445 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
12446 blockage if function uses alloca.
12448 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12450 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
12451 New "uid" fields to hold pretty-print IDs of group and ref.
12452 Memory references are now identified as <group_id>:<ref_id>
12453 instead of using [random] addresses.
12454 (dump_mem_details): Simplify, no functional change.
12455 (dump_mem_ref): Simplify and make output more concise.
12456 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
12457 (find_or_create_group): Initialize group uid.
12458 (record_ref): Initialize ref uid. Improve debug output.
12459 (prune_group_by_reuse, should_issue_prefetch_p,)
12460 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
12461 (mark_nontemporal_store, determine_loop_nest_reuse):
12462 Improve debug output.
12464 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12466 * dbgcnt.def (prefetch): New debug counter.
12467 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
12468 (schedule_prefetches): Stop issueing prefetches if debug counter
12471 2017-06-06 Tom de Vries <tom@codesourcery.com>
12473 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
12474 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
12476 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12478 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
12479 Use aarch64_reg_or_zero predicate for operand 4.
12480 (aarch64_compare_and_swap<mode> define_insn_and_split):
12481 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
12482 (aarch64_store_exclusive<mode>): Likewise for operand 2.
12484 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12486 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
12487 (arm_compute_save_core_reg_mask): This.
12488 (thumb1_compute_save_reg_mask): Rename into ...
12489 (thumb1_compute_save_core_reg_mask): This.
12490 (arm_compute_save_reg0_reg12_mask): Adapt comment.
12491 (arm_compute_frame_layout): Likewise.
12493 2017-06-06 Richard Biener <rguenther@suse.de>
12495 PR tree-optimization/80974
12496 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
12497 keep or clear leaders SSA info.
12499 2017-06-06 Tom de Vries <tom@codesourcery.com>
12501 * config/nvptx/nvptx.c (split_mode_p): New function.
12502 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
12504 2017-06-06 Tom de Vries <tom@codesourcery.com>
12506 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
12508 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12511 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
12514 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12516 * shrink-wrap.c (handle_simple_exit): Update profile.
12517 (try_shrink_wrapping): Upate profile.
12519 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12521 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
12522 (tree_guess_outgoing_edge_probabilities): New.
12523 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
12524 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
12526 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12528 * ipa-split.c (split_function): Initialize return bb profile.
12530 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12532 * profile.c (compute_branch_probabilities): Also initialize
12533 EXIT_BLOCK profile.
12535 2017-06-06 Richard Biener <rguenther@suse.de>
12537 PR tree-optimization/80928
12538 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
12539 (vect_analyze_loop_operations): Properly guard analysis for
12541 (vect_transform_loop): Likewise.
12542 (vect_analyze_loop_2): Also reset SLP type on PHIs.
12543 (vect_model_induction_cost): Do not cost for pure SLP.
12544 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
12545 of induction in inner loop vectorization.
12546 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
12547 (vect_get_and_check_slp_defs): Handle vect_induction_def.
12548 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
12550 (vect_analyze_slp_cost_1): Cost induction.
12551 (vect_detect_hybrid_slp_stmts): Handle PHIs.
12552 (vect_get_slp_vect_defs): Likewise.
12553 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
12554 (vect_transform_stmt): Handle SLP reductions.
12555 * tree-vectorizer.h (vectorizable_induction): Adjust.
12557 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12559 * config/rs6000/rs6000.c (make_resolver_func): Update
12560 init_lowered_empty_function call.
12562 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
12564 * doc/invoke.texi: Document the -fprofile-abs-path option.
12565 * common.opt (fprofile-abs-path): New option.
12566 * gcov-io.h (gcov_write_filename): Declare.
12567 * gcov-io.c (gcov_write_filename): New function.
12568 * coverage.c (coverage_begin_function): Use gcov_write_filename.
12569 * profile.c (output_location): Likewise.
12571 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12573 * shring-wrap.c: Revert accidental commit.
12575 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
12577 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
12579 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12581 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
12583 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
12584 profile in callgraph edge.
12585 * profile-count.h (apply_probability): If THIS is 0, then result is 0
12586 (apply_scale): Likewise.
12587 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
12588 Also scale profile when inlining function with zero profile.
12589 (initialize_cfun): Update exit block profile even when it is zero.
12590 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
12591 when profile is read.
12593 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12595 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
12596 (CLONE_*): New constants to define the processors we can generate
12597 code for with the target_clone attribute.
12598 (rs6000_clone_map): New array to identify which clone processors
12599 the current program is running on.
12600 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
12601 target_clone attribute.
12602 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
12603 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
12604 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
12605 (cpu_expand_builtin): Add support for target_clone attribute.
12606 (rs6000_valid_attribute_p): Allow "default" attribute.
12607 (get_decl_name): New debug function to simplify printing the
12608 current function name in debugging statements.
12609 (rs6000_clone_priority): New functions to support the target_clone
12610 attribute, and be able to generate code to switch between ISA 2.05
12611 through ISA 3.0 (power6 through power9).
12612 (rs6000_compare_version_priority): Likewise.
12613 (rs6000_get_function_versions_dispatcher): Likewise.
12614 (make_resolver_func): Likewise.
12615 (add_condition_to_bb): Likewise.
12616 (dispatch_function_versions): Likewise.
12617 (rs6000_generate_version_dispatcher_body): Likewise.
12618 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
12619 (fusion_gpr_load_p): Fix a spacing issue.
12620 * doc/extend.texi (Common Function Attributes): Document that the
12621 PowerPC supports the target_clone attribute.
12623 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
12625 * config/arm/arm.h: explain F symbol found in description of ARM
12626 register allocation in its legend.
12628 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12630 * config/mips/frame-header-opt.c: Include profile-count.h.
12631 * config/riscv/riscv.c: Include profile-count.h
12633 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12635 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
12637 (sm_set_flag_if_changed): Add bbs field.
12638 (execute_sm_if_changed_flag_set): Pass BBS.
12639 (execute_sm): Update.
12641 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12643 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
12646 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12648 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
12649 (peephole2): New peephole2 to emit the above.
12650 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
12652 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12654 * config/aarch64/aarch64.c (define_peephole2 above
12655 *sub_<shift>_<mode>): New peephole.
12657 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12659 * config/i386/i386.c (make_resolver_func): Update.
12660 * Makefile.in: Add profile-count.h and profile-count.o
12661 * auto-profile.c (afdo_indirect_call): Update to new API.
12662 (afdo_set_bb_count): Update.
12663 (afdo_propagate_edge): Update.
12664 (afdo_propagate_circuit): Update.
12665 (afdo_calculate_branch_prob): Update.
12666 (afdo_annotate_cfg): Update.
12667 * basic-block.h: Include profile-count.h
12668 (struct edge_def): Turn count to profile_count.
12669 (struct basic_block_def): Likewie.
12670 (REG_BR_PROB_BASE): Move to profile-count.h
12671 (RDIV): Move to profile-count.h
12672 * bb-reorder.c (max_entry_count): Turn to profile_count.
12673 (find_traces): Update.
12674 (rotate_loop):Update.
12675 (connect_traces):Update.
12676 (sanitize_hot_paths):Update.
12677 * bt-load.c (migrate_btr_defs): Update.
12678 * cfg.c (RDIV): Remove.
12679 (init_flow): Use alloc_block.
12680 (alloc_block): Uninitialize count.
12681 (unchecked_make_edge): Uninitialize count.
12682 (check_bb_profile): Update.
12683 (dump_edge_info): Update.
12684 (dump_bb_info): Update.
12685 (update_bb_profile_for_threading): Update.
12686 (scale_bbs_frequencies_int): Update.
12687 (scale_bbs_frequencies_gcov_type): Update.
12688 (scale_bbs_frequencies_profile_count): New.
12689 * cfg.h (update_bb_profile_for_threading): Update.
12690 (scale_bbs_frequencies_profile_count): Declare.
12691 * cfgbuild.c (compute_outgoing_frequencies): Update.
12692 (find_many_sub_basic_blocks): Update.
12693 * cfgcleanup.c (try_forward_edges): Update.
12694 (try_crossjump_to_edge): Update.
12695 * cfgexpand.c (expand_gimple_tailcall): Update.
12696 (construct_exit_block): Update.
12697 * cfghooks.c (verify_flow_info): Update.
12698 (dump_bb_for_graph): Update.
12699 (split_edge): Update.
12700 (make_forwarder_block): Update.
12701 (duplicate_block): Update.
12702 (account_profile_record): Update.
12703 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
12704 (get_estimated_loop_iterations): Update.
12705 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
12706 (single_likely_exit): Update.
12707 * cfgloopmanip.c (scale_loop_profile): Update.
12709 (set_zero_probability): Update.
12710 (lv_adjust_loop_entry_edge): Update.
12711 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12712 (purge_dead_edges): Update.
12713 (rtl_account_profile_record): Update.
12714 * cgraph.c (cgraph_node::create): Uninitialize count.
12715 (symbol_table::create_edge): Uninitialize count.
12716 (cgraph_update_edges_for_call_stmt_node): Update.
12717 (cgraph_edge::dump_edge_flags): Update.
12718 (cgraph_node::dump): Update.
12719 (cgraph_edge::maybe_hot_p): Update.
12720 * cgraph.h: Include profile-count.h
12721 (create_clone), create_edge, create_indirect_edge): Update.
12722 (cgraph_node): Turn count to profile_count.
12723 (cgraph_edge0: Likewise.
12724 (make_speculative, clone): Update.
12725 (create_edge): Update.
12726 (init_lowered_empty_function): Update.
12727 * cgraphclones.c (cgraph_edge::clone): Update.
12728 (duplicate_thunk_for_node): Update.
12729 (cgraph_node::create_clone): Update.
12730 * cgraphunit.c (cgraph_node::analyze): Update.
12731 (cgraph_node::expand_thunk): Update.
12732 * final.c (dump_basic_block_info): Update.
12733 * gimple-streamer-in.c (input_bb): Update.
12734 * gimple-streamer-out.c (output_bb): Update.
12735 * graphite.c (print_global_statistics): Update.
12736 (print_graphite_scop_statistics): Update.
12737 * hsa-brig.c: Include basic-block.h.
12738 * hsa-dump.c: Include basic-block.h.
12739 * hsa-gen.c (T sum_slice): Update.
12740 (convert_switch_statements):Update.
12741 * hsa-regalloc.c: Include basic-block.h.
12742 * ipa-chkp.c (chkp_produce_thunks): Update.
12743 * ipa-cp.c (struct caller_statistics): Update.
12744 (init_caller_stats): Update.
12745 (gather_caller_stats): Update.
12746 (ipcp_cloning_candidate_p): Update.
12747 (good_cloning_opportunity_p): Update.
12748 (get_info_about_necessary_edges): Update.
12749 (dump_profile_updates): Update.
12750 (update_profiling_info): Update.
12751 (update_specialized_profile): Update.
12752 (perhaps_add_new_callers): Update.
12753 (decide_about_value): Update.
12754 (ipa_cp_c_finalize): Update.
12755 * ipa-devirt.c (struct odr_type_warn_count): Update.
12756 (struct decl_warn_count): Update.
12757 (struct final_warning_record): Update.
12758 (possible_polymorphic_call_targets): Update.
12759 (ipa_devirt): Update.
12760 * ipa-fnsummary.c (redirect_to_unreachable): Update.
12761 * ipa-icf.c (sem_function::merge): Update.
12762 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
12763 * ipa-inline.c (compute_uninlined_call_time): Update.
12764 (compute_inlined_call_time): Update.
12765 (want_inline_small_function_p): Update.
12766 (want_inline_self_recursive_call_p): Update.
12767 (edge_badness): Update.
12768 (lookup_recursive_calls): Update.
12769 (recursive_inlining): Update.
12770 (inline_small_functions): Update.
12771 (dump_overall_stats): Update.
12772 (dump_inline_stats): Update.
12773 * ipa-profile.c (ipa_profile_generate_summary): Update.
12774 (ipa_propagate_frequency): Update.
12775 (ipa_profile): Update.
12776 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12777 * ipa-utils.c (ipa_merge_profiles): Update.
12778 * loop-doloop.c (doloop_modify): Update.
12779 * loop-unroll.c (report_unroll): Update.
12780 (unroll_loop_runtime_iterations): Update.
12781 * lto-cgraph.c (lto_output_edge): Update.
12782 (lto_output_node): Update.
12783 (input_node): Update.
12784 (input_edge): Update.
12785 (merge_profile_summaries): Update.
12786 * lto-streamer-in.c (input_cfg): Update.
12787 * lto-streamer-out.c (output_cfg): Update.
12788 * mcf.c (create_fixup_graph): Update.
12789 (adjust_cfg_counts): Update.
12790 (sum_edge_counts): Update.
12791 * modulo-sched.c (sms_schedule): Update.
12792 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
12793 * predict.c (maybe_hot_count_p): Update.
12794 (probably_never_executed): Update.
12795 (dump_prediction): Update.
12796 (combine_predictions_for_bb): Update.
12797 (propagate_freq): Update.
12798 (handle_missing_profiles): Update.
12799 (counts_to_freqs): Update.
12800 (rebuild_frequencies): Update.
12801 (force_edge_cold): Update.
12802 * predict.h: Include profile-count.h
12803 (maybe_hot_count_p, counts_to_freqs): UPdate.
12804 * print-rtl-function.c: Do not include cfg.h
12805 * print-rtl.c: Include basic-block.h
12806 * profile-count.c: New file.
12807 * profile-count.h: New file.
12808 * profile.c (is_edge_inconsistent): Update.
12809 (correct_negative_edge_counts): Update.
12810 (is_inconsistent): Update.
12811 (set_bb_counts): Update.
12812 (read_profile_edge_counts): Update.
12813 (compute_frequency_overlap): Update.
12814 (compute_branch_probabilities): Update; Initialize and deinitialize
12816 (branch_prob): Update.
12817 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
12818 (edge_gcov_count): New.
12819 (bb_gcov_count): New.
12820 * shrink-wrap.c (try_shrink_wrapping): Update.
12821 * tracer.c (better_p): Update.
12822 * trans-mem.c (expand_transaction): Update.
12823 (ipa_tm_insert_irr_call): Update.
12824 (ipa_tm_insert_gettmclone_call): Update.
12825 * tree-call-cdce.c: Update.
12826 * tree-cfg.c (gimple_duplicate_sese_region): Update.
12827 (gimple_duplicate_sese_tail): Update.
12828 (gimple_account_profile_record): Update.
12829 (execute_fixup_cfg): Update.
12830 * tree-inline.c (copy_bb): Update.
12831 (copy_edges_for_bb): Update.
12832 (initialize_cfun): Update.
12833 (freqs_to_counts): Update.
12834 (copy_cfg_body): Update.
12835 (expand_call_inline): Update.
12836 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
12837 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
12838 (try_unroll_loop_completely): Update.
12839 (try_peel_loop): Update.
12840 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12841 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
12842 * tree-ssa-loop-split.c (connect_loops): Update.
12843 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
12844 * tree-ssa-reassoc.c (branch_fixup): Update.
12845 * tree-ssa-tail-merge.c (replace_block_by): Update.
12846 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
12847 (compute_path_counts): Update.
12848 (update_profile): Update.
12849 (recompute_probabilities): Update.
12850 (update_joiner_offpath_counts): Update.
12851 (estimated_freqs_path): Update.
12852 (freqs_to_counts_path): Update.
12853 (clear_counts_path): Update.
12854 (ssa_fix_duplicate_block_edges): Update.
12855 (duplicate_thread_path): Update.
12856 * tree-switch-conversion.c (case_bit_test_cmp): Update.
12857 (struct switch_conv_info): Update.
12858 * tree-tailcall.c (decrease_profile): Update.
12859 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
12860 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12861 * value-prof.c (check_counter): Update.
12862 (gimple_divmod_fixed_value): Update.
12863 (gimple_mod_pow2): Update.
12864 (gimple_mod_subtract): Update.
12865 (gimple_ic_transform): Update.
12866 (gimple_stringop_fixed_value): Update.
12867 * value-prof.h (gimple_ic): Update.
12869 2017-06-02 Carl Love <cel@us.ibm.com>
12871 * config/rs6000/rs6000-c: Add support for built-in functions
12872 vector double vec_doublee (vector signed int);
12873 vector double vec_doublee (vector unsigned int);
12874 vector double vec_doublee (vector float);
12875 vector double vec_doubleh (vector signed int);
12876 vector double vec_doubleh (vector unsigned int);
12877 vector double vec_doubleh (vector float);
12878 vector double vec_doublel (vector signed int);
12879 vector double vec_doublel (vector unsigned int);
12880 vector double vec_doublel (vector float);
12881 vector double vec_doubleo (vector signed int);
12882 vector double vec_doubleo (vector unsigned int);
12883 vector double vec_doubleo (vector float);.
12884 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
12885 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
12887 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
12888 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
12889 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
12891 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
12892 vec_doublel, vec_doubleh.
12893 * doc/extend.texi: Update the built-in documentation file for the
12894 new built-in functions.
12896 2017-06-02 David Malcolm <dmalcolm@redhat.com>
12899 * ipa-inline-analysis.c (free_growth_caches): Set
12900 edge_removal_hook_holder to NULL after removing it.
12902 2017-06-02 Sudakshina Das <sudi.das@arm.com>
12904 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
12905 comparision with zero.
12907 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
12908 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12909 for early expansion of vec_min and vec_max builtins.
12910 (builtin_function_type): Add min/max unsigned variants to those
12911 identified as having unsigned arguments.
12913 2017-06-02 Olivier Hainque <hainque@adacore.com>
12915 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
12917 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12919 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
12920 Use VALL_F16 iterator rather than VALL.
12922 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12924 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
12925 Emit CBNZ inside loop when doing a strong exchange and comparing
12926 against zero. Generate the CC flags after the loop.
12928 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
12930 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
12931 (dl_section_ref): New.
12932 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
12933 On AIX, append an expression to subtract the size of the
12934 section length to dl_section_ref.
12936 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
12938 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12939 for early expansion of vector absolute builtins.
12941 2017-06-02 Richard Biener <rguenther@suse.de>
12943 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
12944 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
12946 2017-06-02 Richard Biener <rguenther@suse.de>
12948 PR tree-optimization/80948
12949 * tree-tailcall.c (find_tail_calls): Track stmts to move in
12950 stmt order as well.
12952 2017-06-02 Richard Biener <rguenther@suse.de>
12954 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
12956 * tree-vect-stmts.c (process_use): Do not mark backedge defs
12957 for inductions as relevant.
12959 2017-06-02 Richard Biener <rguenther@suse.de>
12961 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
12962 (vectorizable_induction): ... this. Remove dead code.
12964 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
12966 * builtins. (expand_builtin_alloca): Remove second parameter and
12967 infer its value from the first parameter instead.
12968 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
12970 2017-06-02 Jakub Jelinek <jakub@redhat.com>
12972 PR rtl-optimization/80903
12973 * loop-doloop.c (add_test): Unshare sequence.
12975 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
12977 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
12979 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
12981 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
12983 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
12984 xlogue_layout::get_instance, logue_layout::xlogue_layout,
12985 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
12986 (xlogue_layout::get_stub_rtx): Make static.
12987 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
12988 (xlogue_layout::compute_stub_managed_regs): Rename to...
12989 (xlogue_layout::count_stub_managed_regs): ...this.
12990 (xlogue_layout::is_stub_managed_reg): New function.
12991 (xlogue_layout::m_stub_names): Rename to...
12992 (xlogue_layout::s_stub_names): ...this, make static.
12993 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
12994 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
12995 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
12996 xlogue_layout::s_stub_names): Instantiate statics.
12997 (stub_managed_regs): Remove.
12998 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
12999 (disable_call_ms2sysv_xlogues): Rename to...
13000 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
13001 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
13003 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
13004 change after reload_completed.
13005 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
13007 (ix86_expand_prologue): Likewise.
13008 (ix86_expand_epilogue): Likewise.
13009 (ix86_expand_split_stack_prologue): Likewise.
13010 (ix86_compute_frame_layout): Remove frame parameter ...
13011 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
13012 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
13014 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
13015 (ix86_frame): Move from here ...
13016 * config/i386/i386.h (ix86_frame): ... to here.
13017 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
13018 complete ix86_frame data structure instead. Remove some_ld_name.
13020 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13022 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
13023 symbols that hold a DECL_VALUE_EXPR.
13025 2017-06-01 Martin Jambor <mjambor@suse.cz>
13027 PR tree-optimization/80898
13028 * tree-sra.c (process_subtree_disqualification): Removed.
13029 (disqualify_candidate): Do not acll
13030 process_subtree_disqualification.
13031 (subtree_mark_written_and_enqueue): New function.
13032 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
13033 RHS has been disqualified and re-queue LHS if necessary. Apart
13034 from that, ignore disqualified RHS.
13036 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13038 * config/s390/s390.c (s390_emit_epilogue): Disable early return
13039 address fetch for z10 or later.
13041 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13043 * config/arc/arc.md (tst_movb): Add guard when splitting.
13045 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13047 * config/arc/arc.c (arc_can_eliminate): Test against
13048 arc_frame_pointer_needed.
13050 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13052 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
13053 to prevent store reordering.
13054 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
13055 (type): Add block type.
13056 (stack_tie): Define special instruction to be used in
13059 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13061 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
13062 constraint. It is not valid for the pattern.
13063 (noncommutative_binary_comparison): Likewise.
13065 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13067 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
13070 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13072 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
13073 be used by the reg-alloc.
13075 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13077 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
13078 reg-alloc when having mul64 or mul32x16 instructions.
13079 (mulsidi3): Likewise.
13080 (umulsidi3): Likewise.
13081 (mulsi32x16): New pattern.
13082 (mulsi64): Likewise.
13083 (mulsidi64): Likewise.
13084 (umulsidi64): Likewise.
13085 (MUL32x16_REG): Define.
13086 (mul64_600): Use MUL32x16_REG.
13087 (mac64_600): Likewise.
13088 (umul64_600): Likewise.
13089 (umac64_600): Likewise.
13091 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13093 * config/arc/arc.md (mulsi3_700): Make it commutative.
13095 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
13097 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
13099 (*sign_extendsidi2_insn): Likewise for movstosw.
13101 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13103 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
13104 the type of the input discriminant value. Convert the
13105 discriminant value of signedness vary.
13107 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
13109 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
13110 Add to -Wall section.
13112 2017-06-01 Richard Biener <rguenther@suse.de>
13114 PR middle-end/66313
13115 * fold-const.c (fold_plusminus_mult_expr): If the factored
13116 factor may be zero use a wrapping type for the inner operation.
13117 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
13118 and handle moved defs.
13119 (process_assignment): Properly guard the unary op case. Return a
13120 tri-state indicating that moving the stmt before the call may allow
13121 to continue. Pass through to_move.
13122 (find_tail_calls): Handle moving unrelated defs before
13125 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
13128 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
13129 splitter result in the canonical way.
13131 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
13133 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
13134 also for 32bit target. Update insn attributes.
13135 (zero-extendsidi2 splitter): Allow all registers for operand 1.
13137 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
13139 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
13140 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
13141 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
13142 (_mm_maskz_min_ss): New intrinsics.
13144 2017-05-31 Martin Liska <mliska@suse.cz>
13146 * tree-vect-loop.c (vect_create_epilog_for_reduction):
13147 Change comment style to one we normally use.
13148 (vectorizable_reduction): Likewise.
13149 (vectorizable_induction): Likewise.
13150 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13151 (vectorizable_call): Likewise.
13152 (vectorizable_simd_clone_call): Likewise.
13153 (vectorizable_conversion): Likewise.
13154 (vectorizable_assignment): Likewise.
13155 (vectorizable_shift): Likewise.
13156 (vectorizable_operation): Likewise.
13157 (vectorizable_store): Likewise.
13158 (vectorizable_load): Likewise.
13159 * tree-vectorizer.h: Likewise.
13161 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
13163 * passes.c (emergency_dump_function): New.
13164 * tree-pass.h (emergency_dump_function): Declare.
13165 * plugin.c (plugins_internal_error_function): Remove.
13166 * plugin.h (plugins_internal_error_function): Remove declaration.
13167 * toplev.c (internal_error_function): New static function. Use it...
13168 (general_init): ...here.
13170 2017-05-31 Graham Markall <graham.markall@embecosm.com>
13172 * config/arc/arc.c (arc_print_operand): Handle constant operands.
13173 (arc_rtx_costs): Add costs for new patterns.
13174 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
13175 * config/arc/predicates.md: Add _1_2_3_operand predicate.
13177 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13179 * tree-ssa-strlen.c (get_next_strinfo): New function.
13180 (get_stridx_plus_constant): Use it.
13181 (zero_length_string): Likewise.
13182 (adjust_related_strinfos): Likewise.
13183 (adjust_last_stmt): Likewise.
13185 2017-05-31 Richard Biener <rguenther@suse.de>
13188 * config/i386/i386.c (ix86_expand_builtin): Remove assert
13189 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
13191 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13193 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
13194 loop_vinfo argument and use of dependence distance vectors.
13195 Check instead whether the two references differ only in their
13196 initial value and assume that they have the same alignment if the
13197 difference is a multiple of the vector alignment.
13198 (vect_analyze_data_refs_alignment): Update call accordingly.
13200 2017-05-31 Martin Liska <mliska@suse.cz>
13203 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
13205 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13207 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
13208 (create_intersect_range_checks): Move from ...
13209 * tree-data-ref.c (create_intersect_range_checks_index)
13210 (create_intersect_range_checks): ... to here.
13211 (create_runtime_alias_checks): New function factored from ...
13212 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
13213 here. Call above function.
13214 * tree-data-ref.h (create_runtime_alias_checks): New function.
13216 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13218 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
13219 segment length for dr_b and compute it in wide_int.
13221 2017-05-31 Richard Biener <rguenther@suse.de>
13223 PR tree-optimization/80906
13224 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
13225 and pass through iv_map.
13226 (copy_bb_and_scalar_dependences): Adjust.
13227 (translate_pending_phi_nodes): Likewise.
13228 (copy_loop_close_phi_args): Handle code-generating IVs instead
13231 2017-05-30 David Malcolm <dmalcolm@redhat.com>
13233 * diagnostic-color.c (color_dict): Add "type-diff".
13234 (parse_gcc_colors): Update comment.
13235 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
13236 -fdiagnostics-show-template-tree and -fno-elide-type.
13237 (GCC_COLORS): Add type-diff to example.
13239 (-fdiagnostics-show-template-tree): New.
13240 (-fno-elide-type): New.
13241 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
13242 the pp_format_decoder callback. Call any m_format_postprocessor's
13244 (pretty_printer::pretty_printer): Initialize
13245 m_format_postprocessor.
13246 (pretty_printer::~pretty_printer): Delete any
13247 m_format_postprocessor.
13248 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
13249 (class format_postprocessor): New class.
13250 (struct pretty_printer::format_decoder): Document the new parameters.
13251 (struct pretty_printer::m_format_postprocessor): New field.
13252 * tree-diagnostic.c (default_tree_printer): Update for new
13253 bool and const char ** params.
13254 * tree-diagnostic.h (default_tree_printer): Likewise.
13256 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
13258 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
13259 (lwa_operand): Delete rs6000_gen_cell_microcode test.
13260 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13261 rs6000_gen_cell_microcode code.
13262 (rs6000_final_prescan_insn): Delete.
13263 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
13264 "warn-cell-microcode" entries.
13265 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
13266 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
13267 throughout. Change cc_reg_not_micro_cr0_operand to
13268 cc_reg_not_cr0_operand throughout.
13269 (*extendhi<mode>2_noload): Delete.
13270 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
13271 (mwarn-cell-microcode): Delete.
13272 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
13273 -mgen-cell-microcode and -mwarn-cell-microcode.
13275 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
13278 * config/i386/constraints.md (Yd): New constraint.
13280 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
13281 and (?Yd, r) alternatives. Update insn attributes.
13282 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
13283 and (?*Yd, r) alternatives. Update insn attributes.
13284 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
13286 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
13288 * gimplify.c (gimplify_modify_expr): Don't create a
13289 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
13292 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
13294 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
13296 2017-05-30 Richard Biener <rguenther@suse.de>
13298 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
13299 and reduc_def fields.
13300 (STMT_VINFO_REDUC_TYPE): New define.
13301 (STMT_VINFO_REDUC_DEF): Likewise.
13302 (vect_force_simple_reduction): Adjust prototype.
13303 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
13304 (vect_is_simple_reduction): Remove check_reduction argument.
13305 (vect_force_simple_reduction): Adjust and set
13306 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13307 (vectorizable_reduction): Do not re-do reduction analysis
13308 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13309 * tree-parloops.c (gather_scalar_reductions): Adjust.
13311 2017-05-30 Richard Biener <rguenther@suse.de>
13313 PR middle-end/80901
13314 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
13317 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
13319 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
13320 Introduce unknown_misalignment parameter and remove vf.
13321 (vect_peeling_hash_get_lowest_cost):
13322 Pass unknown_misalignment parameter.
13323 (vect_enhance_data_refs_alignment):
13324 Fix unsupportable data ref treatment.
13326 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13328 * tree-vect-data-refs.c (vect_get_data_access_cost):
13329 Workaround for SLP handling.
13330 (vect_enhance_data_refs_alignment):
13331 Compute costs for doing no peeling at all, compare to the best
13332 peeling costs so far and avoid peeling if cheaper.
13334 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13336 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
13337 Return peeling info and set costs to zero for unlimited cost
13339 (vect_enhance_data_refs_alignment): Also inspect all datarefs
13340 with unknown misalignment. Compute and costs for unknown
13341 misalignment, compare them to the costs for known misalignment
13342 and choose the cheapest for peeling.
13344 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13346 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
13347 (vect_get_peeling_costs_all_drs): Create function.
13348 (vect_peeling_hash_get_lowest_cost):
13349 Use vect_get_peeling_costs_all_drs.
13350 (vect_peeling_supportable): Create function.
13351 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
13353 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13355 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
13356 DR_HAS_NEGATIVE_STEP.
13357 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
13358 (vect_enhance_data_refs_alignment): Use.
13359 (vect_duplicate_ssa_name_ptr_info): Use.
13360 * tree-vectorizer.h (dr_misalignment): Use.
13361 (known_alignment_for_access_p): Use.
13363 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13366 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
13368 (has_section_name): New function.
13370 2017-05-30 Martin Liska <mliska@suse.cz>
13373 * auto-profile.c (get_function_decl_from_block): Fix
13376 2017-05-30 Richard Biener <rguenther@suse.de>
13378 PR middle-end/80876
13379 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
13381 2017-05-30 Martin Liska <mliska@suse.cz>
13383 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
13384 * dumpfile.h (struct dump_file_info): Remove ctors.
13386 2017-05-30 Martin Liska <mliska@suse.cz>
13388 * predict.def: Fix GNU coding style.
13390 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13392 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13393 Mark 'to' argument with ATTRIBUTE_UNUSED.
13395 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13397 * config/xtensa/xtensa.c (xtensa_emit_call): Use
13398 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
13399 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
13402 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
13404 * doc/install.texi (Options specification): Restore entry of
13405 --enable-sjlj-exceptions.
13407 2017-05-27 Michael Eager <eager@eagercon.com>
13410 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
13412 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
13414 * config/microblaze/microblaze.h
13415 (FIXED_REGISTERS): Update in macro.
13416 (CALL_USED_REGISTERS): Update in macro.
13418 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
13420 * doc/install.texi: Add links to macOS binary distributions.
13422 2017-05-27 Jakub Jelinek <jakub@redhat.com>
13426 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13428 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13430 2017-05-26 Martin Liska <mliska@suse.cz>
13432 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
13434 2017-05-26 Martin Liska <mliska@suse.cz>
13436 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
13437 always leading ';; '.
13438 (dump_bb_info): Likewise.
13439 (brief_dump_cfg): Likewise.
13440 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
13441 * dumpfile.c: Remove usage of TDF_VERBOSE.
13442 * dumpfile.h (enum dump_kind): Likewise.
13443 (dump_gimple_bb_header): Do not use TDF_COMMENT.
13444 * print-tree.c (debug_verbose): Remove.
13445 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
13446 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
13447 * tree-diagnostic.c (default_tree_printer): Replace
13448 TDF_DIAGNOSTIC with TDF_SLIM.
13450 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13452 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
13453 in parameter loop, rather than loop_vinfo.
13454 (create_intersect_range_checks): Ditto.
13455 (vect_create_cond_for_alias_checks): Update call to above functions.
13457 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13459 PR tree-optimization/80815
13460 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
13461 for merging runtime alias checks. Handle negative DR_STEPs.
13463 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13465 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
13467 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
13468 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
13469 out code pruning runtime alias checks.
13470 * tree-data-ref.c (prune_runtime_alias_test_list): New function
13471 factored out from above.
13472 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13474 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13476 (prune_runtime_alias_test_list): New decalaration.
13478 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13480 * tree-vect-data-refs.c (compare_tree): Rename and move ...
13481 * tree-data-ref.c (data_ref_compare_tree): ... to here.
13482 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
13483 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
13484 (operator==, comp_dr_with_seg_len_pair): Ditto.
13485 (vect_prune_runtime_alias_test_list): Ditto.
13487 2017-05-26 Martin Liska <mliska@suse.cz>
13490 * params.def: Bound partial-inlining-entry-probability param.
13492 2017-05-26 Marek Polacek <polacek@redhat.com>
13495 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
13498 2017-05-26 Richard Biener <rguenther@suse.de>
13500 PR tree-optimization/80842
13501 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
13504 2017-05-26 Richard Biener <rguenther@suse.de>
13506 PR tree-optimization/80844
13507 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
13509 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
13511 * doc/md.texi (Machine Constraints): Update x86 family
13512 machine constraints section to match 'config/i386/constraints.md'.
13514 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
13516 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
13518 2017-05-25 Nathan Sidwell <nathan@acm.org>
13520 * doc/invoke.texi (--enable-languages): Update documentation.
13522 2017-05-25 Martin Liska <mliska@suse.cz>
13524 * dumpfile.c: Add TDF_FOLDING.
13525 * dumpfile.h (enum dump_kind): Likewise.
13526 * genmatch.c (dt_simplify::gen_1): Use it.
13528 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13530 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
13532 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13534 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13535 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
13537 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13539 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
13540 * match.pd (X == C): Rewrite it here.
13541 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
13542 with_certain_nonzero_bits2): New predicates.
13543 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
13545 2017-05-24 Nathan Sidwell <nathan@acm.org>
13547 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
13550 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
13553 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13555 * config/powerpcspe: New port. Files are copied from the rs6000
13556 port, with "rs6000" in filenames replaced by "powerpcspe".
13558 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
13560 PR rtl-optimization/80754
13561 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
13563 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
13565 * config/sparc/sparc.md (length): Return the correct value for -mflat
13566 sibcalls to match output_sibcall.
13568 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13572 * config/rs6000/rs6000.c (struct machine_function): Add new field
13574 (rs6000_get_separate_components): Init that field, use it.
13575 (rs6000_components_for_bb): Use the field.
13577 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
13579 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
13581 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
13583 PR middle-end/80823
13584 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
13586 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13589 * config/s390/s390.c (s390_check_qrst_address): Check incoming
13590 address against address_operand predicate.
13591 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
13593 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
13595 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
13596 parameters passed indirectly.
13598 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
13600 * config/i386/i386.md (*movdi_internal): Remove SSE4
13601 alternative 18 (?r, *v). Update insn attributes.
13602 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
13603 Update insn attributes.
13604 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
13605 Update insn attributes.
13606 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
13607 alternative 1 (r, v). Remove isa attribute.
13608 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
13609 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
13610 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
13612 2017-05-23 Tom de Vries <tom@codesourcery.com>
13614 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
13617 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13619 * cgraphunit.c (symbol_table::process_new_functions): Update.
13620 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
13621 (inline_generate_summary): Rename to ...
13622 (ipa_fn_summary_generate): ... this one.
13623 (inline_read_summary): Rename to ...
13624 (ipa_fn_summary_read): ... this one.
13625 (inline_write_summary): Rename to ...
13626 (ipa_fn_summary_write): ... this one.
13627 (inline_free_summary): Rename to ...
13628 (ipa_free_fn_summary): ... this one.
13629 (pass_data_local_fn_summary, pass_local_fn_summary,
13630 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
13631 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
13632 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
13633 make_pass_ipa_fn_summary): New.
13634 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
13635 inline_write_summary, inline_free_summary): Remove.
13636 (ipa_free_fn_summary) : New.
13637 * ipa-inline.c (ipa_inline): Update.
13638 (pass_ipa_inline): Do not generate summaries.
13639 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
13641 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
13642 and add pass_ipa_fn_summary.
13643 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
13645 (make_pass_inline_parameters): Remove.
13647 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
13649 * omp-low.c (struct omp_context): Remove "default_kind" member.
13652 * omp-offload.c (execute_oacc_device_lower): Remove the
13653 parallelism dimensions function attributes for unparallelized
13654 OpenACC kernels constructs.
13656 2017-05-23 Martin Liska <mliska@suse.cz>
13658 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
13660 (cgraph_edge::make_speculative): Likewise.
13661 (cgraph_edge::resolve_speculation): Likewise.
13662 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
13663 (cgraph_node::dump): Likewise.
13664 * cgraph.h: Likewise.
13665 * cgraphunit.c (analyze_functions): Likewise.
13666 (symbol_table::compile): Likewise.
13667 * ipa-cp.c (print_all_lattices): Likewise.
13668 (determine_versionability): Likewise.
13669 (initialize_node_lattices): Likewise.
13670 (ipcp_verify_propagated_values): Likewise.
13671 (estimate_local_effects): Likewise.
13672 (update_profiling_info): Likewise.
13673 (create_specialized_node): Likewise.
13674 (perhaps_add_new_callers): Likewise.
13675 (decide_about_value): Likewise.
13676 (decide_whether_version_node): Likewise.
13677 (identify_dead_nodes): Likewise.
13678 (ipcp_store_bits_results): Likewise.
13679 * ipa-devirt.c (dump_targets): Likewise.
13680 (ipa_devirt): Likewise.
13681 * ipa-icf.c (sem_item::dump): Likewise.
13682 (sem_function::equals): Likewise.
13683 (sem_variable::equals): Likewise.
13684 (sem_item_optimizer::read_section): Likewise.
13685 (sem_item_optimizer::execute): Likewise.
13686 (congruence_class::dump): Likewise.
13687 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
13688 (dump_inline_summary): Likewise.
13689 (estimate_node_size_and_time): Likewise.
13690 (inline_analyze_function): Likewise.
13691 * ipa-inline-transform.c (inline_call): Likewise.
13692 * ipa-inline.c (report_inline_failed_reason): Likewise.
13693 (want_early_inline_function_p): Likewise.
13694 (edge_badness): Likewise.
13695 (update_edge_key): Likewise.
13696 (inline_small_functions): Likewise.
13697 * ipa-profile.c (ipa_profile): Likewise.
13698 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
13699 (ipa_make_edge_direct_to_target): Likewise.
13700 (remove_described_reference): Likewise.
13701 (ipa_impossible_devirt_target): Likewise.
13702 (propagate_controlled_uses): Likewise.
13703 (ipa_print_node_params): Likewise.
13704 (ipcp_transform_function): Likewise.
13705 * ipa-pure-const.c (pure_const_read_summary): Likewise.
13706 (propagate_pure_const): Likewise.
13707 * ipa-reference.c (generate_summary): Likewise.
13708 (read_write_all_from_decl): Likewise.
13709 (propagate): Likewise.
13710 (ipa_reference_read_optimization_summary): Likewise.
13711 * ipa-utils.c (ipa_merge_profiles): Likewise.
13712 * ipa.c (walk_polymorphic_call_targets): Likewise.
13713 (symbol_table::remove_unreachable_nodes): Likewise.
13714 (ipa_single_use): Likewise.
13715 * passes.c (execute_todo): Likewise.
13716 * predict.c (drop_profile): Likewise.
13717 * symtab.c (symtab_node::get_dump_name): New function.
13718 (symtab_node::dump_name): Likewise.
13719 (symtab_node::dump_asm_name): Likewise.
13720 (symtab_node::dump_references): Likewise.
13721 (symtab_node::dump_referring): Likewise.
13722 (symtab_node::dump_base): Likewise.
13723 (symtab_node::debug_symtab): Likewise.
13724 * tree-sra.c (convert_callers_for_node): Likewise.
13725 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
13726 * value-prof.c (init_node_map): Likewise.
13728 2017-05-23 Martin Liska <mliska@suse.cz>
13730 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
13731 and symtab_node::debug_symtab to symbol_table::debug.
13732 * cgraphunit.c (analyze_functions): Use the renamed function.
13733 (symbol_table::compile): Likewise.
13734 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
13735 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
13736 * passes.c (execute_todo): Likewise.
13737 * symtab.c (symbol_table::dump): New function.
13738 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
13740 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13742 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
13743 that nonconst implies exec.
13745 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13747 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
13748 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
13749 (inline_edge_summary_vec): Turn into ...
13750 (ipa_call_summaries): ... this one.
13751 (redirect_to_unreachable, edge_set_predicate,
13752 evaluate_properties_for_edge, inline_summary_alloc,
13753 reset_ipa_call_summary, reset_inline_summary,
13754 inline_summary_t::duplicate): Update.
13755 (inline_edge_duplication_hook): Turn to ...
13756 (ipa_call_summary_t::duplicate): ... this one.
13757 (inline_edge_removal_hook): Turn to ...
13758 (ipa_call_summary_t::remove): ... this one.
13759 (dump_inline_edge_summary): Turn to ...
13760 (dump_ipa_call_summary): ... this one.
13761 (estimate_function_body_sizes): Update.
13762 (inline_update_callee_summaries): Update.
13763 (remap_edge_change_prob): Update.
13764 (remap_edge_summaries): Update.
13765 (inline_merge_summary): Update.
13766 (do_estimate_edge_time): Update.
13767 (inline_generate_summary): Update.
13768 (inline_read_section): Update.
13769 (inline_read_summary): Update.
13770 (inline_free_summary): Update.
13771 * ipa-inline.c (can_inline_edge_p): Update.
13772 (compute_inlined_call_time): Update.
13773 (want_inline_small_function_p): Update.
13774 (edge_badness): Update.
13775 (early_inliner): Update.
13776 * ipa-inline.h (inline_edge_summary): Turn to ...
13777 (ipa_call_summary): ... this one.
13778 (ipa_call_summary_t): New class.
13779 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
13780 (ipa_call_summaries): New.
13781 (inline_edge_summary): Remove.
13782 (estimate_edge_growth): Update.
13783 * ipa-profile.c (ipa_propagate_frequency_1): Update.
13784 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13785 * ipa-split.c (execute_split_functions): Update.
13786 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
13788 2017-05-23 Tom de Vries <tom@codesourcery.com>
13790 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13791 attributes): Document rdrand effective target.
13793 2017-05-23 Tom de Vries <tom@codesourcery.com>
13795 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13796 attributes): Sort alphabetically.
13798 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
13800 * config/avr/genmultilib.awk: Use gsub instead of gensub.
13802 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
13805 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
13806 V2DF/V2DI splat into two separate patterns, one that handles
13807 registers, and the other that only handles memory. Drop support
13808 for splatting from a GPR on ISA 2.07 and then splitting the
13809 splat into direct move and splat.
13810 (vsx_splat_<mode>_reg): Likewise.
13811 (vsx_splat_<mode>_mem): Likewise.
13813 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
13815 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
13817 2017-05-22 Jakub Jelinek <jakub@redhat.com>
13819 PR middle-end/80809
13820 * omp-low.c (finish_taskreg_remap): New function.
13821 (finish_taskreg_scan): If unit size of ctx->record_type
13822 is non-constant, unshare the size expression and replace
13823 decls in it with possible outer var refs.
13825 PR middle-end/80809
13826 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
13827 GOVD_SHARED rather than GOVD_PRIVATE with it.
13828 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
13829 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
13831 PR middle-end/80853
13832 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
13833 as last argument to build_outer_var_ref for pointer bases of array
13834 section reductions.
13836 2017-05-19 Martin Sebor <msebor@redhat.com>
13838 * print-tree.c (print_node): Print DECL_READ_P flag.
13840 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13842 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
13843 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
13844 * cgraph.c: Likewise.
13845 * cgraphunit.c: Likewise.
13846 * gengtype.c: Likewise.
13847 * ipa-cp.c: Likewise.
13848 * ipa-devirt.c: Likewise.
13849 * ipa-icf.c: Likewise.
13850 * ipa-predicate.c: Likewise.
13851 * ipa-profile.c: Likewise.
13852 * ipa-prop.c: Likewise.
13853 * ipa-split.c: Likewise.
13855 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
13856 edge_predicate_pool, dump_inline_hints,
13857 inline_summary::account_size_time, redirect_to_unreachable,
13858 edge_set_predicate, set_hint_predicate,
13859 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
13860 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
13861 inline_summary_t::remove, remap_hint_predicate_after_duplication,
13862 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
13863 ipa_call_summary_t::remove, initialize_growth_caches,
13864 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
13865 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
13866 mark_modified, unmodified_parm_1, unmodified_parm,
13867 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
13868 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
13869 compute_bb_predicates, will_be_nonconstant_expr_predicate,
13870 will_be_nonconstant_predicate, record_modified_bb_info,
13871 get_minimal_bb, record_modified, param_change_prob,
13872 phi_result_unknown_predicate, predicate_for_phi_result,
13873 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
13874 estimate_function_body_sizes, compute_inline_parameters,
13875 compute_inline_parameters_for_curren, pass_data_inline_parameters,
13876 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
13877 inline_update_callee_summaries, remap_edge_change_prob,
13878 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
13879 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
13880 inline_analyze_function, inline_summary_t::insert,
13881 inline_generate_summary, read_ipa_call_summary, inline_read_section,
13882 inline_read_summary, write_ipa_call_summary, inline_write_summary,
13883 inline_free_summary): Move to ipa-fnsummary.h
13884 (predicate_t): Remove.
13885 * ipa-fnsummary.c: New file.
13886 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
13887 (enum inline_hints_vals, inline_hints, agg_position_info,
13888 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
13889 inline_summaries, ipa_call_summary, ipa_call_summary_t,
13890 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
13891 dump_inline_summary, dump_inline_hints, inline_generate_summary,
13892 inline_read_summary, inline_write_summary, inline_free_summary,
13893 inline_analyze_function, initialize_inline_failed,
13894 inline_merge_summary, inline_update_overall_summary,
13895 compute_inline_parameters): Move to ipa-fnsummary.h
13896 * ipa-fnsummary.h: New file.
13897 * ipa-inline-transform.h: Include ipa-inline.h.
13898 * ipa-inline.c: LIkewise.
13900 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13902 * ipa-inline.c (edge_badness): Use inlined_time instead of
13903 inline_summaries->get.
13905 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13907 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
13909 2017-05-22 Nathan Sidwell <nathan@acm.org>
13911 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
13912 (fdump-lang): Document 'raw' option.
13913 * dumpfile.h (TDI_tu): Delete.
13914 * dumpfile.c (dump_files): Remove translation-unit.
13915 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
13917 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
13919 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
13920 command option from $(AWK) call.
13921 * config/avr/genmultilib.awk: Simplify and rewrite so that it
13922 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
13923 [FORMAT]: Remove handling of variable.
13924 * config/avr/t-multilib: Regenerate.
13926 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13928 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
13930 (dump_inline_summary): Do not print self_time.
13931 (estimate_function_body_sizes): Do not set self_time.
13932 (compute_inline_parameters): Likewise.
13933 (inline_read_section, inline_write_summary): Do not stream self_time.
13934 * ipa-inline.h (inline_summary): Drop self_time.
13936 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13938 * ipa-inline-analysis.c (account_size_time): Rename to ...
