1 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
3 * sancov.c: Include tm_p.h.
5 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8 * output.h (switch_to_other_text_partition): New declaration.
9 * varasm.c (switch_to_other_text_partition): New function.
10 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
11 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
12 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
13 to the other text partition before emitting LCL label and switch back
14 after emitting the word after it.
16 2017-09-07 Richard Biener <rguenther@suse.de>
18 * passes.def (pass_split_crit_edges): Remove instance before PRE.
19 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
20 critical edges here, after loop init.
21 (pass_data_pre): Remove PROP_no_crit_edges flags.
22 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
23 for valueization of call args to avoid leaking VN_TOP.
24 (visit_use): Assert we do not visit default defs.
25 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
26 Use error_mark_node to more easily detect leaking VN_TOP.
27 All default-defs are varying, not VN_TOP. Mark them visited.
28 (run_scc_vn): Make code match comment.
30 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
32 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
33 OPTION_MASK_FLOAT128_KEYWORD.
34 (POWERPC_MASKS): Likewise.
35 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
36 support for the -mfloat128-type option, and make -mfloat128
37 default on PowerPC Linux systems. Define or undefine
38 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
39 Define __float128 to be __ieee128 if IEEE 128-bit support is
40 enabled, or undefine it.
41 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
42 Delete defining __FLOAT128_TYPE__.
43 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
44 -mfloat128-type option and make -mfloat128 default on PowerPC
46 (TARGET_FLOAT128_TYPE): Likewise.
47 (-mfloat128-type): Likewise.
48 * config/rs6000/rs6000.c (rs6000_option_override_internal):
49 Delete the -mfloat128-type option and make -mfloat128 default on
50 PowerPC Linux systems. Always use __ieee128 to be the keyword for
51 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
52 128-bit floating point is enabled. Change tests from using
53 -mfloat128-type to -mfloat128.
54 (rs6000_mangle_type): Use the correct mangling for the __float128
55 type even if normal long double is restricted to 64-bits.
56 (floatn_mode): Enable the _Float128 type by default on VSX Linux
58 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
59 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
60 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
62 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
63 documentation for -mfloat128.
65 2017-09-06 Olivier Hainque <hainque@adacore.com>
67 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
69 2017-09-06 Wish Wu <wishwu007@gmail.com>
70 Jakub Jelinek <jakub@redhat.com>
72 * asan.c (initialize_sanitizer_builtins): Add
73 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
74 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
75 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
76 BT_FN_VOID_UINT64_PTR variables.
77 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
78 (BT_FN_VOID_UINT16_UINT16): Likewise.
79 (BT_FN_VOID_UINT32_UINT32): Likewise.
80 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
81 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
82 (BT_FN_VOID_UINT64_PTR): Likewise.
83 * common.opt (flag_sanitize_coverage): New variable.
84 (fsanitize-coverage=trace-pc): Remove.
85 (fsanitize-coverage=): Add.
86 * flag-types.h (enum sanitize_coverage_code): New enum.
87 * fold-const.c (fold_range_test): Disable non-short-circuit
88 optimization if flag_sanitize_coverage.
89 (fold_truth_andor): Likewise.
90 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
91 * opts.c (COVERAGE_SANITIZER_OPT): Define.
92 (coverage_sanitizer_opts): New array.
93 (get_closest_sanitizer_option): Add OPTS argument, handle also
94 OPT_fsanitize_coverage_.
95 (parse_sanitizer_options): Adjusted to also handle
96 OPT_fsanitize_coverage_.
97 (common_handle_option): Add OPT_fsanitize_coverage_.
98 * sancov.c (instrument_comparison, instrument_switch): New function.
99 (sancov_pass): Add trace-cmp support.
100 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
101 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
102 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
103 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
104 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
105 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
106 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
107 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
108 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
109 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
111 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
113 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
114 error. Only quit immediately if parsing is complete.
115 (BEGIN): Initialize fatal_err and parse_done.
116 (begin fpu, end fpu): Check number of arguments.
117 (begin arch, end arch): Likewise.
118 (begin cpu, end cpu): Likewise.
119 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
120 (optalias): Likewise.
122 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
124 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
125 * config/arm/arm-isa.h: Delete. Move definitions to ...
126 * arm-cpus.in: ... here. Use new feature and fgroup values.
127 * config/arm/arm.c (arm_option_override): Use lower case for feature
129 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
130 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
131 * config/arm/parsecpu.awk (END): Add new command 'isa'.
133 (print_isa_bits_for): New function.
134 (gen_isa): New function.
135 (gen_comm_data): Use print_isa_bits_for.
136 (define feature): New keyword.
137 (define fgroup): New keyword.
138 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
139 (arm-isa.h): Add rule to generate file.
140 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
141 case for feature bit names.
143 2017-09-06 Richard Biener <rguenther@suse.de>
145 * tree-ssa-pre.c (NECESSARY): Remove.
146 (create_expression_by_pieces): Do not touch pass-local flags.
147 (insert_into_preds_of_block): Likewise.
148 (do_pre_regular_insertion): Likewise.
149 (eliminate_insert): Likewise.
150 (eliminate_dom_walker::before_dom_children): Likewise.