13939 (inline_summary::account_size_time): ... this one.
13940 (reset_ipa_call_summary): Turn to ...
13941 (ipa_call_summary::reset): ... this one.
13942 (reset_inline_summary): Turn to ...
13943 (inline_summary::reset): ... this one.
13944 (inline_summary_t::remove): Update.
13945 (inline_summary_t::duplicate): Update.
13946 (ipa_call_summary_t::remove): Update.
13947 (dump_inline_summary): Update.
13948 (estimate_function_body_sizes): Update.
13949 (compute_inline_parameters): Update.
13950 (estimate_node_size_and_time): Update.
13951 (inline_merge_summary): Update.
13952 (inline_update_overall_summary): Update.
13953 (inline_read_section): Update.
13954 (inline_write_summary): Update.
13955 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
13956 add account_size_time and reset member functions.
13957 (ipa_call_summary): Add reset function.
13958 * ipa-predicate.h (predicate::operator &): Constify.
13960 2017-05-22 Richard Biener <rguenther@suse.de>
13962 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
13964 2017-05-19 Jason Merrill <jason@redhat.com>
13966 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
13968 2017-05-19 Marek Polacek <polacek@redhat.com>
13971 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
13972 TYPE_OVERFLOW_WRAPS checks.
13974 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
13976 * tree-core.h (enum omp_clause_default_kind): Add
13977 "OMP_CLAUSE_DEFAULT_PRESENT".
13978 * tree-pretty-print.c (dump_omp_clause): Handle it.
13979 * gimplify.c (enum gimplify_omp_var_data): Add
13980 "GOVD_MAP_FORCE_PRESENT".
13981 (gimplify_adjust_omp_clauses_1): Map it to
13982 "GOMP_MAP_FORCE_PRESENT".
13983 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
13985 * gimplify.c (oacc_default_clause): Clarify.
13987 2017-05-19 Nathan Sidwell <nathan@acm.org>
13989 LANG_HOOK_REGISTER_DUMPS
13990 * toplev.c (general_init): Call register dump lang hook.
13991 * doc/invoke.texi: Document -fdump-lang option family.
13992 * dumpfile.c (dump_files): Remove class dump here.
13993 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
13994 * dumpfile.h (tree_dump_index): Remove TDI_class.
13995 * langhooks-def.h (lhd_register_dumps): Declare.
13996 (LANG_HOOKS_REGISTER_DUMPS): Define.
13997 (LANG_HOOKS_INITIALIZER): Add it.
13998 * langhooks.c (lhd_register_dumps): Define.
13999 * langhooks.h (struct lang_hooks): Add register_dumps.
14001 2017-05-19 Nathan Sidwell <nathan@acm.org>
14003 * context.h (context::set_passes): New.
14004 * context.c (context::context): Do not create pass manager.
14005 * toplev.c (general_init): Create pass manager here.
14007 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
14009 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
14010 use this splitter if two add or or instructions would also work for
14011 the constant we want to generate.
14013 2017-05-19 Richard Biener <rguenther@suse.de>
14016 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
14017 predicate evaluation.
14019 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14021 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
14023 * ipa-inline.c (want_inline_small_function_p): Do not cast to
14026 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14028 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14029 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14030 (inline_edge_summary_vec): Turn into ...
14031 (ipa_call_summaries): ... this one.
14032 (redirect_to_unreachable, edge_set_predicate,
14033 evaluate_properties_for_edge, inline_summary_alloc,
14034 reset_ipa_call_summary, reset_inline_summary,
14035 inline_summary_t::duplicate): Update.
14036 (inline_edge_duplication_hook): Turn to ...
14037 (ipa_call_summary_t::duplicate): ... this one.
14038 (inline_edge_removal_hook): Turn to ...
14039 (ipa_call_summary_t::remove): ... this one.
14040 (dump_inline_edge_summary): Turn to ...
14041 (dump_ipa_call_summary): ... this one.
14042 (estimate_function_body_sizes): Update.
14043 (inline_update_callee_summaries): Update.
14044 (remap_edge_change_prob): Update.
14045 (remap_edge_summaries): Update.
14046 (inline_merge_summary): Update.
14047 (do_estimate_edge_time): Update.
14048 (inline_generate_summary): Update.
14049 (inline_read_section): Update.
14050 (inline_read_summary): Update.
14051 (inline_free_summary): Update.
14052 * ipa-inline.c (can_inline_edge_p): Update.
14053 (compute_inlined_call_time): Update.
14054 (want_inline_small_function_p): Update.
14055 (edge_badness): Update.
14056 (early_inliner): Update.
14057 * ipa-inline.h (inline_edge_summary): Turn to ...
14058 (ipa_call_summary): ... this one.
14059 (ipa_call_summary_t): New class.
14060 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14061 (ipa_call_summaries): New.
14062 (inline_edge_summary): Remove.
14063 (estimate_edge_growth): Update.
14064 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14065 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14066 * ipa-split.c (execute_split_functions): Update.
14067 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14069 2017-05-19 Richard Biener <rguenther@suse.de>
14071 PR middle-end/80764
14072 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
14074 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
14076 * config/rs6000/rs6000.c (struct machine_function): Add field
14077 fpr_is_wrapped_separately.
14078 (rs6000_get_separate_components): Use 64 components. Handle the
14079 new FPR components.
14080 (rs6000_components_for_bb): Handle the FPR components.
14081 (rs6000_emit_prologue_components): Handle the FPR components.
14082 (rs6000_emit_epilogue_components): Handle the FPR components.
14083 (rs6000_set_handled_components): Handle the FPR components.
14084 (rs6000_emit_prologue): Don't output prologue code for those FPRs
14085 that are already separately shrink-wrapped.
14086 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
14087 that are already separately shrink-wrapped.
14089 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
14092 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
14095 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
14096 (define_peephole2 for Altivec d-form load): Add peepholes to catch
14097 cases where the register allocator uses a move and an offsettable
14098 memory operation to/from a FPR register on ISA 2.06/2.07.
14099 (define_peephole2 for Altivec d-form store): Likewise.
14101 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
14104 * config/i386/mmx.md (*mov<mode>_internal): Enable
14105 alternatives 11, 12, 13 and 14 also for 32bit targets.
14106 Remove alternatives 15, 16, 17 and 18.
14107 * config/i386/sse.md (vec_concatv2di): Change
14108 alternative (!x, *y) to (x, ?!*Yn).
14110 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
14112 * dumpfile.h (enum dump_kind): Remove stray comma.
14114 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14116 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
14117 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
14118 predicate::num_conditions
14119 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
14120 (CHANGED): turn into predicate::changed.
14121 (agg_position_info): Move to ipa-predicate.h
14122 (add_condition, predicate::add_clause, predicate::operator &=,
14123 predicate::or_with, predicate::evaluate, predicate::probability,
14124 dump_condition, dump_clause, predicate::dump,
14125 predicate::remap_after_duplication, predicate::remap_after_inlining,
14126 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
14127 (evaluate_conditions_for_known_args): Update.
14128 (set_cond_stmt_execution_predicate): Update.
14129 * ipa-inline.h: Include ipa-predicate.h
14130 (condition, inline_param_summary, conditions, agg_position_info,
14131 predicate): Move to ipa-predicate.h
14132 * ipa-predicate.c: New file.
14133 * ipa-predicate.h: New file.
14135 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
14137 * final.c (leaf_function_p): Check we are not in a sequence.
14139 2017-05-18 Martin Liska <mliska@suse.cz>
14141 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
14142 * dumpfile.c (dump_register): Use new enum dump_kind.
14143 (get_dump_file_name): Likewise.
14144 (dump_enable_all): Likewise.
14145 (dump_switch_p_1): Likewise.
14146 (enable_rtl_dump_file): Remove usage of TDF_RTL.
14147 * dumpfile.h (enum dump_kind): New enum type.
14148 (struct dump_file_info): Create constructor and
14149 format fields and comments.
14150 * passes.c (pass_manager::register_one_dump_file):
14152 * statistics.c (statistics_early_init): Likewise.
14153 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
14154 TDF_TREE with TDF_SLIM.
14155 (gather_memory_references_ref): Likewise.
14157 2017-05-18 Martin Liska <mliska@suse.cz>
14159 * vec.h (struct vnull): Use it.
14161 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14163 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
14164 (true_predicate, false_predicate, true_predicate_p,
14165 false_predicate_p): Remove.
14166 (single_cond_predicate, not_inlined_predicate): Turn to member function
14168 (add_condition): Update.
14169 (add_clause): Turn to...
14170 (predicate::add_clause): ... this one; update; allow passing NULL
14172 (and_predicates): Turn to ...
14173 (predicate::operator &=): ... this one.
14174 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
14175 (or_predicates): Turn to ...
14176 (predicate::or_with): ... this one.
14177 (evaluate_predicate): Turn to ...
14178 (predicate::evaluate): ... this one.
14179 (predicate_probability): Turn to ...
14180 (predicate::probability): ... this one.
14181 (dump_condition): Update.
14182 (dump_predicate): Turn to ...
14183 (predicate::dump): ... this one.
14184 (account_size_time): Update.
14185 (edge_set_predicate): Update.
14186 (set_hint_predicate): UPdate.
14187 (evaluate_conditions_for_known_args): Update.
14188 (evaluate_properties_for_edge): Update.
14189 (remap_predicate_after_duplication): Turn to...
14190 (predicate::remap_after_duplication): ... this one.
14191 (remap_hint_predicate_after_duplication): Update.
14192 (inline_summary_t::duplicate): UPdate.
14193 (dump_inline_edge_summary): Update.
14194 (dump_inline_summary): Update.
14195 (set_cond_stmt_execution_predicate): Update.
14196 (set_switch_stmt_execution_predicate): Update.
14197 (compute_bb_predicates): Update.
14198 (will_be_nonconstant_expr_predicate): Update.
14199 (will_be_nonconstant_predicate): Update.
14200 (phi_result_unknown_predicate): Update.
14201 (predicate_for_phi_result): Update.
14202 (array_index_predicate): Update.
14203 (estimate_function_body_sizes): Update.
14204 (estimate_node_size_and_time): Update.
14205 (estimate_ipcp_clone_size_and_time): Update.
14206 (remap_predicate): Rename to ...
14207 (predicate::remap_after_inlining): ... this one.
14208 (remap_hint_predicate): Update.
14209 (inline_merge_summary): Update.
14210 (inline_update_overall_summary): Update.
14211 (estimate_size_after_inlining): Update.
14212 (read_predicate): Rename to ...
14213 (predicate::stream_in): ... this one.
14214 (read_inline_edge_summary): Update.
14215 (write_predicate): Rename to ...
14216 (predicate::stream_out): ... this one.
14217 (write_inline_edge_summary): Update.
14218 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
14219 (clause_t): Turn to uint32_t
14220 (predicate): Turn to class; implement constructor and operators
14222 (size_time_entry): Update.
14223 (inline_summary): Update.
14224 (inline_edge_summary): Update.
14226 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
14228 * fold-const.c (fold_binary_loc): Move transformation...
14229 * match.pd (C - X CMP X): ... here.
14231 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
14233 * config/sparc/sparc.c (sparc_option_override): Set function
14234 alignment for -mcpu=niagara7 to 64 to match the I$ line.
14235 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
14237 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
14239 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
14241 2017-05-18 Marek Polacek <polacek@redhat.com>
14244 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
14245 (pass_ubsan::execute): Call gimple_assign_single_p instead of
14246 gimple_assign_load_p.
14248 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
14250 PR middle-end/80692
14251 * real.c (do_compare): Give decimal_do_compare preference over
14252 comparing just the signs.
14254 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
14256 * doc/md.texi (Canonicalization of Instructions): Describe the
14257 canonical form of instructions that inherently set a condition
14260 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
14262 PR middle-end/80775
14263 * tree-cfg.c: Move deletion of unreachable case statements to after
14264 the merging of consecutive case labels.
14266 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
14268 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
14269 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
14270 restoring of callee-saved registers.
14272 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
14274 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
14275 * config/visium/visium.c (single_set_and_flags): Likewise.
14276 * config/visium/visium.md (Substitutions): Likewise.
14278 2017-05-17 Martin Liska <mliska@suse.cz>
14280 * cfg.c: Introduce dump_flags_t type and
14281 use it instead of int type.
14283 * cfghooks.c: Likewise.
14284 * cfghooks.h (struct cfg_hooks): Likewise.
14285 * cfgrtl.c: Likewise.
14286 * cfgrtl.h: Likewise.
14287 * cgraph.c (cgraph_node::get_body): Likewise.
14288 * coretypes.h: Likewise.
14289 * domwalk.c: Likewise.
14290 * domwalk.h: Likewise.
14291 * dumpfile.c (struct dump_option_value_info): Likewise.
14292 (dump_enable_all): Likewise.
14293 (dump_switch_p_1): Likewise.
14294 (opt_info_switch_p): Likewise.
14295 * dumpfile.h (enum tree_dump_index): Likewise.
14296 (struct dump_file_info): Likewise.
14297 * genemit.c: Likewise.
14298 * generic-match-head.c: Likewise.
14299 * gengtype.c (open_base_files): Likewise.
14300 * gimple-pretty-print.c: Likewise.
14301 * gimple-pretty-print.h: Likewise.
14302 * graph.c (print_graph_cfg): Likewise.
14303 * graphite-scop-detection.c (dot_all_sese): Likewise.
14304 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
14305 * loop-unroll.c (report_unroll): Likewise.
14306 * passes.c (pass_manager::register_one_dump_file): Likewise.
14307 * print-tree.c: Likewise.
14308 * statistics.c: Likewise.
14309 * tree-cfg.c: Likewise.
14310 * tree-cfg.h: Likewise.
14311 * tree-dfa.c: Likewise.
14312 * tree-dfa.h: Likewise.
14313 * tree-dump.c (dump_function): Likewise.
14314 * tree-dump.h (struct dump_info): Likewise.
14315 * tree-pretty-print.c: Likewise.
14316 * tree-pretty-print.h: Likewise.
14317 * tree-ssa-live.c: Likewise.
14318 * tree-ssa-live.h: Likewise.
14319 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
14320 * tree-vect-loop.c: Likewise.
14321 * tree-vect-slp.c: Likewise.
14323 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
14324 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14326 PR tree-optimization/80457
14327 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
14328 of all arguments to a statement as scalar_to_vec operations.
14329 (vectorizable_call): Adjust call to vect_model_simple_cost for
14331 (vectorizable_conversion): Likewise.
14332 (vectorizable_assignment): Likewise.
14333 (vectorizable_shift): Likewise.
14334 (vectorizable_operation): Likewise.
14335 (vectorizable_comparison): Likewise.
14336 (vect_is_simple_cond): Record the def types for operands.
14337 (vectorizable_condition): Likewise, call vect_model_simple_cost.
14338 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
14339 for statement argument count.
14341 2017-05-16 Carl Love <cel@us.ibm.com>
14343 * config/rs6000/rs6000-c: Add support for built-in functions
14344 vector unsigned long long vec_bperm (vector unsigned long long,
14345 vector unsigned char)
14346 vector signed long long vec_mule (vector signed int,
14348 vector unsigned long long vec_mule (vector unsigned int,
14349 vector unsigned int)
14350 vector signed long long vec_mulo (vector signed int,
14352 vector unsigned long long vec_mulo (vector unsigned int,
14353 vector unsigned int)
14354 vector signed char vec_sldw (vector signed char,
14355 vector signed char,
14357 vector unsigned char vec_sldw (vector unsigned char,
14358 vector unsigned char,
14360 vector signed short vec_sldw (vector signed short,
14361 vector signed short,
14363 vector unsigned short vec_sldw (vector unsigned short,
14364 vector unsigned short,
14366 vector signed int vec_sldw (vector signed int,
14369 vector unsigned int vec_sldw (vector unsigned int,
14370 vector unsigned int,
14372 vector signed long long vec_sldw (vector signed long long,
14373 vector signed long long,
14375 vector unsigned long long vec_sldw (vector unsigned long long,
14376 vector unsigned long long,
14378 * config/rs6000/rs6000-c: Add support for built-in functions
14379 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
14380 * config/rs6000/altivec.h: Add defintion for vec_sldw.
14381 * doc/extend.texi: Update the built-in documentation for the
14382 new built-in functions.
14384 2017-05-16 Marek Polacek <polacek@redhat.com>
14388 * tree.c (save_expr): Don't fold the expression.
14390 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
14392 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
14393 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
14394 and (?*y,m). Update insn attributes.
14396 2017-05-16 Martin Liska <mliska@suse.cz>
14398 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
14399 flags argument of print_gimple_stmt, print_gimple_expr,
14400 print_generic_stmt and print_generic_expr.
14401 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
14402 * coretypes.h: Likewise.
14403 * except.c (dump_eh_tree): Likewise.
14404 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14405 * gimple-pretty-print.h: Likewise.
14406 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
14407 (backprop::push_to_worklist): Likewise.
14408 (backprop::pop_from_worklist): Likewise.
14409 (backprop::process_use): Likewise.
14410 (backprop::intersect_uses): Likewise.
14411 (note_replacement): Likewise.
14412 * gimple-ssa-store-merging.c
14413 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
14414 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
14415 (pass_store_merging::execute): Likewise.
14416 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
14417 (ssa_base_cand_dump_callback): Likewise.
14418 (dump_incr_vec): Likewise.
14419 (replace_refs): Likewise.
14420 (replace_mult_candidate): Likewise.
14421 (create_add_on_incoming_edge): Likewise.
14422 (create_phi_basis): Likewise.
14423 (insert_initializers): Likewise.
14424 (all_phi_incrs_profitable): Likewise.
14425 (introduce_cast_before_cand): Likewise.
14426 (replace_one_candidate): Likewise.
14427 * gimplify.c (gimplify_expr): Likewise.
14428 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
14429 (set_rename): Likewise.
14430 (rename_uses): Likewise.
14431 (copy_loop_phi_nodes): Likewise.
14432 (add_close_phis_to_merge_points): Likewise.
14433 (copy_loop_close_phi_args): Likewise.
14434 (copy_cond_phi_args): Likewise.
14435 (graphite_copy_stmts_from_block): Likewise.
14436 (translate_pending_phi_nodes): Likewise.
14437 * graphite-poly.c (print_pdr): Likewise.
14438 (dump_gbb_cases): Likewise.
14439 (dump_gbb_conditions): Likewise.
14440 (print_scop_params): Likewise.
14441 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
14442 (build_cross_bb_scalars_use): Likewise.
14443 (gather_bbs::before_dom_children): Likewise.
14444 * hsa-dump.c (dump_hsa_immed): Likewise.
14445 * ipa-cp.c (print_ipcp_constant_value): Likewise.
14446 (get_replacement_map): Likewise.
14447 * ipa-inline-analysis.c (dump_condition): Likewise.
14448 (estimate_function_body_sizes): Likewise.
14449 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
14450 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
14451 * ipa-prop.c (ipa_dump_param): Likewise.
14452 (ipa_print_node_jump_functions_for_edge): Likewise.
14453 (ipa_modify_call_arguments): Likewise.
14454 (ipa_modify_expr): Likewise.
14455 (ipa_dump_param_adjustments): Likewise.
14456 (ipa_dump_agg_replacement_values): Likewise.
14457 (ipcp_modif_dom_walker::before_dom_children): Likewise.
14458 * ipa-pure-const.c (check_stmt): Likewise.
14459 (pass_nothrow::execute): Likewise.
14460 * ipa-split.c (execute_split_functions): Likewise.
14461 * omp-offload.c (dump_oacc_loop_part): Likewise.
14462 (dump_oacc_loop): Likewise.
14463 * trans-mem.c (tm_log_emit): Likewise.
14464 (tm_memopt_accumulate_memops): Likewise.
14465 (dump_tm_memopt_set): Likewise.
14466 (dump_tm_memopt_transform): Likewise.
14467 * tree-cfg.c (gimple_verify_flow_info): Likewise.
14468 (print_loop): Likewise.
14469 * tree-chkp-opt.c (chkp_print_addr): Likewise.
14470 (chkp_gather_checks_info): Likewise.
14471 (chkp_get_check_result): Likewise.
14472 (chkp_remove_check_if_pass): Likewise.
14473 (chkp_use_outer_bounds_if_possible): Likewise.
14474 (chkp_reduce_bounds_lifetime): Likewise.
14475 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
14476 (chkp_mark_completed_bounds): Likewise.
14477 (chkp_register_incomplete_bounds): Likewise.
14478 (chkp_mark_invalid_bounds): Likewise.
14479 (chkp_maybe_copy_and_register_bounds): Likewise.
14480 (chkp_build_returned_bound): Likewise.
14481 (chkp_get_bound_for_parm): Likewise.
14482 (chkp_build_bndldx): Likewise.
14483 (chkp_get_bounds_by_definition): Likewise.
14484 (chkp_generate_extern_var_bounds): Likewise.
14485 (chkp_get_bounds_for_decl_addr): Likewise.
14486 * tree-chrec.c (chrec_apply): Likewise.
14487 * tree-data-ref.c (dump_data_reference): Likewise.
14488 (dump_subscript): Likewise.
14489 (dump_data_dependence_relation): Likewise.
14490 (analyze_overlapping_iterations): Likewise.
14491 * tree-inline.c (expand_call_inline): Likewise.
14492 (tree_function_versioning): Likewise.
14493 * tree-into-ssa.c (dump_defs_stack): Likewise.
14494 (dump_currdefs): Likewise.
14495 (dump_names_replaced_by): Likewise.
14496 (dump_update_ssa): Likewise.
14497 (update_ssa): Likewise.
14498 * tree-object-size.c (pass_object_sizes::execute): Likewise.
14499 * tree-parloops.c (build_new_reduction): Likewise.
14500 (try_create_reduction_list): Likewise.
14501 (ref_conflicts_with_region): Likewise.
14502 (oacc_entry_exit_ok_1): Likewise.
14503 (oacc_entry_exit_single_gang): Likewise.
14504 * tree-pretty-print.h: Likewise.
14505 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
14506 (get_scalar_evolution): Likewise.
14507 (add_to_evolution): Likewise.
14508 (get_loop_exit_condition): Likewise.
14509 (analyze_evolution_in_loop): Likewise.
14510 (analyze_initial_condition): Likewise.
14511 (analyze_scalar_evolution): Likewise.
14512 (instantiate_scev): Likewise.
14513 (number_of_latch_executions): Likewise.
14514 (gather_chrec_stats): Likewise.
14515 (final_value_replacement_loop): Likewise.
14516 (scev_const_prop): Likewise.
14517 * tree-sra.c (dump_access): Likewise.
14518 (disqualify_candidate): Likewise.
14519 (create_access): Likewise.
14520 (reject): Likewise.
14521 (maybe_add_sra_candidate): Likewise.
14522 (create_access_replacement): Likewise.
14523 (analyze_access_subtree): Likewise.
14524 (analyze_all_variable_accesses): Likewise.
14525 (sra_modify_assign): Likewise.
14526 (initialize_constant_pool_replacements): Likewise.
14527 (find_param_candidates): Likewise.
14528 (decide_one_param_reduction): Likewise.
14529 (replace_removed_params_ssa_names): Likewise.
14530 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
14531 * tree-ssa-copy.c (dump_copy_of): Likewise.
14532 (copy_prop_visit_cond_stmt): Likewise.
14533 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
14534 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
14535 (record_equivalences_from_stmt): Likewise.
14536 * tree-ssa-dse.c (compute_trims): Likewise.
14537 (delete_dead_call): Likewise.
14538 (delete_dead_assignment): Likewise.
14539 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
14540 (forward_propagate_into_cond): Likewise.
14541 (pass_forwprop::execute): Likewise.
14542 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14543 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
14545 (move_computations_worker): Likewise.
14546 (execute_sm): Likewise.
14547 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
14548 (remove_exits_and_undefined_stmts): Likewise.
14549 (remove_redundant_iv_tests): Likewise.
14550 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
14551 (adjust_iv_update_pos): Likewise.
14552 * tree-ssa-math-opts.c (bswap_replace): Likewise.
14553 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
14554 (value_replacement): Likewise.
14555 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
14556 * tree-ssa-pre.c (print_pre_expr): Likewise.
14557 (get_representative_for): Likewise.
14558 (create_expression_by_pieces): Likewise.
14559 (insert_into_preds_of_block): Likewise.
14560 (eliminate_insert): Likewise.
14561 (eliminate_dom_walker::before_dom_children): Likewise.
14562 (eliminate): Likewise.
14563 (remove_dead_inserted_code): Likewise.
14564 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
14565 * tree-ssa-reassoc.c (get_rank): Likewise.
14566 (eliminate_duplicate_pair): Likewise.
14567 (eliminate_plus_minus_pair): Likewise.
14568 (eliminate_not_pairs): Likewise.
14569 (undistribute_ops_list): Likewise.
14570 (eliminate_redundant_comparison): Likewise.
14571 (update_range_test): Likewise.
14572 (optimize_range_tests_var_bound): Likewise.
14573 (optimize_vec_cond_expr): Likewise.
14574 (rewrite_expr_tree): Likewise.
14575 (rewrite_expr_tree_parallel): Likewise.
14576 (linearize_expr): Likewise.
14577 (break_up_subtract): Likewise.
14578 (linearize_expr_tree): Likewise.
14579 (attempt_builtin_powi): Likewise.
14580 (attempt_builtin_copysign): Likewise.
14581 (transform_stmt_to_copy): Likewise.
14582 (transform_stmt_to_multiply): Likewise.
14583 (dump_ops_vector): Likewise.
14584 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
14585 (print_scc): Likewise.
14586 (set_ssa_val_to): Likewise.
14587 (visit_reference_op_store): Likewise.
14588 (visit_use): Likewise.
14589 (sccvn_dom_walker::before_dom_children): Likewise.
14590 (run_scc_vn): Likewise.
14591 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14593 (expr_hash_elt::print): Likewise.
14594 (const_and_copies::pop_to_marker): Likewise.
14595 (const_and_copies::record_const_or_copy_raw): Likewise.
14596 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
14597 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
14598 (dump_predicates): Likewise.
14599 (find_uninit_use): Likewise.
14600 (warn_uninitialized_phi): Likewise.
14601 (pass_late_warn_uninitialized::execute): Likewise.
14602 * tree-ssa.c (verify_vssa): Likewise.
14603 (verify_ssa): Likewise.
14604 (maybe_optimize_var): Likewise.
14605 * tree-vrp.c (dump_value_range): Likewise.
14606 (dump_all_value_ranges): Likewise.
14607 (dump_asserts_for): Likewise.
14608 (register_edge_assert_for_2): Likewise.
14609 (vrp_visit_cond_stmt): Likewise.
14610 (vrp_visit_switch_stmt): Likewise.
14611 (vrp_visit_stmt): Likewise.
14612 (vrp_visit_phi_node): Likewise.
14613 (simplify_cond_using_ranges_1): Likewise.
14614 (fold_predicate_in): Likewise.
14615 (evrp_dom_walker::before_dom_children): Likewise.
14616 (evrp_dom_walker::push_value_range): Likewise.
14617 (evrp_dom_walker::pop_value_range): Likewise.
14618 (execute_early_vrp): Likewise.
14620 2017-05-16 Richard Biener <rguenther@suse.de>
14622 * dwarf2out.c (loc_list_from_tree_1): Do not create
14623 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
14625 2017-05-16 Richard Biener <rguenther@suse.de>
14627 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
14629 (note_variable_value_in_expr): If we resolved the decl ref
14630 do not push to the stack.
14632 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
14634 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
14635 operations in fast-math mode.
14636 (vaddq_f16): Likewise.
14637 (vmul_f16): Likewise.
14638 (vmulq_f16): Likewise.
14639 (vsub_f16): Likewise.
14640 (vsubq_f16): Likewise.
14641 * config/arm/neon.md (add<mode>3): New.
14643 (fma:<VH:mode>3): New. Also remove outdated comment.
14646 2017-05-16 Martin Liska <mliska@suse.cz>
14650 * ipa-devirt.c (warn_types_mismatch): Fix typo.
14651 (odr_types_equivalent_p): Likewise.
14653 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
14655 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
14657 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
14660 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
14661 non-interunit SSE move alternatives with '?'.
14662 (zero-extendsidi peephole2): New peephole to skip intermediate
14663 general register in SSE zero-extend sequence.
14665 2017-05-15 Jeff Law <law@redhat.com>
14667 * reorg.c (relax_delay_slots): Create a new variable to hold
14668 the temporary target rather than clobbering TARGET_LABEL.
14670 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
14671 missing argument to extract_bit_field call.
14672 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
14674 2017-05-15 Martin Liska <mliska@suse.cz>
14677 * gcc.c (process_command): Do not allow empty argument of -o option.
14679 2017-05-15 Renlin Li <renlin.li@arm.com>
14681 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
14682 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
14683 * config/aarch64/constraints.md (Usf): Add long call check.
14684 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
14685 (call_value): Likewise.
14686 (sibcall): Likewise.
14687 (sibcall_value): Likewise.
14689 (call_value_insn): New.
14690 (sibcall_insn): Update rtx pattern.
14691 (sibcall_value_insn): Likewise.
14692 (call_internal): Remove.
14693 (call_value_internal): Likewise.
14694 (sibcall_internal): Likewise.
14695 (sibcall_value_internal): Likewise.
14696 (call_reg): Likewise.
14697 (call_symbol): Likewise.
14698 (call_value_reg): Likewise.
14699 (call_value_symbol): Likewise.
14701 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
14704 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
14706 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
14708 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
14709 compatible with CCGOCmode and with CCZmode.
14711 2017-05-14 Martin Sebor <msebor@redhat.com>
14713 PR middle-end/77671
14714 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
14715 (gimple_fold_builtin_snprintf): Same.
14716 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
14717 (gimple_fold_builtin_snprintf): Same.
14718 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
14719 of character types.
14720 (is_call_safe): New function.
14721 (try_substitute_return_value): Call it.
14722 (try_simplify_call): New function.
14723 (pass_sprintf_length::handle_gimple_call): Call it.
14725 2017-05-14 Martin Sebor <msebor@redhat.com>
14727 PR middle-end/80669
14728 * builtins.c (expand_builtin_stpncpy): Simplify.
14730 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
14732 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
14733 * config/i386/i386.h
14734 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
14735 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
14736 (struct machine_function): Add new members call_ms2sysv,
14737 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
14738 (struct machine_frame_state): New fields sp_realigned and
14739 sp_realigned_offset.
14740 * config/i386/i386.c
14741 (enum xlogue_stub): New enum.
14742 (enum xlogue_stub_sets): New enum.
14743 (class xlogue_layout): New class.
14744 (struct ix86_frame): New fields stack_realign_allocate_offset,
14745 stack_realign_offset and outlined_save_offset. Modify comments to
14746 detail stack layout when using out-of-line stubs.
14747 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
14748 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
14749 -mcall-ms2sysv-xlogues.
14750 (stub_managed_regs): New static variable.
14751 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
14752 registers managed by out-of-line stub.
14753 (disable_call_ms2sysv_xlogues): New function.
14754 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
14755 m->call_ms2sysv when appropriate and compute frame layout for
14757 (sp_valid_at, fp_valid_at): New inline functions.
14758 (choose_basereg): New function.
14759 (choose_baseaddr): Add align parameter, use choose_basereg and modify
14761 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
14762 Use align parameter of choose_baseaddr to generated aligned SSE movs
14764 (pro_epilogue_adjust_stack): Modify to track
14765 machine_frame_state::sp_realigned.
14766 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
14767 (ix86_nsaved_sseregs): Likewise.
14768 (ix86_emit_save_regs): Likewise.
14769 (ix86_emit_save_regs_using_mov): Likewise.
14770 (ix86_emit_save_sse_regs_using_mov): Likewise.
14771 (get_scratch_register_on_entry): Likewise.
14772 (gen_frame_set): New function.
14773 (gen_frame_load): Likewise.
14774 (gen_frame_store): Likewise.
14775 (emit_outlined_ms2sysv_save): Likewise.
14776 (emit_outlined_ms2sysv_restore): Likewise.
14777 (ix86_expand_prologue): Modify stack re-alignment code and call
14778 emit_outlined_ms2sysv_save when appropriate.
14779 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
14780 parameter rtx_insn *insn, which allows the function to be used to only
14781 generate the notes.
14782 (ix86_expand_epilogue): Modify validity checks of frame and stack
14783 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
14784 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
14785 * config/i386/predicates.md
14786 (save_multiple): New predicate.
14787 (restore_multiple): Likewise.
14788 * config/i386/sse.md
14789 (save_multiple<mode>): New pattern.
14790 (save_multiple_realign<mode>): Likewise.
14791 (restore_multiple<mode>): Likewise.
14792 (restore_multiple_and_return<mode>): Likewise.
14793 (restore_multiple_leave_return<mode>): Likewise.
14794 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
14796 2017-05-14 Julia Koval <julia.koval@intel.com>
14798 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
14799 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
14800 (__builtin_ia32_xsetbv): New builtins.
14801 * config/i386/i386.c (ix86_expand_special_args_builtin):
14803 (ix86_expand_builtin): Special expand for new intrinsics.
14804 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
14805 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
14806 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
14808 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14810 * cfganal.c (inverted_post_order_compute): Change argument type
14812 * cfganal.h (inverted_post_order_compute): Adjust prototype.
14813 * df-core.c (rest_of_handle_df_initialize): Adjust.
14814 (rest_of_handle_df_finish): Likewise.
14815 (df_analyze_1): Likewise.
14816 (df_analyze): Likewise.
14817 (loop_inverted_post_order_compute): Change argument to be a vec *.
14818 (df_analyze_loop): Adjust.
14819 (df_get_n_blocks): Likewise.
14820 (df_get_postorder): Likewise.
14821 * df.h (struct df_d): Change field to be a vec.
14822 * lcm.c (compute_laterin): Adjust.
14823 (compute_available): Likewise.
14824 * lra-lives.c (lra_create_live_ranges_1): Likewise.
14825 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
14826 * tree-ssa-pre.c (compute_antic): Likewise.
14828 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14830 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
14831 (depth_first_search::depth_first_search): Change structure init
14832 function to this constructor.
14833 (depth_first_search::add_bb): Rename function to this member.
14834 (depth_first_search::execute): Likewise.
14835 (flow_dfs_compute_reverse_finish): Adjust.
14837 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14839 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
14840 (longest_simple_path): Likewise.
14841 * shrink-wrap.c (spread_components): Likewise.
14842 (disqualify_problematic_components): Likewise.
14843 (emit_common_heads_for_components): Likewise.
14844 (emit_common_tails_for_components): Likewise.
14845 (insert_prologue_epilogue_for_components): Likewise.
14847 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14849 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
14852 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14854 * df-core.c (df_set_blocks): Start using auto_bitmap.
14855 (df_compact_blocks): Likewise.
14856 * df-problems.c (df_rd_confluence_n): Likewise.
14857 * df-scan.c (df_insn_rescan_all): Likewise.
14858 (df_process_deferred_rescans): Likewise.
14859 (df_update_entry_block_defs): Likewise.
14860 (df_update_exit_block_uses): Likewise.
14861 (df_entry_block_bitmap_verify): Likewise.
14862 (df_exit_block_bitmap_verify): Likewise.
14863 (df_scan_verify): Likewise.
14864 * lra-constraints.c (lra_constraints): Likewise.
14865 (undo_optional_reloads): Likewise.
14866 (lra_undo_inheritance): Likewise.
14867 * lra-remat.c (calculate_gen_cands): Likewise.
14868 (do_remat): Likewise.
14869 * lra-spills.c (assign_spill_hard_regs): Likewise.
14870 (spill_pseudos): Likewise.
14871 * tree-ssa-pre.c (bitmap_set_and): Likewise.
14872 (bitmap_set_subtract_values): Likewise.
14874 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14876 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
14877 management with auto_bitmap.
14878 (fix_inter_tick): Likewise.
14879 (fix_recovery_deps): Likewise.
14880 * ira.c (add_store_equivs): Likewise.
14881 (find_moveable_pseudos): Likewise.
14882 (split_live_ranges_for_shrink_wrap): Likewise.
14883 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
14884 (rtx_reuse_manager::seen_def_p): Likewise.
14885 (rtx_reuse_manager::set_seen_def): Likewise.
14886 * print-rtl.h (class rtx_reuse_manager): Likewise.
14888 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14890 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
14892 (migrate_btr_def): Likewise.
14893 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
14894 * df-core.c (loop_post_order_compute): Likewise.
14895 (loop_inverted_post_order_compute): Likewise.
14896 * hsa-common.h: Likewise.
14897 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
14898 * init-regs.c (initialize_uninitialized_regs): Likewise.
14899 * ipa-inline.c (resolve_noninline_speculation): Likewise.
14900 (inline_small_functions): Likewise.
14901 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
14902 * ira.c (combine_and_move_insns): Likewise.
14903 (build_insn_chain): Likewise.
14904 * loop-invariant.c (find_invariants): Likewise.
14905 * lower-subreg.c (propagate_pseudo_copies): Likewise.
14906 * predict.c (tree_predict_by_opcode): Likewise.
14907 (predict_paths_leading_to): Likewise.
14908 (predict_paths_leading_to_edge): Likewise.
14909 (estimate_loops_at_level): Likewise.
14910 (estimate_loops): Likewise.
14911 * shrink-wrap.c (try_shrink_wrapping): Likewise.
14912 (spread_components): Likewise.
14913 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
14914 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
14915 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
14916 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
14917 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
14918 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
14919 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
14920 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
14921 (mark_threaded_blocks): Likewise.
14922 (thread_through_all_blocks): Likewise.
14923 * tree-ssa.c (verify_ssa): Likewise.
14924 (execute_update_addresses_taken): Likewise.
14925 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
14927 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14929 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
14931 (post_order_compute): Likewise.
14932 (inverted_post_order_compute): Likewise.
14933 (pre_and_rev_post_order_compute_fn): Likewise.
14935 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14937 * genrecog.c (int_set::int_set): Explicitly construct our
14938 auto_vec base class.
14939 * vec.h (auto_vec::auto_vec): New constructor.
14941 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14943 * bitmap.h (class auto_bitmap): New constructor taking
14944 bitmap_obstack * argument.
14946 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14948 * bitmap.h (class auto_bitmap): Change type of m_bits to
14949 bitmap_head, and adjust ctor / dtor and member operators.
14951 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
14953 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
14954 when returned register mode doesn't match original mode.
14956 2017-05-12 Jeff Law <law@redhat.com>
14957 Jakub Jelinek <jakub@redhat.com>
14959 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
14960 we look for cc setter after the compare-elim changes.
14961 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
14962 within the vector to match what compare-elim now expects.
14963 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
14964 (xorsi3_flags, one_cmplsi2_flags): Likewise.
14966 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
14967 after the compare-elim changes.
14968 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
14969 the vector to match what compare-elim now expects.
14970 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
14971 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
14972 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
14973 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
14974 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
14976 * config/visium/visium.c (single_set_and_flags): Fix where
14977 we look for cc setter after the compare-elim changes.
14978 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
14979 with the vector to match what compare-elim now expects.
14980 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
14981 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
14982 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
14983 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
14984 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
14985 (neg<mode>2_insn_set_overflow): Likewise.
14987 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
14989 PR middle-end/79794
14990 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
14991 maybe_expand_insn call, set ops[0].target. If still set after call,
14992 set alt_rtl. Add extra arg to recursive calls.
14993 (extract_bit_field): Add alt_rtl argument. Pass to
14995 * expmed.h (extract_bit_field): Fix prototype.
14996 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
14997 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
14998 to extract_bit_field_calls.
14999 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
15000 Pass alt_rtl to extract_bit_field calls.
15001 * calls.c (store_unaligned_arguments_into_psuedos)
15002 load_register_parameters): Pass extra NULL to extract_bit_field calls.
15003 * optabs.c (maybe_legitimize_operand): Clear op->target when call
15005 * optabs.h (struct expand_operand): Add target bitfield.
15007 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15009 * compare-elim.c (try_eliminate_compare): Canonicalize
15010 operation with embedded compare to
15011 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
15012 (set (reg) (operation)].
15014 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
15016 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15019 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
15020 cost of adding a carry flag for ADC instruction.
15021 [case MINUS]: Ignore the cost of subtracting a carry flag
15022 for SBB instruction.
15024 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
15026 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
15028 * config/rs6000/bmiintrin.h: New file.
15029 * config/rs6000/bmi2intrin.h: New file.
15030 * config/rs6000/x86intrin.h: New file.
15032 2017-05-12 Jeff Law <law@redhat.com>
15034 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
15037 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
15039 PR middle-end/80707
15040 * tree-cfg.c: Remove cfg edges of unreachable case statements.
15042 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15044 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15045 early expansion of vector divide builtins.
15046 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
15047 builtins identified as having unsigned arguments.
15049 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15051 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
15052 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
15053 expansion of vector logical operations (and, andc, or, xor,
15056 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15058 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
15059 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
15061 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15063 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15064 early GIMPLE expansion of vector multiplies.
15066 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15068 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
15069 TARGET_HAVE_MOVT conditional.
15070 (movt splitter): Likewise.
15072 2017-05-12 Richard Biener <rguenther@suse.de>
15074 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
15075 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15076 Fold all stmts not inplace.
15078 2017-05-12 Richard Biener <rguenther@suse.de>
15080 PR tree-optimization/80713
15081 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
15082 inserted_exprs bit for not removed stmts.
15084 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
15086 PR middle-end/69921
15087 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
15088 parallelized" attribute for parallelized OpenACC kernels.
15089 * omp-offload.c (execute_oacc_device_lower): Use it.
15091 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
15092 Set "oacc kernels" attribute.
15093 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
15094 parameter. Adjust all users.
15095 (oacc_fn_attrib_kernels_p): Remove function.
15096 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
15097 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
15098 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
15099 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
15100 assert "oacc kernels" attribute is set.
15102 2017-05-11 Carl Love <cel@us.ibm.com>
15104 * config/rs6000/rs6000-c: Add support for built-in functions
15105 vector unsigned char vec_popcnt (vector signed char)
15106 vector unsigned char vec_popcnt (vector unsigned char)
15107 vector unsigned short vec_popcnt (vector signed short)
15108 vector unsigned short vec_popcnt (vector unsigned short)
15109 vector unsigned int vec_popcnt (vector signed int)
15110 vector unsigned int vec_popcnt (vector unsigned int)
15111 vector unsigned long long vec_popcnt (vector signed long long)
15112 vector unsigned long long vec_popcnt (vector unsigned long long)
15113 vector signed long long vec_slo (vector signed long long,
15114 vector signed char)
15115 vector signed long long vec_slo (vector signed long long,
15116 vector unsigned char)
15117 vector unsigned long long vec_slo (vector unsigned long long,
15118 vector signed char)
15119 vector unsigned long long vec_slo (vector unsigned long long,
15120 vector unsigned char)
15121 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
15122 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
15123 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
15124 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
15125 * doc/extend.texi: Update the built-in documentation file for the
15126 new built-in functions.
15128 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
15130 * attribs.h (sorted_attr_string): Move machine independent
15131 functions for target clone support from the i386 port to common
15132 code. Rename ix86_function_versions to common_function_versions.
15133 Rename make_name to make_unique_name.
15134 (common_function_versions): Likewise.
15135 (make_unique_name): Likewise.
15136 (make_dispatcher_decl): Likewise.
15137 (is_function_default_version): Likewise.
15138 * attribs.c (attr_strcmp): Likewise.
15139 (sorted_attr_string): Likewise.
15140 (common_function_versions): Likewise.
15141 (make_unique_name): Likewise.
15142 (make_dispatcher_decl): Likewise.
15143 (is_function_default_version): Likewise.
15144 * config/i386/i386.c (attr_strcmp): Likewise.
15145 (sorted_attr_string): Likewise.
15146 (ix86_function_versions): Likewise.
15147 (make_name): Likewise.
15148 (make_dispatcher_decl): Likewise.
15149 (is_function_default_version): Likewise.
15150 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
15152 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15155 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15156 Account for direct move costs for vec_construct of integer
15159 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
15162 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
15163 (UNSPEC_STX_ATOMIC): Ditto.
15164 (loaddi_via_sse): New insn.
15165 (storedi_via_sse): Ditto.
15166 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
15167 Update corresponding peephole2 patterns.
15168 (atomic_storedi_fpu): Ditto.
15170 2017-05-11 Julia Koval <julia.koval@intel.com>
15172 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
15173 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
15175 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
15176 (__builtin_ia32_rsqrt14ss_mask): New builtins.
15177 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
15179 2017-05-11 Nathan Sidwell <nathan@acm.org>
15181 * graphite-poly.c: Include dumpfile.h.
15183 * dumpfle.h (dump_function): Declare here ...
15184 * tree-dump.h (dump_function): ... not here.
15185 * dumpfile.c: #include tree-cfg.h.
15186 (dump_function): Move here from ...
15187 * tree-dump.c (dump_function): ... here.
15188 * gimplify.c: #include splay-tree.h, not tree-dump.h.
15189 * graphite-poly.c: Don't include tree-dump.h.
15190 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
15191 * print-tree.c: Likewise.
15192 * stor-layout.c: Likewise.
15193 * tree-nested.c: Likewise.
15195 * dumpfile.c (dump_start): Use TDF_FLAGS.
15196 (dump_enable_all): Fix TDF_KIND check thinko.
15198 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
15200 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15201 array entries to represent two legal parameterizations of the
15202 overloaded __builtin_cmpb function, as represented by the
15203 P6_OV_BUILTIN_CMPB constant.
15204 (altivec_resolve_overloaded_builtin): Add special case handling
15205 for the __builtin_cmpb function, as represented by the
15206 P6_OV_BUILTIN_CMPB constant.
15207 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
15208 (BU_P6_64BIT_2): New macro.
15209 (BU_P6_OVERLOAD_2): New macro
15210 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
15211 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
15212 (CMPB): Add overload support to represent both 32-bit and 64-bit
15213 compare-bytes function.
15214 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15215 support for TARGET_CMPB.
15216 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
15217 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15218 documentation of the __builtin_cmpb overloaded built-in function.
15220 2017-05-11 Richard Biener <rguenther@suse.de>
15222 PR tree-optimization/80705
15223 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
15224 bases are not vectorizable.
15226 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15228 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
15229 when counting register pressure.
15231 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15233 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
15234 (iv_ca_more_deps): Renamed to ...
15235 (iv_ca_compare_deps): ... this.
15236 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
15238 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15240 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
15242 (determine_group_iv_costs): ... here.
15243 (find_inv_vars_cb): Record inv var if it's not recorded before.
15245 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15247 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
15248 (get_shiftadd_cost): Ditto.
15250 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15252 * tree-ssa-address.c: Include header file.
15253 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
15255 (add_to_parts): Refactor.
15256 (addr_to_parts): New parameter. Update use of move_hint_to_base.
15257 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
15260 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15262 PR tree-optimization/53090
15263 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
15265 (extract_cond_operands): Detect condition with IV on both sides
15266 and return COMP_IV_EXPR_2.
15267 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
15268 (rewrite_use_compare): Simplify by removing call to function
15269 extract_cond_operands.
15271 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15273 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
15274 (extract_cond_operands): Detect condition comparing against non-
15275 invariant bound and return appropriate enum value.
15276 (find_interesting_uses_cond): Update use of extract_cond_operands.
15277 Handle its return value accordingly.
15278 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
15280 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15282 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
15283 nonlinear iv_use computation in loop invariant sensitive way.
15285 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15287 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
15288 (find_iv_candidates): Call relate_compare_use_with_all_cands.
15290 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15292 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
15293 (dump_cand): Support iv_cand.inv_exprs.
15294 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
15296 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
15299 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15301 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
15303 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
15304 as local function. Include necessary header files.
15305 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
15307 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15309 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
15311 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15313 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
15314 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
15315 RSHIFT_EXPR and BIT_NOT_EXPR.
15317 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15319 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
15320 (adjust_setup_cost): New parameter supporting round up adjustment.
15321 (struct address_cost_data): Delete.
15322 (force_expr_to_var_cost): Don't bound cost with spill_cost.
15323 (split_address_cost, ptr_difference_cost): Delete.
15324 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
15325 (struct ainc_cost_data): New struct.
15326 (get_address_cost_ainc): New function.
15327 (get_address_cost, get_computation_cost): Reimplement.
15328 (determine_group_iv_cost_address): Record inv_expr for all uses of
15330 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
15331 (iv_ca_has_deps): Reimplemented to ...
15332 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
15334 (iv_ca_extend): Call iv_ca_more_deps.
15336 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15338 * tree-ssa-address.c (struct mem_address): Move to header file.
15339 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
15340 * tree-ssa-address.h (struct mem_address): Move from C file.
15341 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
15343 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15345 * tree-affine.h (aff_combination_type): New interface.
15346 (aff_combination_zero_p): Remove static.
15347 (aff_combination_const_p): New interface.
15348 (aff_combination_singleton_var_p): New interfaces.
15350 2017-05-11 Richard Biener <rguenther@suse.de>
15352 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15353 Skip unreachable blocks and destinations.
15354 (eliminate): Move stmt removal and fixup ...
15355 (fini_eliminate): ... here. Skip inserted exprs.
15356 (pass_pre::execute): Move fini_pre after fini_eliminate.
15357 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
15358 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
15359 PRE to get rid of dead code that has invalid SSA form and
15360 split critical edges again.
15362 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15364 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15366 2017-05-11 Richard Biener <rguenther@suse.de>
15368 * passes.c (execute_function_todo): Verify loops if they are
15369 said to be up-to-date.
15370 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
15371 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
15373 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
15376 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
15377 handle calling assemble_external ourself.
15380 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
15381 modes with zero size. Enhance comment.
15383 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15385 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
15386 built-ins for vec_xl and vec_xst with short and char pointer
15389 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15391 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
15392 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
15393 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
15394 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
15395 (_mm_maskz_min_round_ss): New intrinsics.
15396 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
15397 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
15398 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
15399 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
15400 (__builtin_ia32_minss_mask_round): New builtins.
15401 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15402 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15403 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
15405 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
15406 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
15408 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
15411 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15413 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
15414 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
15415 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
15416 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
15417 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
15418 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
15419 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
15420 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15421 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
15422 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
15423 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
15424 (__builtin_ia32_mulss_mask_round): New builtins.
15425 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15426 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15427 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
15429 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
15430 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
15432 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
15435 2017-05-10 Julia Koval <julia.koval@intel.com>
15437 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
15438 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
15439 (_mm256_setr_m128i): New intrinsics.
15441 2017-05-10 Julia Koval <julia.koval@intel.com>
15443 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
15444 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
15445 (_mm_maskz_rcp14_ss): New intrinsics.
15446 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
15447 (__builtin_ia32_rcp14ss_mask): New builtins.
15448 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
15450 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
15452 PR tree-optimization/51513
15453 * tree-cfg.c (gimple_seq_unreachable_p): New function.
15454 (assert_unreachable_fallthru_edge_p): Use it.
15455 (group_case_labels_stmt): Likewise.
15456 * tree-cfg.h: Prototype it.
15457 * stmt.c: Include cfghooks.h and tree-cfg.h.
15458 (emit_case_dispatch_table) <gap_label>: New local variable.
15459 Use it to fill dispatch table gaps.
15460 Test for default_label before updating probabilities.
15461 (expand_case) <default_label>: Remove unneeded initialization.
15462 Test for unreachable default case statement and remove its edge.
15463 Set default_label accordingly.
15464 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
15466 2017-05-10 Carl Love <cel@us.ibm.com>
15468 * config/rs6000/rs6000-c: Add support for built-in functions
15469 vector signed char vec_neg (vector signed char)
15470 vector signed short int vec_neg (vector short int)
15471 vector signed int vec_neg (vector signed int)
15472 vector signed long long vec_neg (vector signed long long)
15473 vector float vec_neg (vector float)
15474 vector double vec_neg (vector double)
15475 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
15477 * config/rs6000/altivec.h: Add define for vec_neg
15478 * doc/extend.texi: Update the built-in documentation for the
15479 new built-in functions.
15481 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15483 PR tree-optimization/77644
15484 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
15486 2017-05-10 Nathan Sidwell <nathan@acm.org>
15488 * dumpfile.h (TDI_lang_all): New.
15489 (TDF_KIND): New. Renumber others
15490 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
15492 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
15494 (get_dump_file_name): Adjust suffix generation.
15495 (dump_enable_all): Use TDF_KIND.
15496 * doc/invoke.texi (-fdump-lang-all): Document.
15498 * dumpfile.h: Tabify.
15500 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
15503 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
15504 Move member access before delete.
15506 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
15508 * tree-inline.c (expand_call_inline): Split block at stmt
15511 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
15514 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
15515 are now unused after splitting mov{sf,sd}_hardfloat.
15516 (f32_lr2): Likewise.
15517 (f32_lm): Likewise.
15518 (f32_lm2): Likewise.
15519 (f32_li): Likewise.
15520 (f32_li2): Likewise.
15521 (f32_lv): Likewise.
15522 (f32_sr): Likewise.
15523 (f32_sr2): Likewise.
15524 (f32_sm): Likewise.
15525 (f32_sm2): Likewise.
15526 (f32_si): Likewise.
15527 (f32_si2): Likewise.
15528 (f32_sv): Likewise.
15529 (f32_dm): Likewise.
15530 (f32_vsx): Likewise.
15531 (f32_av): Likewise.
15532 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
15533 For movsf, order stores so the VSX stores occur before the GPR
15534 store which encourages the register allocator to use a traditional
15535 FPR instead of a GPR. For movsd, order the stores so that the GPR
15536 store comes before the VSX stores to allow the power6 to work.
15537 This is due to the power6 not having a 32-bit integer store
15538 instruction from a FPR.
15539 (movsf_hardfloat): Likewise.
15540 (movsd_hardfloat): Likewise.
15542 2017-05-09 Martin Sebor <msebor@redhat.com>
15544 PR translation/80280
15545 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
15548 PR translation/80280
15549 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
15550 data member added in r247778.
15551 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
15553 2017-05-09 Nathan Sidwell <nathan@acm.org>
15555 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
15557 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
15560 2017-05-09 Marek Polacek <polacek@redhat.com>
15562 * doc/invoke.texi: Fix typo.
15564 2017-05-09 Richard Biener <rguenther@suse.de>
15566 * tree-vrp.c (vrp_val_is_max): Adjust comment.
15567 (vrp_val_is_min): Likewise.
15568 (set_value_range_to_value): Likewise.
15569 (set_value_range_to_nonnegative): Likewise.
15570 (gimple_assign_nonzero_p): Likewise.
15571 (gimple_stmt_nonzero_p): Likewise.
15572 (vrp_int_const_binop): Likewise. Remove unreachable case.
15573 (adjust_range_with_scev): Adjust comments.
15574 (compare_range_with_value): Likewise.
15575 (extract_range_from_phi_node): Likewise.
15576 (test_for_singularity): Likewise.
15578 2017-05-09 Richard Biener <rguenther@suse.de>
15580 * tree-vrp.c (get_single_symbol): Add assert that we don't
15581 get overflowed constants as invariant part.
15582 (compare_values_warnv): Add comment before the TREE_NO_WARNING
15583 checks. Use wi::cmp instead of recursing for integer constants.
15584 (compare_values): Just ignore whether we assumed undefined
15585 overflow instead of failing the compare.
15586 (extract_range_for_var_from_comparison_expr): Add comment before the
15587 TREE_NO_WARNING sets.
15588 (test_for_singularity): Likewise.
15589 (extract_range_from_comparison): Do not disable optimization
15590 when we assumed undefined overflow.
15591 (extract_range_basic): Remove init of unused var.
15593 2017-05-09 Richard Biener <rguenther@suse.de>
15595 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
15596 (extract_range_from_multiplicative_op_1): Adjust.
15597 (extract_range_from_binary_expr_1): Use int_const_binop.
15599 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
15602 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
15603 rs6000_store_data_bypass_p in seven define_bypass directives and
15604 in several comments.
15605 * config/rs6000/rs6000-protos.h: Add prototype for
15606 rs6000_store_data_bypass_p function.
15607 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
15608 function implements slightly different (rs6000-specific) semantics
15609 than store_data_bypass_p, returning false rather than aborting
15610 with assertion error when arguments do not satisfy the
15611 requirements of store data bypass.
15612 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
15613 rs6000_store_data_bypass_p.
15615 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
15617 * config/xtensa/xtensa-protos.h
15618 (xtensa_initial_elimination_offset): New declaration.
15619 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15620 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
15621 macro definition, add case for FRAME_POINTER_REGNUM when
15622 FRAME_GROWS_DOWNWARD.
15623 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
15624 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
15625 xtensa_initial_elimination_offset.
15627 2017-05-08 Nathan Sidwell <nathan@acm.org>
15629 * doc/invoke.texi: Alphabetize -fdump options.
15631 2017-05-08 Martin Sebor <msebor@redhat.com>
15633 PR translation/80280
15634 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
15636 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
15638 * target.def (compute_frame_layout): New optional target hook.
15639 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
15640 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
15641 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
15643 * reload1.c (verify_initial_elim_offsets): Likewise.
15644 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
15645 (use_simple_return_p): Call arm_compute_frame_layout if needed.
15646 (arm_get_frame_offsets): Split up into this ...
15647 (arm_compute_frame_layout): ... and this function.
15649 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
15651 * config/aarch64/constraints.md (Usa): New constraint.
15652 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
15654 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
15656 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
15657 with_multilib_list after it has been checked.
15659 2017-05-08 Richard Biener <rguenther@suse.de>
15661 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
15662 (bitmap_set_subtract_values): Likewise.
15664 2017-05-08 Richard Biener <rguenther@suse.de>
15666 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
15667 (gimple_assign_nonzero): ... this and remove strict_overflow_p
15669 (gimple_stmt_nonzero_warnv_p): Rename to ...
15670 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
15672 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
15673 (extract_range_basic): Adjust, do not disable propagation on
15674 strict overflow sensitive simplification.
15675 (vrp_visit_cond_stmt): Likewise.
15677 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
15679 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
15680 body size unconditionally.
15682 2017-05-07 Jeff Law <law@redhat.com>
15685 2017-05-06 Jeff Law <law@redhat.com>
15686 PR tree-optimization/78496
15687 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15690 PR tree-optimization/78496
15691 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15692 (simplify_stmt_using_ranges): Call it.
15693 (vrp_dom_walker::before_dom_children): Extract equivalences
15694 from an ASSERT_EXPR with an equality comparison against a
15697 2017-05-06 Jeff Law <law@redhat.com>
15699 PR tree-optimization/78496
15700 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15703 PR tree-optimization/78496
15704 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15705 (simplify_stmt_using_ranges): Call it.
15706 (vrp_dom_walker::before_dom_children): Extract equivalences
15707 from an ASSERT_EXPR with an equality comparison against a
15710 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15712 * lra-constraints.c (lra_copy_reg_equiv): New function.
15713 (split_reg): Use it to copy equivalence information from the
15714 original register to the spill register.
15716 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15718 PR rtl-optimization/75964
15719 * simplify-rtx.c (simplify_const_relational_operation): Remove
15720 invalid handling of comparisons of integer ABS.
15722 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
15724 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
15725 initialize to zero.
15726 (init_regs): Remove declaration.
15727 (function_arg_advance_32): Initialize error_p as boolean variable.
15729 2017-05-05 Nathan Sidwell <nathan@acm.org>
15731 * store-motion.c (remove_reachable_equiv_notes): Reformat long
15732 lines. Use for (;;).
15734 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15736 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
15737 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
15738 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
15739 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
15740 VF=2 that require versioning.
15742 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15744 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
15747 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15749 * diagnostic.h (diagnostic_override_option_index): Convert from
15750 macro to inline function.
15752 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15754 * diagnostic.c (last_module_changed_p): New function.
15755 (set_last_module): New function.
15756 (diagnostic_report_current_module): Convert macro usage to
15757 the above functions.
15758 * diagnostic.h (diagnostic_context::last_module): Strengthen
15759 from const line_map * to const line_map_ordinary *.
15760 (diagnostic_last_module_changed): Delete macro.
15761 (diagnostic_set_last_module): Delete macro.
15763 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15765 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
15766 with diagnostic_report_diagnostic.
15767 (diagnostic_n_impl_richloc): Likewise.
15768 * diagnostic.h (report_diagnostic): Delete macro.
15769 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
15770 with diagnostic_report_diagnostic.
15771 * substring-locations.c (format_warning_va): Likewise.
15773 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15775 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
15776 save/restor of format_spec. Move option-printing code to...
15777 (print_option_information): ...this new function, and
15778 reimplement by simply printing to the pretty_printer,
15779 rather than appending to the format string.
15781 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15783 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
15784 handling logic into...
15785 (update_effective_level_from_pragmas): ...this new function.
15787 2017-05-04 Andrew Waterman <andrew@sifive.com>
15789 * config/riscv/riscv.opt (mstrict-align): New option.
15790 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
15791 (SLOW_UNALIGNED_ACCESS): Define.
15792 (riscv_slow_unaligned_access): Declare.
15793 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
15795 (riscv_slow_unaligned_access): New variable.
15796 (rocket_tune_info): Set slow_unaligned_access to true.
15797 (optimize_size_tune_info): Set slow_unaligned_access to false.
15798 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
15799 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
15800 (riscv_option_override): Set riscv_slow_unaligned_access.
15801 * doc/invoke.texi: Add -mstrict-align to RISC-V.
15803 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
15805 * config/riscv/riscv.md: Unify indentation.
15807 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15812 * config/rs6000/rs6000.md (u code attribute): Add FIX and
15814 (extendsi<mode>2): Add support for doing sign extension via
15815 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
15816 don't have ISA 3.0 instructions.
15817 (extendsi<mode>2 splitter): Likewise.
15818 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
15819 generate the normal insns since SImode can now go in vector
15820 registers. Disallow the special UNSPECs needed for previous
15821 machines to hide SImode being used. Add new insns
15822 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
15823 (fix_trunc<mode>si2_stfiwx): Likewise.
15824 (fix_trunc<mode>si2_internal): Likewise.
15825 (fixuns_trunc<mode>si2): Likewise.
15826 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15827 (fctiw<u>z_<mode>_smallint): Likewise.
15828 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
15829 of floating point to 32-bit integer from doing a direct move to
15830 the GPR registers to do a store.
15831 (fctiwz_<mode>): Break long line.
15833 2017-05-05 Bin Cheng <bin.cheng@arm.com>
15835 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
15836 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
15837 (addr_list, addr_offset_valid_p): New.
15838 (split_address_groups): Check offset validity with above function.
15839 (gt-tree-ssa-loop-ivopts.h): Include header file.
15841 2017-05-05 Nathan Sidwell <nathan@acm.org>
15843 * config.gcc (arm*-*-*): Add missing 'fi'.
15845 2017-05-05 Steve Ellcey <sellcey@cavium.com>
15847 * doc/invoke.texi (-fopt-info): Explicitly say order of options
15848 included in -fopt-info does not matter.
15849 * doc/optinfo.texi (-fopt-info): Fix description of default
15850 behavour. Explicitly say order of options included in -fopt-info
15853 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
15855 * config.gcc: Allow combinations of aprofile and rmprofile values for
15856 --with-multilib-list.
15857 * config/arm/t-multilib: New file.
15858 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
15859 variables. Remove setting of ISA and floating-point ABI in
15860 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
15861 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
15862 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
15863 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
15865 * config/arm/t-rmprofile: Likewise except for the matches changes.
15866 * doc/install.texi (--with-multilib-list): Document the combination of
15867 aprofile and rmprofile values and warn about pitfalls in doing that.
15869 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
15871 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
15872 (movdi_aarch64): Likewise.
15874 2017-05-05 Jakub Jelinek <jakub@redhat.com>
15876 PR tree-optimization/80632
15877 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
15879 (build_arrays): Initialize it for virtual phis.
15880 (fix_phi_nodes): Use it for virtual phis.
15882 PR tree-optimization/80558
15883 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
15884 [x, y] op z into [x op, y op z] for op & or | if conditions
15887 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
15888 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15891 * config/arm/arm.md (use_literal_pool): Remove.
15892 (64-bit immediate split): No longer takes cost into consideration
15893 if arm_disable_literal_pool is enabled.
15894 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
15895 used when arm_disable_literal_pool is enabled.
15896 (arm_max_const_double_inline_cost): Remove use of
15897 arm_disable_literal_pool.
15898 (push_minipool_fix): Add assert.
15899 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
15900 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
15901 (no_literal_pool_sf_immediate): New.
15903 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15905 PR tree-optimization/80613
15906 * tree-ssa-dce.c (propagate_necessity): Remove cases for
15907 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
15909 2017-05-05 Richard Biener <rguenther@suse.de>
15911 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
15913 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
15915 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
15916 of this flag from insn conditions due to removal from r247495.
15918 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
15920 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
15922 (arm_early_store_addr_dep_ptr): Likewise.
15923 * config/arm/aarch-common-protos.h
15924 (arm_early_load_addr_dep_ptr): Add prototype.
15925 (arm_early_store_addr_dep_ptr): Likewise.
15926 * config/arm/cortex-a53.md: Add new bypasses.
15928 2017-05-05 Jakub Jelinek <jakub@redhat.com>
15930 * tree.c (next_type_uid): Change type to unsigned.
15931 (type_hash_canon): Decrement back next_type_uid if
15932 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
15933 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
15936 2017-05-04 Martin Sebor <msebor@redhat.com>
15938 * builtins.c: Fix a trivial typo in a comment.
15940 PR middle-end/79234
15941 * builtins.c (check_sizes): Adjust to handle reading past the end.
15942 Avoid printing excessive upper bound of ranges. Use %E to print
15943 tree nodes instead of converting them to %wu.
15944 (expand_builtin_memchr): New function.
15945 (compute_dest_size): Rename...
15946 (compute_objsize): ...to this.
15947 (expand_builtin_memcpy): Adjust.
15948 (expand_builtin_mempcpy): Adjust.
15949 (expand_builtin_strcat): Adjust.
15950 (expand_builtin_strcpy): Adjust.
15951 (check_strncat_sizes): Adjust.
15952 (expand_builtin_strncat): Adjust.
15953 (expand_builtin_strncpy): Adjust and simplify.
15954 (expand_builtin_memset): Adjust.
15955 (expand_builtin_bzero): Adjust.
15956 (expand_builtin_memcmp): Adjust.
15957 (expand_builtin): Handle memcmp.
15958 (maybe_emit_chk_warning): Check strncat just once.
15960 2017-05-04 Martin Sebor <msebor@redhat.com>
15962 PR preprocessor/79214
15963 PR middle-end/79222
15964 PR middle-end/79223
15965 * builtins.c (check_sizes): Add inlining context and issue
15966 warnings even when -Wno-system-headers is set.
15967 (check_strncat_sizes): Same.
15968 (expand_builtin_strncat): Same.
15969 (expand_builtin_memmove): New function.
15970 (expand_builtin_stpncpy): Same.
15971 (expand_builtin): Handle memmove and stpncpy.
15973 2017-05-04 Bin Cheng <bin.cheng@arm.com>
15975 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
15976 which is not used any more.
15978 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15980 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
15982 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15984 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
15985 (cortexa53_tunings): Likewise.
15986 (cortexa57_tunings): Likewise.
15987 (cortexa72_tunings): Likewise.
15988 (cortexa73_tunings): Likewise.
15990 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15992 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
15993 Set loop alignment to 8.
15995 2017-05-04 Martin Sebor <msebor@redhat.com>
15997 PR translation/80280
15998 * builtins.c (expand_builtin_object_size): Add missing quoting to
15999 %D and like directives.
16000 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
16001 (hsa_type_for_tree_type): Same.
16002 (verify_function_arguments): Same.
16003 * symtab.c (symbol_table::change_decl_assembler_name): Same.
16004 * varasm.c (get_section): Same.
16007 2017-05-04 Martin Sebor <msebor@redhat.com>
16009 PR translation/80280
16010 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
16012 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16014 * config/aarch64/aarch64.c (generic_addrcost_table):
16015 Change HI/TI mode setting.
16017 2017-05-04 Martin Jambor <mjambor@suse.cz>
16019 PR tree-optimization/80622
16020 * tree-sra.c (comes_initialized_p): New function.
16021 (build_accesses_from_assign): Only set write lazily when
16022 comes_initialized_p is false.
16023 (analyze_access_subtree): Use comes_initialized_p.
16024 (propagate_subaccesses_across_link): Assert !comes_initialized_p
16025 instead of testing for PARM_DECL.
16027 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16029 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
16030 constraint on operand 0 to allow more general addressing modes.
16031 Adjust output template.
16032 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
16034 * config/aarch64/aarch64-protos.h
16035 (aarch64_address_valid_for_prefetch_p): Declare prototype.
16036 * config/aarch64/constraints.md (Dp): New address constraint.
16037 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
16040 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
16042 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
16043 update use of estimate_ipcp_clone_size_and_time.
16044 (estimate_local_effects): Update use of
16045 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
16046 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
16047 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
16048 Return nonspecialized time.
16050 2017-05-04 Richard Biener <rguenther@suse.de>
16052 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
16053 for the last VUSE which def dominates the PHI. Directly call
16055 (get_continuation_for_phi_1): Remove.
16057 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
16059 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
16060 to explain the use of truncating division. Cap the number of
16061 iterations to the maximum given by nb_iterations_upper_bound,
16064 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16066 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
16067 * configure: Regenerate.
16068 * config.in: Regenerate.
16069 * config/i386/driver-mingw32.c: new file.
16070 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
16071 * config.host: Link driver-mingw32.o on MinGW host.
16072 * doc/install.texi: Document new --enable-mingw-wildcard configure
16075 2017-05-04 Marek Polacek <polacek@redhat.com>
16077 PR tree-optimization/80612
16078 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
16080 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16081 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
16083 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
16084 (movt splitter): Likewise.
16085 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
16086 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
16087 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
16088 block for Thumb-1 with MOVT.
16089 (thumb2_legitimate_address_p): Move code block ...
16090 (can_avoid_literal_pool_for_label_p): ... into this new function.
16091 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
16093 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
16094 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
16095 "M-profile targets with the MOVT instruction".
16097 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16099 * config/arm/arm-builtins.c (arm_init_builtins): Rename
16100 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
16101 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
16103 2017-05-04 Martin Liska <mliska@suse.cz>
16105 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
16106 variable cond_code.
16108 2017-05-04 Richard Biener <rguenther@suse.de>
16110 * tree.c (array_at_struct_end_p): Handle arrays at struct
16111 end with flexarrays more conservatively. Refactor and treat
16112 arrays of arrays or aggregates more strict. Fix
16113 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
16114 * tree.c (array_at_struct_end_p): Adjust prototype.
16115 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
16116 * gimple-fold.c (get_range_strlen): Likewise.
16117 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
16119 2017-05-04 Richard Biener <rguenther@suse.de>
16121 PR tree-optimization/31130
16122 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
16124 (supports_overflow_infinity): Likewise.
16125 (is_negative_overflow_infinity): Likewise.
16126 (is_positive_overflow_infinity): Likewise.
16127 (is_overflow_infinity): Likewise.
16128 (stmt_overflow_infinity): Likewise.
16129 (overflow_infinity_range_p): Likewise.
16130 (usable_range_p): Remove as always returning true.
16131 (make_overflow_infinity): Remove.
16132 (negative_overflow_infinity): Likewise.
16133 (positive_overflow_infinity): Likewise.
16134 (avoid_overflow_infinity): Likewise.
16135 (set_value_range): Adjust accordingly.
16136 (set_value_range_to_nonnegative): Likewise, remove now unused
16137 overflow_infinity arg.
16138 (vrp_operand_equal_p): Adjust.
16139 (update_value_range): Likewise.
16140 (range_int_cst_singleton_p): Likewise.
16141 (operand_less_p): Likewise.
16142 (compare_values_warnv): Likewise.
16143 (extract_range_for_var_from_comparison_expr): Likewise.
16144 (vrp_int_const_binop): Likewise.
16145 (zero_nonzero_bits_from_vr): Likewise.
16146 (extract_range_from_multiplicative_op_1): Likewise.
16147 (extract_range_from_binary_expr_1): Likewise.
16148 (extract_range_from_unary_expr): Likewise.
16149 (extract_range_from_comparison): Likewise.
16150 (extract_range_basic): Likewise.
16151 (adjust_range_with_scev): Likewise.
16152 (compare_ranges): Likewise.
16153 (compare_range_with_value): Likewise.
16154 (dump_value_range): Likewise.
16155 (test_for_singularity): Likewise, remove strict_overflow_p parameter
16157 (simplify_cond_using_ranges): Adjust.
16159 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16161 * brig-builtins.def: Added a builtin for class_f64.
16162 * builtin-types.def: Added a builtin type needed by class_f64.
16164 2017-05-03 Jason Merrill <jason@redhat.com>
16166 * timevar.def: Add TV_CONSTEXPR.
16168 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16170 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
16172 2017-05-03 Martin Jambor <mjambor@suse.cz>
16174 * ipa-prop.c (ipa_update_after_lto_read): Removed.
16175 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
16176 * ipa-cp.c (ipcp_propagate_stage): Do not call
16177 ipa_update_after_lto_read.
16178 * ipa-inline.c (ipa_inline): Likewise.
16180 2017-05-03 Martin Jambor <mjambor@suse.cz>
16182 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
16183 tag. Added a default constructor and a destructor.
16184 (ipa_edge_args_sum_t): New class;
16185 (ipa_edge_args_sum): Declare.
16186 (ipa_edge_args_vector): Remove declaration.
16187 (IPA_EDGE_REF): Use ipa_edge_args_sum.
16188 (ipa_free_edge_args_substructures): Remove declaration.
16189 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
16190 (ipa_edge_args_info_available_for_edge_p): Likewise.
16191 * ipa-prop.c (ipa_edge_args_vector): Removed.
16192 (edge_removal_hook_holder): Likewise.
16193 (edge_duplication_hook_holder): Likewise.
16194 (ipa_edge_args_sum): New variable.
16195 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
16196 ipa_edge_args_vector.
16197 (ipa_free_edge_args_substructures): Likewise.
16198 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
16199 ipa_edge_args_vector.
16200 (ipa_edge_removal_hook): Turned into method
16201 ipa_edge_args_sum_t::remove.
16202 (ipa_edge_duplication_hook): Turned into method
16203 ipa_edge_args_sum_t::duplicate.
16204 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
16205 registering edge hooks.
16206 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
16207 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
16208 ipa_edge_args_sum instead of ipa_edge_args_vector.
16209 * ipa-profile.c (ipa_profile): Likewise.
16211 2017-05-03 Martin Jambor <mjambor@suse.cz>
16213 * symbol-summary.h (function_summary): New method exists.
16214 (function_summary::symtab_removal): Deallocate through release.
16215 (call_summary): New class.
16216 (gt_ggc_mx): New overload.
16217 (gt_pch_nx): Likewise.
16218 (gt_pch_nx): Likewise.
16220 2017-05-03 Jeff Law <law@redhat.com>
16222 PR tree-optimization/78496
16223 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
16224 from simplify_cond_using_ranges. Split off code to walk
16225 backwards through casts into ...
16226 (simplify_cond_using_ranges_2): New function.
16227 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
16228 (execute_vrp): After identifying jump threads, call
16229 simplify_cond_using_ranges_2.
16231 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
16234 * ipa-inline.h (inline_summary): Add ctor.
16235 (create_ggc): Do not use ggc_cleared_alloc.
16237 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
16238 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16240 * gcc.c (handle_braces): Support escaping in switch matching
16242 * doc/invoke.texi (Spec Files): Document it.
16243 Remove superfluous @code markup in items.
16245 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16247 * diagnostic-show-locus.c (struct column_range): New struct.
16248 (get_affected_columns): New function.
16249 (get_printed_columns): New function.
16250 (struct correction): New struct.
16251 (correction::ensure_capacity): New function.
16252 (correction::ensure_terminated): New function.
16253 (struct line_corrections): New struct.
16254 (line_corrections::~line_corrections): New dtor.
16255 (line_corrections::add_hint): New function.
16256 (layout::print_trailing_fixits): Reimplement in terms of the new
16258 (selftest::test_overlapped_fixit_printing): New function.
16259 (selftest::diagnostic_show_locus_c_tests): Call it.
16261 2017-05-03 Nathan Sidwell <nathan@acm.org>
16263 Canonicalize canonical type hashing
16264 * tree.h (type_hash_canon_hash): Declare.
16265 * tree.c (type_hash_list, attribute_hash_list): Move into
16266 type_hash_canon_hash.
16267 (build_type_attribute_qual_variant): Break out hash code calc into
16268 type_hash_canon_hash.
16269 (type_hash_canon_hash): New. Generic type hash computation.
16270 (build_range_type_1, build_array_type_1, build_function_type,
16271 build_method_type_directly, build_offset_type, build_complex_type,
16272 make_vector_type): Call it.
16274 2017-05-03 Richard Biener <rguenther@suse.de>
16276 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
16277 When all DRs have unknown misaligned do not always peel
16278 when there is a store but apply the same costing model as if
16279 there were only loads.
16281 2017-05-03 Richard Biener <rguenther@suse.de>
16284 PR tree-optimization/80492
16285 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16286 compare_base_decls returning dont-know properly.
16288 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
16290 * config/arm/iterators.md (CCSI): New mode iterator.
16291 (arch): New mode attribute.
16292 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
16293 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
16294 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
16295 code iterator for success result mode.
16296 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
16297 the corresponding new insn generators.
16299 2017-05-03 Bin Cheng <bin.cheng@arm.com>
16302 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16303 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16305 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
16307 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
16308 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
16309 (DDR_A): Wrap DDR argument in brackets.
16310 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
16311 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
16312 (DDR_REVERSED_P): Likewise.
16314 2017-05-03 Jakub Jelinek <jakub@redhat.com>
16316 PR tree-optimization/79472
16317 * tree-switch-conversion.c (struct switch_conv_info): Add
16318 contiguous_range and default_case_nonstandard fields.
16319 (collect_switch_conv_info): Compute contiguous_range and
16320 default_case_nonstandard fields, don't clear final_bb if
16321 contiguous_range and only the default case doesn't have the required
16323 (check_all_empty_except_final): Set default_case_nonstandard instead
16324 of failing if contiguous_range and the default case doesn't have empty
16326 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
16327 and only the default case doesn't have the required constants. Skip
16329 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
16330 if default_case_nonstandard.
16331 (build_constructors): Build constant 1 just once. Assert that default
16332 values aren't inserted in between cases if contiguous_range. Skip
16334 (build_arrays): Skip virtual phis.
16335 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
16336 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
16337 Handle virtual phis.
16338 (gen_inbound_check): Handle default_case_nonstandard case.
16339 (process_switch): Adjust check_final_bb caller. Call
16340 gather_default_values with the first non-default case instead of
16341 default case if default_case_nonstandard.
16343 2017-05-02 Nathan Sidwell <nathan@acm.org>
16345 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
16346 check. Fix formatting.
16348 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
16350 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
16351 errors when comparing specialized and unspecialized times.
16353 2017-05-02 David Malcolm <dmalcolm@redhat.com>
16355 * diagnostic-show-locus.c
16356 (layout::should_print_annotation_line_p): Make private.
16357 (layout::print_annotation_line): Make private.
16358 (layout::annotation_line_showed_range_p): Make private.
16359 (layout::show_ruler): Make private.
16360 (layout::print_source_line): Make private. Pass in line and
16361 line_width, rather than calling location_get_source_line. Drop
16363 (layout::print_leading_fixits): New method.
16364 (layout::print_any_fixits): Rename to...
16365 (layout::print_trailing_fixits): ...this, and make private.
16366 Don't print newline fixits.
16367 (diagnostic_show_locus): Move logic for printing one row into...
16368 (layout::print_line): ...this new function. Move the
16369 location_get_source_line call and error-handling from
16370 print_source_line to here. Call print_leading_fixits, and rename
16371 print_any_fixits to print_trailing_fixits.
16372 (selftest::test_fixit_insert_containing_newline): Update now that
16373 newlines are partially supported.
16374 (selftest::test_fixit_insert_containing_newline_2): New test.
16375 (selftest::test_fixit_replace_containing_newline): Update comments.
16376 (selftest::diagnostic_show_locus_c_tests): Call the new test.
16377 * edit-context.c (class added_line): New class.
16378 (class edited_line): Describe newline handling in comment.
16379 (edited_line::actually_edited_p): New method.
16380 (edited_line::print_content): Delete redundant decl.
16381 (edited_line::m_predecessors): New field.
16382 (edited_file::print_content): Call edited_line::print_content.
16383 (edited_file::print_diff): Update to support newlines.
16384 (edited_file::print_diff_hunk): Likewise.
16385 (edited_file::print_run_of_changed_lines): New function.
16386 (edited_file::print_diff_line): Convert to...
16387 (print_diff_line): ...this.
16388 (edited_file::get_effective_line_count): New function.
16389 (edited_line::edited_line): Initialize new field m_predecessors.
16390 (edited_line::~edited_line): Clean up m_predecessors.
16391 (edited_line::apply_fixit): Handle newlines.
16392 (edited_line::get_effective_line_count): New function.
16393 (edited_line::print_content): New function.
16394 (edited_line::print_diff_lines): New function.
16395 (selftest::test_applying_fixits_insert_containing_newline): New
16397 (selftest::test_applying_fixits_replace_containing_newline): New
16399 (selftest::insert_line): New function.
16400 (selftest::test_applying_fixits_multiple_lines): Add example of
16402 (selftest::edit_context_c_tests): Call the new tests.
16404 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16406 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
16407 parameter cand. Update dump information.
16408 (get_computation_cost): Update uses.
16410 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16412 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
16413 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
16414 (get_computation_at, rewrite_use_address): Update use of
16415 get_computation_aff.
16417 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16419 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
16420 (get_computation): Delete.
16421 (get_computation_cost): Implement like get_computation_cost_at.
16422 Use get_computation_at.
16423 (get_computation_cost_at): Delete.
16424 (rewrite_use_nonlinear_expr): Use get_computation_at.
16425 (rewrite_use_compare, remove_unused_ivs): Ditto.
16427 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16429 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
16431 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16433 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
16434 (ivopts_global_cost_for_size): Rename parameter and update uses.
16435 (iv_ca_recount_cost): Update uses.
16436 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
16437 candidates seperately in n_invs and n_cands.
16438 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
16440 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16442 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
16443 (find_inv_vars_cb): New.
16444 (find_depends): Renamed to ...
16445 (find_inv_vars): ... this.
16446 (add_candidate_1, force_var_cost): Call find_inv_vars.
16447 (split_address_cost, determine_group_iv_cost_cond): Ditto.
16449 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16451 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
16452 inv_vars. Add inv_exprs.
16453 (struct iv_cand): Rename depends_on to inv_vars.
16454 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
16455 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
16456 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
16457 (dump_cand): Dump inv_vars.
16458 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
16459 (record_invariant, find_depends, add_candidate_1): Ditto.
16460 (set_group_iv_cost, force_var_cost): Ditto.
16461 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
16462 (get_computation_cost_at, get_computation_cost): Ditto.
16463 (determine_group_iv_cost_generic): Ditto.
16464 (determine_group_iv_cost_address): Ditto.
16465 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
16466 (determine_group_iv_costs): Ditto.
16467 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
16468 (iv_ca_set_remove_invariants): Renamed to ...
16469 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
16470 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
16471 (iv_ca_set_add_invariants): Renamed to ...
16472 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
16473 (iv_ca_set_cp): Use iv_ca_set_add_invs.
16474 (iv_ca_has_deps): Support inv_vars and inv_exprs.
16475 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
16476 (create_new_ivs): Remove useless dump.
16478 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16480 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
16482 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
16483 (iv_ca_set_no_cp, create_new_iv): Ditto.
16485 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16487 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16489 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16491 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
16492 function tree_check2.
16494 2017-05-02 Martin Liska <mliska@suse.cz>
16496 * doc/gcov.texi: Add missing preposition.
16497 * gcov.c (function_info::function_info): Properly fill up
16498 all member variables.
16500 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16502 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
16504 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16506 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
16508 2017-05-02 Martin Liska <mliska@suse.cz>
16511 * lto-streamer-in.c (lto_read_tree_1): Remove
16512 LTO_STREAMER_DEBUG.
16513 * lto-streamer.c (struct tree_hash_entry): Likewise.
16514 (struct tree_entry_hasher): Likewise.
16515 (tree_entry_hasher::hash): Likewise.
16516 (tree_entry_hasher::equal): Likewise.
16517 (lto_streamer_init): Likewise.
16518 (lto_orig_address_map): Likewise.
16519 (lto_orig_address_get): Likewise.
16520 (lto_orig_address_remove): Likewise.
16521 * lto-streamer.h: Likewise.
16522 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
16523 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
16525 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
16527 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
16528 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
16529 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
16530 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
16531 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
16532 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
16533 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
16534 (mm_maskz_sub_ss): New intrinsics.
16535 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16536 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16537 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
16538 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
16539 (__builtin_ia32_subss_mask_round): New builtins.
16540 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16541 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16542 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
16544 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
16545 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16547 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16550 2017-05-02 Martin Jambor <mjambor@suse.cz>
16552 PR tree-optimization/78687
16553 * tree-sra.c (access): New field parent.
16554 (process_subtree_disqualification): New function.
16555 (disqualify_candidate): Call it.
16556 (build_accesses_from_assign): Reset write flag if creating an
16558 (build_access_subtree): Fill in parent field and also prpagate
16559 down grp_write flag.
16560 (create_artificial_child_access): New parameter set_grp_write, set
16561 grp_write to its value.
16562 (propagate_subaccesses_across_link): Also propagate grp_write flag
16564 (propagate_all_subaccesses): Push the closest parent back to work
16565 queue if add_access_to_work_queue returned true.
16567 2017-05-02 Richard Biener <rguenther@suse.de>
16569 * common.opt (fstrict-overflow): Alias negative to fwrapv.
16570 * doc/invoke.texi (fstrict-overflow): Remove all traces of
16571 -fstrict-overflow documentation.
16572 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
16573 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
16574 flag_strict_overflow.
16575 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
16576 * lto-opts.c (lto_write_options): Do not stream it.
16577 * lto-wrapper.c (merge_and_complain): Do not handle it.
16578 * opts.c (default_options_table): Do not set -fstrict-overflow.
16579 (finish_options): Likewise do not clear it when sanitizing.
16580 * simplify-rtx.c (simplify_const_relational_operation): Do not
16581 test flag_strict_overflow.
16583 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16585 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
16586 using enabled attribute.
16587 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
16588 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
16589 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
16590 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
16591 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
16592 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
16593 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
16594 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
16595 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
16596 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
16598 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16600 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
16602 2017-05-02 Richard Biener <rguenther@suse.de>
16604 PR tree-optimization/80591
16606 2017-04-10 Richard Biener <rguenther@suse.de>
16608 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16611 2017-05-02 Richard Biener <rguenther@suse.de>
16613 PR tree-optimization/80549
16614 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
16615 (cleanup_tree_cfg_noloop): Create forwarders to known loop
16616 headers if they do not have a preheader.
16618 2017-05-02 Martin Liska <mliska@suse.cz>
16621 * common.opt: Fix typo.
16622 * doc/invoke.texi: Likewise.
16624 2017-05-01 Jan Beulich <jbeulich@suse.com>
16626 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
16627 swapping, add (x,x,m,x,n) alternative.
16629 2017-05-01 Nathan Sidwell <nathan@acm.org>
16631 * calls.c (combine_pending_stack_adjustment_and_call): Remove
16632 unnecessary unadjusted_alignment check.
16634 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16637 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
16639 * gimplify.c (gimplify_cilk_detach): New function.
16640 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
16641 * tree-core.h: Document EXPR_CILK_SPAWN.
16642 * tree.h (EXPR_CILK_SPAWN): Define.
16644 2017-05-01 David Malcolm <dmalcolm@redhat.com>
16646 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
16647 to use new fixit_hint representation, using the "replace" logic.
16648 (get_line_span_for_fixit_hint): Likewise.
16649 (layout::print_any_fixits): Likewise.
16650 (selftest::test_one_liner_many_fixits): Rename to...
16651 (selftest::test_one_liner_many_fixits_1): ...this, and update
16652 comment and expected output to reflect that the multiple fix-it
16653 hints are now consolidated into one insertion.
16654 (selftest::test_one_liner_many_fixits_2): New test.
16655 (selftest::test_diagnostic_show_locus_one_liner): Update for
16657 (selftest::test_fixit_consolidation): Update for fix-it API
16659 * diagnostic.c (print_parseable_fixits): Likewise.
16660 * edit-context.c (edited_line::m_line_events): Convert from
16661 auto_vec <line_event *> to auto_vec <line_event>.
16662 (class line_event): Convert from abstract base class to a concrete
16663 class, taking over the role of replace_event.
16664 (class insert_event): Delete.
16665 (class replace_event): Rename to class line_event. Convert to
16667 (edit_context::add_fixits): Reimplement.
16668 (edit_context::apply_insert): Delete.
16669 (edit_context::apply_replace): Rename to...
16670 (edit_context::apply_fixit): ...this. Convert to half-open range.
16671 (edited_file::apply_insert): Delete.
16672 (edited_file::apply_replace): Rename to...
16673 (edited_file::apply_fixit): ...this.
16674 (edited_line::~edited_line): Drop deletion of events.
16675 (edited_line::apply_insert): Delete.
16676 (edited_line::apply_replace): Rename to...
16677 (edited_line::apply_fixit): ...this. Convert to half-open range.
16678 Update for change to type of m_line_events.
16679 * edit-context.h (edit_context::apply_insert): Delete.
16680 (edit_context::apply_replace): Rename to...
16681 (edit_context::apply_fixit): ...this.
16683 2017-05-01 Martin Sebor <msebor@redhat.com>
16685 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
16688 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
16691 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
16692 __get_cpuid_max returns 0.
16693 (__get_cpuid_count): Ditto.
16695 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
16697 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
16698 replacement expression is another instance of one of its arguments.
16700 2017-05-01 Jakub Jelinek <jakub@redhat.com>
16703 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
16704 check for stack push/pop autoinc.
16705 * config/i386/i386.c (ix86_agi_dependent): Return false
16706 if the only reason why modified_in_p returned true is that
16707 addr is SP based and set_insn is a push or pop.
16709 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16711 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
16714 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16717 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
16718 (account_size_time): Use two predicates - exec_pred and
16720 (evaluate_conditions_for_known_args): Compute both clause and
16722 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
16723 (inline_summary_t::duplicate): Update.
16724 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
16726 (compute_inline_parameters): Likewise.
16727 (estimate_edge_size_and_time): Update caluclation of time.
16728 (estimate_node_size_and_time): Compute both time and nonspecialized
16730 (estimate_ipcp_clone_size_and_time): Update.
16731 (inline_merge_summary): Update.
16732 (do_estimate_edge_time): Update.
16733 (do_estimate_edge_size): Update.
16734 (do_estimate_edge_hints): Update.
16735 (inline_read_section, inline_write_summary): Stream both new predicates.
16736 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
16738 (compute_inlined_call_time): Cleanup.
16739 (big_speedup_p): Update.
16740 (edge_badness): Update.
16741 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
16742 (size_time_entry): Replace predicate by exec_predicate and
16743 nonconst_predicate.
16744 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
16745 (estimate_edge_time): Return also nonspec_time.
16746 (reset_edge_growth_cache): Update.
16748 2017-04-29 Jakub Jelinek <jakub@redhat.com>
16750 PR rtl-optimization/80491
16751 * ifcvt.c (noce_process_if_block): When looking for x setter
16752 with missing else_bb, don't check only the insn right before
16753 cond_earliest, but look for the last insn that x is modified in
16754 within the same bb.
16756 PR rtl-optimization/80491
16757 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
16759 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
16761 PR tree-optimization/80487
16762 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
16764 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16766 PR tree-optimization/79697
16767 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
16768 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
16769 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
16771 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
16772 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
16774 2017-04-28 Martin Sebor <msebor@redhat.com>
16776 PR tree-optimization/80523
16777 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
16778 (init_target_to_host_charmap, target_to_host, target_strtol10): New
16780 (maybe_warn, format_directive, parse_directive): Use new functions.
16781 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
16783 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
16785 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
16787 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
16789 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
16790 target_header_dir): Set correctly.
16791 * configure: Regenerated.
16792 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
16793 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
16794 instead of SYSTEM_HEADER_DIR.
16796 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
16798 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
16799 (estimate_local_effects): Likewise.
16800 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
16801 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
16802 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
16803 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
16804 do_estimate_edge_time, estimate_edge_time): Likewise.
16805 * ipa-inline-analysis.c (estimate_node_size_and_time,
16806 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
16807 (estimate_time_after_inlining): Remove.
16809 2017-04-28 Martin Liska <mliska@suse.cz>
16811 * doc/gcov.texi: Enhance documentation of gcov.
16813 2017-04-28 Martin Liska <mliska@suse.cz>
16815 * doc/gcov.texi: Sort options in alphabetic order.
16816 * doc/gcov-dump.texi: Likewise.
16817 * doc/gcov-tool.texi: Likewise.
16818 * gcov.c (print_usage): Likewise.
16819 * gcov-dump.c (print_usage): Likewise.
16820 * gcov-tool.c (print_merge_usage_message): Likewise.
16821 (print_rewrite_usage_message): Likewise.
16822 (print_overlap_usage_message): Likewise.
16824 2017-04-28 Martin Liska <mliska@suse.cz>
16826 PR gcov-profile/53915
16827 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
16829 2017-04-28 Martin Liska <mliska@suse.cz>
16831 PR gcov-profile/79891
16832 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
16833 is marked by compiler as living on a line.
16834 (get_cycles_count): Remove usage of the union.
16835 (output_intermediate_file): Likewise.
16836 (find_source): Fix GNU coding style.
16837 (accumulate_line_counts): Remove old non-all block mode.
16838 (output_lines): Remove usage of the union.
16839 * profile.c (output_location): Include all BBs, even if
16840 belonging to a same line (and file) as a previous BB.
16842 2017-04-28 Martin Liska <mliska@suse.cz>
16844 * gcov.c (process_args): Handle new argument 'w'.
16845 (read_graph_file): Assign ID to BBs.
16846 (output_branch_count): Display BB # if verbose flag is set.
16847 (output_lines): Likewise for arcs.
16848 (print_usage): Add '--verbose' option help.
16849 * doc/gcov.texi: Document --verbose (-w) option.
16851 2017-04-28 Martin Liska <mliska@suse.cz>
16853 * gcov.c (struct block_location_info): New struct.
16854 (process_file): Fill up the new structure.
16855 (read_graph_file): Replace usage of encoding by the newly added
16857 (add_line_counts): Likewise.
16858 (accumulate_line_counts): Remove usage of the union.
16859 (function_info::function_info): New function.
16860 (function_info::~function_info): Likewise.
16861 (process_file): Call delete instead of release_function.
16862 (release_function): Release the function.
16863 (release_structures): Call delete instead of release_function.
16864 (solve_flow_graph): Replace usage of num_blocks.
16865 (find_exception_blocks): Likewise.
16866 (output_lines): Fix GNU coding style.
16868 2017-04-28 Martin Liska <mliska@suse.cz>
16871 * coverage.c (coverage_remove_note_file): New function.
16872 * coverage.h: Declare the function.
16873 * toplev.c (finalize): Clean if an error has been seen.
16875 2017-04-28 Martin Liska <mliska@suse.cz>
16877 PR gcov-profile/80031
16878 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
16879 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
16880 * gcov.c (read_graph_file): Read just number of blocks.
16881 * profile.c (branch_prob): Do not stream 0 flags per a basic
16884 2017-04-28 Martin Liska <mliska@suse.cz>
16886 * gcov-dump.c (tag_*): Add new argument to declarations.
16887 (dump_gcov_file): Likewise.
16888 (tag_blocks): Add and use new argument depth.
16889 (tag_arcs): Likewise.
16890 (tag_lines): Likewise.
16891 (tag_counters): Likewise.
16892 (tag_summary): Likewise.
16893 (dump_working_sets): Use depth to do a proper indentation.
16895 2017-04-28 Jakub Jelinek <jakub@redhat.com>
16898 * cgraph.h (symtab_node::debug_symtab): No longer inline.
16899 * symtab.c (symtab_node::debug_symtab): Move definition here.
16901 2017-04-28 Richard Biener <rguenther@suse.de>
16903 * lto-streamer.h (LTO_major_version): Bump to 7.
16905 2017-04-28 Richard Biener <rguenther@suse.de>
16907 * tree-vrp.c (assert_info): New struct.
16908 (add_assert_info): New helper.
16909 (register_edge_assert_for_2): Refactor to add asserts to a vector
16911 (register_edge_assert_for_1): Likewise.
16912 (register_edge_assert_for): Likewise.
16913 (finish_register_edge_assert_for): New helper actually registering
16914 asserts where live on edge.
16915 (find_conditional_asserts): Adjust.
16916 (find_switch_asserts): Likewise.
16917 (evrp_dom_walker::try_find_new_range): Generalize.
16918 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
16920 2017-04-27 Marek Polacek <polacek@redhat.com>
16923 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
16924 arg10 and arg11 to itype.
16926 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
16928 * doc/extend.texi (Object Size Checking): Improve grammar.
16930 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
16933 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
16934 that the logic for permitting reciprocal estimates matches that
16937 2017-04-27 Jakub Jelinek <jakub@redhat.com>
16940 * tree.c (type_cache_hasher::equal): Only compare
16941 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
16942 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
16943 non-aggregate element types.
16944 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
16945 about the flag on ARRAY_TYPEs in the comment, formatting fix.
16947 2017-04-27 Richard Biener <rguenther@suse.de>
16949 PR middle-end/80533
16950 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
16951 stripping ARRAY_REFs from MEM_EXPR make sure we're not
16952 keeping a reference to a trailing array.
16954 2017-04-27 Richard Biener <rguenther@suse.de>
16956 PR middle-end/80539
16957 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
16958 being in loop-closed SSA form conservatively.
16959 (chrec_fold_multiply_poly_poly): Likewise.
16961 2017-04-27 Tamar Christina <tamar.christina@arm.com>
16963 PR middle-end/79665
16964 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
16965 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
16967 2017-04-27 Jakub Jelinek <jakub@redhat.com>
16970 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
16971 (aarch64_function_arg_alignment): Return unsigned int again, but still
16972 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
16973 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
16974 Don't emit -Wpsabi note.
16975 (aarch64_function_arg_boundary): Likewise.
16976 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
16979 2017-04-26 Nathan Sidwell <nathan@acm.org>
16981 * tree.h (crc32_unsigned_n): Declare.
16982 (crc32_unsigned, crc32_unsigned): Make inline.
16983 * tree.c (crc32_unsigned_bits): Replace with ...
16984 (crc32_unsigned_n): ... this.
16985 (crc32_unsigned, crc32_byte): Remove.
16986 (crc32_string): Remove unnecessary braces.
16988 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
16990 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
16991 * ipa-inline-analysis.c (MAX_TIME): Remove.
16992 (account_size_time): Use sreal for time.
16993 (dump_inline_summary): Update.
16994 (estimate_function_body_sizes): Update.
16995 (estimate_edge_size_and_time): Update.
16996 (estimate_calls_size_and_time): Update.
16997 (estimate_node_size_and_time): Update.
16998 (inline_merge_summary): Update.
16999 (inline_update_overall_summary): Update.
17000 (estimate_time_after_inlining): Update.
17001 (inline_read_section): Update.
17002 (inline_write_summary): Update.
17003 * ipa-inline.c (compute_uninlined_call_time): Update.
17004 (compute_inlined_call_time): Update.
17005 (recursive_inlining): Update.
17006 (inline_small_functions): Update.
17007 (dump_overall_stats): Update.
17008 * ipa-inline.h: Include sreal.h.
17009 (size_time_entry): Turn time to sreal.
17010 (inline_summary): Turn self_time nad time to sreal.
17012 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17014 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
17016 (sreal::stream_out, sreal::stream_in): New.
17017 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
17019 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17021 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
17024 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
17027 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
17029 (dimode_scalar_chain::compute_convert_gain): Ditto.
17030 (dimode_scalar_chain::make_vector_copies): Ditto.
17031 (dimode_scalar_chain::convert_reg): Ditto.
17032 (dimode_scalar_chain::convert_insn): Ditto.
17033 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
17034 (VI248_AVX512BW_1): New mode iterator.
17035 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
17036 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
17039 2017-04-25 Martin Sebor <msebor@redhat.com>
17041 PR tree-optimization/80497
17042 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
17043 constants are representable in HOST_WIDE_INT.
17044 (parse_directive): Ditto.
17046 2017-04-25 Martin Sebor <msebor@redhat.com>
17049 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
17050 (new_zero_array): Adjust signature.
17051 (dom_info::dom_init): Used unsigned rather that size_t.
17052 (dom_info::dom_info): Same.
17054 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17055 Jakub Jelinek <jakub@redhat.com>
17058 * config/arm/arm.c: Include gimple.h.
17059 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17060 returns negative, increment ncrn only if it returned positive.
17061 (arm_needs_doubleword_align): Return int instead of bool,
17062 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
17063 members, but if there is any such non-FIELD_DECL
17064 > PARM_BOUNDARY aligned decl, return -1 instead of false.
17065 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17066 returns negative, increment nregs only if it returned positive.
17067 (arm_setup_incoming_varargs): Likewise.
17068 (arm_function_arg_boundary): Emit -Wpsabi note if
17069 arm_needs_doubleword_align returns negative, return
17070 DOUBLEWORD_ALIGNMENT only if it returned positive.
17072 2017-04-25 Marek Polacek <polacek@redhat.com>
17075 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17076 first argument to type.
17078 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
17081 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
17082 type checks to test for compatibility instead of equality.
17084 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17085 Jakub Jelinek <jakub@redhat.com>
17088 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
17090 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
17091 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
17092 the alignment computation, but return their maximum in warn_alignment.
17093 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17094 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
17096 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
17097 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17100 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17102 * config/arc/simdext.md (dmpyh): Fix typo.
17104 2017-04-25 Richard Biener <rguenther@suse.de>
17106 PR tree-optimization/80492
17107 * alias.c (compare_base_decls): Handle registers with asm
17108 specification conservatively.
17109 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17110 compare_base_decls returning dont-know properly.
17112 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17114 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
17115 (legitimate_offset_address_p): New function.
17116 (arc_legitimate_address_p): Use above function.
17118 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17120 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
17122 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17124 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
17125 ACCH registers whenever they are available.
17127 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17129 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
17130 double regs fix when not used.
17132 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17134 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
17136 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
17137 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
17139 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17141 * config/arc/arc.c (arc_output_addsi): Check for h-register class
17142 when emitting short ADD instructions.
17144 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17146 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
17148 (cmpsi_cc_c_insn): Likewise.
17149 (cbranchsi4_scratch): Compute proper instruction length using
17150 compact_hreg_operand.
17151 * config/arc/predicates.md (compact_hreg_operand): New predicate.
17153 2017-04-25 Richard Biener <rguenther@suse.de>
17155 PR middle-end/80509
17156 * passes.c (pass_manager::pass_manager): Initialize
17157 m_name_to_pass_map.
17159 2017-04-25 Richard Biener <rguenther@suse.de>
17161 PR tree-optimization/79201
17162 * tree-ssa-sink.c (statement_sink_location): Handle calls.
17164 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17167 * config/s390/vector.md: Split MEM->GPR vector moves for
17168 non-s_operand addresses.
17170 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17173 * config/s390/predicates.md (reload_const_wide_int_operand): New
17175 * config/s390/s390.md ("movti"): Remove d/P alternative.
17176 ("movti_bigconst"): New pattern definition.
17178 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17181 * s390-protos.h (s390_expand_cs_hqi): Removed.
17182 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
17183 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
17184 modes as well as CCZ1mode and CCZmode.
17185 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
17186 signature of s390_emit_compare_and_swap.
17187 (s390_expand_cs_hqi): Likewise, make static.
17188 (s390_expand_cs_tdsi): Generate an explicit compare before trying
17189 compare-and-swap, in some cases.
17190 (s390_expand_cs): Wrapper function.
17191 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
17193 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
17194 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
17195 patterns for small and large integers. Forbid symref memory operands.
17196 Move expander to s390.c. Require cc register.
17197 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
17198 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
17199 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
17200 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
17201 symref memory operands. Remove CC mode and call s390_match_ccmode
17203 ("atomic_exchange<mode>"): Allow and implement all integer modes.
17205 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17207 * config/s390/s390.md (define_peephole2): New peephole to help
17208 combining the load-and-test pattern with volatile memory.
17210 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17212 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
17213 with CCZmode for TARGET_Z196.
17215 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17217 PR rtl-optimization/80501
17218 * combine.c (make_compound_operation_int): Set subreg_code to SET
17219 even for AND with mask of the sign bit of mode.
17221 PR rtl-optimization/80500
17222 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
17223 sum's initial value.
17225 2017-04-25 Julian Brown <julian@codesourcery.com>
17226 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17228 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
17230 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
17232 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
17234 2017-04-25 Julian Brown <julian@codesourcery.com>
17235 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17237 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
17238 (thunderx2t99_sha): New Reservation.
17240 2017-04-25 Julian Brown <julian@codesourcery.com>
17241 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17243 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
17244 type for 1-element load.
17246 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
17248 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
17250 2017-04-24 Martin Jambor <mjambor@suse.cz>
17252 PR tree-optimization/80293
17253 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
17254 char arrays not totally scalarizable if it is false.
17255 (analyze_all_variable_accesses): Pass correct value in the new
17256 parameter. Add a statistics counter.
17258 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
17260 PR middle-end/79931
17261 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
17263 2017-04-24 Richard Biener <rguenther@suse.de>
17265 PR tree-optimization/80494
17266 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
17267 out for complex types.
17269 2017-04-24 Richard Biener <rguenther@suse.de>
17271 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
17272 * tree-ssa-sccvn.c (print_scc): Print SCC size.
17273 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
17274 (DFS): Adjust and never fail.
17275 (sccvn_dom_walker::fail): Remove.
17276 (sccvn_dom_walker::before_dom_children): Adjust.
17277 (run_scc_vn): Likewise and never fail.
17278 * tree-ssa-pre.c (pass_pre::execute): Adjust.
17279 (pass_fre::execute): Likewise.
17281 2017-04-24 Richard Biener <rguenther@suse.de>
17283 PR tree-optimization/79725
17284 * tree-ssa-sink.c (statement_sink_location): Return whether
17285 failure reason was zero uses. Move that check later.
17286 (sink_code_in_bb): Deal with zero uses by removing the stmt
17289 2017-04-24 Richard Biener <rguenther@suse.de>
17292 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
17293 pointer-based references.
17295 2017-04-24 Richard Biener <rguenther@suse.de>
17298 * pass_manager.h (pass_manager::operator new): Remove.
17299 (pass_manager::operator delete): Likewise.
17300 * passes.c (pass_manager::operator new): Remove.
17301 (pass_manager::operator delete): Likewise.
17302 (pass_manager::pass_manager): Zero individual pass members.
17304 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
17307 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
17308 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
17309 Check "XEXP (src, 1)" operand here.
17310 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
17311 Check "XEXP (src, 1)" operand here.
17312 (dimode_scalar_chain::make_vector_copies): Detect count register
17313 of a shift instruction. Zero extend count register from QImode
17314 to DImode to satisfy vector shift pattern count operand predicate.
17315 Substitute vector shift count operand with a DImode copy.
17316 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
17319 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
17321 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
17322 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17323 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
17324 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17325 (UNSPEC_NOREX_MEM): Remove definition.
17327 2017-04-21 Richard Biener <rguenther@suse.de>
17329 PR tree-optimization/79547
17330 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17331 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
17332 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
17333 without any constraints.
17335 2017-04-21 Richard Biener <rguenther@suse.de>
17337 PR tree-optimization/78847
17338 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
17340 2017-04-21 Richard Biener <rguenther@suse.de>
17342 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
17343 (build_distinct_type_copy): Likewise.
17344 (build_variant_type_copy): Likewise.
17345 * tree.c (build_qualified_type): Pass down mem-stat info.
17346 (build_distinct_type_copy): Likewise.
17347 (build_variant_type_copy): Likewise.
17349 2017-04-21 Richard Biener <rguenther@suse.de>
17351 PR tree-optimization/80237
17352 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
17354 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
17355 for a simplified result.
17357 2016-04-21 Richard Biener <rguenther@suse.de>
17359 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
17360 sth as strict as a simple_iv but a chrec without symbols and an
17361 operand defined in the loop we are peeling (and not some subloop).
17362 (propagate_constants_for_unrolling): Propagate all constants.
17364 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17367 * config/i386/i386.c (print_reg): Remove assert for disalowed
17368 regno values, call output_operand_lossage instead.
17370 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17373 * config/i386/constraints.md (Yc): New register constraint.
17374 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
17375 Use Yc constraint for alternative 2 of operand 0. Remove
17376 preferred_for_speed attribute.
17378 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17380 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
17381 lastprivate clauses in SIMT case.
17383 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
17385 * doc/invoke.texi (-Wextra-semi): Document new warning option.
17387 2017-04-20 Richard Biener <rguenther@suse.de>
17389 PR tree-optimization/57796
17390 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
17391 as N scalar stores.
17392 (vect_model_load_cost): Cost gathers as N scalar loads.
17394 2017-04-20 Richard Biener <rguenther@suse.de>
17396 * ggc-page.c (ggc_allocated_p): Rename to ...
17397 (safe_lookup_page_table_entry): ... this and return the lookup
17399 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
17401 2017-04-20 Richard Biener <rguenther@suse.de>
17403 PR tree-optimization/80453
17404 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
17405 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
17406 from the conditions.
17407 (vn_phi_eq): Pass them down.
17408 (vn_phi_lookup): Record them.
17409 (vn_phi_insert): Likewise.
17411 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
17413 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
17414 uninitialized variable warning to avoid buffer overrun.
17416 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17419 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
17420 is suppressed for '{ 0 }' in C.
17422 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17424 * BASE-VER: Set to 8.0.0.
17426 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
17428 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
17429 priority .init_array and .fini_array section with SECTION_NOTYPE
17432 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17434 PR middle-end/80423
17435 * tree.h (build_array_type): Add typeless_storage default argument.
17436 * tree.c (type_cache_hasher::equal): Also compare
17437 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
17438 (build_array_type): Add typeless_storage argument, set
17439 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
17441 (build_nonshared_array_type): Adjust build_array_type_1 caller.
17442 (build_array_type): Likewise. Add typeless_storage argument.
17444 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17445 Jakub Jelinek <jakub@redhat.com>
17447 PR tree-optimization/80426
17448 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
17449 operation on symbolic operands, also compute the overflow for the
17450 invariant part when the operation degenerates into a negation.
17452 2017-04-19 Jakub Jelinek <jakub@redhat.com>
17455 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
17456 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
17459 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
17461 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
17464 * config/avr/avr.c (tree.h): Include it.
17465 (cgraph.h): Include it.
17466 (avr_encode_section_info): Don't warn for uninitialized progmem
17467 variable if it's just an alias.
17469 2017-04-19 Richard Biener <rguenther@suse.de>
17472 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
17473 when needed by AutoPGO.
17475 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
17478 * doc/lto.texi: Remove an extra 'that'.
17480 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
17482 PR rtl-optimization/80429
17483 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
17484 are only used in debug insns.
17486 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17487 Vladimir Makarov <vmakarov@redhat.com>
17489 * config/sparc/predicates.md (input_operand): Add comment. Return
17490 true for any memory operand when LRA is in progress.
17491 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
17493 2017-04-18 Jeff Law <law@redhat.com>
17496 * mips.md ({return,simple_return}_internal): Do not overwrite
17499 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17501 PR tree-optimization/80443
17502 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
17503 instead of adding 1, subtract -1 and similarly instead of subtracting
17506 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
17508 PR rtl-optimization/80357
17509 * haifa-sched.c (tmp_bitmap): New variable.
17510 (model_recompute): Handle duplicate use records.
17511 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
17512 (free_global_sched_pressure_data): Free it.
17514 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
17517 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
17518 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17519 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17520 instead of SYSTEM_HEADER_DIR.
17522 2017-04-18 Jeff Law <law@redhat.com>
17524 PR middle-end/80422
17525 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
17526 predecessors after walking up the insn chain.
17528 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17531 * dwarf2out.c (modified_type_die): Try harder not to emit internal
17532 sizetype type into debug info.
17534 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17537 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
17538 unneeded test for TARGET_UPPER_REGS_SF.
17539 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
17541 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17544 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
17545 instead of gsi_after_labels.
17547 2017-04-18 Jeff Law <law@redhat.com>
17549 * regcprop.c (maybe_mode_change): Avoid creating copies of the
17553 2017-04-13 Jeff Law <law@redhat.com>
17554 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17555 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17557 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
17560 * config/avr/avr.c (intl.h): Include it.
17561 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
17563 2017-04-18 Martin Liska <mliska@suse.cz>
17565 PR gcov-profile/78783
17566 * gcov-tool.c (gcov_output_files): Validate that destination
17567 file is either removed by the tool or by a user.
17569 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17570 Guy Benyei <guybe@mellanox.com>
17572 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
17573 block, and do not negate it, the stored id is already negative.
17575 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17577 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
17579 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17582 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
17583 masks of options that should be turned off if the VSX vector
17584 options are turned off.
17585 (OTHER_P8_VECTOR_MASKS): Likewise.
17586 (OTHER_VSX_VECTOR_MASKS): Likewise.
17587 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
17588 rs6000_disable_incompatible_switches to validate no type switches
17590 (rs6000_incompatible_switch): New function to disallow turning on
17591 other vector options if -mno-vsx, -mno-power8-vector, or
17592 -mno-power9-vector are specified.
17594 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17596 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
17598 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17600 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
17601 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
17602 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
17603 (ARG_POINTER_CFA_OFFSET): Likewise.
17605 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17607 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
17608 conditions to take advantage of various optimizations.
17610 2017-04-13 Jeff Law <law@redhat.com>
17612 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17613 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17614 (zero_extendsidi2_dext): Likewise.
17616 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17619 * fold-const.c (fold_ternary_loc): Revert
17620 use op0 instead of fold_convert_loc (loc, type, arg0) part of
17623 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
17625 PR rtl-optimization/80343
17626 * lra-remat.c (update_scratch_ops): Assign original hard reg to
17627 new scratch pseudo.
17629 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
17632 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
17633 to ubsan_encode_value.
17635 2017-04-13 Jeff Law <law@redhat.com>
17637 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
17638 appearing in DEBUG_INSNs.
17640 2017-04-13 Martin Liska <mliska@suse.cz>
17642 PR gcov-profile/80413
17643 * gcov-io.c (gcov_write_string): Copy to buffer just when
17644 allocated size is greater than zero.
17646 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17649 * dwarf2out.c (decls_for_scope): Ignore declarations of
17650 current_function_decl in BLOCK_NONLOCALIZED_VARS.
17652 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
17655 * ipa-visibility.c (non_local_p): Fix typos.
17656 (localize_node): When localizing symbol in same comdat group,
17657 dissolve the group only when we know external symbols are going
17659 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
17661 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17663 PR tree-optimization/79390
17664 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
17665 order does not result in usable sequence, retry with reversed operand
17671 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
17672 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
17673 op0 instead of fold_convert_loc (loc, type, arg0).
17675 2017-04-12 Jeff Law <law@redhat.com>
17677 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
17678 has a delay slot in the generated code.
17680 * config/cris/cris.md (cris_preferred_reload_class): Return
17681 GENNONACR_REGS rather than GENERAL_REGS.
17683 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17686 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
17687 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
17688 signedness of the result type.
17690 2017-04-12 Richard Biener <rguenther@suse.de>
17691 Jeff Law <law@redhat.com>
17693 PR tree-optimization/80359
17694 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
17695 trim stores to TARGET_MEM_REFs.
17697 2017-04-12 Richard Biener <rguenther@suse.de>
17699 PR tree-optimization/79390
17700 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
17701 threading case even more.
17703 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
17706 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
17707 for quad_address_p for TImode, instead of just not indexed_address.
17709 2017-04-12 Richard Biener <rguenther@suse.de>
17710 Bernd Edlinger <bernd.edlinger@hotmail.de>
17712 PR middle-end/79671
17713 * alias.c (component_uses_parent_alias_set_from): Handle
17714 TYPE_TYPELESS_STORAGE.
17715 (get_alias_set): Likewise.
17716 * tree-core.h (tree_type_common): Add typeless_storage flag.
17717 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
17718 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
17719 for types containing members with TYPE_TYPELESS_STORAGE.
17720 (place_field): Likewise.
17721 (layout_type): Likewise for ARRAY_TYPE.
17722 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
17723 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
17724 TYPE_TYPELESS_STORAGE.
17725 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
17727 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17730 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17731 first argument to type.
17733 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17737 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
17738 CONST0_RTX (mode) rather than const0_rtx where appropriate.
17739 (rs6000_expand_binop_builtin): Likewise.
17740 (rs6000_expand_ternop_builtin): Likewise; also add missing
17741 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
17742 vshasigma built-ins.
17743 * doc/extend.texi: Document that vec_xxpermdi's third argument
17744 must be a constant.
17746 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
17748 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
17749 Use shift_const cost parameter when calculating gain of STV shifts.
17751 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
17753 PR rtl-optimization/70478
17754 * lra-constraints.c (process_alt_operands): Check memory for
17755 disfavoring memory insn operand.
17757 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17759 PR middle-end/80100
17760 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
17761 left shift in unsigned HOST_WIDE_INT type.
17763 PR rtl-optimization/80385
17764 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
17765 (not (neg X)) into (plus X -1) for complex or non-integral modes.
17768 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
17769 if they have any depend clauses.
17771 2017-04-11 Martin Liska <mliska@suse.cz>
17774 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
17775 * ipa-split.c (split_function): Create a local comdat symbol
17776 if caller is in a comdat group.
17778 2017-04-11 Martin Liska <mliska@suse.cz>
17781 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
17784 2017-04-11 Martin Sebor <msebor@redhat.com>
17786 PR middle-end/80364
17787 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
17788 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
17790 (directive::set_width, directive::set_precision, format_character):
17792 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
17795 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
17798 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
17799 conflict, set target->arch_name instead of target->cpu_name.
17801 2017-04-11 Richard Biener <rguenther@suse.de>
17803 PR tree-optimization/80374
17804 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
17805 build_zero_cst, remove fold_convertible_p check again.
17807 2017-04-11 Martin Liska <mliska@suse.cz>
17810 * ubsan.c (instrument_object_size): Do not instrument register
17813 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17816 * config/i386/i386-builtin-types.def
17817 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
17818 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
17819 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
17820 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
17821 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
17822 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
17823 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
17824 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
17825 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
17826 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
17827 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
17828 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
17829 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
17830 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
17831 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
17832 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
17833 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
17834 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
17835 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
17836 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
17837 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
17838 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
17839 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
17840 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
17841 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
17842 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
17843 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
17844 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
17845 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
17846 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
17847 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
17848 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
17849 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
17850 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
17851 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
17852 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
17853 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
17854 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
17855 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
17856 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
17857 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
17858 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
17859 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
17860 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
17861 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
17862 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
17864 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
17865 flag to second_arg_count, handle 4 argument function type _COUNT
17866 aliases, handle second_arg_count on second argument rather than last.
17868 2017-04-10 Jeff Law <law@redhat.com>
17870 PR tree-optimization/80374
17871 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
17872 record anything if we can not convert integer_zero_node to the
17875 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
17878 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17879 Enhance special handling given to the TARGET_P9_MINMAX option in
17880 relation to certain other options.
17882 2017-04-10 Bin Cheng <bin.cheng@arm.com>
17884 PR tree-optimization/80153
17885 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
17886 remove POINTER_PLUS_EXPR's base part directly, rather than through
17889 2017-04-10 Richard Biener <rguenther@suse.de>
17890 Bin Cheng <bin.cheng@arm.com>
17892 PR tree-optimization/80153
17893 * tree-affine.c (aff_combination_to_tree): Get base pointer from
17894 the first element of pointer type aff_tree. Build result expr in
17896 (add_elt_to_tree): Convert to type unconditionally. Remove other
17897 fold_convert calls.
17898 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
17899 (rewrite_use_nonlinear_expr): Check invariant using iv information.
17901 2017-04-10 Richard Biener <rguenther@suse.de>
17903 * tree-ssa-structalias.c (find_func_aliases): Properly handle
17906 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
17908 PR rtl-optimization/70478
17909 * lra-constraints.c (curr_small_class_check): New.
17910 (update_and_check_small_class_inputs): New.
17911 (process_alt_operands): Update curr_small_class_check. Disfavor
17912 alternative insn memory operands. Check available regs for small
17915 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
17918 * config/mips/mips.opt (-mvirt): Update description.
17919 * doc/invoke.texi (-mvirt): Likewise.
17921 2017-04-10 Richard Biener <rguenther@suse.de>
17923 PR middle-end/80362
17924 * fold-const.c (fold_binary_loc): Look at unstripped ops when
17925 looking for NEGATE_EXPR in -A / -B to A / B folding.
17927 2017-04-10 Martin Liska <mliska@suse.cz>
17929 PR gcov-profile/80224
17930 * gcov.c (print_usage): Fix usage string.
17931 (get_gcov_intermediate_filename): Remove.
17932 (output_gcov_file): Use both for normal and intermediate format.
17933 (generate_results): Do not initialize special file for
17934 intermediate format.
17936 2017-04-10 Richard Biener <rguenther@suse.de>
17938 PR tree-optimization/80304
17939 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
17942 2017-04-10 Nathan Sidwell <nathan@acm.org>
17945 * config/rs6000/rs6000.c (rs6000_vector_type): New.
17946 (rs6000_init_builtins): Use it.
17948 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17950 * config/arm/arm.md (<mrc>): Add mode to SET source.
17951 (<mrrc>): Likewise.
17953 2017-04-10 Richard Biener <rguenther@suse.de>
17955 PR middle-end/80344
17956 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
17958 2017-04-10 Jakub Jelinek <jakub@redhat.com>
17961 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
17962 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
17963 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
17964 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
17965 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
17966 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
17967 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
17968 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
17969 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
17970 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
17971 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
17972 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
17973 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
17974 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
17975 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
17976 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
17977 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
17978 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
17979 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
17980 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
17981 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
17982 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
17983 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
17985 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
17987 PR rtl-optimization/70478
17988 * lra-constraints.c: Reverse the last patch.
17990 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
17992 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
17993 Add comment for WCHAR_T.
17995 2017-04-08 Martin Liska <mliska@suse.cz>
17998 2017-04-07 Martin Liska <mliska@suse.cz>
18001 * ipa-split.c (split_function): Add function part to a same comdat
18004 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18007 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
18009 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
18011 * rs6000/rs6000.c (vec_load_pendulum): Rename...
18012 (vec_pairing): ...to this.
18013 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
18014 (rs6000_sched_init): Adjust for name change.
18015 (struct rs6000_sched_context): Likewise.
18016 (rs6000_init_sched_context): Likewise.
18017 (rs6000_set_sched_context): Likewise.
18019 2017-04-07 Jakub Jelinek <jakub@redhat.com>
18025 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
18027 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
18028 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
18029 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
18031 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
18033 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
18035 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18037 PR rtl-optimization/70703
18038 * ira-color.c (update_conflict_hard_regno_costs): Use
18039 int64_t instead of HOST_WIDE_INT.
18041 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18043 PR rtl-optimization/70478
18044 * lra-constraints.c (process_alt_operands): Disfavor alternative
18045 insn memory operands.
18047 2017-04-07 Jeff Law <law@redhat.com>
18049 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
18050 CALL and NOTE_INSN_CALL_ARG_LOCATION.
18052 2017-04-07 Martin Liska <mliska@suse.cz>
18055 * config/aarch64/aarch64.c (aarch64_process_target_attr):
18056 Show error message instead of an ICE.
18058 2017-04-07 Martin Liska <mliska@suse.cz>
18061 * ipa-split.c (split_function): Add function part to a same comdat
18064 2017-04-07 Richard Biener <rguenther@suse.de>
18066 PR middle-end/80341
18067 * tree.c (get_unwidened): Also handle ! for_type case for
18069 * convert.c (do_narrow): Split out from ...
18070 (convert_to_integer_1): ... here. Do not pass final truncation
18071 type to get_unwidened for TRUNC_DIV_EXPR.
18073 2017-04-07 Richard Biener <rguenther@suse.de>
18075 * tree-affine.c (wide_int_ext_for_comb): Take type rather
18077 (aff_combination_const): Adjust.
18078 (aff_combination_scale): Likewise.
18079 (aff_combination_add_elt): Likewise.
18080 (aff_combination_add_cst): Likewise.
18081 (aff_combination_convert): Likewise.
18082 (add_elt_to_tree): Likewise. Remove unused argument.
18083 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
18085 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
18087 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
18089 * config/arm/arm.c (arm_default_short_enums): Use
18090 ARM_DEFAULT_SHORT_ENUMS.
18091 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
18093 2017-04-06 Jakub Jelinek <jakub@redhat.com>
18096 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
18097 members with redundant out-of-class redeclaration.
18099 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18102 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
18103 * config/i386/i386.md (*zero_extendsidi2):
18104 Add (?*x,*x) and (?*v,*v) alternatives.
18106 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18109 * config/i386/i386.c (ix86_expand_builtin)
18110 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
18111 mode from insn data. Convert operands to insn operand mode.
18112 Copy operands that don't satisfy insn predicate to a register.
18114 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
18116 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
18119 2017-04-06 Richard Biener <rguenther@suse.de>
18121 PR tree-optimization/80334
18122 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
18123 preserve alignment of accesses.
18125 2017-04-06 Richard Biener <rguenther@suse.de>
18127 PR tree-optimization/80262
18128 * tree-sra.c (build_ref_for_offset): Preserve address-space
18130 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
18131 Drop useless address-space information on MEM_REF offsets.
18133 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
18135 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
18137 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18139 PR rtl-optimization/70703
18140 * ira-color.c (update_conflict_hard_regno_costs): Use
18141 HOST_WIDE_INT instead of long.
18143 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
18146 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
18147 not defined for x86_64 target. Add -mmmx target option when __SSE2__
18149 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
18150 for x86_64 target. Handle -m3dnowa option.
18152 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18154 PR rtl-optimization/70703
18155 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
18156 (update_conflict_hard_regno_costs): Use long instead of unsigned
18157 arithmetic for cost calculation.
18159 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18160 Bernd Edlinger <bernd.edlinger@hotmail.de>
18163 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
18166 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
18169 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
18170 ptr_mode with Pmode throughout.
18171 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
18172 into probe_stack_range and use DImode.
18174 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
18177 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
18178 call_eh_return is true.
18180 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18182 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
18183 Initialize last_match_fntype_index.
18185 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18188 * tree-nvr.c: Include internal-fn.h.
18189 (pass_return_slot::execute): Ignore internal calls without
18192 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18193 Richard Biener <rguenther@suse.de>
18196 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
18197 captures used multiple times, except for the last use.
18198 * generic-match-head.c: Include gimplify.h.
18200 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18202 PR tree-optimization/79390
18203 * target.h (struct noce_if_info): Declare.
18204 * targhooks.h (default_noce_conversion_profitable_p): Declare.
18205 * target.def (noce_conversion_profitable_p): New target hook.
18206 * ifcvt.h (struct noce_if_info): New type, moved from ...
18207 * ifcvt.c (struct noce_if_info): ... here.
18208 (noce_conversion_profitable_p): Renamed to ...
18209 (default_noce_conversion_profitable_p): ... this. No longer
18211 (noce_try_store_flag_constants, noce_try_addcc,
18212 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
18213 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
18214 instead of noce_conversion_profitable_p.
18215 * config/i386/i386.c: Include ifcvt.h.
18216 (ix86_option_override_internal): Don't override
18217 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
18218 (ix86_noce_conversion_profitable_p): New function.
18219 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
18220 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
18221 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
18222 * doc/tm.texi: Regenerated.
18224 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18226 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
18229 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18232 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
18233 instructions for small multiply cores.
18235 2017-04-04 Jeff Law <law@redhat.com>
18237 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
18239 (mips_expand_vec_perm_const): Initialize elements in orig_perm
18240 that are not set by the loop over the elements.
18242 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18245 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
18246 int mode, convert_modes it to mode as unsigned, otherwise use
18247 lowpart_subreg to mode rather than SImode.
18248 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
18249 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
18250 Use DImode instead of SImode for the shift count operand.
18251 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
18254 2017-04-04 Richard Biener <rguenther@suse.de>
18256 PR middle-end/80281
18257 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
18258 arithmetic done for the negate or the plus. Simplify.
18259 (A - (-B) -> A + B): Likewise.
18260 * fold-const.c (split_tree): Make sure to not negate pointers.
18262 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
18264 PR rtl-optimization/60818
18265 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
18266 a compare of comparisons with the thing compared if this results
18267 in a different machine mode.
18269 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
18271 * alias.c (base_alias_check): Fix typo in comment.
18272 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
18273 * cgraphunit.c (symbol_table::compile): Likewise.
18274 * collect2.c (maybe_run_lto_and_relink): Likewise.
18275 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
18276 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
18277 * config/avr/avr.c (avr_map_op_t): Likewise.
18278 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
18279 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
18280 * config/epiphany/epiphany.md (movcc): Likewise.
18281 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
18282 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
18284 * config/mips/mips.c (mips_save_restore_reg): Likewise.
18285 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
18286 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
18287 * config/sh/sh.c (sh_rtx_costs): Likewise.
18288 * fold-const.c (fold_truth_andor): Likewise.
18289 * genautomata.c (collapse_flag): Likewise.
18290 * gengtype.h (struct type::u::s): Likewise.
18291 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
18292 * input.c (FORMAT_AMOUNT): Likewise.
18293 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
18294 (known_aggs_to_agg_replacement_list): Likewise.
18295 * ipa-inline-analysis.c: Likewise.
18296 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
18297 * ipa-polymorphic-call.c
18298 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
18299 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
18300 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
18302 * modulo-sched.c (apply_reg_moves): Likewise.
18303 * omp-expand.c (build_omp_regions_1): Likewise.
18304 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
18305 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
18306 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
18307 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
18308 * value-prof.c: Likewise.
18309 * var-tracking.c (val_reset): Likewise.
18311 2017-04-03 Richard Biener <rguenther@suse.de>
18313 PR tree-optimization/80275
18314 * fold-const.c (split_address_to_core_and_offset): Handle
18317 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
18319 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
18320 descriptors is at least equal to that of functions.
18322 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18324 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
18326 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18329 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
18330 (mov<IMOD4:mode>): New expander.
18331 (*mov<IMOD4:mode>_internal): New insn and split pattern.
18333 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
18335 PR rtl-optimization/79405
18336 * fwprop.c (propagations_left): New variable.
18337 (forward_propagate_into): Decrement it.
18338 (fwprop_init): Initialize it.
18339 (fw_prop): If the variable has reached zero, stop propagating.
18340 (fwprop_addr): Ditto.
18342 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18345 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
18346 a FUNCTION_DECL, pass it as decl instead of origin to
18349 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
18351 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
18354 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
18357 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
18358 TARGET_VSX_SMALL_INTEGER.
18360 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18362 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18363 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
18365 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18367 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
18368 extraction from odd-numbered MSA register.
18370 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18372 PR middle-end/80173
18373 * expmed.c (store_bit_field_1): Don't attempt to create
18374 a word subreg out of hard registers wider than word if they
18375 have HARD_REGNO_NREGS of 1 for their mode.
18377 PR middle-end/80163
18378 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
18379 conversions to integer types wider than word and pointer.
18382 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
18383 (rtx_equal_for_cselib_p): Pass 0 to it.
18384 * cselib.c (cselib_hasher::equal): Likewise.
18385 (rtx_equal_for_cselib_1): Add depth argument. If depth
18386 is 128, don't look up VALUE locs and punt. Increment
18387 depth in recursive calls when walking VALUE locs.
18389 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
18391 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
18392 (make_gcov_file_name): Use the canonical path name for generating
18394 (read_line): Fix handling of files with ascii null bytes.
18396 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
18398 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
18399 to initialise a vector register instead
18400 of using a const_int.
18402 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18404 PR translation/80189
18405 * gimplify.c (omp_default_clause): Use %qs instead of %s in
18406 diagnostic messages.
18408 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
18411 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
18412 (dfp_diex_<mode>): Update mode of operand 1.
18413 * doc/extend.texi (dxex, dxexq): Document change to return type.
18414 (diex, diexq): Document change to argument type.
18416 2017-03-30 Martin Jambor <mjambor@suse.cz>
18419 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
18420 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
18421 it reflects the signature changes performed at the callee side.
18422 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
18423 to cgraph_build_function_type_skip_args.
18424 (build_function_decl_skip_args): Adjust call to the above function.
18426 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18429 * config/i386/sse.md
18430 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
18431 register as dest whenever it is a MEM not rtx_equal_p to the
18432 corresponding dup operand, and when forcing into reg move the
18433 reg into the memory afterwards.
18434 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
18435 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
18436 for the force_reg mode.
18437 (avx512vl_vextractf128<mode>): Use register as dest either
18438 always when a MEM, or when it is a MEM not rtx_equal_p to the
18439 corresponding dup operand, or even not when it is a CONST_VECTOR
18440 depending on the mode and lo vs. hi.
18441 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
18443 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
18444 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
18445 Likewise. Require that operands[2] is even.
18446 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
18447 Remove extraneous parens. Require that operands[2] is a multiple
18449 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
18450 operands[0] is a MEM if <mask_applied>, the predicates/constraints
18451 disallow memory then.
18453 2017-03-30 Richard Biener <rguenther@suse.de>
18455 PR tree-optimization/77498
18456 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
18457 to non-constants over backedges.
18459 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
18461 PR rtl-optimization/80233
18462 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
18463 as last_combined_insn. Do not test for BARRIER_P separately.
18465 2017-03-29 Andreas Schwab <schwab@suse.de>
18468 * calls.c (prepare_call_address): Convert funexp to Pmode before
18469 copying to temp reg.
18471 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18473 PR tree-optimization/80158
18474 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
18475 Handle possible future case of more than one alternate
18477 (replace_rhs_if_not_dup): Likewise.
18478 (replace_one_candidate): Likewise.
18480 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
18482 PR rtl-optimization/80193
18483 * ira.c (ira): Do not check allocation for LRA.
18485 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
18487 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
18488 (nvptx_output_simt_exit): Declare.
18489 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
18490 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
18491 (init_softstack_frame): Move initialization of crtl->is_leaf to...
18492 (nvptx_declare_function_name): ...here. Emit declaration of local
18493 memory space buffer for omp_simt_enter insn.
18494 (nvptx_output_unisimt_switch): New.
18495 (nvptx_output_softstack_switch): New.
18496 (nvptx_output_simt_enter): New.
18497 (nvptx_output_simt_exit): New.
18498 * config/nvptx/nvptx.h (struct machine_function): New fields
18499 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
18500 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
18501 (UNSPECV_SIMT_EXIT): Ditto.
18502 (omp_simt_enter_insn): New insn.
18503 (omp_simt_enter): New expansion.
18504 (omp_simt_exit): New insn.
18505 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
18507 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
18508 (expand_GOMP_SIMT_ENTER_ALLOC): New.
18509 (expand_GOMP_SIMT_EXIT): New.
18510 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
18511 (GOMP_SIMT_ENTER_ALLOC): Ditto.
18512 (GOMP_SIMT_EXIT): Ditto.
18513 * target-insns.def (omp_simt_enter): New insn.
18514 (omp_simt_exit): Ditto.
18515 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
18517 (lower_rec_simd_input_clauses): Implement SIMT privatization.
18518 (lower_rec_input_clauses): Likewise.
18519 (lower_lastprivate_clauses): Handle SIMT privatization.
18521 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
18522 (ompdevlow_adjust_simt_enter): New.
18523 (find_simtpriv_var_op): New.
18524 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
18525 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
18527 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
18528 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
18529 (copy_decl_for_dup_finish): Ditto.
18531 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
18533 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
18536 * config/i386/i386.c (ix86_option_override_internal): Always
18537 allow -mpreferred-stack-boundary=3 for 64-bit targets.
18539 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18541 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
18543 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18545 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
18546 mark new edge's irreducible flag accordign to it.
18547 (vect_do_peeling): Check loop preheader edge's irreducible flag
18548 and pass it to function slpeel_add_loop_guard.
18550 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
18552 PR tree-optimization/80218
18553 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18554 Update block frequencies and counts.
18556 2017-03-28 Richard Biener <rguenther@suse.de>
18558 PR tree-optimization/78644
18559 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
18560 of a simplification result we may not use it at all.
18562 2017-03-28 Richard Biener <rguenther@suse.de>
18565 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
18566 without arguments, generate default definition of a SSA name.
18568 2017-03-28 Richard Biener <rguenther@suse.de>
18570 PR middle-end/80222
18571 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
18572 TYPE_REF_CAN_ALIAS_ALL references.
18573 * fold-const.c (fold_indirect_ref_1): Likewise.
18575 2017-03-28 Martin Liska <mliska@suse.cz>
18578 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
18579 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
18581 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18582 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
18584 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
18585 (EXTRA_SPECS): Define.
18586 (SUBTARGET_EXTRA_SPECS): Likewise.
18587 (SUBTARGET_CPP_SPEC): Likewise.
18588 * config/arc/elf.h (EXTRA_SPECS): Renamed to
18589 SUBTARGET_EXTRA_SPECS.
18590 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
18592 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18594 * config/arc/simdext.md (vst64_insn): Update pattern.
18595 (vld32wh_insn): Likewise.
18596 (vld32wl_insn): Likewise.
18597 (vld64_insn): Likewise.
18598 (vld32_insn): Likewise.
18600 2017-03-28 Marek Polacek <polacek@redhat.com>
18603 * fold-const.c (fold_comparison): Use protected_set_expr_location
18604 instead of SET_EXPR_LOCATION.
18606 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
18608 * tree.c (add_expr): Avoid name lookup warning.
18610 2017-03-27 Jeff Law <law@redhat.com>
18612 PR tree-optimization/80216
18613 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
18614 function name. Limit recursion depth.
18615 (record_temporary_equivalences): Corresponding changes.
18617 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
18619 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
18622 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18625 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
18627 * cfgcleanup.c (reg_note_cfa_p): New array.
18628 (insns_have_identical_cfa_notes): New function.
18629 (old_insns_match_p): Don't cross-jump in between /f
18630 and non-/f instructions. If both i1 and i2 are frame related,
18631 verify all CFA notes, their order and content.
18633 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
18636 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
18637 HImode and SImode with zero extend to DImode to one insn.
18638 (bswap<mode>2_extenddi): Likewise.
18639 (bswapsi2_extenddi): Likewise.
18640 (bswaphi2_extendsi): Likewise.
18641 (bswaphi2): Combine bswap HImode and SImode into one insn.
18642 Separate memory insns from swapping register.
18643 (bswapsi2): Likewise.
18644 (bswap<mode>2): Likewise.
18645 (bswaphi2_internal): Delete, no longer used.
18646 (bswapsi2_internal): Likewise.
18647 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
18648 store, and gpr<-gpr swap insns.
18649 (bswap<mode>2_store): Likewise.
18650 (bswaphi2_reg): Register only splitter, combine with the splitter.
18651 (bswaphi2 splitter): Likewise.
18652 (bswapsi2_reg): Likewise.
18653 (bswapsi2 splitter): Likewise.
18654 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
18655 the insns into load, store, and register/register insns.
18656 (bswapdi2_ldbrx): Likewise.
18657 (bswapdi2_load): Likewise.
18658 (bswapdi2_store): Likewise.
18659 (bswapdi2_reg): Likewise.
18661 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
18663 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
18664 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
18666 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
18669 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
18671 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18672 special handling for target option conflicts between dform
18673 options (-mpower9-dform, -mpower9-dform-vector,
18674 -mpower9-dform-scalar) and -mno-direct-move.
18676 2017-03-27 Richard Biener <rguenther@suse.de>
18678 PR tree-optimization/80181
18679 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
18681 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18683 * config/arc/predicates.md (move_double_src_operand): Replace the
18684 call to move_double_src_operand with a call to address_operand.
18686 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18688 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
18689 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
18690 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
18692 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18694 * config/arc/predicates.md (long_immediate_loadstore_operand):
18695 Consider scaled addresses cases.
18697 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18699 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
18700 restored when in interrupt.
18701 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
18702 doesn't have delay slot.
18704 2017-03-27 Richard Biener <rguenther@suse.de>
18707 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
18708 inlined thunk clones.
18710 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18713 * asan.c (instrument_derefs): Copy over last operand from
18714 original COMPONENT_REF to the new COMPONENT_REF with
18715 DECL_BIT_FIELD_REPRESENTATIVE.
18716 * ubsan.c (instrument_object_size): Likewise.
18718 2017-03-27 Richard Biener <rguenther@suse.de>
18720 PR tree-optimization/80170
18721 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
18722 sure DR/SCEV didnt fold in constants we do not see when looking
18723 at the reference base alignment.
18725 2017-03-27 Richard Biener <rguenther@suse.de>
18727 PR middle-end/80171
18728 * gimple-fold.c (fold_ctor_reference): Properly guard against
18729 NULL return value from canonicalize_constructor_val.
18731 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
18734 * config/i386/i386.c (ix86_expand_builtin)
18735 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
18736 flags reg setting and flags reg using instructions.
18737 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
18738 clobbering instructions to zero extend op2.
18740 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
18742 * doc/install.texi (Configuration) <--with-aix-soname>:
18743 Update link to AIX ld.
18745 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
18747 PR rtl-optimization/80160
18748 PR rtl-optimization/80159
18749 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
18750 reg_alternate_class into account.
18752 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
18755 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
18756 to consider in curr_insn_transform.
18758 2017-03-24 Jakub Jelinek <jakub@redhat.com>
18760 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
18761 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
18762 and emit_mode_inner.
18764 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18766 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
18767 argument to the overloaded builtin variants. Use the new flag to
18768 deprecate certain builtin variants.
18769 * config/s390/s390-builtin-types.def: Add new builtin types.
18770 * config/s390/s390-builtins.h: Support new flags field for
18771 overloaded builtins.
18772 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
18773 (s390_macro_to_expand): Enable vector float data type.
18774 (s390_cpu_cpp_builtins_internal): Indicate support of the new
18775 builtins by incrementing the __VEC__ version number.
18776 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
18778 (s390_resolve_overloaded_builtin): Emit error messages depending
18779 on the builtin flags.
18780 * config/s390/s390.c (s390_expand_builtin): Support additional
18781 flags argument. Change error message to match the messages
18782 emitted in s390-c.c.
18783 * config/s390/s390.md: New UNSPEC_* constants.
18784 (op_type): Add new instruction types.
18785 * config/s390/vecintrin.h: Add new builtins and test data class
18787 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
18788 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
18789 (VEC_INEXACT, VEC_NOINEXACT): New constants.
18790 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
18791 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
18792 ("vec_mergel<mode>"): V_HW -> VEC_HW.
18794 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
18795 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
18796 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
18797 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
18799 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
18800 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
18801 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
18802 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
18804 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
18805 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
18806 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
18807 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
18808 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
18809 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
18810 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
18812 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
18813 ("vec_scatter_element<V_HW_4:mode>_DI")
18814 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
18815 ("vec_fpint<mode>", "vflls")
18816 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
18817 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
18818 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
18819 ("*vec_cmphe<mode>_cc"): ... these.
18821 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
18822 mode constant instead of magic value.
18824 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18826 * config/s390/s390.c (s390_expand_vec_compare): Support other
18827 vector floating point modes than just V2DF.
18828 (s390_expand_vcond): Likewise.
18829 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
18830 (s390_cannot_change_mode_class): Prevent mode changes between TF
18831 and V1TF in vector registers.
18832 * config/s390/s390.md (DF, SF): New mode attributes.
18833 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
18834 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
18835 SFmode support for VRs.
18836 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
18838 (VFT, VF_HW): New mode iterators.
18839 (vw, sdx): New mode attributes.
18840 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
18841 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
18842 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
18843 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
18844 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
18845 also the new vector floating point modes. Renaming to ...
18847 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
18848 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
18849 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
18850 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
18851 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
18852 ("vec_unordered<mode>"): ... these.
18854 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
18855 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
18856 ("*vec_extendv2df"): New insn definitions.
18858 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18860 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
18861 ("mulditi3_2", "*muldi3_sign"): New patterns.
18862 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
18863 rename the pattern definition.
18865 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18867 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
18869 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
18871 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18873 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
18874 instruction if possible.
18875 * config/s390/vector.md (vec_halfnumelts): New mode
18877 ("*vec_vllezlf<mode>"): New pattern.
18879 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18881 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
18882 ("popcountv4si2", "popcountv2di2"): Rename to ...
18883 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
18884 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
18886 ("popcount<mode>2_vxe"): New pattern.
18888 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18890 * common/config/s390/s390-common.c (processor_flags_table): Add
18892 * config.gcc: Add arch12.
18893 * config/s390/driver-native.c (s390_host_detect_local_cpu):
18894 Default to arch12 for unknown CPU model numbers.
18895 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
18896 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
18897 PROCESSOR_max sanity check.
18898 * config/s390/s390-opts.h (enum processor_type): Add
18900 * config/s390/s390.c (processor_table): Add arch12.
18901 (s390_expand_builtin): Add check for B_VXE flag.
18902 (s390_issue_rate): Add PROCESSOR_ARCH12.
18903 (s390_get_sched_attrmask): Likewise.
18904 (s390_get_unit_mask): Likewise.
18905 (s390_sched_score): Enable z13 scheduling for arch12.
18906 (s390_sched_reorder): Likewise.
18907 (s390_sched_variable_issue): Likewise.
18908 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
18910 (s390_tune_attr): Use z13 scheduling also for arch12.
18911 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
18912 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
18913 (TARGET_VXE_P): New macros.
18914 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
18915 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
18916 * config/s390/s390.opt: Add arch12 as processor_type.
18918 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18920 * config/s390/s390.md
18921 ("fixuns_truncdddi2", "fixuns_trunctddi2")
18922 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
18923 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
18925 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
18926 Rename expanders to ...
18928 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
18929 ("fixuns_truncdddi2_emu"): ... these.
18931 ("fixuns_trunc<mode>si2_emu"): New expander.
18933 ("*fixuns_truncdfdi2_z13"): Rename to ...
18934 ("*fixuns_truncdfdi2_vx"): ... this.
18936 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18938 * config/s390/2964.md: Remove the single element vector compare
18939 instructions which are no longer used.
18940 * config/s390/s390.c (s390_select_ccmode): Remove handling of
18942 (s390_canonicalize_comparison): Remove handling of DFmode
18944 (s390_expand_vec_compare_scalar): Remove function.
18945 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
18946 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
18948 ("*cmp<mode>_ccs"): Add wfcdb instruction.
18950 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18952 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
18954 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
18955 will anyway by matched by mov<mode>_64dfp.
18957 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18959 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
18960 vlef/vstef. Add missing operand to vleif.
18962 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18964 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
18965 pair for all vector types with 64 bit elements.
18966 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
18967 * config/s390/vector.md (V_HW_64): ... here.
18968 (V_128_NOSINGLE): New mode iterator.
18969 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
18970 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
18971 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
18972 ("*vec_load_pairv2di"): Change to ...
18973 ("*vec_load_pair<mode>"): ... this one.
18975 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18977 * config/s390/constraints.md: Add comments.
18978 (jKK): Reject element sizes > 8 bytes.
18979 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
18981 * config/s390/s390.md: Add the s_operand checks formerly in
18982 s390_split_ok_p to various splitters where they are still
18984 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
18985 for 128 bit vectors. Plus two splitters.
18987 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18989 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
18992 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18995 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
18996 error if the boundary argument is not constant.
18998 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19000 PR rtl-optimization/80112
19001 * loop-doloop.c (doloop_condition_get): Don't check condition
19002 if cmp isn't SET with IF_THEN_ELSE src.
19004 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19006 PR tree-optimization/80158
19007 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
19008 replacing a candidate statement, also replace it for the
19009 candidate's alternate interpretation.
19010 (replace_rhs_if_not_dup): Likewise.
19011 (replace_one_candidate): Likewise.
19013 2017-03-24 Richard Biener <rguenther@suse.de>
19015 PR tree-optimization/80167
19016 * graphite-isl-ast-to-gimple.c
19017 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
19019 (translate_isl_ast_to_gimple::get_rename): Likewise.
19021 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
19023 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
19024 handling of certain combinations of target options, including the
19025 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
19026 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
19028 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19031 * config/arm/arm.md (*load_multiple): Add reload_completed to
19032 matching condition.
19034 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19035 Richard Biener <rguenth@suse.de>
19037 PR tree-optimization/79908
19038 PR tree-optimization/80136
19039 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19040 been cast away, gimplify_and_add suffices.
19042 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
19044 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
19046 2017-03-23 Richard Biener <rguenther@suse.de>
19048 PR tree-optimization/80032
19049 * gimplify.c (gimple_push_cleanup): Forced unconditional
19050 cleanups still have to go to the conditional_cleanups
19053 2017-03-22 Jakub Jelinek <jakub@redhat.com>
19055 PR tree-optimization/80072
19056 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
19058 (next_operand_entry_id): Change type to unsigned int.
19059 (sort_by_operand_rank): Make sure to return the right return value
19060 even if unsigned fields are bigger than INT_MAX.
19061 (struct oecount): Change cnt and id type to unsigned int.
19062 (oecount_hasher::equal): Formatting fix.
19063 (oecount_cmp): Make sure to return the right return value
19064 even if unsigned fields are bigger than INT_MAX.
19065 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
19068 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
19069 TREE_READONLY on result if writing it more than once.
19072 * doc/invoke.texi (-fsanitize=thread): Document that with
19073 -fnon-call-exceptions atomics are not able to throw
19077 * tsan.c: Include tree-eh.h.
19078 (instrument_builtin_call): Call maybe_clean_eh_stmt or
19079 maybe_clean_or_replace_eh_stmt where needed.
19080 (instrument_memory_accesses): Add cfg_changed argument.
19081 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
19082 if it returned true.
19083 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
19085 PR rtl-optimization/63191
19086 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
19087 wrapper function, moved the whole old content into ...
19088 (ix86_delegitimize_address_1): ... this. New inline function.
19089 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
19090 true as last argument instead of ix86_delegitimize_address.
19092 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19094 * config/aarch64/aarch64.c (generic_branch_cost): Copy
19095 cortexa57_branch_cost.
19097 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19099 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
19101 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19104 * doc/md.texi (Constraints): Document wA constraint.
19105 * config/rs6000/constraints.md (wA): New.
19106 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
19107 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
19108 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
19109 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
19111 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
19114 * gimplify.c (is_oacc_declared): New function.
19115 (oacc_default_clause): Use it to set default flags for acc declared
19116 variables inside parallel regions.
19117 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
19118 declared variables.
19119 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
19120 declare attribute to any decl as necessary.
19122 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
19125 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
19126 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
19127 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
19128 (arm_arch_lpae): This.
19129 * config/arm/arm.c (arm_arch7ve): Rename into ...
19130 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
19131 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
19134 2017-03-22 Martin Liska <mliska@suse.cz>
19137 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
19138 error message instead of an ICE.
19140 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19142 * doc/extend.texi (6.11 Additional Floating Types): Revise.
19144 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
19146 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19148 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19151 2017-03-21 Martin Sebor <msebor@redhat.com>
19153 * doc/extend.texi: Use "cannot" instead of "can't."
19154 * doc/hostconfig.texi: Same.
19155 * doc/install.texi: Same.
19156 * doc/invoke.texi: Same.
19157 * doc/loop.texi: Same.
19158 * doc/md.texi: Same.
19159 * doc/objc.texi: Same.
19160 * doc/rtl.texi: Same.
19161 * doc/tm.texi: Same.
19162 * doc/tm.texi.in: Same.
19163 * doc/trouble.texi: Same.
19165 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
19168 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
19169 (collect_checksum_attributes): Set it.
19170 (die_checksum_ordered): Use it.
19172 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19174 PR tree-optimization/79908
19175 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
19176 change: For a VA_ARG whose LHS has been cast away, use
19177 force_gimple_operand to construct the side effects.
19179 2017-03-21 David Malcolm <dmalcolm@redhat.com>
19181 PR translation/80001
19182 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
19183 more amenable to translation.
19184 (oacc_loop_auto_partitions): Likewise.
19186 2017-03-21 Marek Polacek <polacek@redhat.com>
19187 Martin Sebor <msebor@redhat.com>
19189 PR tree-optimization/80109
19190 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
19191 on INTEGRAL_TYPE_P.
19193 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19194 Segher Boessenkool <segher@kernel.crashing.org>
19197 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
19198 check reg_used_between_p between insn and one of succ or succ2
19199 depending on if succ is artificial insn not inserted into insn
19202 2017-03-21 Martin Liska <mliska@suse.cz>
19204 PR gcov-profile/80081
19205 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
19206 * doc/gcc.texi: Include gcov-dump stuff.
19207 * doc/gcov-dump.texi: New file.
19209 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
19211 PR rtl-optimization/79150
19212 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
19213 conditional jump, if the jump is the last insn of the loop.
19215 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19216 Richard Biener <rguenth@suse.de>
19218 PR tree-optimization/79908
19219 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19220 been cast away, use force_gimple_operand to construct the side
19223 2017-03-21 Martin Liska <mliska@suse.cz>
19225 PR libfortran/79956
19226 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
19229 2017-03-21 Brad Spengler <spender@grsecurity.net>
19232 * plugin.c (htab_hash_plugin): New function.
19233 (add_new_plugin): Use it and adjust.
19234 (parse_plugin_arg_opt): Adjust.
19235 (init_one_plugin): Likewise.
19237 2017-03-21 Richard Biener <rguenther@suse.de>
19239 PR tree-optimization/80032
19240 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
19241 if set force the cleanup to happen unconditionally.
19242 (gimplify_target_expr): Push inserted clobbers with force_uncond
19243 to avoid them being removed by control-dependent DCE.
19245 2017-03-21 Richard Biener <rguenther@suse.de>
19247 PR tree-optimization/80122
19248 * tree-inline.c (copy_bb): Do not expans va-arg packs or
19249 va_arg_pack_len when the inlined call stmt requires pack
19251 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
19253 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19256 * tsan.c (instrument_builtin_call): If the memory model argument
19257 is not a constant, assume it is valid.
19260 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
19263 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19265 PR rtl-optimization/79910
19266 * combine.c (can_combine_p): Do not allow combining an I0 or I1
19267 if its dest is used by an insn before I2 (other than the combined
19268 insns themselves, which are properly handled already).
19270 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19273 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19275 * combine.c (record_used_regs): New static function.
19276 (try_combine): Handle situations where there is an additional
19277 instruction between I2 and I3 which needs to have a LOG_LINK
19281 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19283 * combine.c (try_combine): Delete redundant i1 test. Call
19284 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19286 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19289 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
19290 alternatives 13/14.
19292 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19294 PR tree-optimization/80054
19295 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
19296 the optimization if a PHI or any of its arguments is not dominated
19297 by the candidate's basis. Use gphi* rather than gimple* as
19299 (replace_profitable_candidates): Clean up a gimple* variable that
19300 should be a gphi* variable.
19302 2017-03-20 Martin Sebor <msebor@redhat.com>
19305 * doc/extend.texi (attribute constructor): Document present limitation.
19307 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
19310 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
19311 __POWER9_VECTOR__ #ifdef control, change template definition to
19312 use Power9-specific built-in function.
19313 (vec_any_eq): Likewise.
19314 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
19315 to control outcomes from this test.
19316 (vector_ae_<mode>p): For VEC_F modes, likewise.
19318 2017-03-20 Ian Lance Taylor <iant@google.com>
19320 * config/i386/i386.c (ix86_function_regparm): Save an extra
19321 register for -fsplit-stack with DECL_STATIC_CHAIN.
19323 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19326 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
19327 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
19329 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19331 * config/riscv/riscv.c (riscv_print_operand): Use "fence
19333 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
19336 2017-03-20 Marek Polacek <polacek@redhat.com>
19339 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
19341 2017-03-20 Richard Biener <rguenther@suse.de>
19343 PR tree-optimization/80113
19344 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
19345 allocate extra SSA name for PHI def.
19346 (add_close_phis_to_outer_loops): Likewise.
19347 (add_close_phis_to_merge_points): Likewise.
19348 (copy_loop_close_phi_args): Likewise.
19349 (copy_cond_phi_nodes): Likewise.
19351 2017-03-20 Martin Liska <mliska@suse.cz>
19353 PR middle-end/79753
19354 * tree-chkp.c (chkp_build_returned_bound): Do not build
19355 returned bounds for a LHS that's not a BOUNDED_P type.
19357 2017-03-20 Martin Liska <mliska@suse.cz>
19361 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
19362 COMPLEX_CST and VECTOR_CST.
19364 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19367 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
19368 target operand. A new splitter adds the clobber statement in case
19369 the target operand is dead anyway.
19371 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
19373 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
19374 to age-old versions of binutils and glibc.
19376 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
19378 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
19380 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19382 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
19384 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19386 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
19387 requirement for binutils 2.13.
19389 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19391 * combine.c (try_combine): Delete redundant i1 test. Call
19392 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19394 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
19396 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
19397 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
19399 <riscv64-*-elf>: Re-arrange section
19400 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
19401 <riscv32-*-linux>: Likewise.
19402 <riscv64-*-elf>: Likewise
19403 <riscv64-*-linux>: Likewise.
19405 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
19408 * aarch64.opt(verbose-cost-dump): Fix typo.
19410 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
19413 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
19414 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
19416 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19418 * reload.c (find_reloads): When reloading a nonoffsettable address,
19419 use RELOAD_OTHER for it and its address reloads.
19421 PR rtl-optimization/79910
19422 * combine.c (record_used_regs): New static function.
19423 (try_combine): Handle situations where there is an additional
19424 instruction between I2 and I3 which needs to have a LOG_LINK
19427 2017-03-17 Jeff Law <law@redhat.com>
19429 PR tree-optimization/71437
19430 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
19431 conditional in the hash table first.
19432 (vrp_dom_walker::before_dom_children): Extract condition from
19433 ASSERT_EXPR. Record condition, its inverion and any implied
19434 conditions as well.
19436 2017-03-17 Marek Polacek <polacek@redhat.com>
19437 Markus Trippelsdorf <markus@trippelsdorf.de>
19439 PR tree-optimization/80079
19440 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
19443 2017-03-17 Richard Biener <rguenther@suse.de>
19445 PR middle-end/80075
19446 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
19447 Properly verify the LHS before the RHS possibly claims to be
19449 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
19452 2017-03-17 Martin Jambor <mjambor@suse.cz>
19454 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
19455 (List of -O2 options): Likewise.
19456 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
19459 2017-03-17 Tom de Vries <tom@codesourcery.com>
19461 * gcov-dump.c (print_usage): Print bug_report_url.
19463 2017-03-17 Richard Biener <rguenther@suse.de>
19465 PR middle-end/80050
19466 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
19467 (parser::peek): Likewise.
19469 2017-03-17 Richard Biener <rguenther@suse.de>
19471 PR tree-optimization/80048
19472 * sese.c (free_sese_info): Properly release rename_map and
19473 copied_bb_map elements.
19475 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
19477 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
19478 Add linked-list forward and backlinks. Insert on
19479 construction, remove on destruction.
19480 (class pass_store_merging): Add m_stores_head field.
19481 (pass_store_merging::terminate_and_process_all_chains):
19482 Iterate over m_stores_head list.
19483 (pass_store_merging::terminate_all_aliasing_chains):
19485 (pass_store_merging::execute): Check for debug stmts first.
19486 Push new chains onto the m_stores_head stack.
19488 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
19491 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
19492 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
19493 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
19495 2017-03-16 Jeff Law <law@redhat.com>
19497 PR tree-optimization/71437
19498 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
19499 member function. Implementation moved into after_dom_children
19500 member function and into the threader's thread_outgoing_edges
19502 (dom_opt_dom_walker::after_dom_children): Simplify by moving
19503 some code into new thread_outgoing_edges.
19504 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
19505 definition. Simplify marker handling (do it here). Assume we always
19506 have the available expression and the const/copies tables.
19507 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
19509 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
19510 * tree-vrp.c (equiv_stack): No longer file scoped.
19511 (vrp_dom_walker): New class.
19512 (vrp_dom_walker::before_dom_children): New member function.
19513 (vrp_dom_walker::after_dom_children): Likewise.
19514 (identify_jump_threads): Setup domwalker. Use it rather than
19515 walking edges in a random order by hand. Simplify setup/finalization.
19516 (finalize_jump_threads): Remove.
19517 (vrp_finalize): Do not call identify_jump_threads here.
19518 (execute_vrp): Do it here instead and call thread_through_all_blocks
19521 PR tree-optimization/71437
19522 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
19524 (simplify_stmt_for_jump_threading): Add basic_block argument. All
19526 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
19527 (dom_opt_dom_walker::thread_across_edge): Remove
19528 handle_dominating_asserts argument. All callers changed.
19529 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
19530 changes. Remove calls to lhs_of_dominating_assert. Other
19531 uses of handle_dominating_asserts turn into unconditional code
19532 (simplify_control_stmt_condition_1): Likewise.
19533 (simplify_control_stmt_condition): Likewise.
19534 (thread_through_normal_block, thread_across_edge): Likewise.
19535 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
19536 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
19537 object if it is not an SSA_NAME.
19538 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
19539 before calling into the VRP specific simplifiers.
19540 (identify_jump_threads): Remove handle_dominating_asserts
19543 2017-03-16 Jakub Jelinek <jakub@redhat.com>
19546 * tree-diagnostic.c (default_tree_printer): No longer static.
19547 * tree-diagnostic.h (default_tree_printer): New prototype.
19549 2017-03-16 Tamar Christina <tamar.christina@arm.com>
19551 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
19552 Change ins into fmov.
19554 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19556 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
19557 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
19558 Use h_con constraint for operand 1.
19559 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
19560 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
19562 2017-03-15 Jeff Law <law@redhat.com>
19564 PR tree-optimization/71437
19565 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
19566 (record_temporary_equivalences): Use it.
19568 PR tree-optimization/71437
19569 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
19570 tree-ssa-scopedtables.
19571 (lookup_avail_expr, build_and_record_new_cond): Likewise.
19572 (record_conditions, record_cond, vuse_eq): Likewise.
19573 (record_edge_info): Adjust to API tweak of record_conditions.
19574 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
19575 (record_temporary_equivalences, optimize_stmt): Likewise.
19576 (eliminate_redundant_computations): Likewise.
19577 (record_equivalences_from_stmt): Likewise.
19578 * tree-ssa-scopedtables.c: Include options.h and params.h.
19579 (vuse_eq): New function, moved from tree-ssa-dom.c
19580 (build_and_record_new_cond): Likewise.
19581 (record_conditions): Likewise. Accept vector of conditions rather
19582 than edge_equivalence structure for first argument.
19583 for the first argument.
19584 (avail_exprs_stack::lookup_avail_expr): New member function, moved
19585 from tree-ssa-dom.c.
19586 (avail_exprs_stack::record_cond): Likewise.
19587 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
19588 from tree-ssa-dom.c.
19589 (avail_exprs_stack): Add new member functions lookup_avail_expr
19591 (record_conditions): Declare.
19593 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
19596 * lra-constraints.c (process_alt_operands): Increase reject for
19597 reloading an input/output operand.
19599 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
19602 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
19603 insns to convert from signed/unsigned char/short to IEEE 128-bit
19605 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
19607 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
19610 * config/i386/i386.c (ix86_vector_duplicate_value): Create
19611 subreg of inner mode for values already in registers.
19613 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
19615 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
19616 iteration reg is used after the loop.
19618 2017-03-14 Martin Sebor <msebor@redhat.com>
19620 PR tree-optimization/79800
19621 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
19622 precision in negative-positive range.
19623 (format_floating): Call non-const overload with adjusted precision.
19625 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19628 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
19631 2017-03-14 Martin Sebor <msebor@redhat.com>
19633 PR middle-end/80020
19634 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
19635 * builtins.def (aligned_alloc): Use it.
19638 * Makefile.in (GTFILES): Add calls.c.
19639 * calls.c: Include "gt-calls.h".
19641 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
19643 PR rtl-optimization/79728
19644 * regs.h (struct target_regs): New field
19645 x_contains_allocatable_regs_of_mode.
19646 (contains_allocatable_regs_of_mode): New macro.
19647 * reginfo.c (init_reg_sets_1): Initialize it, and change
19648 contains_reg_of_mode so it includes global regs as well.
19649 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
19650 rather than contains_regs_of_mode.
19652 2017-03-14 Martin Liska <mliska@suse.cz>
19654 * doc/invoke.texi: Document options that can't be combined with
19655 -fcheck-pointer-bounds.
19657 2017-03-14 Martin Liska <mliska@suse.cz>
19659 PR middle-end/79831
19660 * doc/invoke.texi (-Wchkp): Document the option.
19662 2017-03-14 Martin Liska <mliska@suse.cz>
19664 * Makefile.in: Install gcov-dump.
19666 2017-03-14 Martin Liska <mliska@suse.cz>
19668 * multiple_target.c (expand_target_clones): Bail out for
19669 an invalid attribute.
19671 2017-03-14 Richard Biener <rguenther@suse.de>
19673 * alias.c (struct alias_set_entry): Pack properly.
19674 * cfgloop.h (struct loop): Likewise.
19675 * cse.c (struct set): Likewise.
19676 * ipa-utils.c (struct searchc_env): Likewise.
19677 * loop-invariant.c (struct invariant): Likewise.
19678 * lra-remat.c (struct cand): Likewise.
19679 * recog.c (struct change_t): Likewise.
19680 * rtl.h (struct address_info): Likewise.
19681 * symbol-summary.h (function_summary): Likewise.
19682 * tree-loop-distribution.c (struct partition): Likewise.
19683 * tree-object-size.c (struct object_size_info): Likewise.
19684 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
19685 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
19686 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
19687 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
19688 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
19689 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
19690 (struct _stmt_vec_info): Likewise.
19692 2017-03-14 Martin Liska <mliska@suse.cz>
19695 * multiple_target.c (create_dispatcher_calls): Check that
19696 a target can create a function dispatcher.
19698 2017-03-14 Martin Liska <mliska@suse.cz>
19701 * multiple_target.c (expand_target_clones): Drop local.local
19702 flag for default implementation.
19704 2017-03-14 Richard Biener <rguenther@suse.de>
19706 PR tree-optimization/80030
19707 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
19709 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
19711 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
19712 gcc_fallthrough() instead of __attribute__((fallthrough));
19714 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19716 * doc/gcc.texi: Remove "up" link to (DIR).
19717 * doc/gccint.texi: Ditto.
19719 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19721 * doc/install.texi (Specific) <avr>: Remove reference to
19724 2017-03-13 Jeff Law <law@redhat.com>
19726 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
19727 attribute rather than comments.
19729 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
19730 match_scratch operand is highest.
19732 2017-03-13 Martin Liska <mliska@suse.cz>
19734 PR middle-end/78339
19735 * ipa-pure-const.c (warn_function_noreturn): If the declarations
19736 is a CHKP clone, use original declaration.
19738 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19740 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
19741 (arc_conditional_register_usage): Use a different allocation order
19742 when optimizing for size.
19743 * common/config/arc/arc-common.c (arc_option_optimization_table):
19744 Section anchors default on when optimizing for size.
19746 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19748 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
19750 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19752 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
19753 * config/arc/arc.md (cpu_facility): Add cd variant.
19754 (*movqi_insn): Add code density variant.
19755 (*movhi_insn): Likewise.
19756 (*movqi_insn): Likewise.
19757 (*addsi3_mixed): Likewise.
19758 (subsi3_insn): Likewise.
19760 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19762 * config/arc/arc.md (movsi_cond_exec): Update constraint.
19764 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19766 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
19767 expressions with MINUS and UNARY ops.
19769 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19772 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
19774 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
19775 between vec_select and vector argument.
19776 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
19777 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
19778 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
19779 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
19780 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
19781 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
19783 2017-03-13 Richard Biener <rguenther@suse.de>
19786 * params.def (vect-max-peeling-for-alignment): Fix typo.
19788 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19790 * doc/install.texi (Specific) <mips-*-*>: Remove description of
19791 issue that only occurred with binutils below 2.18.
19793 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19795 * doc/install.texi (Specific) <cris-axis-elf>: No longer
19796 refer to binutils 2.11/2.12 minimum.
19798 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19800 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
19801 ftp.kernel.org and simplify binutils requirement.
19803 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
19805 * doc/invoke.texi (Warning Options): Fix spelling of link-time
19807 (Optimize Options): Ditto. Also remove redundancy.
19809 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19811 PR translation/79848
19812 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
19814 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
19815 to G_ to avoid double translation.
19817 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19819 PR translation/79923
19820 * auto-profile.c (get_combined_location): Convert leading
19821 character of diagnostics to lower case and remove trailing period.
19822 (read_profile): Likewise for various diagnostics.
19823 * config/arm/arm.c (arm_option_override): Remove trailing period
19824 from various diagnostics.
19825 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
19826 (msp430_expand_delay_cycles): Likewise.
19828 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19831 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
19832 full command-line argument, rather than just "str".
19833 (aarch64_validate_march): Likewise.
19834 (aarch64_validate_mtune): Likewise.
19836 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
19838 PR rtl-optimization/78911
19839 * lra-assigns.c (must_not_spill_p): New function.
19840 (spill_for): Use it.
19842 2017-03-10 Jakub Jelinek <jakub@redhat.com>
19844 PR tree-optimization/79981
19845 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
19846 ATOMIC_COMPARE_EXCHANGE ifn result.
19847 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
19848 IFN_ATOMIC_COMPARE_EXCHANGE.
19850 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19853 * opts.c (parse_sanitizer_options): Add missing question mark to
19854 "did you mean" message.
19856 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19858 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
19860 (VMULEUH_UNS): Likewise.
19861 (VMULOUB_UNS): Likewise.
19862 (VMULOUH_UNS): Likewise.
19863 * config/rs6000/rs6000.c (builtin_function_type): Remove
19864 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
19866 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19869 * read-rtl-function.c (function_reader::read_rtx_operand): Update
19870 x with result of extra_parsing_for_operand_code_0.
19871 (function_reader::extra_parsing_for_operand_code_0): Convert
19872 return type from void to rtx, returning x. When reading
19873 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
19874 larger size containing struct block_symbol.
19876 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
19878 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
19879 -mfloat128-hardware without -m64.
19881 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
19884 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
19885 entries to the case statement that marks unsigned arguments to
19886 overloaded functions.
19888 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19890 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
19891 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
19893 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
19896 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
19897 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
19899 2017-03-10 Martin Liska <mliska@suse.cz>
19903 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
19904 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
19907 2017-03-10 Olivier Hainque <hainque@adacore.com>
19909 * tree-switch-conversion (array_value_type): Start by resetting
19910 candidate type to it's main variant.
19912 2017-03-10 Jakub Jelinek <jakub@redhat.com>
19914 PR rtl-optimization/79909
19915 * combine.c (try_combine): Use simplify_replace_rtx on individual
19916 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
19917 of the whole CALL_INSN_FUNCTION_USAGE.
19919 PR tree-optimization/79972
19920 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
19921 get_range_info on SSA_NAMEs. Formatting fixes.
19923 2017-03-10 Richard Biener <rguenther@suse.de>
19924 Jakub Jelinek <jakub@redhat.com>
19926 PR tree-optimization/77975
19927 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
19928 edge to be constant.
19929 (get_val_for): For constant x return it. Formatting fix.
19930 (loop_niter_by_eval): Avoid pointless looping if the next iteration
19931 would use the same bases as the current one.
19933 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19935 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
19936 instead of vec_select for V1TImode.
19937 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
19939 (VSX_LE_128): Add V1TI to this mode iterator.
19940 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
19941 (*vsx_le_perm_store_<mode>): Likewise.
19942 (pre-reload splitter for VSX stores): Likewise.
19943 (post-reload splitter for VSX stores): Likewise.
19944 (*vsx_xxpermdi2_le_<mode>): Likewise.
19945 (*vsx_lxvd2x2_le_<mode>): Likewise.
19946 (*vsx_stxvd2x2_le_<mode>): Likewise.
19948 2017-03-09 Michael Eager <eager@eagercon.com>
19950 Correct failures with --enable-checking=yes,rtl.
19952 * config/microblaze/microblaze.c (microblaze_expand_shift):
19953 Replace GET_CODE test with CONST_INT_P and INTVAL test with
19954 test for const0_rtx.
19955 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
19956 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
19958 2017-03-09 Richard Biener <rguenther@suse.de>
19960 PR tree-optimization/79977
19961 * graphite-scop-detection.c (scop_detection::merge_sese):
19962 Handle the case of extra exits to blocks dominating the entry.
19964 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
19966 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
19969 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
19971 PR rtl-optimization/79949
19972 * lra-constraints.c (process_alt_operands): Check memory when
19973 trying to predict a cycle. Print about the overall increase.
19975 2017-03-09 Richard Biener <rguenther@suse.de>
19977 PR middle-end/79971
19978 * gimple-expr.c (useless_type_conversion_p): Preserve
19979 TYPE_SATURATING for fixed-point types.
19981 2017-03-09 Richard Biener <rguenther@suse.de>
19984 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
19985 alignment of BLKmode params.
19987 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19990 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
19991 (VALL_NO_V2Q): Likewise.
19993 * config/aarch64/aarch64-simd.md
19994 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
19996 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
19997 VALL_NO_V2Q mode iterator.
19998 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
20000 2017-03-09 Martin Liska <mliska@suse.cz>
20002 PR tree-optimization/79631
20003 * tree-chkp-opt.c (chkp_is_constant_addr): Call
20004 tree_int_cst_sign_bit just for INTEGER constants.
20006 2017-03-09 Martin Liska <mliska@suse.cz>
20010 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
20013 2017-03-09 Marek Polacek <polacek@redhat.com>
20016 * tree.c (inchash::add_expr): Handle TREE_VEC.
20018 2017-03-09 Martin Liska <mliska@suse.cz>
20021 (chkp_narrow_size_and_offset): New function.
20022 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
20023 (void chkp_parse_bit_field_ref): New function.
20024 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
20025 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
20027 2017-03-09 Martin Liska <mliska@suse.cz>
20030 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
20031 (chkp_find_bounds_1): Remove gcc_unreachable.
20033 2017-03-09 Jakub Jelinek <jakub@redhat.com>
20036 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
20037 BUILT_IN_SYNC*, determine the access type from the size suffix and
20038 always build a MEM_REF with that type. Handle forgotten
20039 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
20042 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
20043 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
20044 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
20045 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
20046 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
20047 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
20048 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
20049 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
20050 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
20051 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
20052 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
20053 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
20054 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
20055 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
20056 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
20057 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
20058 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
20059 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
20060 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
20061 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
20062 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
20063 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
20064 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
20065 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
20066 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
20067 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
20068 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
20069 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
20070 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
20071 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
20072 definitions outside of __OPTIMIZE__ guarded section.
20075 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
20076 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
20077 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
20078 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
20081 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20083 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
20084 ("vfenez<mode>"): Add missing constraints.
20086 2017-03-08 Martin Sebor <msebor@redhat.com>
20089 * config/nds32/nds32.c (nds32_option_override):
20090 Fix misspelled diagnostic.
20092 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20095 * gimplify.c (gimplify_omp_for): Replace index var in outer
20096 taskloop statement with an artificial variable and add
20097 OMP_CLAUSE_PRIVATE clause for it.
20099 2017-03-08 Richard Biener <rguenther@suse.de>
20101 PR tree-optimization/79955
20102 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
20103 for accesses that are completely outside of the variable.
20105 2017-03-08 Andrew Haley <aph@redhat.com>
20107 PR tree-optimization/79943
20108 * tree-ssa-loop-split.c (compute_new_first_bound): When
20109 calculating the new upper bound, (END-BEG) should be added, not
20112 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20114 * config/avr/avr.md (setmemhi): Make sure match_dup
20115 operand number comes before match_scratch.
20117 2017-03-08 Richard Biener <rguenther@suse.de>
20119 PR tree-optimization/79920
20120 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
20121 with ncopies == 1 to ...
20122 (vect_transform_slp_perm_load): ... here. Properly compute
20123 all element loads by iterating VF times over the group. Do
20124 not handle ncopies (computed in a broken way) in
20125 vect_create_mask_and_perm.
20127 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20130 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
20131 is a uniform vector, use uniform_vector_p return value instead of
20132 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
20134 2017-03-07 Marek Polacek <polacek@redhat.com>
20136 PR middle-end/79809
20137 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
20138 (alloca_call_type): Likewise.
20140 2017-03-07 Martin Liska <mliska@suse.cz>
20142 * gcov.c (process_args): Put comment to correct location.
20144 2017-03-07 Martin Liska <mliska@suse.cz>
20146 PR middle-end/68270
20147 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
20148 Use array_at_struct_end_p instead of DECL_CHAIN (field).
20149 (chkp_narrow_bounds_for_field): Likewise.
20150 (chkp_parse_array_and_component_ref): Pass one more argument to
20153 2017-03-07 Richard Biener <rguenther@suse.de>
20155 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
20158 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
20160 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
20161 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
20163 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20166 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
20167 to end of description.
20168 (PARAM_MAX_STORES_TO_MERGE): Likewise.
20170 2017-03-07 Jakub Jelinek <jakub@redhat.com>
20172 PR rtl-optimization/79901
20173 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
20175 (*avx512f_<code><mode>3<mask_name>): ... this.
20176 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
20177 iterator instead of VI8_AVX2_AVX512BW.
20179 PR rtl-optimization/79901
20180 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
20181 min/max expander, expand it using expand_vec_cond_expr.
20184 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
20187 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20190 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
20191 to void * for PCH reasons.
20192 * dwarf2out.c (output_loc_operands, output_die): Cast
20193 v.val_vec.array to unsigned char *.
20195 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
20198 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
20201 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
20203 PR rtl-optimization/79571
20204 * lra-constraints.c (process_alt_operands): Calculate static
20205 reject and subtract it from overall when only addresses will be
20208 2017-03-06 Julia Koval <julia.koval@intel.com>
20211 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
20212 incoming stack boundary to 128 for 64-bit targets.
20214 2017-03-06 Richard Biener <rguenther@suse.de>
20216 PR tree-optimization/79894
20217 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
20218 to NULL after folding it.
20220 2017-03-06 Richard Biener <rguenther@suse.de>
20222 PR tree-optimization/79824
20223 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
20224 check disabling peeling for gaps.
20226 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
20228 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20229 attributes): Document gettimeofday.
20231 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20233 * config/s390/s390.c (s390_option_override_internal): Set
20234 PARAM_MIN_VECT_LOOP_BOUND
20236 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20238 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
20239 * config/s390/s390.md: Likewise.
20241 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20244 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
20245 (<avx2_avx512>_perm<mode>): Rename to ...
20246 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
20248 (<avx512>_perm<mode>_mask): Rename to ...
20249 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
20251 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
20252 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
20253 instead of VI8F_256_512.
20254 (avx512f_perm<mode>): New define_expand.
20255 (avx512f_perm<mode>_mask): Likewise.
20256 (avx512f_perm<mode>_1<mask_name>): New define_insn.
20257 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
20259 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20261 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
20262 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
20264 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
20266 2017-03-06 Martin Liska <mliska@suse.cz>
20269 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
20270 when having a SSA NAME w/o VAR_DECL assigned to it.
20272 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20274 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
20275 msa_dpsub_<su>_d): Fix MODE for vec_select.
20277 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20279 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
20281 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
20283 2017-03-06 Richard Biener <rguenther@suse.de>
20285 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
20286 * plugin.c (register_plugin_info): Likewise.
20287 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
20289 2017-03-05 Jakub Jelinek <jakub@redhat.com>
20291 * config/i386/sse.md (sse_storehps, sse_storelps,
20292 avx_<castmode><avxsizesuffix>_<castmode>,
20293 avx512f_<castmode><avxsizesuffix>_<castmode>,
20294 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
20295 in condition that at least one operand is not a MEM.
20297 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20299 PR middle-end/79805
20300 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
20301 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
20303 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
20304 gimple_call_nothrow_p flag based on whether original builtin can throw.
20305 If it can, emit following stmts on the fallthrough edge.
20306 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
20307 don't create new bb if inserting just debug stmts on the edge, try to
20308 insert them on the fallthru bb or just reset debug stmts.
20310 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
20313 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
20314 restore recog_data (including the operand rtxes inside it) around
20315 the call to get_insn_template.
20317 2017-03-03 Martin Sebor <msebor@redhat.com>
20319 PR tree-optimization/79699
20320 * context.c (context::~context): Free MPFR caches to avoid
20321 a memory leak on program exit.
20323 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20325 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
20326 Use wide_int::ulow () instead of .elt (0).
20328 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20330 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
20331 (*pushxf): Limit oF constraint to 32bit targets and add oC
20332 constraint for 64bit targets.
20333 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
20334 (*pushdf): Change rmF constraint to rmC.
20336 2017-03-03 Martin Liska <mliska@suse.cz>
20338 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
20339 Remove unused variable.
20341 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20344 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
20345 is a memory operand, increase num_memory.
20346 (ix86_expand_args_builtin): Likewise.
20348 2017-03-03 Jan Hubicka <jh@suse.cz>
20351 * ipa-devirt.c (maybe_record_node): Properly handle
20352 __cxa_pure_virtual visibility.
20354 2017-03-03 Martin Liska <mliska@suse.cz>
20356 PR tree-optimization/79803
20357 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
20359 (pass_loop_prefetch::execute): Disabled optimization if an
20360 assumption about L1 cache size is not met.
20362 2017-03-03 Martin Liska <mliska@suse.cz>
20364 PR rtl-optimization/79574
20365 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
20366 (hash_scan_set): Likewise.
20367 (dump_hash_table): Likewise.
20368 (hoist_code): Likewise.
20370 2017-03-03 Richard Biener <rguenther@suse.de>
20372 * fixed-value.c (fixed_from_string): Restore use of elt (1)
20373 in place of uhigh ().
20374 (fixed_convert_from_real): Likewise.
20376 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20379 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
20381 2017-03-03 Richard Biener <rguenther@suse.de>
20383 PR middle-end/79818
20384 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
20385 TYPE_OVERFLOW_UNDEFINED check.
20387 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20389 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
20391 (vector_ae_<mode>_p): Likewise.
20392 (vector_nez_<mode>_p): Likewise.
20393 (vector_ne_v2di_p): Likewise.
20394 (vector_ae_v2di_p): Likewise.
20395 (vector_ne_<mode>_p): Likewise.
20396 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
20398 (vsx_tsqrt<mode>2_fe): Likewise.
20400 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
20403 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
20405 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20407 PR rtl-optimization/79780
20408 * cprop.c (one_cprop_pass): When second and further conditional trap
20409 in a single basic block is turned into an unconditional trap, turn it
20410 into a deleted note to avoid RTL verification failures.
20412 2017-03-02 Richard Biener <rguenther@suse.de>
20414 * fold-const.c (const_binop): Use ulow () instead of elt (0).
20416 2017-03-02 Richard Biener <rguenther@suse.de>
20418 PR tree-optimization/79345
20420 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
20421 param and abort the walk, returning -1 if it is hit.
20422 (walk_aliased_vdefs): Take a limit param and pass it on.
20423 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
20424 defaulting to 0 and return a signed int.
20425 * tree-ssa-uninit.c (struct check_defs_data): New struct.
20426 (check_defs): New helper.
20427 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
20428 about uninitialized memory.
20429 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
20430 bogus uninitialized warning.
20431 (fixed_convert_from_real): Likewise.
20433 2017-03-02 Bin Cheng <bin.cheng@arm.com>
20435 PR tree-optimization/66768
20436 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
20437 iv_use if base object can't be determined.
20439 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20441 PR tree-optimization/79345
20442 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
20443 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
20444 (get_pattern_stats): Initialize it.
20445 * genemit.c (gen_expand): Verify match_scratch numbers come after
20446 match_operand/match_dup numbers.
20447 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
20448 match_scratch numbers.
20449 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
20451 * config/s390/s390.md (trunctdsd2): Likewise.
20453 2017-03-02 Richard Biener <rguenther@suse.de>
20455 * wide-int.h (wide_int_storage::operator=): Implement in terms
20458 2017-03-02 Richard Biener <rguenther@suse.de>
20460 PR tree-optimization/79777
20461 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
20462 the to insert expression to sth existing.
20464 2017-03-01 Martin Sebor <msebor@redhat.com>
20466 PR middle-end/79692
20467 * gimple-ssa-sprintf.c
20468 (directive::known_width_and_precision): New function.
20469 (format_integer): Use it.
20470 (get_mpfr_format_length): Consider the full range of precision
20471 when computing %g output with the # flag. Set the likely byte
20472 count to 3 rather than 1 when precision is indeterminate.
20473 (format_floating): Correct the lower bound of precision.
20475 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20477 * doc/invoke.texi: Document default code model for 64-bit Linux.
20479 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20482 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
20483 udiv rather than div since input pattern is unsigned.
20485 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
20487 * config/i386/i386.c (print_reg): Warn for values of
20488 unsupported size in integer register.
20490 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
20493 * config/rs6000/predicates.md (current_file_function_operand): Do
20494 not allow self calls to be local if the function is replaceable.
20496 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
20499 * config/rs6000/altivec.h (vec_ctz and others): Change the
20500 preprocessor macro that controls conditional compilation from
20501 _ARCH_PWR9 to __POWER9_VECTOR__.
20502 (vec_all_ne): Change parameterization of __altivec_scalar_pred
20503 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
20504 control (instead of _ARCH_PWR9 control) so that template
20505 definition uses power9-specific function.
20506 (vec_any_eq): Likewise.
20507 (vec_all_ne): Change macro definition to use a power9-specific
20508 expansion under #ifdef __POWER9_VECTOR__ control (instead of
20509 _ARCH_PWR9 control).
20510 (vec_any_eq) Likewise.
20511 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
20512 expansion for CMPNEF to remove support for xvcmpnesp instruction.
20513 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
20514 support for xvcmpnedp instruction.
20515 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
20516 macro expansion so that Power9 implementation of vec_all_ne does
20517 not use the AltiVec predicate framework.
20518 (VCMPNEH_P): Likewise.
20519 (VCMPNEW_P): Likewise.
20520 (VCMPNED_P): Likewise.
20521 (VCMPNEFP_P): Likewise.
20522 (VCMPNEDP_P): Likewise.
20523 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
20524 implementation of vec_any_eq to not use AltiVec predicate
20526 (VCMPAEH_P): Likewise.
20527 (VCMPAEW_P): Likewise.
20528 (VCMPAED_P): Likewise.
20529 (VCMPAEFP_P): Likewise.
20530 (VCMPAEDP_P): Likewise.
20531 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
20532 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
20533 not use the AltiVec predicate framework.
20534 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
20535 of vec_any_eq to not use AltiVec predicate framework.
20536 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20537 support for predefined __POWER9_VECTOR__ macro to indicate that
20538 Power9 instruction selection is enabled.
20539 (altivec_overloaded_builtins): Remove extraneous
20540 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
20541 function argument types RS6000_BTI_bool_V16QI and
20542 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
20543 entry for overloaded function argument types RS6000_BTI_bool_V4SI
20544 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
20545 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
20546 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
20547 Power9 for implementations of vec_cmpne. Change the signature for
20548 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
20549 (representing vec_all_ne) to remove the previously described first
20550 argument of type RS6000_BTI_INTSI, as this was an artifact of
20551 reliance on the AltiVec predicate framework, which is no longer
20552 used in the implementation of these functions. Add
20553 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
20554 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
20555 since, unlike the AltiVec predicate framework implementation, we
20556 do not share function descriptors between vec_alle and vec_anyeq.
20557 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
20558 set of modes that receive special treatment even when
20559 TARGET_P9_VECTOR is true. The special treatment emits code that
20560 does not depend on Power9 instructions.
20561 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
20562 define_expand to not rely on AltiVec predicate framework.
20563 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20565 (vector_ne_v2di_p): Change this define_expand to not rely on
20566 AltiVec predicate framework.
20567 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
20569 (vector_ne_<mode>_p): Change this define_expand to not rely on
20570 AltiVec predicate framework.
20571 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20573 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
20574 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
20575 define_insn pattern.
20576 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
20577 define_insn pattern because the xvcmpne<VSs>. instruction is not
20579 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
20580 instruction is not supported.
20582 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20584 * config/nvptx/nvptx.c: Include intl.h.
20586 2017-03-01 Martin Jambor <mjambor@suse.cz>
20589 * ipa-prop.h (ipa_bits): Removed field known.
20590 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
20591 to pointers. Adjusted their comments to warn about their sharing.
20592 (ipcp_transformation_summary): Change bits to a vector of pointers.
20593 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
20594 (ipa_get_ipa_bits_for_value): Declare.
20595 * tree-vrp.h (value_range): Mark as GTY((for_user)).
20596 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
20597 (ipa_bits_hash_table): Likewise.
20598 (ipa_vr_ggc_hash_traits): Likewise.
20599 (ipa_vr_hash_table): Likewise.
20600 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
20601 being pointers and vr_known being removed.
20602 (ipa_set_jf_unknown): Likewise.
20603 (ipa_get_ipa_bits_for_value): New function.
20604 (ipa_set_jfunc_bits): Likewise.
20605 (ipa_get_value_range): New overloaded functions.
20606 (ipa_set_jfunc_vr): Likewise.
20607 (ipa_compute_jump_functions_for_edge): Use the above functions to
20608 construct bits and vr parts of jump functions.
20609 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
20610 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20612 (ipcp_grow_transformations_if_necessary): Also allocate
20613 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20615 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
20616 them. Fix too long lines.
20617 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
20618 vr_known being removed.
20619 (ipa_read_jump_function): Use new setter functions to construct bits
20620 and vr parts of jump functions or set them to NULL.
20621 (write_ipcp_transformation_info): Adjust for bits being pointers.
20622 (read_ipcp_transformation_info): Likewise.
20623 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
20625 Include gt-ipa-prop.h.
20626 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
20628 (ipcp_store_bits_results): Likewise.
20629 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
20630 Do not write to existing jump functions but use a temporary instead.
20632 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20635 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
20636 attempt to use its first operand as BIT_FIELD_REF base.
20638 2017-03-01 Richard Biener <rguenther@suse.de>
20640 PR middle-end/79721
20641 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
20642 interpolating formula in wrapping arithmetic.
20643 (chrec_apply): Convert chrec_evaluate return value to wanted type.
20645 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20647 PR tree-optimization/79734
20648 * tree-vect-generic.c (expand_vector_condition): Optimize
20649 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
20650 Handle VEC_COND_EXPR where comparison has different inner width from
20651 type's inner width.
20653 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
20655 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
20656 markup, and similar issues. Remove @opindex entries for things
20657 that aren't options. Add missing -mmpy-option entries.
20659 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20661 PR tree-optimization/79737
20662 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
20663 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
20664 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
20665 instead of byte_size. Formatting fix.
20666 (shift_bytes_in_array_right): Formatting fix.
20668 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
20671 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
20672 condition on optimize for the leaf function test.
20674 2017-02-28 Martin Liska <mliska@suse.cz>
20677 * read-rtl-function.c (function_reader::handle_unknown_directive):
20678 Bail out when one uses -flto.
20680 2017-02-28 Martin Liska <mliska@suse.cz>
20682 * common.opt: Replace space with tabular for options of <number>
20684 * config/i386/i386.opt: Show <number> value for
20685 -mlarge-data-threshold.
20686 * opts.c (print_filtered_help): Do not display number in hexadecimal
20689 2017-02-28 Martin Liska <mliska@suse.cz>
20691 * common.opt: Fix --help=option -Q for options which are of
20694 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
20696 * config/i386/i386.c (print_reg): Error out for values
20697 of 8-bit size in invalid integer register.
20699 2017-02-28 Martin Sebor <msebor@redhat.com>
20701 PR tree-optimization/79691
20702 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
20704 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20707 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
20708 gcc_unreachable with output_operand_lossage.
20710 2017-02-28 Richard Biener <rguenther@suse.de>
20712 PR tree-optimization/79740
20713 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
20715 (visit_nary_op): Insert the nary into the hashtable if we
20716 pattern-matched sth.
20717 * tree-ssa-pre.c (eliminate_insert): Robustify.
20719 2017-02-28 Richard Biener <rguenther@suse.de>
20721 PR middle-end/79731
20722 * fold-const.c (decode_field_reference): Reject out-of-bound
20725 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20727 * config/i386/i386.c: Include intl.h.
20728 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
20729 instead of just cond ? "..." : "...".
20730 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
20731 * coverage.c (read_counts_file): Likewise.
20732 * omp-offload.c: Include intl.h.
20733 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
20734 of just cond ? "..." : "...".
20735 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
20736 of just cond ? "..." : "...".
20738 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
20741 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
20743 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
20744 'tune for' CPU name.
20745 * config/arm/arm-cpu-data.h: Regenerated.
20747 2017-02-28 Richard Biener <rguenther@suse.de>
20749 PR tree-optimization/79732
20750 * tree-inline.c (expand_call_inline): Do not shadow var.
20752 2017-02-28 Richard Biener <rguenther@suse.de>
20754 PR tree-optimization/79723
20755 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
20756 address-space properly.
20758 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
20760 * doc/optinfo.texi (Optimization groups): Fix option used for
20762 * doc/invoke.texi (-fopt-info): Document "omp".
20763 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
20764 (OPTGROUP_ALL): Add OPTGROUP_OMP.
20765 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
20766 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
20767 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
20769 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
20771 * dumpfile.c (optgroup_options): Instead of "openmp", associate
20772 OPTGROUP_OMP with "omp".
20774 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
20777 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
20778 for arithmetic shift of unsigned V2DI.
20780 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
20782 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
20783 arc/linux.h headers.
20784 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
20785 (LINK_SPEC): Likewise.
20786 (ARC_TLS_EXTRA_START_SPEC): Likewise.
20787 (EXTRA_SPECS): Likewise.
20788 (STARTFILE_SPEC): Likewise.
20789 (ENDFILE_SPEC): Likewise.
20790 (LIB_SPEC): Likewise.
20791 (TARGET_SDATA_DEFAULT): Likewise.
20792 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
20793 (MULTILIB_DEFAULTS): Likewise.
20794 (DWARF2_UNWIND_INFO): Likewise.
20795 * config/arc/big.h: New file.
20796 * config/arc/elf.h: Likewise.
20797 * config/arc/linux.h: Likewise.
20798 * config/arc/t-uClibc: Remove.
20800 2017-02-27 Bin Cheng <bin.cheng@arm.com>
20802 PR tree-optimization/77536
20803 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
20804 (tree_transform_and_unroll_loop): Use above function to compute the
20805 estimated niter of unrolled loop and use it when scaling profile.
20806 Also use count info rather than frequency if it's non-zero.
20807 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
20808 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
20809 (vect_transform_loop): Call above function.
20811 2017-02-27 Richard Biener <rguenther@suse.de>
20813 PR tree-optimization/45397
20814 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
20815 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
20816 (visit_nary_op): Add pattern matching for CSEing sign-changed
20817 or truncated operations with wider ones.
20819 2017-02-27 Richard Biener <rguenther@suse.de>
20821 PR tree-optimization/79690
20822 * tree-vect-stmts.c (vectorizable_store): Use vector type
20823 built from the DR with address-space.
20825 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
20827 * doc/invoke.texi (Optimize Options): Refine the description
20828 of asan-use-after-return.
20830 2017-02-25 Alan Modra <amodra@gmail.com>
20832 PR rtl-optimization/79584
20833 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
20834 base, not ad->base_term, the reg within base. Remove assertion
20835 that ad->base == ad->base_term. Replace gen_int_mode using
20836 bogus mode with const0_rtx.
20838 2017-02-25 Jakub Jelinek <jakub@redhat.com>
20840 PR middle-end/79396
20841 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
20842 FMA_EXPR like tcc_binary or tcc_unary.
20844 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
20847 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
20849 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
20850 (output_loc_operands): Handle DW_OP_call_ref and
20851 DW_OP_GNU_variable_value.
20852 (struct variable_value_struct): New type.
20853 (struct variable_value_hasher): Likewise.
20854 (variable_value_hash): New variable.
20855 (string_types): Remove.
20856 (copy_loc_descr): New function.
20857 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
20858 (prepend_loc_descr_to_each): New function.
20859 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
20860 instead of add_loc_descr_to_each if the first argument is single
20861 location list and the second has multiple.
20862 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
20863 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
20864 when looking for variable value which doesn't have other location info.
20865 (loc_list_from_tree): Formatting fix.
20866 (gen_array_type_die): Simplify DW_AT_string_length handling.
20867 (adjust_string_types): Remove.
20868 (gen_subprogram_die): Don't call adjust_string_types nor test/set
20869 string_types. Call resolve_variable_values.
20870 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
20871 (resolve_addr_in_expr): Likewise. Add A argument.
20872 (copy_deref_exprloc): Remove deref argument. Adjust for the
20873 original expression being DW_OP_GNU_variable_value with optionally
20874 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
20875 optionally after it.
20876 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
20877 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
20878 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
20879 (variable_value_hasher::hash, variable_value_hasher::equal): New
20881 (resolve_variable_value_in_expr, resolve_variable_value,
20882 resolve_variable_values, note_variable_value_in_expr,
20883 note_variable_value): New functions.
20884 (dwarf2out_early_finish): Call note_variable_value on all toplevel
20887 2017-02-24 Jakub Jelinek <jakub@redhat.com>
20890 * opts.h (handle_generated_option): Add GENERATED_P argument.
20891 * opts-common.c (handle_option): Adjust function comment.
20892 (handle_generated_option): Add GENERATED_P argument, pass it to
20894 (control_warning_option): Pass false to handle_generated_option
20896 * opts.c (maybe_default_option): Pass true to handle_generated_option
20898 * optc-gen.awk: Likewise.
20900 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20902 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
20903 a REG, look at the REG it is a SUBREG of.
20904 (splitter for cmpeqsi_t): Ditto.
20906 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20908 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
20909 the special USEs with the pattern of the insn, not the insn itself.
20911 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
20914 * doc/invoke.texi: Document -mload-store-pairs.
20916 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20917 Sandra Loosemore <sandra@codesourcery.com>
20919 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
20920 argument isn't a CONST_INT.
20921 (nios2_alternate_compare_const): Assert op is a CONST_INT.
20922 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
20923 (nios2_validate_compare): Bypass alternate compare logic if *op2
20924 is not a CONST_INT.
20925 (ldstwm_operation_p): Return false if first_base is not a REG or
20926 if first_offset is not a CONST_INT.
20928 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20930 * config/cris/cris.md: Use correct operand in a define_peephole2.
20932 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20934 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
20936 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20938 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
20939 this_insn if it is an INSN or JUMP_INSN.
20940 (force_offsettable): Look at base, not at addr.
20941 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
20942 on things that aren't necessarily CONST_INTs.
20944 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
20946 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
20947 -mfpmath=sse is the default also for x86-32 targets with SSE2
20948 instruction set when @option{-ffast-math} is enabled
20950 2017-02-24 Jeff Law <law@redhat.com>
20952 PR rtl-optimizatoin/79286
20953 * ira.c (update_equiv_regs): Drop may_trap_p exception to
20956 2017-02-24 Richard Biener <rguenther@suse.de>
20958 PR tree-optimization/79389
20959 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
20962 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
20964 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
20965 function comment to reflect reality.
20966 (loop_exits_before_overflow): Fix typo in function description.
20968 2017-02-24 Richard Biener <rguenther@suse.de>
20970 PR tree-optimization/79389
20971 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
20972 properly that a threading opportunity exists. Detect conditional
20973 copy/constant propagation opportunities.
20975 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
20977 * config/visium/visium.md (type): Add trap.
20978 (b): New mode attribute.
20979 (*btst): Rename into...
20980 (*btst<mode>): ...this and adjust.
20981 (*cbranchsi4_btst_insn): Rename into...
20982 (*cbranch<mode>4_btst_insn): ...this and adjust.
20983 (trap): New define_insn.
20985 2017-02-23 Jakub Jelinek <jakub@redhat.com>
20987 PR tree-optimization/79389
20988 * ifcvt.c (struct noce_if_info): Add rev_cond field.
20989 (noce_reversed_cond_code): New function.
20990 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
20991 reversed_comparison_code. Formatting fix.
20992 (noce_try_store_flag): Test rev_cond != NULL in addition to
20993 reversed_comparison_code.
20994 (noce_try_store_flag_constants): Likewise.
20995 (noce_try_store_flag_mask): Likewise.
20996 (noce_try_addcc): Use rev_cond if non-NULL instead of
20997 reversed_comparison_code.
20998 (noce_try_cmove_arith): Likewise. Formatting fixes.
20999 (noce_try_minmax, noce_try_abs): Clear rev_cond.
21000 (noce_find_if_block): Initialize rev_cond.
21001 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
21002 instead of false as last argument never attempt to reverse it
21005 2017-02-23 Bin Cheng <bin.cheng@arm.com>
21007 PR tree-optimization/79663
21008 * tree-predcom.c (combine_chains): Process refs in reverse order
21009 only for ZERO length chains, and add explaining comment.
21011 2017-02-23 Jeff Law <law@redhat.com>
21013 PR tree-optimization/79578
21014 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
21015 in call to operand_equal_p.
21017 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
21020 * config/i386/cpuid.h: Fix another undefined behavior.
21022 2017-02-23 Richard Biener <rguenther@suse.de>
21024 PR tree-optimization/79683
21025 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
21026 vector types for data-refs.
21028 2017-02-23 Martin Liska <mliska@suse.cz>
21030 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
21032 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21034 PR middle-end/79665
21035 * internal-fn.c (get_range_pos_neg): Moved to ...
21036 * tree.c (get_range_pos_neg): ... here. No longer static.
21037 * tree.h (get_range_pos_neg): New prototype.
21038 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
21039 are known to be in between 0 and signed maximum inclusive, try to
21040 expand both unsigned and signed divmod and use the cheaper one from
21043 2017-02-22 Jeff Law <law@redhat.com>
21045 PR tree-optimization/79578
21046 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
21047 to compare base operands.
21049 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
21052 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
21053 gpc_reg_operand instead of fpr_reg_operand.
21055 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
21057 * config/mips/mips.c (mips_return_in_memory): Force FP
21058 vector types to be returned in memory for o32 ABI.
21060 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21062 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
21063 instead of DW_TAG_member for static data member declarations and don't
21064 set no_linkage_name for static inline data members.
21065 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
21068 2017-02-22 Martin Liska <mliska@suse.cz>
21070 * doc/invoke.texi: Replace inequality signs with square brackets
21073 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21075 PR tree-optimization/68644
21076 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
21078 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
21081 * lra-constraints.c (simplify_operand_subreg): Handle
21082 WORD_REGISTER_OPERATIONS targets.
21084 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21087 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
21088 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
21089 elimination by swapping fld*.
21091 2017-02-22 Richard Biener <rguenther@suse.de>
21093 PR tree-optimization/79673
21094 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
21095 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
21096 irrelevant address-space qualifiers and avoiding a
21097 ADDR_SPACE_CONVERT_EXPR from fold_convert.
21099 2017-02-22 Richard Biener <rguenther@suse.de>
21101 PR tree-optimization/79666
21102 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
21103 to not symbolically negate if that may introduce undefined
21106 2017-02-22 Martin Liska <mliska@suse.cz>
21109 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
21110 * data-streamer-out.c (streamer_write_gcov_count_stream):
21112 * value-prof.c (stream_out_histogram_value): Make assert more
21113 precise based on type of counter.
21115 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
21118 * config/i386/i386.md (standard_x87sse_constant_load splitter):
21119 Use nonimmediate_operand instead of memory_operand for operand 1.
21120 (float-extend standard_x87sse_constant_load splitter): Ditto.
21122 2017-02-21 Jeff Law <law@redhat.com>
21124 PR tree-optimization/79621
21125 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
21126 blocks with edges to themselves.
21128 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21131 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
21132 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
21133 Use gimple_call_builtin_p.
21136 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
21137 on temporarily removed DEBUG_INSNs.
21139 PR tree-optimization/79649
21140 * tree-loop-distribution.c (classify_partition): Give up on
21141 non-generic address space loads/stores.
21143 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
21145 * doc/loop.texi (Loop manipulation): Remove nonexistent
21146 tree_ssa_loop_version from the documentation.
21147 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
21149 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21152 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
21153 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
21154 * config/rs6000/rs6000.c: Include except.h.
21155 (rs6000_expand_split_stack_prologue): Call
21156 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
21158 2017-02-21 Martin Jambor <mjambor@suse.cz>
21161 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
21162 have been analyzed.
21164 2017-02-21 Martin Jambor <mjambor@suse.cz>
21166 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
21167 for backward compatibility only.
21168 * doc/invoke.texi (Option Summary): Remove all references to
21169 -fipa-cp-alignment.
21171 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
21175 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21177 * lra-constraints.c (curr_insn_transform): Handle
21178 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21180 2017-02-21 Martin Liska <mliska@suse.cz>
21182 * config/i386/i386.opt: Replace -masm-dialect with -masm.
21184 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
21186 PR translation/79638
21187 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
21189 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
21192 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
21193 (arm_function_ok_for_sibcall): Return false for an indirect call by
21194 descriptor if all the argument registers are used.
21195 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
21196 alignment of the function.
21198 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21200 PR tree-optimization/61441
21201 * simplify-rtx.c (simplify_const_unary_operation): For
21202 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
21203 the sNaN unmodified.
21205 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21207 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21208 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21209 instead of SYSTEM_HEADER_DIR.
21211 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
21212 Martin Liška <mliska@suse.cz>
21214 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
21215 Fix typos and grammar, use active voice, and clarify.
21217 2017-02-20 Marek Polacek <polacek@redhat.com>
21219 PR middle-end/79537
21220 * gimplify.c (gimplify_expr): Handle unused *&&L;.
21223 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
21225 2017-02-20 Jakub Jelinek <jakub@redhat.com>
21228 * config/i386/i386.c (ix86_expand_builtin): Handle
21229 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
21230 ix86_builtins_isa[fcode].isa as a requirement of those
21231 flags and any other flag in the bitmask.
21232 (ix86_init_mmx_sse_builtins): Use 0 instead of
21233 ~OPTION_MASK_ISA_64BIT as mask.
21234 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
21235 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
21236 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
21237 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
21239 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21242 * lra-constraints.c (split_reg): Check requested split mode
21243 is supported by the register.
21245 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21247 * lra-constraints.c (simplify_operand_subreg): Remove early
21250 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21253 * lra-constraints.c (curr_insn_transform): Tighten condition
21254 for converting SUBREG reloads from OP_OUT to OP_INOUT.
21256 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21259 * lra-constraints.c (curr_insn_transform): Handle
21260 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21262 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
21265 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
21267 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
21269 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
21272 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
21275 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21277 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
21280 2017-02-18 Jakub Jelinek <jakub@redhat.com>
21282 * final.c (last_columnnum, override_columnnum): New variables.
21283 (final_start_function): Set last_columnnum, pass it to begin_prologue
21284 hook and pass 0 to dwarf2out_begin_prologue.
21285 (final_scan_insn): Update override_columnnum. Pass last_columnnum
21286 to source_line debug hook.
21287 (notice_source_line): Compute last_columnnum and for debug_column_info
21288 return true on column changes.
21289 * debug.h (struct gcc_debug_hooks): Add column argument to
21290 source_line and begin_prologue hooks.
21291 (debug_nothing_int_charstar_int_bool): Remove prototype.
21292 (debug_nothing_int_int_charstar,
21293 debug_nothing_int_int_charstar_int_bool): New prototypes.
21294 (dwarf2out_begin_prologue): Add column argument.
21295 * debug.c (do_nothing_debug_hooks): Adjust source_line and
21296 begin_prologue hooks.
21297 (debug_nothing_int_charstar_int_bool): Remove.
21298 (debug_nothing_int_int_charstar,
21299 debug_nothing_int_int_charstar_int_bool): New functions.
21300 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
21301 through to dwarf2out_source_line.
21302 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
21303 (dwarf2out_source_line): Add column argument, emit it if requested.
21304 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
21306 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21307 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21308 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
21309 through to dwarf2out_begin_prologue.
21310 (vmsdbgout_source_line): Add column argument, pass it through to
21311 dwarf2out_source_line.
21312 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
21313 dbxout_source_line caller.
21314 (dbxout_source_line): Add column argument.
21316 * common.opt (gno-column-info, gcolumn-info): New options.
21317 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
21318 (check_die): Also test for multiple DW_AT_decl_column attributes.
21319 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
21320 DW_AT_decl_column if requested.
21321 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
21323 (gen_variable_die): Likewise.
21324 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
21325 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
21328 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
21329 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
21330 (ix86_handle_option): Handle OPT_m3dnowa.
21331 * doc/invoke.texi (-m3dnowa): Document.
21332 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
21333 -m3dnowa instead of -m3dnow -march=athlon.
21336 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
21337 instead of gcc_assert for K, r and R code checks. Formatting fixes.
21339 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21342 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
21343 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
21344 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
21345 generator for vsx_xxpermdi_<mode>_be.
21346 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
21347 force big-endian semantics.
21348 (vsx_xxpermdi_<mode>_be): New define_expand with same
21349 implementation as previous version of vsx_xxpermdi_<mode>.
21351 2017-02-17 Jakub Jelinek <jakub@redhat.com>
21353 PR tree-optimization/79327
21354 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
21355 variable, its initialization and use.
21357 2017-02-17 Julia Koval <julia.koval@intel.com>
21359 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
21360 (OPTION_MASK_ISA_PKU_UNSET): New.
21361 (ix86_handle_option): Handle -mrdpid.
21362 * config/i386/cpuid.h (bit_RDPID): New.
21363 * config/i386/driver-i386.c (host_detect_local_cpu):
21364 Detect RDPID feature.
21365 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
21366 * config/i386/i386-c.c (ix86_target_macros_internal):
21368 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
21369 (ix86_valid_target_attribute_inner_p): Add "rdpid".
21370 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
21371 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
21372 * config/i386/i386.md (define_insn "rdpid"): New.
21373 * config/i386/i386.opt Add -mrdpid.
21374 * config/i386/immintrin.h (_rdpid_u32): New.
21376 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
21378 PR rtl-optimization/79541
21379 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
21380 instead of transforming it into USE.
21382 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
21384 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
21385 If HONOR_SNANS (SFmode) force the input to a register.
21386 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
21387 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
21388 an frsp or similar insn.
21390 2017-02-17 Martin Liska <mliska@suse.cz>
21392 PR rtl-optimization/79577
21393 * params.def (selsched-max-sched-times): Increase minimum to 1.
21395 2017-02-17 Martin Liska <mliska@suse.cz>
21397 PR rtl-optimization/79574
21398 * gcse.c (want_to_gcse_p): Prevent integer overflow.
21400 2017-02-17 Martin Liska <mliska@suse.cz>
21402 PR tree-optimization/79529
21403 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
21404 ssa_defined_default_def_p to handle cases which are implicitly
21406 * tree-ssa.c (ssa_defined_default_def_p): New function.
21407 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
21408 which are implicitly defined.
21409 * tree-ssa.h (ssa_defined_default_def_p): Declare.
21411 2017-02-17 Richard Biener <rguenther@suse.de>
21413 PR middle-end/79576
21414 * params.def (max-ssa-name-query-depth): Limit to 10.
21416 2017-02-17 Richard Biener <rguenther@suse.de>
21418 PR tree-optimization/79552
21419 * tree-ssa-structalias.c (visit_loadstore): Properly verify
21422 2017-02-17 Richard Biener <rguenther@suse.de>
21425 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
21427 2017-02-17 Marek Polacek <polacek@redhat.com>
21429 PR middle-end/79536
21430 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
21431 (fold_negate_expr): New wrapper.
21433 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
21435 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
21436 Correct terminology and de-emphasize pre-standard behavior.
21438 2017-02-16 Alan Modra <amodra@gmail.com>
21440 PR rtl-optimization/79286
21441 * ira.c (def_dominates_uses): New function.
21442 (update_equiv_regs): Don't create an equivalence for insns that
21443 may trap where the register def does not dominate the use.
21445 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
21447 PR rtl-optimization/78127
21448 * lra.c (lra): Call lra_eliminate before finish the loop after
21451 2017-02-16 Richard Biener <rguenther@suse.de>
21453 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
21455 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
21456 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
21457 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
21458 (isl_val_int_from_wi): New function.
21459 (extract_affine_gmp): Rename to ...
21460 (extract_affine_wi): ... this, take a widest_int.
21461 (extract_affine_int): Just wrap extract_affine_wi.
21462 (add_param_constraints): Use isl_val_int_from_wi.
21463 (add_loop_constraints): Likewise, and extract_affine_wi.
21465 2017-02-15 Jeff Law <law@redhat.com>
21467 PR middle-end/79521
21468 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
21469 ira_init_register_move_cost_if_necessary.
21471 2017-02-15 Martin Sebor <msebor@redhat.com>
21473 PR middle-end/32003
21474 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
21475 removed in a prior commit.
21477 2017-02-15 Bin Cheng <bin.cheng@arm.com>
21479 PR tree-optimization/79347
21480 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
21481 counters during peeling.
21483 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
21485 * Makefile.in (site.exp): Remove "set ISLVER".
21487 2017-02-15 Jakub Jelinek <jakub@redhat.com>
21490 * real.c (real_from_integer): Call real_convert even for decimal.
21492 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
21495 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
21497 2017-02-14 Andrew Pinski <apinski@cavium.com>
21499 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
21500 cores and change the partno/implementer to be correct.
21501 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
21502 the 'B" as the implementer.
21503 * config/aarch64/aarch64-tune.md: Regenerate.
21505 2017-02-14 Carl Love <cel@us.ibm.com>
21507 * config/rs6000/rs6000.c: Add case statement entry to make the
21508 xvcvuxdsp built-in argument unsigned.
21509 * config/rs6000/vsx.md: Fix the source and return operand types so they
21510 match the instruction definitions from the ISA document. Fix typo
21511 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
21514 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
21517 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
21518 member early_clobber_alts.
21519 * lra-lives.c (reg_early_clobber_p): New.
21520 (process_bb_lives): Use it.
21521 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
21522 (debug_operand_data): Initialize early_clobber_alts.
21523 (setup_operand_alternative): Set up early_clobber_alts.
21524 (collect_non_operand_hard_regs): Ditto. Pass early clobber
21525 alternatives to new_insn_reg.
21526 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
21528 (lra_update_insn_regno_info): Pass the new arg.
21530 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21532 PR middle-end/79505
21533 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
21534 (new_oacc_loop_raw): Don't clear already cleared fields.
21537 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
21538 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
21539 _mm512_prefetch_i64gather_ps): New inline functions and macros.
21541 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
21544 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
21546 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
21549 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
21550 the extra instruction to the right place to store 128-bit constant
21553 2017-02-14 Martin Sebor <msebor@redhat.com>
21555 PR middle-end/79448
21556 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
21557 warning for strings of unknown length.
21559 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
21561 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
21563 2017-02-14 Jeff Law <law@redhat.com>
21566 * ira-costs.c (scan_one_insn): Initialize register move costs
21567 for pseudos seen in USE/CLOBBER insns.
21569 PR tree-optimization/79095
21570 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
21571 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
21572 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
21573 if the operands are known to be not equal, then the resulting range
21575 (intersect_ranges): If the new range is ~[0,0] and the old range is
21576 wide, then prefer ~[0,0].
21577 * tree-vrp.c (overflow_comparison_p_1): New function.
21578 (overflow_comparison_p): New function.
21579 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
21580 if NAME is used in an overflow test.
21581 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
21582 overflow check that can be expressed as an equality test, then adjust
21583 ops to be that equality test.
21585 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21587 * config/s390/s390-builtin-types.def: Remove flags argument.
21588 * config/s390/s390.c (s390_init_builtins): Likewise.
21590 2017-02-14 Martin Liska <mliska@suse.cz>
21592 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
21593 vector. Fix trailing white spaces.
21595 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
21597 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
21600 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21602 PR rtl-optimization/68664
21603 * config/arm/arm.c (arm_sched_can_speculate_insn):
21604 New function. Declare prototype.
21605 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21607 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21609 PR rtl-optimization/68664
21610 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
21612 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21614 2017-02-14 Amit Pawar <amit.pawar@amd.com>
21616 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
21617 max skip bytes for function, loop and jump.
21619 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21621 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
21622 ABS_EXPR for gimple dump.
21624 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21627 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
21629 PR tree-optimization/79408
21630 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
21631 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
21632 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
21633 also if rhs1 is INTEGER_CST.
21635 2017-02-14 Richard Biener <rguenther@suse.de>
21637 PR middle-end/79432
21638 * tree-into-ssa.c (insert_phi_nodes): When the function can
21639 have abnormal edges rewrite SSA names with broken use-def
21640 dominance out of SSA and register them for PHI insertion.
21642 2017-02-13 Martin Sebor <msebor@redhat.com>
21644 PR middle-end/79496
21645 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
21646 clearing info.nowrite flag when snprintf size argument is a range.
21648 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21650 * cprop.c (cprop_jump): Add missing space in string literal.
21651 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
21652 (get_constraint_for_component_ref): Likewise.
21653 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
21654 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
21655 * lra-constraints.c (process_alt_operands): Likewise.
21656 * ipa-inline.c (inline_small_functions): Likewise.
21657 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
21658 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21659 * trans-mem.c (diagnose_tm_1_op): Likewise.
21660 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
21661 (grid_parallel_clauses_gridifiable): Likewise.
21663 * config/nvptx/mkoffload.c (process): Add space in between
21666 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
21667 "MOD4_SSE_REGS" and "ALL_REGS".
21669 * spellcheck.c (test_data): Add , in between "foo" and "food".
21671 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21674 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
21675 boundary crossing check and subsequent code generation agree.
21677 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21679 * config/aarch64/aarch64.c (has_memory_op): Delete.
21680 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
21683 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21685 PR rtl-optimization/79388
21686 PR rtl-optimization/79450
21687 * combine.c (distribute_notes): When removing TEM_INSN for which
21688 corresponding dest has last value recorded, invalidate that last
21691 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21693 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
21694 of explicit '@'. Add missing assembly comment marker on branch costs
21697 2017-02-13 Nathan Sidwell <nathan@acm.org>
21699 * gengtype-lex.l (<in_struct>): Add '/'.
21701 2017-02-13 Martin Liska <mliska@suse.cz>
21704 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
21706 2017-02-13 Richard Biener <rguenther@suse.de>
21708 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
21710 * configure: Re-generate.
21711 * config.in: Likewise.
21712 * graphite-dependences.c: Simplify as if
21713 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
21714 * graphite-isl-ast-to-gimple.c: Likewise.
21715 * graphite-optimize-isl.c: Likewise.
21716 * graphite-poly.c: Likewise.
21717 * graphite-sese-to-poly.c: Likewise.
21718 * graphite.h: Likewise.
21719 * toplev.c: Include isl/version.h and use isl_version () for
21720 printing the ISL version.
21721 * doc/install.texi: Update ISL requirement.
21723 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21725 * doc/standards.texi (Standards): Update reference to
21728 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21730 * doc/extend.texi (Named Address Spaces): sourceware.org now
21732 * doc/install.texi (Binaries): Ditto.
21735 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21737 * doc/cpp.texi: Replace "stringify"/"stringification" with C
21738 standard terminology "stringize"/"stringizing" throughout.
21739 * doc/cppinternals.texi: Likewise.
21741 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21743 * doc/extend.texi: Fix some spelling mistakes and typos.
21744 * doc/invoke.texi: Likewise.
21746 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21749 * params.def (inline-min-speedup) Change from 10 to 8.
21751 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21753 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
21756 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21759 * ipa-inline-analysis.c (get_minimal_bb): New function.
21760 (record_modified): Use it.
21761 (remap_edge_change_prob): Handle also ancestor functions.
21763 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
21765 * doc/contrib.texi (Contributors): Remove broken link into
21766 the Mauve CVS repository.
21768 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21770 PR middle-end/79454
21771 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
21772 result computation whenever lhs doesn't have vector mode, not
21773 just when it has BLKmode.
21775 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21777 * doc/makefile.texi (profiledbootstrap): Refer to the
21778 installation instructions only in textual form.
21780 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21783 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
21785 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21787 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
21788 (Specific): Update mingw-w64 reference.
21790 (Specific): Remove broken link to Renesas RX processor.
21792 2017-02-10 Richard Biener <rguenther@suse.de>
21794 * toplev.c (process_options): Do not mention obsolete graphite
21795 options when printing sorry message about missing graphite support.
21796 Mention -floop-nest-optimize.
21798 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
21800 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
21801 (vtst_p16): Likewise.
21802 (vtstq_p8): Likewise.
21803 (vtstq_p16): Likewise.
21805 (vtstq_p64): Likewise.
21806 * config/arm/arm_neon.h (vgetq_lane_p64): New.
21807 (vset_lane_p64): New.
21808 (vsetq_lane_p64): New.
21810 2017-02-10 Jakub Jelinek <jakub@redhat.com>
21812 PR tree-optimization/79411
21813 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
21814 stmt operands are SSA_NAMEs used in abnormal phis.
21815 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
21818 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21821 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
21824 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21826 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
21828 2017-02-09 Jakub Jelinek <jakub@redhat.com>
21830 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
21834 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
21835 not arbitrary TREE_CONSTANT.
21838 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
21839 "omp declare target link" attribute unless is_global_var.
21840 * omp-offload.c (find_link_var_op): Likewise.
21842 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
21843 Chung-Lin Tang <cltang@codesourcery.com>
21845 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
21847 (gimplify_adjust_omp_clauses): Don't delete TILE.
21848 (gimplify_omp_for): Deal with TILE.
21849 * internal-fn.c (expand_GOACC_TILE): New function.
21850 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
21852 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
21853 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
21855 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
21856 avoid DIV for outermost collapse var.
21857 (expand_oacc_for): Insert tile element loop as needed. Adjust.
21858 Remove out of date comments, fix whitespace.
21859 * omp-general.c (omp_extract_for_data): Deal with tiling.
21860 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
21861 adjust OLF_DIM_BASE value.
21862 (struct omp_for_data): Add tiling field.
21863 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
21864 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
21865 for auto loops. Remove default auto determining, moved to
21866 oacc_loop_fixed_partitions.
21867 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
21868 stmts, add e_mask field.
21869 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
21870 (oacc_thread_numbers): Use oacc_dim_call.
21871 (oacc_xform_tile): New.
21872 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
21873 (finish_oacc_loop): Adjust for ifns vector.
21874 (oacc_loop_discover_walk): Append loop abstraction sites to list,
21875 add case for GOACC_TILE fns.
21876 (oacc_loop_xform_loop): Delete.
21877 (oacc_loop_process): Iterate over call list directly, and add
21878 handling for GOACC_TILE fns.
21879 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
21881 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
21882 vector partitioning to outer loops. Assign 2 partitions to loops
21883 when available. Add TILE handling.
21884 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
21885 (execite_oacc_device_lower): Process GOACC_TILE fns,
21886 ignore unknown specs.
21887 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
21888 * tree.c (omp_clause_num_ops): Adjust TILE ops.
21889 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
21891 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
21893 * configure.ac (ACX_BUGURL): Update.
21894 * configure: Regenerate.
21896 2017-02-09 Richard Biener <rguenther@suse.de>
21898 PR tree-optimization/69823
21899 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
21900 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
21902 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
21904 * config/arc/arc-c.def: Add __NPS400__ definition.
21905 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
21906 (TARGET_NPS400): Define.
21908 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
21910 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
21912 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
21913 pointer, arch_info.
21914 (arc_cpu_types): Fill the arch_info field with a pointer into the
21915 arc_arch_types table.
21916 (arc_selected_cpu): Declare.
21917 * config/arc/arc.c (arc_selected_cpu): Make global.
21918 (arc_selected_arch): Delete.
21919 (arc_base_cpu): Delete.
21920 (arc_override_options): Remove references to deleted variables,
21921 update access to arch information.
21922 (ARC_OPT): Update access to arch information.
21923 (ARC_OPTX): Likewise.
21924 * config/arc/arc.h (arc_base_cpu): Remove declaration.
21925 (TARGET_ARC600): Update access to arch information.
21926 (TARGET_ARC601): Likewise.
21927 (TARGET_ARC700): Likewise.
21928 (TARGET_EM): Likewise.
21929 (TARGET_HS): Likewise.
21930 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
21933 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
21936 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
21937 condition/operands for integer GE/LE/GEU/LEU operations.
21939 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
21941 PR translation/79397
21942 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
21945 2017-02-08 Martin Jambor <mjambor@suse.cz>
21948 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
21949 whether allocation happened.
21950 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
21951 nothing was allocated.
21953 2017-02-08 Jakub Jelinek <jakub@redhat.com>
21955 PR tree-optimization/79408
21956 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
21957 constant, but SSA_NAME with a known integer range, use the minimum
21958 of that range instead of op1 to determine if modulo can be replaced
21959 with its first operand.
21961 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21963 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
21965 2017-02-08 Richard Biener <rguenther@suse.de>
21967 PR tree-optimization/71824
21968 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
21969 Check all loops contained in the merged region.
21971 2017-02-07 Andrew Pinski <apinski@cavium.com>
21973 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
21975 2017-02-07 Andrew Pinski <apinski@cavium.com>
21977 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
21978 (thunderxt88): Likewise.
21979 (thunderxt81): Disable LSE and change v8.1 to v8.
21980 (thunderxt83): Likewise.
21982 2017-02-07 Jakub Jelinek <jakub@redhat.com>
21983 Richard Biener <rguenther@suse.de>
21985 PR middle-end/79399
21986 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
21987 type from int to size_t.
21988 * ira-costs.c (struct_costs_size): Change type from int to size_t.
21990 2017-02-07 Jakub Jelinek <jakub@redhat.com>
21992 PR rtl-optimization/79386
21993 * cprop.c (bypass_conditional_jumps): Initialize
21994 bypass_last_basic_block already before splitting bbs after
21995 unconditional traps...
21996 (bypass_conditional_jumps): ... rather than here.
21999 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
22000 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
22001 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
22002 fix -masm=intel patterns.
22004 2017-02-07 Richard Biener <rguenther@suse.de>
22006 PR tree-optimization/79256
22007 PR middle-end/79278
22008 * builtins.c (get_object_alignment_2): Use min_align_of_type
22009 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
22010 and ADJUST_FIELD_ALIGN.
22012 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
22014 * doc/tm.texi: Regenerate.
22015 * stor-layout.c (layout_decl): Adjust.
22016 (update_alignment_for_field): Likewise.
22017 (place_field): Likewise.
22018 (min_align_of_type): Likewise.
22019 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
22020 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
22021 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
22022 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
22023 * config/frv/frv.c (frv_adjust_field_align): Likewise.
22024 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
22025 * config/i386/i386.c (x86_field_alignment): Likewise.
22026 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
22027 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
22028 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
22029 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
22030 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
22031 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
22035 2017-01-30 Richard Biener <rguenther@suse.de>
22037 PR tree-optimization/79256
22038 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22039 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22042 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
22044 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
22045 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
22046 builtins to SImode and emit a zero-extend, if necessary.
22048 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22050 * docs/invoke.texi (RISC-V Options): Alphabetize.
22052 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22054 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
22057 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22059 * config/riscv/riscv.c: New file.
22060 * common/config/riscv/riscv-common.c: Likewise.
22061 * config.gcc: Likewise.
22062 * config/riscv/constraints.md: Likewise.
22063 * config/riscv/elf.h: Likewise.
22064 * config/riscv/generic.md: Likewise.
22065 * config/riscv/linux.h: Likewise.
22066 * config/riscv/multilib-generator: Likewise.
22067 * config/riscv/peephole.md: Likewise.
22068 * config/riscv/pic.md: Likewise.
22069 * config/riscv/predicates.md: Likewise.
22070 * config/riscv/riscv-builtins.c: Likewise.
22071 * config/riscv/riscv-c.c: Likewise.
22072 * config/riscv/riscv-ftypes.def: Likewise.
22073 * config/riscv/riscv-modes.def: Likewise.
22074 * config/riscv/riscv-opts.h: Likewise.
22075 * config/riscv/riscv-protos.h: Likewise.
22076 * config/riscv/riscv.h: Likewise.
22077 * config/riscv/riscv.md: Likewise.
22078 * config/riscv/riscv.opt: Likewise.
22079 * config/riscv/sync.md: Likewise.
22080 * config/riscv/t-elf-multilib: Likewise.
22081 * config/riscv/t-linux: Likewise.
22082 * config/riscv/t-linux-multilib: Likewise.
22083 * config/riscv/t-riscv: Likewise.
22084 * configure.ac: Likewise.
22085 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
22086 Waterman as RISC-V maintainers.
22087 * doc/install.texi: Add RISC-V entries.
22088 * doc/invoke.texi: Add RISC-V options section.
22089 * doc/md.texi: Add RISC-V constraints section.
22090 * configure: Regenerated.
22092 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
22095 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
22096 false values to be constant vectors with all 0 or all 1 bits set.
22097 (vcondu<mode><mode>): Likewise.
22098 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
22100 (fpmask_comparison_operator): Update comment.
22101 (vecint_comparison_operator): New predicate.
22102 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
22103 vector conditionals when the true and false values are constant
22104 vectors with all 0 bits or all 1 bits set.
22106 2017-02-06 Martin Sebor <msebor@redhat.com>
22108 PR tree-optimization/79376
22109 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
22111 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
22113 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
22114 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
22115 to simplify split condition.
22117 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22119 * omp-expand.c (oxpand_omp_atomic_fetch_op,
22120 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
22123 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
22125 PR rtl-optimization/68664
22126 * target.def (can_speculate_insn): New hook.
22127 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
22128 * doc/tm.texi: Regenerate.
22129 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
22130 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
22131 (rs6000_sched_can_speculate_insn): New function.
22133 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22135 PR tree-optimization/79284
22136 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
22137 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
22138 vectorizable_mask_load_store, vectorizable_operation,
22139 vect_is_simple_cond, get_same_sized_vectype): Use it instead
22140 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
22141 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
22142 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
22143 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
22144 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
22145 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
22146 is_gimple_assign (stmt). Replace another such test with
22147 is_gimple_assign (stmt).
22149 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22152 * config/avr/avr.c (rtl-iter.h): Include it.
22153 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
22154 (avr_legitimate_combined_insn): ...and implementation.
22156 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
22158 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
22159 * config/s390/s390.c (s390_const_operand_ok)
22160 (s390_canonicalize_comparison, s390_extract_part)
22161 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
22162 (s390_contiguous_bitmask_p, s390_rtx_costs)
22163 (legitimize_pic_address): Likewise.
22164 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
22165 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
22166 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
22167 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
22168 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
22170 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22172 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
22173 REGNO($0) == REGNO($1).
22175 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22177 * config/s390/linux.h(SIZE_TYPE): Add comment.
22179 2017-02-06 Julian Brown <julian@codesourcery.com>
22180 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22181 Virendra Pathak <virendra.pathak@broadcom.com>
22183 * config/aarch64/aarch64-cores.def: Change the scheduler
22185 * config/aarch64/aarch64.md: Include thunderx2t99.md.
22186 * config/aarch64/thunderx2t99.md: New file.
22188 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22190 * doc/standards.texi (Go Language): Update link to language
22193 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22195 * tree-eh.c (lower_resx): Sanitize profile.
22196 (cleanup_empty_eh_move_lp): Likewise.
22198 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22201 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
22203 * cfgloopmanip.h (loop_version): Update prototype.
22204 * modulo-sched.c (sms_schedule): Update call of loop_version.
22205 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
22206 * tree-parloops.c (gen_parallel_loop): Likewise.
22207 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
22208 * tree-ssa-loop-split.c (split_loop): Likewise.
22209 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
22210 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22212 2017-02-05 Martin Liska <mliska@suse.cz>
22215 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
22217 (print_operand_address): Initialize a struct to zero.
22219 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22221 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
22222 garbage collector only in textual form.
22224 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22226 * doc/extend.texi (x86 specific memory model extensions for
22227 transactional memory): Simplify a phrase.
22229 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
22232 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
22233 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
22234 (atomic_storedi_1): Likewise.
22236 2017-02-04 Jakub Jelinek <jakub@redhat.com>
22238 PR tree-optimization/79338
22239 * tree-parloops.c (gather_scalar_reductions): Don't call
22240 vect_analyze_loop_form for loop->inner before destroying loop's
22243 2017-02-03 Martin Sebor <msebor@redhat.com>
22245 PR tree-optimization/79327
22246 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
22247 when precision has resulted in leading zeros.
22248 (format_integer): Adjust the likely counter to assume an unknown
22249 argument that may be zero is non-zero.
22251 2017-02-03 Jason Merrill <jason@redhat.com>
22254 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
22255 avoid copying non-taken branch.
22257 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22259 PR tree-optimization/79340
22260 * tree-vect-loop.c (vectorizable_reduction): Release
22261 vec_defs elements after safe_splicing them into other vectors.
22264 PR tree-optimization/79327
22265 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
22266 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
22268 (format_integer): Use wide_int_to_tree instead of build_int_cst
22269 + to_?hwi. If argmin is NULL, just set argmin and argmax to
22270 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
22271 of shortest and longest sequence.
22273 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
22275 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
22276 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
22278 2017-02-03 Walter Lee <walt@tilera.com>
22281 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
22282 after initial stackframe link reg save.
22283 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
22285 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22288 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
22289 wu for stxssp alternative.
22291 2017-02-03 Martin Sebor <msebor@redhat.com>
22293 PR tree-optimization/79352
22294 * gimple-fold.c (get_range_strlen): Add argument.
22295 (get_range_strlen): Change return type to bool.
22296 (get_maxval_strlen): Pass in a dummy argument.
22297 * gimple-fold.h (get_range_strlen): Change return type to bool.
22298 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
22299 * tree.h (array_at_struct_end_p): Add argument.
22300 * tree.c (array_at_struct_end_p): Handle it.
22302 2017-02-03 Martin Liska <mliska@suse.cz>
22305 * multiple_target.c (create_dispatcher_calls): Redirect edge
22306 from a caller of a dispatcher.
22307 (expand_target_clones): Make the clones local.
22308 (ipa_target_clone): Do both target clones and resolvers.
22309 (ipa_dispatcher_calls): Remove the pass.
22310 (pass_dispatcher_calls::gate): Likewise.
22311 (make_pass_dispatcher_calls): Likewise.
22312 * passes.def (pass_target_clone): Put as very first IPA early
22315 2017-02-03 Martin Liska <mliska@suse.cz>
22317 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
22318 in case of a function with ifunc attribute.
22320 2017-02-03 Martin Liska <mliska@suse.cz>
22322 * cgraph.c (cgraph_node::dump): Dump function version info.
22323 * symtab.c (symtab_node::dump_base): Add missing new line.
22325 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22327 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
22328 (ifcombine_ifandif): Use it.
22330 2017-02-03 Martin Liska <mliska@suse.cz>
22332 * doc/invoke.texi: Document default value for
22333 use-after-scope-direct-emission-threshold.
22335 2017-02-03 Martin Liska <mliska@suse.cz>
22337 PR tree-optimization/79339
22338 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
22339 (format_floating): Likewise.
22341 2017-02-03 Martin Liska <mliska@suse.cz>
22344 * ipa-prop.c (ipa_node_params_t::insert): Remove current
22346 (ipa_node_params_t::remove): Likewise.
22347 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
22348 initialization from removed ipa_node_params_t::insert.
22349 (ipa_node_params::~ipa_node_params): Move from removed
22350 ipa_node_params_t::release.
22351 * symbol-summary.h (symbol_summary::m_released): New member.
22352 Do not release a summary twice. Do not allow to call finalizer
22353 for types of a summary that live in GGC memory.
22355 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22357 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
22360 2017-02-02 Martin Sebor <msebor@redhat.com>
22362 PR middle-end/79275
22363 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
22364 (format_string): Tighten up the range of output for non-constant
22365 strings and correct the expected range for wide non-constant strings.
22367 2017-02-02 Martin Sebor <msebor@redhat.com>
22369 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
22371 PR middle-end/32003
22372 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
22374 (-fdump-tree-@var): Add to index and document how to come up
22375 with pass-specific option and dump file names.
22376 (-fdump-passes): Clarify where to look for output.
22378 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22380 PR middle-end/77445
22381 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
22382 statistics of the analyzed path; allow threading for speed when
22383 any of BBs along the path are optimized for speed.
22385 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
22387 PR middle-end/78468
22388 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
22389 settings of the virtual registers.
22392 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
22394 * explow.c (get_dynamic_stack_size): Take known alignment of stack
22395 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
22398 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22400 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
22401 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
22403 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22405 * config/s390/s390.md: Add missing comments with the expanded
22407 * config/s390/vector.md: Likewise.
22408 * config/s390/vx-builtins.md: Likewise.
22410 2017-02-02 Jakub Jelinek <jakub@redhat.com>
22413 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
22414 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
22415 conditions on a single line.
22417 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22419 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22420 __S390_VX__ to __VX__.
22422 2017-02-01 Andrew Pinski <apinski@cavium.com>
22424 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
22425 stmt_info to record_stmt_cost.
22426 (vect_get_known_peeling_cost): Pass stmt_info if known to
22428 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
22429 cpu_vector_cost field into
22430 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
22431 field into vec_int_stmt_cost and vec_fp_stmt_cost.
22432 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
22433 splitting of scalar_stmt_cost and vec_stmt_cost.
22434 (thunderx_vector_cost): Likewise.
22435 (cortexa57_vector_cost): LIkewise.
22436 (exynosm1_vector_cost): Likewise.
22437 (xgene1_vector_cost): Likewise.
22438 (thunderx2t99_vector_cost): Improve after the splitting of the two
22440 (aarch64_builtin_vectorization_cost): Update for the splitting of
22441 scalar_stmt_cost and vec_stmt_cost.
22443 2017-02-01 Torvald Riegel <triegel@redhat.com>
22444 Richard Henderson <rth@redhat.com>
22446 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
22447 conditional on existance of a fast atomic load.
22448 * optabs-query.c (can_atomic_load_p): New function.
22449 * optabs-query.h (can_atomic_load_p): Declare it.
22450 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
22451 no fast atomic load is available for the particular size of access.
22452 (expand_atomic_compare_and_swap): Likewise.
22453 (expand_atomic_load): Likewise.
22454 (expand_atomic_store): Likewise.
22455 (expand_atomic_fetch_op): Likewise.
22456 * testsuite/lib/target-supports.exp
22457 (check_effective_target_sync_int_128): Remove x86 because it provides
22458 no fast atomic load.
22459 (check_effective_target_sync_int_128_runtime): Likewise.
22461 2017-02-01 Richard Biener <rguenther@suse.de>
22463 * graphite.c: Include tree-vectorizer.h for find_loop_location.
22464 (graphite_transform_loops): Provide opt-info for optimized nests.
22465 * tree-parloop.c (parallelize_loops): Provide opt-info for
22466 parallelized loops.
22468 2017-02-01 Richard Biener <rguenther@suse.de>
22470 PR middle-end/79315
22471 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
22472 was not set before.
22474 2017-02-01 Richard Biener <rguenther@suse.de>
22476 PR tree-optimization/71824
22477 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22478 Verify the loops are valid in the merged SESE region.
22479 (scop_detection::can_represent_loop_1): Check analyzing the
22480 evolution of the number of iterations in the region succeeds.
22482 2017-01-31 Ian Lance Taylor <iant@golang.org>
22484 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
22485 REG_ARGS_SIZE note to 32-bit push insns and call insn.
22487 2017-01-31 David Malcolm <dmalcolm@redhat.com>
22489 PR preprocessor/79210
22490 * input.c (get_substring_ranges_for_loc): Replace line_width
22491 assertion with error-handling.
22493 2017-01-31 Richard Biener <rguenther@suse.de>
22495 PR tree-optimization/77318
22496 * graphite-sese-to-poly.c (extract_affine): Fix assert.
22497 (create_pw_aff_from_tree): Take loop parameter.
22498 (add_condition_to_pbb): Pass loop of the condition to
22499 create_pw_aff_from_tree.
22501 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22503 * config/s390/s390.c (s390_asan_shadow_offset): New function.
22504 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
22506 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
22510 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
22512 (convert_int_to_float128): Likewise.
22513 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
22514 (convert_int_to_float128): Likewise.
22515 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
22516 (UNSPEC_IEEE128_CONVERT): Likewise.
22517 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
22518 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
22519 Use local variables for IBM extended format.
22520 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
22521 (fix_trunc<mode>si2_fprs): Likewise.
22522 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
22523 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
22524 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
22525 to know that we can now have integers of all sizes in vector
22527 (fix<uns>_<mode>di2_hw): Likewise.
22528 (float<uns>_<mode>si2_hw): Likewise.
22529 (fix_<mode>si2_hw): Likewise.
22530 (fixuns_<mode>si2_hw): Likewise.
22531 (float<uns>_<mode>di2_hw): Likewise.
22532 (float_<mode>di2_hw): Likewise.
22533 (float_<mode>si2_hw): Likewise.
22534 (floatuns_<mode>di2_hw): Likewise.
22535 (floatuns_<mode>si2_hw): Likewise.
22536 (xscvqp<su>wz_<mode>): Delete, no longer used.
22537 (xscvqp<su>dz_<mode>): Likewise.
22538 (xscv<su>dqp_<mode>): Likewise.
22539 (ieee128_mfvsrd_64bit): Likewise.
22540 (ieee128_mfvsrd_32bit): Likewise.
22541 (ieee128_mfvsrwz): Likewise.
22542 (ieee128_mtvsrw): Likewise.
22543 (ieee128_mtvsrd_64bit): Likewise.
22544 (ieee128_mtvsrd_32bit): Likewise.
22546 2017-01-31 Martin Liska <mliska@suse.cz>
22549 * ipa-prop.c (ipa_free_all_node_params): Call release method
22550 instead of ~sumbol_summary to not to trigger double times
22553 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
22555 PR tree-optimization/71691
22556 * bitmap.h (class auto_bitmap): New.
22557 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
22558 is_maybe_undefined instead of ssa_undefined_value_p.
22560 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22562 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22563 __S390_ARCH_LEVEL__ to __ARCH__.
22565 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22567 PR tree-optimization/79267
22568 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
22569 if should_remove_lhs_p is true.
22571 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
22574 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
22575 (add_alignment_attribute): New.
22576 (base_type_die): Add alignment attribute.
22577 (subrange_type_die): Likewise.
22578 (modified_type_die): Likewise.
22579 (gen_array_type_die): Likewise.
22580 (gen_descr_array_type_die: Likewise.
22581 (gen_enumeration_type_die): Likewise.
22582 (gen_subprogram_die): Likewise.
22583 (gen_variable_die): Likewise.
22584 (gen_field_die): Likewise.
22585 (gen_ptr_to_mbr_type_die): Likewise.
22586 (gen_struct_or_union_type_die): Likewise.
22587 (gen_subroutine_type_die): Likewise.
22588 (gen_typedef_die): Likewise.
22589 (base_type_cmp): Compare alignment attribute.
22591 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22594 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
22595 (setb_unsigned) New pattern for setb with CCUNS.
22596 * config/rs6000/rs6000.c (expand_block_compare): Use a different
22597 subfc./subfe sequence to avoid overflow problems. Generate a
22598 shorter sequence with cmpld/setb for power9.
22599 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
22600 for generating subfc. instruction.
22601 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
22602 now uses this instruction.
22604 2017-01-30 Ian Lance Taylor <iant@google.com>
22607 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
22608 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
22610 2017-01-30 Martin Sebor <msebor@redhat.com>
22612 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
22613 Move constant to the right of a relational operator.
22614 (get_mpfr_format_length, format_character, format_string): Ditto.
22615 (should_warn_p, maybe_warn): Same.
22617 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
22619 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
22622 * asan.c (get_translation_unit_decl): Remove function.
22623 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
22625 2017-01-30 Martin Liska <mliska@suse.cz>
22627 PR gcov-profile/79259
22628 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
22629 -fprofile-generate.
22631 2017-01-30 Martin Liska <mliska@suse.cz>
22634 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
22635 Initialize variables with NULL value.
22637 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
22640 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
22642 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
22644 2017-01-30 Richard Biener <rguenther@suse.de>
22646 PR tree-optimization/79276
22647 * tree-vrp.c (process_assert_insertions): Properly adjust common
22648 when removing a duplicate.
22650 * gcc.dg/torture/pr79276.c: New testcase.
22652 2017-01-30 Richard Biener <rguenther@suse.de>
22654 PR tree-optimization/79256
22655 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22656 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22658 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
22660 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
22663 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
22664 ("*r<noxa>sbg_<mode>_sll_bitmask")
22665 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
22666 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
22667 Use contiguous_bitmask_nowrap_operand.
22669 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22672 * config/rs6000/altivec.h (vec_xl): Revise #define.
22673 (vec_xst): Likewise.
22675 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
22677 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
22679 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
22681 PR rtl-optimization/79194
22682 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
22683 traps before call to bypass_conditional_jumps.
22685 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22687 PR tree-optimization/71374
22688 * lra-constraints.c (check_conflict_input_operands): New.
22689 (match_reload): Use it.
22691 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22694 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
22695 account to calculate conflict_set.
22697 2017-01-27 Bin Cheng <bin.cheng@arm.com>
22699 PR rtl-optimization/78559
22700 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
22701 other_insn in combine.
22703 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22705 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
22706 uint16_type_node for BT_UINT16.
22708 2017-01-27 David Malcolm <dmalcolm@redhat.com>
22710 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
22711 "RTL Tests" to menu.
22712 (GIMPLE Tests): New node.
22713 (RTL Tests): New node.
22715 2017-01-27 Richard Biener <rguenther@suse.de>
22717 PR tree-optimization/79245
22718 * tree-loop-distribution.c (distribute_loop): Apply cost
22719 modeling also to detected patterns.
22721 2017-01-27 Richard Biener <rguenther@suse.de>
22723 PR tree-optimization/71433
22724 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
22725 (compare_assert_loc): New function.
22726 (process_assert_insertions): Sort and optimize assert locations
22727 to remove duplicates and push down identical assertions on
22728 edges to their destination block.
22730 2017-01-27 Richard Biener <rguenther@suse.de>
22732 PR tree-optimization/79244
22733 * tree-vrp.c (remove_range_assertions): Forcefully propagate
22734 out SSA names even if abnormal.
22736 2017-01-27 Jakub Jelinek <jakub@redhat.com>
22738 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
22739 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
22740 instead of MPFR_RNDN.
22742 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
22745 * arm.c (arm_option_override): Don't call build_target_option_node
22746 until after doing all option overrides.
22747 (arm_valid_target_attribute_tree): Likewise.
22749 2017-01-27 Martin Liska <mliska@suse.cz>
22751 * doc/invoke.texi (-fprofile-arcs): Document profiling support
22752 for {cd}tors and C++ {cd}tors.
22754 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
22756 * config/s390/s390.md ("*setmem_long_and")
22757 ("*setmem_long_and_31z"): Use zero_extend instead of and.
22759 2017-01-26 Martin Sebor <msebor@redhat.com>
22761 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
22764 2017-01-26 Martin Sebor <msebor@redhat.com>
22766 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
22767 HAVE_DFmode before using XFmode or DFmode.
22768 (parse_directive): Avoid using the z length modifier to avoid
22769 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
22771 PR middle-end/78703
22772 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
22773 to accept adjustment as an array.
22774 (get_int_range): New function.
22775 (struct directive): Make width and prec arrays.
22776 (directive::set_width, directive::set_precision): Call get_int_range.
22777 (format_integer, format_floating): Handle width and precision ranges.
22778 (format_string, parse_directive): Same.
22780 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22783 * dwarf2out.c (generate_skeleton_bottom_up): For children with
22784 comdat_type_p set, just clone them, but keep the children in the
22788 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
22789 which have direct callers with -fvar-tracking-assignments enabled
22791 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
22792 inside of type units.
22794 2017-01-26 Martin Sebor <msebor@redhat.com>
22796 PR middle-end/78703
22797 * gimple-ssa-sprintf.c (struct result_range): Add likely and
22799 (struct format_result): Replace number_chars, number_chars_min,
22800 and number_chars_max with a single member of struct result_range.
22802 (format_result::operator+=): Adjust.
22803 (struct fmtresult): Remove bounded. Handle likely and unlikely
22805 (fmtresult::adjust_for_width_or_precision): New function.
22806 (fmtresult:type_max_digits): New function.
22807 (bytes_remaining): Handle likely and unlikely counters.
22808 (min_bytes_remaining): Remove.
22809 (format_percent): Simplify.
22810 (format_integer, format_floating): Set likely and unlikely counters.
22811 (get_string_length, format_character, format_string): Same.
22812 (format_plain, should_warn_p): New function.
22813 (maybe_warn): Call should_warn_p. Update diagnostic messages
22814 and handle those for all directives, including plain strings.
22815 (format_directive): Handle likely and unlikely counters.
22816 Remove unnecessary quoting from diagnostics. Add an informational
22818 (add_bytes): Remove.
22819 (pass_sprintf_length::compute_format_length): Simplify.
22820 (try_substitute_return_value): Handle likely and unlikely counters.
22822 2017-01-26 Carl Love <cel@us.ibm.com>
22824 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
22825 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
22827 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
22830 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
22831 endianess for subregs into account.
22832 * lra-constraints.c (lra_constraints): Do risky transformations
22833 always on the first iteration.
22834 * lra-lives.c (check_pseudos_live_through_calls): Add arg
22835 last_call_used_reg_set.
22836 (process_bb_lives): Define and use last_call_used_reg_set.
22837 * lra.c (lra): Always continue after lra_constraints on the first
22840 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
22842 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
22844 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
22846 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22848 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
22849 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
22850 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
22851 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
22852 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
22853 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
22854 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
22855 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
22856 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
22858 2017-01-26 Marek Polacek <polacek@redhat.com>
22861 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
22862 for the third operand.
22864 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22866 PR middle-end/79236
22867 * omp-low.c (struct omp_context): Add simt_stmt field.
22868 (scan_omp_for): Return omp_context *.
22869 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
22870 context to the _simt_ SIMD stmt.
22871 (lower_omp_for): For combined SIMD with sibling _simt_
22872 SIMD, make sure to use the same decls in _looptemp_
22873 clauses as in the sibling.
22875 2017-01-26 David Sherwood <david.sherwood@arm.com>
22877 PR middle-end/79212
22878 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
22881 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22884 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
22885 emit fld b; fld a; if possible.
22887 * brig-builtins.def: Update copyright years.
22888 * config/arm/arm_acle_builtins.def: Update copyright years.
22890 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
22893 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
22894 constraint instead of o for the stxsd instruction.
22896 2017-01-25 Carl Love <cel@us.ibm.com>
22898 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
22899 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
22901 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
22903 * doc/invoke.texi (C++ Dialect Options): Fix typo.
22905 2017-01-25 Richard Biener <rguenther@suse.de>
22907 PR tree-optimization/69264
22908 * target.def (vector_alignment_reachable): Improve documentation.
22909 * doc/tm.texi: Regenerate.
22910 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
22912 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
22913 earlier changes with respect to TYPE_USER_ALIGN.
22914 (vector_alignment_reachable_p): Likewise. Improve dumping.
22916 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22919 * config/arm/arm.md (xordi3): Force constant operand into a register
22922 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22924 * doc/invoke.texi (-fstore-merging): Correct default optimization
22925 levels at which it is enabled.
22926 (-O): Move -fstore-merging from list to...
22929 2017-01-25 Richard Biener <rguenther@suse.de>
22932 * omp-expand.c: Include debug.h.
22933 (expand_omp_taskreg): Make sure to generate early debug before
22934 outlining anything from a function.
22935 (expand_omp_target): Likewise.
22936 (grid_expand_target_grid_body): Likewise.
22938 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
22941 * asan.c (get_translation_unit_decl): New function.
22942 (asan_add_global): Extract modules file name from globals
22943 TRANSLATION_UNIT_DECL name.
22945 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
22948 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
22949 for long calls with APCS frame and VFP.
22951 2017-01-24 David Malcolm <dmalcolm@redhat.com>
22953 * cfg.c (original_copy_tables_initialized_p): New function.
22954 * cfg.h (original_copy_tables_initialized_p): New decl.
22955 * cfgrtl.c (relink_block_chain): Guard the call to
22956 free_original_copy_tables with a call to
22957 original_copy_tables_initialized_p.
22958 * cgraph.h (symtab_node::native_rtl_p): New decl.
22959 * cgraphunit.c (symtab_node::native_rtl_p): New function.
22960 (symtab_node::needed_p): Don't assert for early assembly output
22961 for __RTL functions.
22962 (cgraph_node::finalize_function): Set "force_output" for __RTL
22964 (cgraph_node::analyze): Bail out early for __RTL functions.
22965 (analyze_functions): Update assertion to support __RTL functions.
22966 (cgraph_node::expand): Bail out early for __RTL functions.
22967 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
22969 * function.h (struct function): Update comment for field
22971 * gimple-expr.c: Include "tree-pass.h".
22972 (gimple_has_body_p): Return false for __RTL functions.
22973 * Makefile.in (OBJS): Add run-rtl-passes.o.
22974 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
22976 (gcc::pass_manager::get_clean_slate): New accessor.
22977 * passes.c: Include "insn-addr.h".
22978 (should_skip_pass_p): Add logging. Update logic for running
22979 "expand" to be compatible with both __GIMPLE and __RTL. Guard
22980 property-provider override so it is only done for gimple passes.
22982 (skip_pass): New function.
22983 (execute_one_pass): Call skip_pass when skipping passes.
22984 * read-md.c (md_reader::read_char): Support filtering
22985 the input to a subset of line numbers.
22986 (md_reader::md_reader): Initialize fields
22987 m_first_line and m_last_line.
22988 (md_reader::read_file_fragment): New function.
22989 * read-md.h (md_reader::read_file_fragment): New decl.
22990 (md_reader::m_first_line): New field.
22991 (md_reader::m_last_line): New field.
22992 * read-rtl-function.c (function_reader::create_function): Only
22993 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
22994 curr_properties. Set DECL_INITIAL to a dummy block.
22995 (read_rtl_function_body_from_file_range): New function.
22996 * read-rtl-function.h (read_rtl_function_body_from_file_range):
22998 * run-rtl-passes.c: New file.
22999 * run-rtl-passes.h: New file.
23001 2017-01-24 Jeff Law <law@redhat.com>
23003 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
23006 2017-01-24 Bin Cheng <bin.cheng@arm.com>
23008 PR tree-optimization/79159
23009 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
23010 (record_nonwrapping_iv): Improve boundary using above function if no
23011 value range information.
23013 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
23014 Martin Jambor <mjambor@suse.cz>
23016 * brig-builtins.def: New file.
23017 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
23018 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
23019 (DEF_HSAIL_SAT_BUILTIN): Likewise.
23020 (DEF_HSAIL_INTR_BUILTIN): Likewise.
23021 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
23022 * builtin-types.def (BT_INT8): New.
23023 (BT_INT16): Likewise.
23024 (BT_UINT8): Likewise.
23025 (BT_UINT16): Likewise.
23026 (BT_FN_ULONG): Likewise.
23027 (BT_FN_UINT_INT): Likewise.
23028 (BT_FN_UINT_ULONG): Likewise.
23029 (BT_FN_UINT_LONG): Likewise.
23030 (BT_FN_UINT_PTR): Likewise.
23031 (BT_FN_ULONG_PTR): Likewise.
23032 (BT_FN_INT8_FLOAT): Likewise.
23033 (BT_FN_INT16_FLOAT): Likewise.
23034 (BT_FN_UINT32_FLOAT): Likewise.
23035 (BT_FN_UINT16_FLOAT): Likewise.
23036 (BT_FN_UINT8_FLOAT): Likewise.
23037 (BT_FN_UINT64_FLOAT): Likewise.
23038 (BT_FN_UINT16_UINT32): Likewise.
23039 (BT_FN_UINT32_UINT16): Likewise.
23040 (BT_FN_UINT16_UINT16_UINT16): Likewise.
23041 (BT_FN_INT_PTR_INT): Likewise.
23042 (BT_FN_UINT_PTR_UINT): Likewise.
23043 (BT_FN_LONG_PTR_LONG): Likewise.
23044 (BT_FN_ULONG_PTR_ULONG): Likewise.
23045 (BT_FN_VOID_UINT64_UINT64): Likewise.
23046 (BT_FN_UINT8_UINT8_UINT8): Likewise.
23047 (BT_FN_INT8_INT8_INT8): Likewise.
23048 (BT_FN_INT16_INT16_INT16): Likewise.
23049 (BT_FN_INT_INT_INT): Likewise.
23050 (BT_FN_UINT_FLOAT_UINT): Likewise.
23051 (BT_FN_FLOAT_UINT_UINT): Likewise.
23052 (BT_FN_ULONG_UINT_UINT): Likewise.
23053 (BT_FN_ULONG_UINT_PTR): Likewise.
23054 (BT_FN_ULONG_ULONG_ULONG): Likewise.
23055 (BT_FN_UINT_UINT_UINT): Likewise.
23056 (BT_FN_VOID_UINT_PTR): Likewise.
23057 (BT_FN_UINT_UINT_PTR: Likewise.
23058 (BT_FN_UINT32_UINT64_PTR): Likewise.
23059 (BT_FN_INT_INT_UINT_UINT): Likewise.
23060 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
23061 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
23062 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
23063 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
23064 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
23065 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
23066 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
23067 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
23068 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
23069 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
23070 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
23071 * doc/frontends.texi: List BRIG FE.
23072 * doc/install.texi (Testing): Add BRIG tesring requirements.
23073 * doc/invoke.texi (Overall Options): Mention BRIG.
23074 * doc/standards.texi (Standards): Doucment BRIG HSA version.
23076 2017-01-24 Richard Biener <rguenther@suse.de>
23078 PR translation/79208
23079 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
23081 2017-01-24 Martin Jambor <mjambor@suse.cz>
23084 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
23085 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
23086 and known_contexts.
23088 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
23090 PR middle-end/79123
23091 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
23092 casts from signed to unsigned really don't have a range.
23094 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
23096 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
23097 GMP_RNDx for compatiblity.
23099 2017-01-24 Martin Liska <mliska@suse.cz>
23102 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
23103 that would prevent us to call alloca with -1 as argument.
23105 2017-01-24 Jakub Jelinek <jakub@redhat.com>
23107 * dwarf2out.c (output_compilation_unit_header, output_file_names):
23108 Avoid -Wformat-security warning.
23110 2017-01-23 Andrew Pinski <apinski@cavium.com>
23112 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
23115 2017-01-23 Martin Sebor <msebor@redhat.com>
23117 PR middle-end/78703
23118 * gimple-ssa-sprintf.c (warn_level): New global.
23119 (format_integer): Use it here and throughout the rest of the file.
23120 Use the same switch to compute sign as base.
23121 (maybe_warn): New function.
23122 (format_directive): Factor out warnings into maybe_warn.
23123 Add debugging output. Use warn_level.
23124 (add_bytes): Use warn_level.
23125 (pass_sprintf_length::compute_format_length): Add debugging output.
23126 (try_substitute_return_value): Same.
23127 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
23129 PR middle-end/78703
23130 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
23131 (struct fmtresult, format_integer, format_floating): Adjust.
23132 (fmtresult::fmtresult): Set max correctly in two argument ctor.
23133 (get_string_length, format_string,format_directive): Same.
23134 (pass_sprintf_length::compute_format_length): Same.
23135 (try_substitute_return_value): Simplify slightly.
23137 PR middle-end/78703
23138 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
23139 (fmtresult::operator+=): Outlined.
23140 (struct fmtresult): Add ctors.
23141 (struct conversion_spec): Rename...
23142 (struct directive): ...to this. Add and remove data members.
23143 (directive::set_width, directive::set_precision): New functions.
23144 (format_percent): Use fmtresult ctor.
23145 (get_width_and_precision): Remove.
23146 (format_integer): Make naming changes. Avoid computing width and
23148 (format_floating): Same. Adjust indentation.
23149 (format_character, format_none): New functions.
23150 (format_string): Moved character handling to format_character.
23151 (format_directive): Remove arguments, change return type.
23152 (parse_directive): New function.
23153 (pass_sprintf_length::compute_format_length): Move directive
23154 parsing to parse_directive.
23156 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23158 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
23159 (assign_assembler_name_if_needed): ... this.
23160 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
23161 (assign_assembler_name_if_needed): ... this.
23162 (free_lang_data_in_cgraph): Adjust callers.
23163 * cgraphunit.c (cgraph_node::analyze): Likewise.
23164 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
23167 2017-01-23 Richard Biener <rguenther@suse.de>
23169 PR tree-optimization/79088
23170 PR tree-optimization/79188
23171 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
23172 resetting loop bounds after last path deletion. Reset loop
23173 bounds of the target loop, make code match the comments.
23174 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
23175 Make sure loops need no fixups.
23177 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23179 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
23180 exponent support with double type for first argument.
23181 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
23182 type returned by __builtin_vec_extract_sig,
23183 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
23184 functions from "vector int" to "vector unsigned int" or from
23185 "vector long long int" to "vector unsigned long long int".
23186 Changed type returned by __builtin_vec_extract_exp,
23187 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
23188 functions from "vector int" to "vector unsigned int" or from
23189 "vector long long int" to "vector unsigned long long int".
23190 Changed return type of __builtin_vec_test_data_class,
23191 __builtin_vec_test_data_class_sp, and
23192 __builtin_vec_test_data_class_dp from "vector int" to
23193 "vector bool int" or from "vector long long int" to "vector bool
23194 long long int" and changed second argument type from "unsigned
23195 int" to "int". Added new overloaded function forms "vector float
23196 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
23197 "vector float __builtin_vec_insert_exp_sp (vector float, vector
23198 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
23199 double, vector unsigned long long int)" and "vector double
23200 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
23201 long int)". Changed return type of
23202 __builtin_scalar_test_data_class and
23203 __builtin_scalar_test_data_class_sp and
23204 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
23205 int" and changed second argument from "unsigned int" to "int".
23206 Changed type returned by __builtin_scalar_test_neg,
23207 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
23208 from "int" to "bool int". Added new overloaded function form
23209 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
23210 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
23211 exponent double-precision with floating point first argument.
23212 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
23213 documentation of scalar_test_data_class, scalar_test_neg,
23214 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
23215 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
23216 vec_test_data_class built-in functions to reflect refinements in
23217 their type signatures.
23219 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
23221 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
23223 (aarch64_elf_asm_destructor): Likewise.
23225 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
23227 PR rtl-optimization/78634
23228 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
23229 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
23230 * ifcvt.c (noce_try_cmove): Add missing cost check.
23232 PR rtl-optimization/71724
23233 * combine.c (if_then_else_cond): Look for situations where it is
23234 beneficial to undo the work of one of the recursive calls.
23236 2017-01-23 Bin Cheng <bin.cheng@arm.com>
23238 PR tree-optimization/70754
23239 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
23240 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
23241 combined stmt before it if not NULL.
23242 (combine_chains): Process refs reversely and compute dominance point
23245 2017-01-23 Martin Liska <mliska@suse.cz>
23247 PR tree-optimization/79196
23248 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
23249 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
23251 (strlen_optimize_stmt): Call the renamed function.
23253 2017-01-23 Michael Matz <matz@suse.de>
23255 PR tree-optimization/78384
23256 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
23258 2017-01-23 Richard Biener <rguenther@suse.de>
23260 PR tree-optimization/79186
23261 * tree-vrp.c (register_new_assert_for): Make sure we've seen
23262 both incoming edges before moving an assert.
23264 2017-01-23 Martin Jambor <mjambor@suse.cz>
23266 * ipa-prop.c (load_from_param_1): Removed.
23267 (load_from_unmodified_param): Bits from load_from_param_1 put back
23269 (load_from_param): Removed.
23270 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
23271 with stmt. Reverted back to use of load_from_unmodified_param.
23273 2017-01-23 Martin Jambor <mjambor@suse.cz>
23276 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
23277 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
23278 field a pointer to garbage collected vector, mark lattices and
23279 ipcp_orig_node with GTY((skip)).
23280 (ipa_get_param_count): Adjust to descriptors being a pointer.
23281 (ipa_get_param): Likewise.
23282 (ipa_get_type): Likewise.
23283 (ipa_get_param_move_cost): Likewise.
23284 (ipa_set_param_used): Likewise.
23285 (ipa_get_controlled_uses): Likewise.
23286 (ipa_set_controlled_uses): Likewise.
23287 (ipa_is_param_used): Likewise.
23288 (ipa_node_params_t): Move into garbage collector. New methods insert
23290 (ipa_node_params_sum): Annotate wth GTY(()).
23291 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
23293 (ipa_load_from_parm_agg): Adjust declaration.
23294 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
23295 * ipa-profile.c (ipa_profile): Likewise.
23296 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
23297 (ipa_populate_param_decls): Make descriptors parameter garbage
23299 (ipa_dump_param): Adjust to descriptors being a pointer.
23300 (ipa_alloc_node_params): Likewise.
23301 (ipa_initialize_node_params): Likewise.
23302 (load_from_param_1): Make descriptors parameter garbage collected.
23303 (load_from_unmodified_param): Likewise.
23304 (load_from_param): Likewise.
23305 (ipa_load_from_parm_agg): Likewise.
23306 (ipa_node_params::~ipa_node_params): Removed.
23307 (ipa_free_all_node_params): Remove call to delete operator.
23308 (ipa_node_params_t::insert): New.
23309 (ipa_node_params_t::remove): Likewise.
23310 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
23311 copy known_csts and known_contexts vectors.
23312 (ipa_read_node_info): Adjust to descriptors being a pointer.
23313 (ipcp_modif_dom_walker): Make m_descriptors field garbage
23315 (ipcp_transform_function): Make descriptors variable garbage
23318 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
23320 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
23321 * config/i386/avx512dqintrin.h: Ditto.
23322 * config/i386/avx512fintrin.h: Ditto.
23323 * config/i386/i386.c: Handle new builtins.
23324 * config/i386/i386-builtin.def: Add new builtins.
23325 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
23326 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
23328 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23329 Martin Liska <mliska@suse.cz>
23331 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
23332 * asan.c (asan_expand_poison_ifn): Support stores and use
23333 appropriate ASAN report function.
23334 * internal-fn.c (expand_ASAN_POISON_USE): New function.
23335 * internal-fn.def (ASAN_POISON_USE): Declare.
23336 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
23337 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
23338 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
23339 ASAN_POISON calls w/o LHS.
23340 * tree-ssa.c (execute_update_addresses_taken): Create clobber
23341 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
23342 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
23343 * gimplify.c (asan_poison_variables): Add attribute
23344 use_after_scope_memory to variables that really needs to live
23346 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
23347 having the attribute.
23349 2017-01-23 Martin Liska <mliska@suse.cz>
23351 * asan.c (create_asan_shadow_var): New function.
23352 (asan_expand_poison_ifn): Likewise.
23353 * asan.h (asan_expand_poison_ifn): New declaration.
23354 * internal-fn.c (expand_ASAN_POISON): Likewise.
23355 * internal-fn.def (ASAN_POISON): New builtin.
23356 * sanopt.c (pass_sanopt::execute): Expand
23357 asan_expand_poison_ifn.
23358 * tree-inline.c (copy_decl_for_dup_finish): Make function
23360 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
23361 * tree-ssa.c (is_asan_mark_p): New function.
23362 (execute_update_addresses_taken): Rewrite local variables
23363 (identified just by use-after-scope as addressable) into SSA.
23365 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23367 * doc/install.texi (Specific): opensource.apple.com uses https
23368 now. Remove trailing slash.
23370 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23372 * README.Portability: Remove note on an Irix compatibility issue.
23374 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
23376 * gcov.c (INCLUDE_ALGORITHM): Define.
23377 (INCLUDE_VECTOR): Define.
23378 No longer include <vector> and <algorithm> directly.
23380 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
23382 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
23384 * doc/invoke.texi (Code Gen Options): Ditto.
23386 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
23389 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
23391 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
23393 rtl-optimization/79125
23394 * cprop.c (local_cprop_pass): Handle cases where we make an
23395 unconditional trap.
23397 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
23401 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
23402 read from, for big endian.
23404 2017-01-20 Jiong Wang <jiong.wang@arm.com>
23406 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
23407 register pauth builtins for LP64 only.
23409 2017-01-20 Marek Polacek <polacek@redhat.com>
23412 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
23415 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23417 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
23419 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
23420 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
23421 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
23423 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23426 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
23427 in big-endian mode when they are not a single duplicated value.
23429 2017-01-20 Richard Biener <rguenther@suse.de>
23431 * BASE-VER: Bump to 7.0.1.
23433 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23435 * omp-low.c (omplow_simd_context): New struct. Use it...
23436 (lower_rec_simd_input_clauses): ...here and...
23437 (lower_rec_input_clauses): ...here to hold common data. Adjust all
23438 references to idx, lane, max_vf, is_simt.
23440 2017-01-20 Graham Markall <graham.markall@embecosm.com>
23442 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
23445 2017-01-20 Martin Jambor <mjambor@suse.cz>
23447 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
23448 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
23450 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
23451 (GTFILES): Rename hsa.c to hsa-common.c.
23452 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
23453 * hsa-dump.c: Likewise.
23454 * hsa-gen.c: Likewise.
23455 * hsa-regalloc.c: Likewise.
23456 * ipa-hsa.c: Likewise.
23457 * omp-expand.c: Likewise.
23458 * omp-low.c: Likewise.
23459 * toplev.c: Likewise.
23461 2017-01-20 Marek Polacek <polacek@redhat.com>
23464 * doc/invoke.texi: Document -Wduplicated-branches.
23465 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
23466 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
23467 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
23468 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
23469 return 0 only when not OEP_LEXICOGRAPHIC.
23470 (fold_build_cleanup_point_expr): Use the expression
23471 location when building CLEANUP_POINT_EXPR.
23472 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
23473 * tree.c (add_expr): Handle error_mark_node.
23475 2017-01-20 Martin Liska <mliska@suse.cz>
23478 * tree-profile.c (init_ic_make_global_vars): Do not call
23480 (gimple_init_gcov_profiler): Likewise.
23482 2017-01-20 Martin Liska <mliska@suse.cz>
23485 * cgraph.h (maybe_create_reference): Remove argument and
23487 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
23489 * ipa-cp.c (create_specialized_node): Likewise.
23490 * symtab.c (symtab_node::maybe_create_reference): Handle
23491 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
23493 2017-01-20 Martin Liska <mliska@suse.cz>
23495 * read-rtl-function.c (function_reader::create_function): Use
23496 build_decl instread of build_decl_stat.
23498 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
23500 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
23501 * config/i386/avx512dqintrin.h: Ditto.
23502 * config/i386/avx512fintrin.h: Ditto.
23503 * config/i386/i386-builtin-types.def: Add new types.
23504 * config/i386/i386.c: Handle new types.
23505 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
23506 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
23507 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
23508 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
23509 (__builtin_ia32_kshiftridi): New.
23510 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
23512 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
23516 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
23517 define to rs6000_init_stack_protect_guard.
23518 (rs6000_init_stack_protect_guard): New function.
23520 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23521 Yunqiang Su <yunqiang.su@imgtec.com>
23523 * config.gcc (supported_defaults): Add madd4.
23524 (with_madd4): Add validation.
23525 (all_defaults): Add madd4.
23526 * config/mips/mips.opt (mmadd4): New option.
23527 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23529 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23531 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
23532 (ISA_HAS_FUSED_MADD4): Likewise.
23533 * doc/invoke.texi (-mmadd4): Document the new option.
23534 * doc/install.texi (--with-madd4): Document the new option.
23536 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23538 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
23539 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
23540 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
23541 (aarch64_init_pauth_hint_builtins): New.
23542 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
23543 (aarch64_expand_builtin): Expand new builtins.
23545 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23547 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
23548 * combine-stack-adj.c (no_unhandled_cfa): Handle
23549 REG_CFA_TOGGLE_RA_MANGLE.
23550 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
23551 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
23552 info for return address signing.
23553 (aarch64_expand_epilogue): Likewise.
23555 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23557 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
23558 * config/aarch64/aarch64-protos.h
23559 (aarch64_return_address_signing_enabled): New declaration.
23560 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
23562 (aarch64_expand_prologue): Sign return address before it's pushed onto
23564 (aarch64_expand_epilogue): Authenticate return address fetched from
23566 (aarch64_override_options): Sanity check for ILP32 and ISA level.
23567 (aarch64_attributes): New function attributes for "sign-return-address".
23568 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
23569 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
23570 ("*do_return"): Generate combined instructions according to key index.
23571 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
23572 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
23574 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
23575 * config/aarch64/aarch64.opt (msign-return-address=): New.
23576 * doc/extend.texi (AArch64 Function Attributes): Documents
23577 "sign-return-address=".
23578 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
23580 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23582 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
23583 overall option summary.
23585 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23587 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
23588 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
23589 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
23590 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
23592 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
23594 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
23595 -mpower9-minmax by default for -mcpu=power9.
23596 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
23597 128-bit floating point.
23599 2017-01-20 Alan Modra <amodra@gmail.com>
23601 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
23602 optimizing for size.
23604 2017-01-20 Alan Modra <amodra@gmail.com>
23607 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
23608 for strcmp and strncmp from corresponding builtin decl.
23610 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23612 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
23613 instead of i386/rtems-64.h.
23614 * config/i386/rtems-64.h: Remove.
23616 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23620 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
23622 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
23624 2017-01-19 Tamar Christina <tamar.christina@arm.com>
23626 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
23627 Change int to HOST_WIDE_INT.
23628 * config/aarch64/aarch64-protos.h
23629 (aarch64_simd_gen_const_vector_dup): Likewise.
23630 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
23632 2017-01-19 David Malcolm <dmalcolm@redhat.com>
23634 * langhooks-def.h (lhd_type_for_size): New decl.
23635 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
23636 * langhooks.c (lhd_type_for_size): New function, taken from
23639 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
23641 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
23642 define_bypass for CR latency.
23643 (power9-cracked-alu): Update bypass latency and remove power9-branch.
23644 (power9-alu2): Add define_bypass for CR latency.
23646 (power9-mul): Update insn latency.
23647 (power9-mul-compare): Update insn latency, bypass latency and remove
23650 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23652 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
23654 * config/aarch64/aarch64.md
23655 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
23656 aarch64_nopcrelative_literal_loads.
23657 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
23659 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
23661 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
23662 TARGET_LOONGSON_3A.
23663 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
23665 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
23668 * config.gcc (supported_defaults): Add lxc1-sxc1.
23669 (with_lxc1_sxc1): Add validation.
23670 (all_defaults): Add lxc1-sxc1.
23671 * config/mips/mips.opt (mlxc1-sxc1): New option.
23672 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23674 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23675 __mips_no_lxc1_sxc1.
23676 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
23677 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
23678 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
23680 2017-01-19 Richard Biener <rguenther@suse.de>
23682 PR tree-optimization/72488
23683 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
23684 sure to restore SSA info.
23685 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
23687 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
23689 PR rtl-optimization/79121
23690 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
23691 of the inner type when shifting an extended value.
23693 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
23696 * symtab.c (symtab_node::equal_address_to): Fix comparing of
23697 interposable aliases.
23699 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
23702 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
23703 Use the evmergelohi instruction.
23704 (mov_si<mode>_e500_subreg4_2_le): Likewise.
23705 (mov_sitf_e500_subreg8_2_be): Likewise.
23706 (mov_sitf_e500_subreg12_2_le): Likewise.
23707 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
23708 (mov_si<mode>_e500_subreg4_2_be): Likewise.
23709 (mov_sitf_e500_subreg8_2_le): Likewise.
23710 (mov_sitf_e500_subreg12_2_be): Likewise.
23712 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23714 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
23715 attribute from vecsimple to vecperm.
23716 (altivec_vbpermq2): Likewise.
23718 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23721 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
23723 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23724 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
23725 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
23726 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
23727 case where N arg is SIZE_MAX.
23728 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
23729 (cmpstrsi): Add pattern.
23731 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23733 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23734 __builtin_vec_revb builtins.
23735 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
23736 built-in functions to support generation of the ISA 3.0 XXBR<x>
23737 vector byte reverse instructions.
23738 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
23739 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
23740 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
23741 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
23742 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
23743 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
23744 (P9V_BUILTIN_VEC_REVB): Likewise.
23745 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
23746 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
23747 (p9_xxbrq_v16qi): Likewise.
23748 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
23749 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
23750 (p9_xxbrh_v8hi): Likewise.
23751 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
23752 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
23753 vec_revb built-in functions.
23755 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
23757 PR rtl-optimization/78952
23758 * config/i386/i386.md (any_extract): New code iterator.
23759 (*insvqi_2): Use any_extract for source operand.
23760 (*insvqi_3): Use any_shiftrt for source operand.
23762 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
23764 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
23766 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
23768 2017-01-18 Matthias Klose <doko@ubuntu.com>
23770 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
23772 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23774 * config/rs6000/altivec.h (vec_bperm): Change #define.
23775 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
23776 (altivec_vbpermq2): New define_insn.
23777 (altivec_vbpermd): Likewise.
23778 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
23779 function interface.
23780 (VBPERMD): Likewise.
23781 (VBPERM): New polymorphic function interface.
23782 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
23783 Add entries for P9V_BUILTIN_VEC_VBPERM.
23784 * doc/extend.texi: Add interfaces for vec_bperm.
23786 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23788 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
23789 first letter of error messages.
23790 (s390_resolve_overloaded_builtin): Likewise.
23791 * config/s390/s390.c (s390_expand_builtin): Likewise.
23792 (s390_invalid_arg_for_unprototyped_fn): Likewise.
23793 (s390_valid_target_attribute_inner_p): Likewise.
23794 * config/s390/s390.md ("tabort"): Likewise.
23796 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
23798 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
23799 (ISA_AVOID_DIV_HILO): New macro.
23800 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
23801 (ISA_HAS_DDIV): Likewise.
23803 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23805 * doc/invoke.texi (fabi-version): Correct number of occurrences.
23807 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23809 * doc/invoke.texi (fabi-version): Spelling fix.
23811 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23814 * doc/invoke.texi (fabi-version): Mention mangling fix for
23817 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23820 * doc/invoke.texi (fabi-version): Document discriminator mangling.
23822 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
23825 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
23826 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
23828 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
23830 (stack_protect_test): Ditto.
23831 * config/rs6000/rs6000.opt (mstack-protector-guard=,
23832 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
23834 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
23835 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
23836 -mstack-protector-guard-offset=.
23837 (RS/6000 and PowerPC Options): Ditto.
23839 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
23841 * config/i386/i386.h (MASK_CLASS_P): New define.
23842 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
23843 there are no registers from different register sets also when
23844 mask registers are used. Update function comment.
23845 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
23846 to (*k/*r) and (*k/*km) alternatives.
23848 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
23850 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
23851 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
23852 (EH_RETURN_HANDLER_RTX): New define.
23853 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
23854 Force frame pointer in EH return functions.
23855 (aarch64_expand_epilogue): Add barrier for eh_return.
23856 (aarch64_final_eh_return_addr): Remove.
23857 (aarch64_eh_return_handler_rtx): New function.
23858 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
23860 (aarch64_eh_return_handler_rtx): New prototype.
23862 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23864 * config/rs6000/altivec.h (vec_rlmi): New #define.
23865 (vec_vrlnm): Likewise.
23866 (vec_rlnm): Likewise.
23867 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
23868 (UNSPEC_VRLNM): Likewise.
23869 (VIlong): New mode iterator.
23870 (altivec_vrl<VI_char>mi): New define_insn.
23871 (altivec_vrl<VI_char>nm): Likewise.
23872 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
23874 (VRLDNM): Likewise.
23875 (RLNM): New polymorphic function entry.
23876 (VRLWMI): New monomorphic function entry.
23877 (VRLDMI): Likewise.
23878 (RLMI): New polymorphic function entry.
23879 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
23880 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
23881 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
23884 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23887 * dwarf2out.c (field_byte_offset): Restore the
23888 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
23889 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
23890 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
23893 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
23896 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
23898 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23901 * dwarf2out.c (add_data_member_location_attribute): For constant
23902 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
23903 instead of DW_AT_data_member_location, DW_AT_bit_offset and
23904 DW_AT_byte_size attributes.
23906 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
23908 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
23909 after forcing to constant memory when the code model is medium.
23911 2017-01-17 Julia Koval <julia.koval@intel.com>
23914 * config/i386/avx512fintrin.h
23915 (_mm512_i32gather_ps): Change __addr type to void const*.
23916 (_mm512_mask_i32gather_ps): Ditto.
23917 (_mm512_i32gather_pd): Ditto.
23918 (_mm512_mask_i32gather_pd): Ditto.
23919 (_mm512_i64gather_ps): Ditto.
23920 (_mm512_mask_i64gather_ps): Ditto.
23921 (_mm512_i64gather_pd): Ditto.
23922 (_mm512_mask_i64gather_pd): Ditto.
23923 (_mm512_i32gather_epi32): Ditto.
23924 (_mm512_mask_i32gather_epi32): Ditto.
23925 (_mm512_i32gather_epi64): Ditto.
23926 (_mm512_mask_i32gather_epi64): Ditto.
23927 (_mm512_i64gather_epi32): Ditto.
23928 (_mm512_mask_i64gather_epi32): Ditto.
23929 (_mm512_i64gather_epi64): Ditto.
23930 (_mm512_mask_i64gather_epi64): Ditto.
23931 (_mm512_i32scatter_ps): Change __addr type to void*.
23932 (_mm512_mask_i32scatter_ps): Ditto.
23933 (_mm512_i32scatter_pd): Ditto.
23934 (_mm512_mask_i32scatter_pd): Ditto.
23935 (_mm512_i64scatter_ps): Ditto.
23936 (_mm512_mask_i64scatter_ps): Ditto.
23937 (_mm512_i64scatter_pd): Ditto.
23938 (_mm512_mask_i64scatter_pd): Ditto.
23939 (_mm512_i32scatter_epi32): Ditto.
23940 (_mm512_mask_i32scatter_epi32): Ditto.
23941 (_mm512_i32scatter_epi64): Ditto.
23942 (_mm512_mask_i32scatter_epi64): Ditto.
23943 (_mm512_i64scatter_epi32): Ditto.
23944 (_mm512_mask_i64scatter_epi32): Ditto.
23945 (_mm512_i64scatter_epi64): Ditto.
23946 (_mm512_mask_i64scatter_epi64): Ditto.
23947 * config/i386/avx512pfintrin.h
23948 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
23949 (_mm512_mask_prefetch_i32gather_ps): Ditto.
23950 (_mm512_mask_prefetch_i64gather_pd): Ditto.
23951 (_mm512_mask_prefetch_i64gather_ps): Ditto.
23952 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
23953 (_mm512_prefetch_i32scatter_ps): Ditto.
23954 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
23955 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
23956 (_mm512_prefetch_i64scatter_pd): Ditto.
23957 (_mm512_prefetch_i64scatter_ps): Ditto.
23958 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
23959 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
23960 * config/i386/avx512vlintrin.h
23961 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
23962 (_mm_mmask_i32gather_ps): Ditto.
23963 (_mm256_mmask_i32gather_pd): Ditto.
23964 (_mm_mmask_i32gather_pd): Ditto.
23965 (_mm256_mmask_i64gather_ps): Ditto.
23966 (_mm_mmask_i64gather_ps): Ditto.
23967 (_mm256_mmask_i64gather_pd): Ditto.
23968 (_mm_mmask_i64gather_pd): Ditto.
23969 (_mm256_mmask_i32gather_epi32): Ditto.
23970 (_mm_mmask_i32gather_epi32): Ditto.
23971 (_mm256_mmask_i32gather_epi64): Ditto.
23972 (_mm_mmask_i32gather_epi64): Ditto.
23973 (_mm256_mmask_i64gather_epi32): Ditto.
23974 (_mm_mmask_i64gather_epi32): Ditto.
23975 (_mm256_mmask_i64gather_epi64): Ditto.
23976 (_mm_mmask_i64gather_epi64): Ditto.
23977 (_mm256_i32scatter_ps): Change __addr type to void*.
23978 (_mm256_mask_i32scatter_ps): Ditto.
23979 (_mm_i32scatter_ps): Ditto.
23980 (_mm_mask_i32scatter_ps): Ditto.
23981 (_mm256_i32scatter_pd): Ditto.
23982 (_mm256_mask_i32scatter_pd): Ditto.
23983 (_mm_i32scatter_pd): Ditto.
23984 (_mm_mask_i32scatter_pd): Ditto.
23985 (_mm256_i64scatter_ps): Ditto.
23986 (_mm256_mask_i64scatter_ps): Ditto.
23987 (_mm_i64scatter_ps): Ditto.
23988 (_mm_mask_i64scatter_ps): Ditto.
23989 (_mm256_i64scatter_pd): Ditto.
23990 (_mm256_mask_i64scatter_pd): Ditto.
23991 (_mm_i64scatter_pd): Ditto.
23992 (_mm_mask_i64scatter_pd): Ditto.
23993 (_mm256_i32scatter_epi32): Ditto.
23994 (_mm256_mask_i32scatter_epi32): Ditto.
23995 (_mm_i32scatter_epi32): Ditto.
23996 (_mm_mask_i32scatter_epi32): Ditto.
23997 (_mm256_i32scatter_epi64): Ditto.
23998 (_mm256_mask_i32scatter_epi64): Ditto.
23999 (_mm_i32scatter_epi64): Ditto.
24000 (_mm_mask_i32scatter_epi64): Ditto.
24001 (_mm256_i64scatter_epi32): Ditto.
24002 (_mm256_mask_i64scatter_epi32): Ditto.
24003 (_mm_i64scatter_epi32): Ditto.
24004 (_mm_mask_i64scatter_epi32): Ditto.
24005 (_mm256_i64scatter_epi64): Ditto.
24006 (_mm256_mask_i64scatter_epi64): Ditto.
24007 (_mm_i64scatter_epi64): Ditto.
24008 (_mm_mask_i64scatter_epi64): Ditto.
24009 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
24010 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
24011 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
24012 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
24013 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
24014 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
24015 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
24016 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
24017 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
24018 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
24019 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
24020 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
24021 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
24022 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
24023 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
24024 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
24025 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
24026 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
24027 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
24028 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
24029 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
24030 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
24031 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
24032 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
24033 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
24034 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
24035 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
24036 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
24037 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
24038 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
24039 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
24040 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
24041 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
24042 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
24043 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
24044 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
24045 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
24046 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
24047 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
24048 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
24049 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
24050 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
24051 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
24052 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
24053 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
24054 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
24055 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
24056 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
24057 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
24058 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
24059 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
24060 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
24061 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
24062 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
24063 definitions accordingly.
24065 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
24066 Kuan-Lin Chen <kuanlinchentw@gmail.com>
24069 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
24072 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
24075 * ira-conflicts.c (ira_build_conflicts): Update total conflict
24076 hard regs for inner regno.
24078 2017-01-17 Martin Liska <mliska@suse.cz>
24081 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
24082 assumption and add comment.
24084 2017-01-17 Nathan Sidwell <nathan@acm.org>
24086 * ipa-visibility.c (localize_node): New function, broken out of ...
24087 (function_and_variable_visibility): ... here. Call it.
24089 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24091 PR middle-end/77445
24092 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
24093 correctly set frequency of oudgoing edge.
24094 (duplicate_thread_path): Fix profile updating.
24096 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24099 * configure.ac: Add GCC_BASE_VER.
24100 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
24101 version from BASE-VER file.
24102 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
24103 (gcc.o): Depend on $(BASEVER).
24104 * common.opt (dumpfullversion): New option.
24105 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
24106 * doc/invoke.texi: Document -dumpfullversion.
24107 * doc/install.texi: Document --with-gcc-major-version-only.
24108 * configure: Regenerated.
24110 2017-01-17 Richard Biener <rguenther@suse.de>
24112 PR tree-optimization/71433
24113 * tree-vrp.c (register_new_assert_for): Merge same asserts
24114 on all incoming edges.
24115 (process_assert_insertions_for): Handle insertions at the
24118 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
24120 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
24121 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
24123 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
24126 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
24129 2017-01-17 Alan Modra <amodra@gmail.com>
24132 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
24133 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
24134 symbolic stack limit when pic.
24136 2017-01-16 Martin Sebor <msebor@redhat.com>
24138 PR tree-optimization/78608
24139 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
24141 2017-01-16 Jeff Law <law@redhat.com>
24144 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
24145 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
24146 for several include directories that may be relative to sysroot.
24147 * config/i386/x-mingw32 (gplus_includedir): Define.
24148 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
24149 (native_system_includedir): Likewise.
24150 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
24151 override if TARGET_SYSTEM_ROOT is defined.
24152 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
24154 PR tree-optimization/79090
24155 PR tree-optimization/33562
24156 PR tree-optimization/61912
24157 PR tree-optimization/77485
24158 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
24159 and computed trims into the dump file.
24161 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24163 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
24165 2017-01-16 Jakub Jelinek <jakub@redhat.com>
24168 * gimplify.c (gimplify_init_constructor): If want_value and
24169 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
24173 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
24174 sequence. Formatting fixes.
24175 (doloop_optimize): Formatting fixes.
24178 * gcc.c (debug_level_greater_than_spec_func): New function.
24179 (static_spec_functions): Add debug-level-gt spec function.
24180 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
24182 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24183 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24184 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
24185 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
24186 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
24187 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
24189 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
24191 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
24192 QImode fixups to general and mask registers only.
24194 2017-01-16 Carl Love <cel@us.ibm.com>
24196 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
24197 for built-in functions
24198 vector signed char vec_nabs (vector signed char)
24199 vector signed short vec_nabs (vector signed short)
24200 vector signed int vec_nabs (vector signed int)
24201 vector signed long long vec_nabs (vector signed long long)
24202 vector float vec_nabs (vector float)
24203 vector double vec_nabs (vector double)
24204 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
24206 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
24207 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
24208 * doc/extend.texi: Update the documentation file for the new built-in
24211 2017-01-16 Martin Sebor <msebor@redhat.com>
24213 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
24216 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24218 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
24219 UNSPEC_VSX__XXSPLTD to require special splat handling.
24221 2017-01-16 David Malcolm <dmalcolm@redhat.com>
24224 * system.h: Poison strndup.
24226 2017-01-16 Alan Modra <amodra@gmail.com>
24229 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
24232 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
24234 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
24236 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
24238 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
24239 call recog here. Assert that INSN_CODE (insn) is non-negative.
24241 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24244 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
24246 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
24247 in the currently scheduled RTL fragment.
24249 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24251 PR rtl-optimization/78751
24252 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
24255 2017-01-14 Jeff Law <law@redhat.com>
24257 PR tree-optimization/79090
24258 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
24259 variable length stores.
24260 (compute_trims): Delete dead assignment to *trim_tail.
24261 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
24264 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
24266 PR rtl-optimization/78626
24267 PR rtl-optimization/78727
24268 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
24269 of a block, and split such blocks after everything else is finished.
24271 2017-01-14 Alan Modra <amodra@gmail.com>
24274 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
24275 target legitimate_combined_insn.
24276 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
24277 (rs6000_legitimate_combined_insn): New function.
24278 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
24280 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
24281 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
24282 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
24284 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
24286 * doc/frontends.texi (G++ and GCC): Remove references to Java.
24288 2017-01-13 Jeff Law <law@redhat.com>
24290 PR tree-optimization/33562
24291 PR tree-optimization/61912
24292 PR tree-optimization/77485
24293 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
24295 (delete_dead_assignment): Likewise.
24296 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
24297 statement to delete_dead_call and delete_dead_assignment.
24299 2017-01-13 David Malcolm <dmalcolm@redhat.com>
24302 * substring-locations.c (format_warning_va): Strengthen case 1 so
24303 that both endpoints of the substring must be within the format
24304 range for just the substring to be printed.
24306 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
24308 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
24309 * config/i386/i386.c (ix86_target_string): Add missing options
24310 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
24311 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
24312 flags_other and ix86_target_other to flags2_other. Display unknown
24314 (ix86_valid_target_attribute_inner_p): Add missing options and
24315 reorder options by implied ISAs, as in ix86_target_string.
24317 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24319 * hash-table.h (hash_table::too_empty_p): New function.
24320 (hash_table::expand): Use it.
24321 (hash_table::traverse): Likewise.
24322 (hash_table::empty_slot): Use sizeof (value_type) instead of
24323 sizeof (PTR) to convert bytes to elements. Shrink the table
24324 if the current size is excessive for the current number of
24327 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24329 * ira-costs.c (record_reg_classes): Break from the inner loop
24330 early once alt_fail is known to be true. Update outer loop
24331 handling accordingly.
24333 2017-01-13 Jeff Law <law@redhat.com>
24335 * tree-ssa-dse.c (decrement_count): New function.
24336 (increment_start_addr, maybe_trim_memstar_call): Likewise.
24337 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
24338 when we know the partially dead statement is a mem* function.
24340 PR tree-optimization/61912
24341 PR tree-optimization/77485
24342 * tree-ssa-dse.c: Include expr.h.
24343 (maybe_trim_constructor_store): New function.
24344 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
24346 PR tree-optimization/33562
24347 PR tree-optimization/61912
24348 PR tree-optimization/77485
24349 * doc/invoke.texi: Document new dse-max-object-size param.
24350 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
24351 * tree-ssa-dse.c: Include params.h.
24352 (dse_store_status): New enum.
24353 (initialize_ao_ref_for_dse): New, partially extracted from
24355 (valid_ao_ref_for_dse, normalize_ref): New.
24356 (setup_live_bytes_from_ref, compute_trims): Likewise.
24357 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
24358 (maybe_trim_partially_dead_store): Likewise.
24359 (maybe_trim_complex_store): Likewise.
24360 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
24361 Track what bytes live from the original store. Return tri-state
24362 for dead, partially dead or live.
24363 (dse_dom_walker): Add constructor, destructor and new private members.
24364 (delete_dead_call, delete_dead_assignment): New extracted from
24366 (dse_optimize_stmt): Make a member of dse_dom_walker.
24367 Use initialize_ao_ref_for_dse.
24369 PR tree-optimization/33562
24370 PR tree-optimization/61912
24371 PR tree-optimization/77485
24372 * sbitmap.h (bitmap_count_bits): Prototype.
24373 (bitmap_clear_range, bitmap_set_range): Likewise.
24374 * sbitmap.c (bitmap_clear_range): New function.
24375 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
24377 2017-01-13 Martin Liska <mliska@suse.cz>
24380 * function.c (set_cfun): Add new argument force.
24381 * function.h (set_cfun): Likewise.
24382 * ipa-inline-transform.c (inline_call): Use the function when
24383 strict alising from is dropped for function we inline to.
24385 2017-01-13 Richard Biener <rguenther@suse.de>
24387 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
24388 for dumping GIMPLE INTEGER_CSTs.
24390 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24392 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
24393 to 201112L since C++17.
24395 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24398 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
24399 if -fsanitize=kernel-address is present.
24401 2017-01-13 Richard Biener <rguenther@suse.de>
24403 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
24404 as _Literal ( type ) number in case usual suffixes do not
24405 preserve all information.
24407 2017-01-13 Richard Biener <rguenther@suse.de>
24409 PR tree-optimization/77283
24410 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
24411 and ssa-iterators.h.
24412 (is_feasible_trace): Implement a cost model based on joiner
24415 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24418 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
24419 char or short to __float128/_Float128 directly.
24421 2017-01-12 Martin Sebor <msebor@redhat.com>
24423 to -Wformat-overflow.
24424 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
24425 (min_bytes_remaining): Same.
24426 (get_string_length): Same.
24427 (format_string): Same.
24428 (format_directive): Same.
24430 (pass_sprintf_length::handle_gimple_call): Same.
24432 2017-01-12 Jakub Jelinek <jakub@redhat.com>
24434 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
24435 info.nowrite calls with no lhs that can't throw. Return bool
24436 whether gsi_remove has been called or not.
24437 (pass_sprintf_length::handle_gimple_call): Return bool whether
24438 try_substitute_return_value called gsi_remove. Formatting fix.
24439 (pass_sprintf_length::execute): Don't use gsi_remove if
24440 handle_gimple_call returned true.
24443 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
24444 be removed due to side-effects, don't remove following barrier nor
24445 turn the successor edge into fallthru edge.
24447 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24450 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
24451 element-reversing loads and stores as not swappable.
24453 2017-01-12 Nathan Sidwell <nathan@acm.org>
24454 Nicolai Stange <nicstange@gmail.com>
24456 * combine.c (try_combine): Don't ignore result of overlap checking
24457 loop. Combine overlap & asm check into single loop.
24459 2017-01-12 Richard Biener <rguenther@suse.de>
24461 * tree-pretty-print.c (dump_generic_node): Provide -gimple
24462 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
24464 2017-01-12 Richard Biener <rguenther@suse.de>
24466 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
24467 and TS_TARGET_OPTION directly derive from TS_BASE.
24468 * tree-core.h (tree_optimization_option): Derive from tree_base.
24469 (tree_target_option): Likewise.
24471 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
24473 * config/i386/i386.c (memory_address_length): Increase len
24474 only when rip_relative_addr_p returns false.
24476 2017-01-11 Julia Koval <julia.koval@intel.com>
24478 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
24479 (OPTION_MASK_ISA_SGX_SET): New.
24480 (ix86_handle_option): Handle OPT_msgx.
24481 * config.gcc: Added sgxintrin.h.
24482 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
24483 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
24484 * config/i386/i386.c (ix86_target_string): Add -msgx.
24486 (ix86_option_override_internal): Handle new options.
24487 (ix86_valid_target_attribute_inner_p): Add sgx.
24488 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
24489 * config/i386/i386.opt: Add msgx.
24490 * config/i386/sgxintrin.h: New file.
24491 * config/i386/x86intrin.h: Add sgxintrin.h.
24493 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24496 * fold-const.c (maybe_nonzero_address): Return 1 for function
24498 (tree_single_nonzero_warnv_p): Don't handle function local objects
24502 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
24505 2017-01-11 David Malcolm <dmalcolm@redhat.com>
24508 * opts.c: Include "spellcheck.h"
24509 (struct string_fragment): New struct.
24510 (struct edit_distance_traits<const string_fragment &>): New
24512 (get_closest_sanitizer_option): New function.
24513 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
24515 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24517 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
24519 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
24520 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
24521 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
24522 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
24523 for initial die_offset if dwarf_split_debug_info.
24524 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
24525 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
24527 (output_skeleton_debug_sections): Formatting fix. Use
24528 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
24529 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
24531 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
24533 * config/arm/cortex-a53.md: Add bypasses for
24534 cortex_a53_r2f_cvt.
24535 (cortex_a53_r2f): Only use for transfers.
24536 (cortex_a53_f2r): Likewise.
24537 (cortex_a53_r2f_cvt): Add reservation for conversions.
24538 (cortex_a53_f2r_cvt): Likewise.
24540 2017-01-11 Tamar Christina <tamar.christina@arm.com>
24542 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
24543 to all inlined functions, change static to extern.
24545 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
24548 * config/arm/arm.c (legitimize_pic_address): Handle reference to
24550 (arm_assemble_integer): Likewise.
24552 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
24554 * config.gcc: Use new awk script to check CPU, FPU and architecture
24555 parameters for --with-... options.
24556 * config/arm/parsecpu.awk: New file
24557 * config/arm/arm-cpus.in: New file.
24558 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
24560 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
24562 * config/arm/t-arm: Update dependency rules.
24563 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
24564 of processing .def files.
24565 * config/arm/genopt.sh: Deleted.
24566 * config/arm/gentune.sh: Deleted.
24567 * config/arm/arm-cores.def: Deleted.
24568 * config/arm/arm-arches.def: Deleted.
24569 * config/arm/arm-fpus.def: Deleted.
24570 * config/arm/arm-tune.md: Regenerated.
24571 * config/arm/arm-tables.opt: Regenerated.
24572 * config/arm/arm-cpu.h: New generated file.
24573 * config/arm/arm-cpu-data.h: New generated file.
24574 * config/arm/arm-cpu-cdata.h: New generated file.
24576 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
24579 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
24581 (input_varpool_node): Unpack dynamically_initialized bit.
24583 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
24585 PR rtl-optimization/79032
24586 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
24587 the alignment of the adjusted memory reference against that of MODE,
24588 instead of the alignment of the original memory reference.
24590 2017-01-11 Martin Jambor <mjambor@suse.cz>
24592 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
24594 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
24595 decorated functions.
24597 2017-01-11 Richard Biener <rguenther@suse.de>
24599 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
24600 set range/nonnull info for PHI results. Do not set it on
24601 stmts marked for removal.
24603 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
24605 * expr.c (store_field): In the bitfield case, fetch the return value
24606 from the registers before applying a single big-endian adjustment.
24607 Always do a final load for a BLKmode value not larger than a word.
24609 2017-01-10 David Malcolm <dmalcolm@redhat.com>
24612 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
24613 that we correctly handle column numbers greater than
24614 LINE_MAP_MAX_COLUMN_NUMBER.
24616 2017-01-10 Martin Sebor <msebor@redhat.com>
24618 PR middle-end/78245
24619 * gimple-ssa-sprintf.c (get_destination_size): Call
24620 {init,fini}object_sizes.
24621 * tree-object-size.c (addr_object_size): Adjust.
24622 (pass_through_call): Adjust.
24623 (pass_object_sizes::execute): Adjust.
24624 * tree-object-size.h (fini_object_sizes): Declare.
24626 2017-01-10 Martin Sebor <msebor@redhat.com>
24628 PR tree-optimization/78775
24629 * builtins.c (get_size_range): Move...
24630 * calls.c: ...to here.
24631 (alloc_max_size): Accept zero argument.
24632 (operand_signed_p): Remove.
24633 (maybe_warn_alloc_args_overflow): Call get_size_range.
24634 * calls.h (get_size_range): Declare.
24636 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
24638 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
24639 from TI's devices.csv file as of September 2016.
24640 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
24642 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
24644 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
24645 * doc/invoke.texi: Likewise.
24646 * doc/md.texi: Likewise.
24647 * doc/objc.texi: Likewise.
24649 2017-01-10 Joshua Conner <joshconner@google.com>
24651 * config/arm/fuchsia-elf.h: New file.
24652 * config/fuchsia.h: New file.
24653 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
24654 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
24656 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
24658 2016-01-10 Richard Biener <rguenther@suse.de>
24660 PR tree-optimization/79034
24661 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24662 Propagate out degenerate PHIs in the joiner.
24664 2017-01-10 Martin Liska <mliska@suse.cz>
24666 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
24667 (sort_congruence_classes_by_decl_uid): Likewise.
24668 (sort_congruence_class_groups_by_decl_uid): Likewise.
24669 (sem_item_optimizer::merge_classes): Sort class, groups in these
24670 classes and members in the groups by DECL_UID of declarations.
24671 This would make merge operations stable.
24673 2017-01-10 Martin Liska <mliska@suse.cz>
24675 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
24676 usage of m_classes_vec.
24677 (sem_item_optimizer::~sem_item_optimizer): Likewise.
24678 (sem_item_optimizer::get_group_by_hash): Likewise.
24679 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
24680 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
24681 (sem_item_optimizer::verify_classes): Likewise.
24682 (sem_item_optimizer::process_cong_reduction): Likewise.
24683 (sem_item_optimizer::dump_cong_classes): Likewise.
24684 (sem_item_optimizer::merge_classes): Likewise.
24685 * ipa-icf.h (congruence_class_hash): Rename from
24686 congruence_class_group_hash. Remove declaration of m_classes_vec.
24688 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
24690 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
24691 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
24692 * config.gcc: Add avx512vpopcntdqintrin.h.
24693 * config/i386/avx512vpopcntdqintrin.h: New.
24694 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
24695 * config/i386/i386-builtin-types.def: Add new types.
24696 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
24697 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
24698 __builtin_ia32_vpopcountq_v8di_mask): New.
24699 * config/i386/i386-c.c (ix86_target_macros_internal): Define
24700 __AVX512VPOPCNTDQ__.
24701 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
24702 (PTA_AVX512VPOPCNTDQ): Define.
24703 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
24704 TARGET_AVX512VPOPCNTDQ_P): Define.
24705 * config/i386/i386.opt: Add mavx512vpopcntdq.
24706 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
24707 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
24709 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24711 PR middle-end/77484
24712 * predict.def (PRED_CALL): Set to 67.
24714 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
24716 * expr.c (store_field): In the bitfield case, if the value comes from
24717 a function call and is of an aggregate type returned in registers, do
24718 not modify the field mode; extract the value in all cases if the mode
24719 is BLKmode and the size is not larger than a word.
24721 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
24724 * config/i386/cpuid.h: Fix undefined behavior.
24726 2017-01-04 Jeff Law <law@redhat.com>
24728 PR tree-optimization/79007
24729 PR tree-optimization/67955
24730 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
24731 conservative for pt.null when flag_non_call_exceptions is on.
24733 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24735 PR translation/79019
24736 PR translation/79020
24737 * params.def (PARAM_INLINE_MIN_SPEEDUP,
24738 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
24739 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
24741 * config/avr/avr.opt (maccumulate-args): Likewise.
24742 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
24743 * common.opt (freport-bug): Likewise.
24744 * cif-code.def (CIF_FINAL_ERROR): Likewise.
24745 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
24746 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
24747 translatable string.
24748 * config/i386/i386.c (function_value_32): Likewise.
24749 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
24750 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
24752 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
24753 * common/config/msp430/msp430-common.c (msp430_handle_option):
24755 * symtab.c (symtab_node::verify_base): Likewise.
24756 * opts.c (set_debug_level): Likewise.
24757 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
24758 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
24759 missing whitespace to translatable strings.
24760 * config/avr/avr.md (bswapsi2): Fix typo in comment.
24761 * config/sh/superh.h: Likewise.
24762 * config/i386/xopintrin.h: Likewise.
24763 * config/i386/znver1.md: Likewise.
24764 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
24765 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
24766 * double-int.h (struct double_int): Likewise.
24767 * double-int.c (div_and_round_double): Likewise.
24768 * wide-int.cc: Likewise.
24769 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
24770 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
24771 * cfgcleanup.c (crossjumps_occured): Renamed to ...
24772 (crossjumps_occurred): ... this.
24773 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
24776 PR tree-optimization/78899
24777 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
24778 returning bool return struct loop *, NULL for failure and the new
24780 (versionable_outer_loop_p): Don't version outer loop if it has
24781 dont_vectorized bit set.
24782 (tree_if_conversion): When versioning outer loop, ensure
24783 tree_if_conversion is performed also on the inner loop of the
24784 non-vectorizable outer loop copy.
24785 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
24786 LOOP_VECTORIZED in inner loop of the scalar outer loop and
24787 prevent vectorization of it.
24788 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
24789 the outer loop vectorization of the non-scalar version is attempted
24790 before vectorization of the inner loop in scalar version. If
24791 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
24792 vectorization of its inner loop.
24793 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
24794 has 2 inner loops, rename also on edges from bb whose single pred
24795 is outer_loop->header. Fix typo in function comment.
24797 2017-01-09 Martin Sebor <msebor@redhat.com>
24800 * asan.c (asan_emit_stack_protection): Increase local buffer size
24801 to avoid snprintf truncation warning.
24803 2017-01-09 Andrew Pinski <apinski@cavium.com>
24805 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
24806 to reference thunderx2t99 for the tuning structure
24807 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
24809 (thunderx2t99_extra_costs): This.
24810 * config/aarch64/aarch64-tune.md: Regenerate.
24811 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
24812 (vulcan_addrcost_table): This.
24813 (vulcan_regmove_cost): Rename to ...
24814 (thunderx2t99_regmove_cost): This.
24815 (vulcan_vector_cost): Rename to ...
24816 (thunderx2t99_vector_cost): this.
24817 (vulcan_branch_cost): Rename to ...
24818 (thunderx2t99_branch_cost): This.
24819 (vulcan_tunings): Rename to ...
24820 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
24821 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
24823 2017-01-09 Martin Jambor <mjambor@suse.cz>
24827 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
24828 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
24829 (propagate_vr_accross_jump_function): Use the above function for all
24830 value range computations for pass-through jump functions and type
24831 converasion from explicit value range values.
24832 (ipcp_propagate_stage): Do not attempt to deduce types of formal
24833 parameters from TYPE_ARG_TYPES.
24834 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
24835 (ipa_write_node_info): Stream type of the actual argument.
24836 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
24838 2017-01-09 Martin Liska <mliska@suse.cz>
24841 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
24842 (lookup_compiler): Do not show error message with have_E.
24844 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24846 PR tree-optimization/78938
24847 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
24848 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
24849 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
24850 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
24853 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24855 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
24858 2017-01-09 Richard Biener <rguenther@suse.de>
24860 PR tree-optimization/78997
24861 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
24862 name condition properly.
24864 2017-01-09 Richard Biener <rguenther@suse.de>
24867 * dwarf2out.c (is_cxx): New overload with context.
24868 (is_naming_typedef_decl): Use it.
24870 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24872 * invoke.texi (Option Summary): Correct spacing in option lists
24873 and add line breaks to fix over-long lines.
24875 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24877 PR middle-end/17660
24879 * extend.texi (Common Variable Attributes): Add xref to GCC
24880 Internals manual to explain mode attribute keywords.
24882 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24885 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
24886 and Preprocessor Options.
24887 (Options for Linking): Document -pthread here....
24888 (RS/6000 and PowerPC Options): ...not here.
24889 (Solaris 2 Options): ...or here.
24890 * doc/cppopts.texi: Document -pthread.
24892 2017-01-08 Martin Sebor <msebor@redhat.com>
24894 PR middle-end/77708
24895 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
24896 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
24897 New member functions.
24898 (format_directive): Used them.
24900 (pass_sprintf_length::handle_gimple_call): Same.
24901 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
24902 to avoid truncation for any argument.
24903 (extract_affine_mul): Same.
24904 * tree.c (get_file_function_name): Same.
24906 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24908 PR middle-end/77484
24909 * predict.def (PRED_INDIR_CALL): Set to 86.
24911 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24913 PR preprocessor/54124
24914 * doc/cppopts.texi: Reformat -d subtable to list the full name
24915 of the options. Add cross-reference to the docs for the general
24916 compiler -d options.
24917 * doc/invoke.texi (Developer Options): Add cross-reference to the
24918 preprocessor-specific -d option documentation.
24920 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24922 PR preprocessor/13498
24923 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
24924 redudant material, and reflect new command-line options.
24925 (System Headers): Likewise.
24927 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24929 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
24930 -isystem, and -idirafter. Copy-edit.
24931 * doc/cppopts.texi: Copy-edit. Remove contradiction about
24932 default for -ftrack-macro-expansion. Delete obsolete and
24933 badly-formatted implementation details about -fdebug-cpp output.
24934 * doc/cppwarnopts.texi: Copy-edit.
24936 2017-01-07 David Malcolm <dmalcolm@redhat.com>
24939 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
24940 that the transition from a max line width >= 1<<10 to narrower
24941 lines works correctly.
24943 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
24945 * doc/options.texi (PerFunction): New.
24946 * opt-functions.awk (switch_flags): Map both Optimization and
24947 PerFunction to CL_OPTIMIZATION.
24948 * opth-gen.awk: Test for PerFunction flag along with
24950 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
24951 it only when the latter is present. Skip those that don't in
24952 the hash function generator.
24953 * common.opt (fvar-tracking): Mark as PerFunction instead of
24955 (fvar-tracking-assignments): Likewise.
24956 (fvar-tracking-assignments-toggle): Likewise.
24957 (fvar-tracking-uninit): Likewise.
24959 2017-01-07 Jakub Jelinek <jakub@redhat.com>
24961 PR translation/79018
24962 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
24965 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
24968 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
24969 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
24970 TARGET_LONG_JUMP_TABLE_OFFSETS.
24971 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
24972 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
24973 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
24974 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
24975 * config/m68k/m68k.md (tablejump expander): Likewise.
24976 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
24977 TARGET_LONG_JUMP_TABLE_OFFSETS.
24978 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
24979 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
24981 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
24982 David Holsgrove <david.holsgrove@xilinx.com>
24984 * common/config/microblaze/microblaze-common.c
24985 (TARGET_EXCEPT_UNWIND_INFO): Remove.
24986 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
24988 * config/microblaze/microblaze.c (microblaze_must_save_register)
24989 (microblaze_expand_epilogue, microblaze_return_addr): Handle
24991 (microblaze_eh_return): New function.
24992 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
24993 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
24994 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
24995 * config/microblaze/microblaze.md (eh_return): New pattern.
24997 2017-01-06 Jakub Jelinek <jakub@redhat.com>
24999 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
25000 GCC_DIAGNOSTIC_STRINGIFY): Define.
25002 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
25004 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25006 * config/arm/arm.md (<mcrr>): New.
25008 * config/arm/arm.c (arm_arch5te): New.
25009 (arm_option_override): Set arm_arch5te.
25010 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
25012 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
25013 (arm_mcrr_qualifiers): ... this. New.
25014 (MRRC_QUALIFIERS): Define to...
25015 (arm_mrrc_qualifiers): ... this. New.
25016 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
25018 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
25019 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
25020 (MRRCI, mrrc, MRRC): New.
25021 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
25022 VUNSPEC_MRRC2): New.
25024 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25026 * config/arm/arm.md (<mcr>): New.
25028 * config/arm/arm.c (arm_coproc_builtin_available): Add
25029 support for mcr, mrc, mcr2 and mrc2.
25030 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
25031 (arm_mcr_qualifiers): ... this. New.
25032 (MRC_QUALIFIERS): Define to ...
25033 (arm_mrc_qualifiers): ... this. New.
25034 (MCR_QUALIFIERS): Define to ...
25035 (arm_mcr_qualifiers): ... this. New.
25036 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
25038 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
25039 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
25040 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
25041 VUNSPEC_MRC2): New.
25043 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25045 * config/arm/arm.md (*ldc): New.
25049 * config/arm/arm.c (arm_coproc_builtin_available): Add
25050 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
25051 (arm_coproc_ldc_stc_legitimate_address): New.
25052 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
25053 'qualifier_const_pointer'.
25054 (LDC_QUALIFIERS): Define to...
25055 (arm_ldc_qualifiers): ... this. New.
25056 (STC_QUALIFIERS): Define to...
25057 (arm_stc_qualifiers): ... this. New.
25058 * config/arm/arm-protos.h
25059 (arm_coproc_ldc_stc_legitimate_address): New.
25060 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
25061 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
25062 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
25063 stc2, stcl, stc2l): New.
25064 * config/arm/constraints.md (Uz): New.
25065 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
25066 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
25067 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
25068 VUNSPEC_STC2L): New.
25070 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25072 * config/arm/arm.md (<cdp>): New.
25073 * config/arm/arm.c (neon_const_bounds): Rename this ...
25074 (arm_const_bounds): ... this.
25075 (arm_coproc_builtin_available): New.
25076 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
25077 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
25078 (CDP_QUALIFIERS): Define to...
25079 (arm_cdp_qualifiers): ... this. New.
25081 (arm_expand_builtin_args): Add case for 6 arguments.
25082 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
25083 (arm_const_bounds): ... this.
25084 (arm_coproc_builtin_available): New.
25085 * config/arm/arm_acle.h (__arm_cdp): New.
25087 * config/arm/arm_acle_builtins.def (cdp): New.
25089 * config/arm/iterators.md (CDPI,CDP,cdp): New.
25090 * config/arm/neon.md: Rename all 'neon_const_bounds' to
25091 'arm_const_bounds'.
25092 * config/arm/types.md (coproc): New.
25093 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
25094 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
25095 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
25096 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
25098 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25100 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
25101 (UBINOP_QUALIFIERS): New.
25103 (acle_builtin_data): New. Change comment.
25104 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
25105 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
25106 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
25107 arm_acle_builtins.def.
25108 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
25109 (arm_init_acle_builtins): New.
25110 (CRC32_BUILTIN): Remove.
25111 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
25112 crc32cb, crc32ch and crc32cw.
25113 (arm_init_crc32_builtins): Remove.
25114 (arm_init_builtins): Use arm_init_acle_builtins rather
25115 than arm_init_crc32_builtins.
25116 (arm_expand_acle_builtin): New.
25117 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
25118 * config/arm/arm_acle_builtins.def: New.
25120 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25122 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
25123 (arm_builtin_datum): ... this.
25124 (arm_init_neon_builtin): Rename to ...
25125 (arm_init_builtin): ... this. Add a new parameters PREFIX
25126 and USE_SIG_IN_NAME.
25127 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
25128 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
25129 'arm_builtin_datum'.
25130 (arm_init_vfp_builtins): Likewise.
25131 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
25132 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
25133 (arm_expand_neon_args): Rename to ...
25134 (arm_expand_builtin_args): ... this. Rename builtin_arg
25135 enum values and differentiate between ARG_BUILTIN_MEMORY
25136 and ARG_BUILTIN_NEON_MEMORY.
25137 (arm_expand_neon_builtin_1): Rename to ...
25138 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
25139 values, arm_expand_builtin_args and add bool parameter NEON.
25140 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
25141 (arm_expand_vfp_builtin): Likewise.
25142 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
25144 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25146 PR middle-end/77484
25147 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
25148 * predict.c (tree_estimate_probability_bb): Reverse direction of
25149 polymorphic call predictor.
25151 2017-01-06 David Malcolm <dmalcolm@redhat.com>
25153 * passes.c (execute_one_pass): Split out pass-skipping logic into...
25154 (determine_pass_name_match): ...this new function and...
25155 (should_skip_pass_p): ...this new function.
25157 2017-01-06 Nathan Sidwell <nathan@acm.org>
25159 * ipa-visibility.c (function_and_variable_visibility): Reformat
25160 comments and long lines. Remove extrneous if.
25161 * symtab.c (symtab_node::make_decl_local): Fix code format.
25162 (symtab_node::set_section_for_node): Fix comment typo.
25164 2017-01-06 Martin Liska <mliska@suse.cz>
25167 * lra-constraints.c: Rename invariant to lra_invariant.
25168 * predict.c (set_even_probabilities): Initialize e to NULL.
25170 2017-01-05 Martin Sebor <msebor@redhat.com>
25172 PR tree-optimization/78910
25173 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
25174 (format_integer): Correct off-by-one error in the handling
25175 of precision with negative numbers in signed conversions..
25177 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
25179 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
25181 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25183 PR tree-optimization/71016
25184 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
25185 factor_out_conditional_conversion. Formatting fix.
25186 (factor_out_conditional_conversion): Add cond_stmt argument.
25187 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
25188 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
25191 2017-01-05 David Malcolm <dmalcolm@redhat.com>
25193 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
25194 read-rtl-function.o, and selftest-rtl.o.
25195 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
25196 (selftest::aarch64_test_loading_full_dump): New function.
25197 (selftest::aarch64_run_selftests): New function.
25198 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
25199 selftest::aarch64_run_selftests.
25200 * config/i386/i386.c
25201 (selftest::ix86_test_loading_dump_fragment_1): New function.
25202 (selftest::ix86_test_loading_call_insn): New function.
25203 (selftest::ix86_test_loading_full_dump): New function.
25204 (selftest::ix86_test_loading_unspec): New function.
25205 (selftest::ix86_run_selftests): Call the new functions.
25206 * emit-rtl.c (maybe_set_max_label_num): New function.
25207 * emit-rtl.h (maybe_set_max_label_num): New decl.
25208 * function.c (instantiate_decls): Guard call to
25209 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
25210 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
25212 * gensupport.c (gen_reader::gen_reader): Pass "false"
25213 for new "compact" param of rtx_reader.
25214 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
25215 rather than an empty string for NULL strings.
25216 * read-md.c: Potentially include config.h rather than bconfig.h.
25217 Wrap include of errors.h with #ifdef GENERATOR_FILE.
25218 (have_error): New global, copied from errors.c.
25219 (md_reader::read_name): Rename to...
25220 (md_reader::read_name_1): ...this, adding "out_loc" param,
25221 and converting "missing name or number" to returning false, rather
25223 (md_reader::read_name): Reimplement in terms of read_name_1.
25224 (md_reader::read_name_or_nil): New function.
25225 (md_reader::read_string): Handle "(nil)" by returning NULL.
25226 (md_reader::md_reader): Add new param "compact".
25227 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
25228 (md_reader::read_file): New method.
25229 * read-md.h (md_reader::md_reader): Add new param "compact".
25230 (md_reader::read_file): New method.
25231 (md_reader::is_compact): New accessor.
25232 (md_reader::read_name): Convert return type from void to file_location.
25233 (md_reader::read_name_or_nil): New decl.
25234 (md_reader::read_name_1): New decl.
25235 (md_reader::m_compact): New field.
25236 (noop_reader::noop_reader): Pass "false" for new "compact" param
25238 (rtx_reader::rtx_reader): Add new "compact" param.
25239 (rtx_reader::read_rtx_operand): Make virtual and convert return
25240 type from void to rtx.
25241 (rtx_reader::read_until): New decl.
25242 (rtx_reader::handle_any_trailing_information): New virtual function.
25243 (rtx_reader::postprocess): New virtual function.
25244 (rtx_reader::finalize_string): New virtual function.
25245 (rtx_reader::m_in_call_function_usage): New field.
25246 (rtx_reader::m_reuse_rtx_by_id): New field.
25247 * read-rtl-function.c: New file.
25248 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
25249 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
25250 (selftest::verify_three_block_rtl_cfg): New decl.
25251 * read-rtl-function.h: New file.
25252 * read-rtl.c: Potentially include config.h rather than bconfig.h.
25253 For host, include function.h, memmodel.h, and emit-rtl.h.
25254 (one_time_initialization): New function.
25255 (struct compact_insn_name): New struct.
25256 (compact_insn_names): New array.
25257 (find_code): Handle insn codes in compact dumps.
25258 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
25259 (bind_subst_iter_and_attr): Likewise.
25260 (add_condition_to_string): Likewise.
25261 (add_condition_to_rtx): Likewise.
25262 (apply_attribute_uses): Likewise.
25263 (add_current_iterators): Likewise.
25264 (apply_iterators): Likewise.
25265 (initialize_iterators): Guard usage of apply_subst_iterator with
25266 #ifdef GENERATOR_FILE.
25267 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
25268 (md_reader::read_mapping): Likewise.
25269 (add_define_attr_for_define_subst): Likewise.
25270 (add_define_subst_attr): Likewise.
25271 (read_subst_mapping): Likewise.
25272 (check_code_iterator): Likewise.
25273 (rtx_reader::read_rtx): Likewise. Move one-time initialization
25275 (one_time_initialization): New function.
25276 (rtx_reader::read_until): New method.
25277 (read_flags): New function.
25278 (parse_reg_note_name): New function.
25279 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
25280 Handle reuse_rtx ids.
25281 Wrap iterator lookup within #ifdef GENERATOR_FILE.
25282 Add parsing support for RTL dumps, mirroring the special-cases in
25283 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
25284 values, and calling handle_any_trailing_information.
25285 (rtx_reader::read_rtx_operand): Convert return type from void
25286 to rtx, returning return_rtx. Handle case 'e'. Call
25287 finalize_string on XSTR and XTMPL fields.
25288 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
25289 "(nil)" values were omitted. Call the postprocess vfunc on the
25291 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
25292 class ctor. Initialize m_in_call_function_usage. Call
25293 one_time_initialization.
25294 * rtl-tests.c (selftest::test_uncond_jump): Call
25295 set_new_first_and_last_insn.
25296 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
25297 * selftest-rtl.c: New file.
25298 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
25299 (selftest::get_insn_by_uid): New decl.
25300 * selftest-run-tests.c (selftest::run_tests): Call
25301 read_rtl_function_c_tests.
25302 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
25303 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
25306 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
25308 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
25309 operands in a special way. Assert that pos+len <= mode precision.
25311 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25313 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
25314 3 argument Alias with unlimited for the negative form.
25315 (fno-vect-cost-model): Removed.
25317 2017-01-05 Martin Liska <mliska@suse.cz>
25319 * hsa-gen.c (gen_hsa_divmod): New function.
25320 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
25322 2017-01-05 Martin Liska <mliska@suse.cz>
25325 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
25328 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25330 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
25331 small constant length operands.
25333 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25335 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
25336 between loop iterations.
25338 2017-01-05 Martin Liska <mliska@suse.cz>
25341 * gimplify.c (gimplify_decl_expr): Compare to
25342 asan_poisoned_variables instread of checking flags.
25343 (gimplify_target_expr): Likewise.
25344 (gimplify_expr): Likewise.
25345 (gimplify_function_tree): Conditionally initialize
25346 asan_poisoned_variables.
25348 2017-01-04 Jeff Law <law@redhat.com>
25350 PR tree-optimizatin/78812
25351 * rtl.h (contains_mem_rtx_p): Prototype.
25352 * ifcvt.c (containts_mem_rtx_p): Move from here to...
25353 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
25354 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
25355 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
25356 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
25358 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25360 * input.c (assert_char_at_range): Default-initialize actual_range.
25362 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25364 * df-scan.c (df_ref_create_structure): Make regno unsigned,
25365 to match the caller.
25367 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25369 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
25370 insns after final jump in test to emit dummy move.
25372 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25374 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
25375 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
25377 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25379 * multiple_target.c (create_dispatcher_calls): Init e_next.
25380 * tree-ssa-loop-split.c (split_loop): Init border.
25381 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
25384 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
25389 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
25390 (altivec_register_operand): Do not return true if the operand
25391 contains a SUBREG mixing SImode and SFmode.
25392 (vsx_register_operand): Likewise.
25393 (vsx_reg_sfsubreg_ok): New predicate.
25394 (vfloat_operand): Do not return true if the operand contains a
25395 SUBREG mixing SImode and SFmode.
25396 (vint_operand): Likewise.
25397 (vlogical_operand): Likewise.
25398 (gpc_reg_operand): Likewise.
25399 (int_reg_operand): Likewise.
25400 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
25401 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
25402 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
25404 (rs6000_emit_move_si_sf_subreg): New helper function.
25405 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
25406 fixup SUBREGs involving SImode and SFmode.
25407 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
25408 numbers for the new peephole2 optimization.
25409 (peephole2 for SFmode unions): New peephole2 to optimize cases in
25410 the GLIBC math library that do AND/IOR/XOR operations on single
25411 precision floating point.
25412 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
25413 target macros to say whether we need to avoid SUBREGs mixing
25415 (TARGET_ALLOW_SF_SUBREG): Likewise.
25416 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
25417 (UNSPEC_SI_FROM_SF): Likewise.
25418 (iorxor): Change spacing.
25419 (and_ior_xor): New iterator for AND, IOR, and XOR.
25420 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
25421 (movdi_from_sf_zero_ext): Likewise.
25422 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
25423 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
25424 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
25425 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
25426 (fms<mode>4): Likewise.
25427 (fnma<mode>4): Likewise.
25428 (fnms<mode>4): Likewise.
25429 (nfma<mode>4): Likewise.
25430 (nfms<mode>4): Likewise.
25432 2017-01-04 Marek Polacek <polacek@redhat.com>
25435 * doc/invoke.texi: Document -Wpointer-compare.
25437 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25439 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
25442 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
25443 descriptions for -gdwarf-5 and emit them as uleb128 instead of
25446 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
25449 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
25450 documentation of the powerpc_popcntb_ok attribute.
25451 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25452 code to issue warning messages if a requested CPU configuration is
25453 not supported by the binary (assembler and loader) toolchain.
25454 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
25455 made to define a built-in function that has been disabled.
25456 (paired_init_builtins): Add assertion to prevent ICE if attempt is
25457 made to define a built-in function that has been disabled.
25458 (altivec_init_builtins): Add comment explaining why definition
25459 of the DST built-in functions is not preceded by an assertion
25460 check. Add assertions to prevent ICE if attempts are made to
25461 define an altivec predicate or an abs* built-in function that has
25463 (htm_init_builtins): Add comment explaining why definition of the
25464 htm built-in functions is not preceded by an assertion check.
25466 2017-01-04 Jeff Law <law@redhat.com>
25468 PR tree-optimizatin/67955
25469 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
25470 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
25471 the points-to solution does not include pt_null. Use DECL_PT_UID
25474 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
25476 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
25477 Use gen_int_mode instead of gen_lopwart for const_int operands.
25479 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25481 PR tree-optimization/71563
25482 * match.pd: Simplify X << Y into X if Y is known to be 0 or
25483 out of range value - has low bits known to be zero.
25485 2017-01-04 Alan Modra <amodra@gmail.com>
25487 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
25488 * configure: Regenerate.
25489 * config.in: Regenerate.
25491 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25494 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
25495 a substring of the message, but strcmp with the whole message. Ifdef
25496 ENABLE_NLS, translate the message first using dgettext.
25498 2017-01-03 Jeff Law <law@redhat.com>
25500 PR tree-optimizatin/78856
25501 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
25502 (mark_threaded_blocks): Remove code to truncate thread paths that
25503 cross multiple loop headers. Instead invalidate the cached loop
25504 iteration information and handle case of a thread path walking
25505 into an irreducible region.
25507 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
25510 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
25511 assertions. Add support for doing the signbit if the IEEE 128-bit
25512 floating point value is in a GPR.
25513 * config/rs6000/rs6000.md (Fsignbit): Delete.
25514 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
25515 Update the length attribute if the value is in a GPR.
25516 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
25517 the sign or zero extension instruction, since the value is always 0/1.
25518 (signbit<mode>2_dm2): Delete using <Fsignbit>.
25521 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
25522 extracting SImode to a GPR register so that we can generate a
25523 store, limit the vector to be in a traditional Altivec register
25524 for the vextuwrx instruction.
25526 2017-01-03 Ian Lance Taylor <iant@google.com>
25528 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
25530 2017-01-03 Martin Sebor <msebor@redhat.com>
25532 PR tree-optimization/78696
25533 * gimple-ssa-sprintf.c (format_floating): Correct handling of
25534 precision. Use MPFR for %f for greater fidelity. Correct handling
25536 (pass_sprintf_length::compute_format_length): Set width and precision
25537 specified by asrerisk to void_node for vararg functions.
25538 (try_substitute_return_value): Adjust dump output.
25540 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
25542 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
25544 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
25546 * doc/invoke.texi (SPARC options): Document -mlra as the default.
25547 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
25548 -mlra/-mno-lra was passed to the compiler.
25550 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
25552 PR rtl-optimization/65618
25553 * emit-rtl.c (try_split): Move initialization of "before" and
25554 "after" to just before the call to emit_insn_after_setloc.
25556 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
25558 * doc/md.texi (Standard Names): Remove reference to Java frontend.
25560 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
25562 * dwarf2out.c (gen_enumeration_type_die): When
25563 -gno-strict-dwarf, add a DW_AT_encoding attribute.
25565 2017-01-03 Jakub Jelinek <jakub@redhat.com>
25567 PR tree-optimization/78965
25568 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
25569 Change first argument from const call_info & to call_info &. For %n
25570 set info.nowrite to false.
25572 PR middle-end/78901
25573 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
25574 possibly throwing calls.
25576 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
25577 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
25578 and fns handling, rather than in a separate case SSA_NAME.
25580 2017-01-02 Jeff Law <law@redhat.com>
25582 * config/darwin-driver.c (darwin_driver_init): Const-correctness
25583 fixes for first_period and second_period variables.
25585 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
25588 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
25589 (*insvqi_1): New insn pattern.
25590 (*insvqi_1_mem_rex64): Ditto.
25591 (*insvqi_2): Ditto.
25592 (*insvqi_3): Rename from *insvqi.
25594 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
25596 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
25598 * doc/cfg.texi (Edges): Remove reference to Java.
25599 (Maintaining the CFG): Ditto.
25601 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25603 PR middle-end/77674
25604 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
25605 transparent aliases.
25607 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25609 PR middle-end/77484
25610 * predict.def (PRED_CALL): Update hitrate.
25611 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
25612 * predict.c (tree_estimate_probability_bb): Split CALL predictor
25613 into direct/indirect/polymorphic variants.
25615 2017-01-01 Jakub Jelinek <jakub@redhat.com>
25617 Update copyright years.
25619 * gcc.c (process_command): Update copyright notice dates.
25620 * gcov-dump.c (print_version): Ditto.
25621 * gcov.c (print_version): Ditto.
25622 * gcov-tool.c (print_version): Ditto.
25623 * gengtype.c (create_file): Ditto.
25624 * doc/cpp.texi: Bump @copying's copyright year.
25625 * doc/cppinternals.texi: Ditto.
25626 * doc/gcc.texi: Ditto.
25627 * doc/gccint.texi: Ditto.
25628 * doc/gcov.texi: Ditto.
25629 * doc/install.texi: Ditto.
25630 * doc/invoke.texi: Ditto.
25632 Copyright (C) 2017 Free Software Foundation, Inc.
25634 Copying and distribution of this file, with or without modification,
25635 are permitted in any medium without royalty provided the copyright
25636 notice and this notice are preserved.