151 (fini_eliminate): Do not look at inserted_exprs.
152 (mark_operand_necessary): Remove.
153 (remove_dead_inserted_code): Replace with simple work-list
154 algorithm based on inserted_exprs and SSA uses.
155 (pass_pre::execute): Re-order fini_eliminate and
156 remove_dead_inserted_code.
158 2017-09-06 Olivier Hainque <hainque@adacore.com>
160 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
161 for VxWorks 7. Adjust surrounding comments.
163 2017-09-06 Richard Biener <rguenther@suse.de>
165 * gimple-ssa-strength-reduction.c
166 (find_candidates_dom_walker::before_dom_children): Also allow
169 2017-09-06 Richard Biener <rguenther@suse.de>
171 PR tree-optimization/82108
172 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
173 for gap in the non-permutation SLP case.
175 2017-09-06 Martin Jambor <mjambor@suse.cz>
177 PR tree-optimization/82078
178 * tree-sra.c (sort_and_splice_var_accesses): Move call to
179 add_access_to_work_queue...
180 (build_accesses_from_assign): ...here.
181 (propagate_all_subaccesses): Make sure racc is the group
182 representative, if there is one.
184 2017-09-06 Jakub Jelinek <jakub@redhat.com>
187 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
190 2017-09-06 Richard Biener <rguenther@suse.de>
192 * gimple-ssa-strength-reduction.c
193 (find_candidates_dom_walker::before_doom_children): Use a
194 type and not a mode check.
196 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
199 * config/arm/predicates.md (arm_general_adddi_operand): Create new
201 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
203 2017-09-05 Jeff Law <law@redhat.com>
205 PR tree-optimization/64910
206 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
207 cases where we have 3 or more operands.
209 2017-09-05 Jakub Jelinek <jakub@redhat.com>
212 * omp-low.c (lower_omp_for): Recompute tree invariant if
213 gimple_omp_for_initial/final is ADDR_EXPR.
216 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
217 into gimple val before gimplification fo the COND_EXPR.
219 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
221 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
222 REGION_COPY argument.
223 (thread_through_all_blocks): Remove unused argument to
224 duplicate_thread_path.
226 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
227 Alan Hayward <alan.hayward@arm.com>
228 David Sherwood <david.sherwood@arm.com>
230 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
231 Take a scalar_mode rather than a machine_mode.
232 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
233 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
234 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
235 (aarch64_gen_adjusted_ldpstp): Likewise.
236 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
238 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
239 Alan Hayward <alan.hayward@arm.com>
240 David Sherwood <david.sherwood@arm.com>
242 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
243 Take a scalar_int_mode instead of a machine_mode.
244 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
245 (aarch64_output_scalar_simd_mov_immediate): Likewise.
246 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
247 (aarch64_simd_attr_length_rglist): Delete.
248 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
249 a scalar_int_mode instead of a machine_mode.
250 (aarch64_add_offset): Likewise.
251 (aarch64_internal_mov_immediate): Likewise
252 (aarch64_add_constant_internal): Likewise.
253 (aarch64_add_constant): Likewise.
254 (aarch64_movw_imm): Likewise.
255 (aarch64_rtx_arith_op_extract_p): Likewise.
256 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
257 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
258 Remove assert that the mode isn't a vector.
259 (aarch64_output_scalar_simd_mov_immediate): Likewise.
260 (aarch64_expand_mov_immediate): Update calls after above changes.
261 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
262 (aarch64_and_bitmask_imm): Check for scalar integer modes.
263 (aarch64_move_imm): Likewise.
264 (aarch64_can_const_movi_rtx_p): Likewise.
265 (aarch64_strip_extend): Likewise.
266 (aarch64_extr_rtx_p): Likewise.
267 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
268 a CONST_INT when the mode parameter is VOIDmode.
269 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
271 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
273 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
274 * stor-layout.c (bitwise_mode_for_mode): Likewise.
275 (bitwise_type_for_mode): Update accordingly.
277 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
279 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
280 * stor-layout.c (mode_for_size_tree): Likewise.
281 (mode_for_array): Update accordingly.
282 (layout_decl): Likewise.
283 (compute_record_mode): Likewise. Only set the mode once.
285 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
287 * target.def (get_mask_mode): Change return type to opt_mode.
289 * doc/tm.texi: Regenerate.
290 * targhooks.h (default_get_mask_mode): Return an opt_mode.
291 * targhooks.c (default_get_mask_mode): Likewise.
292 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
293 * optabs-query.c (can_vec_mask_load_store_p): Update use of
294 targetm.get_mask_mode.
295 * tree.c (build_truth_vector_type): Likewise.
297 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
299 * machmode.h (mode_for_vector): Return an opt_mode.
300 * stor-layout.c (mode_for_vector): Likewise.
301 (mode_for_int_vector): Update accordingly.
302 (layout_type): Likewise.
303 * config/i386/i386.c (emit_memmov): Likewise.
304 (ix86_expand_set_or_movmem): Likewise.
305 (ix86_expand_vector_init): Likewise.
306 (ix86_get_mask_mode): Likewise.
307 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
309 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
310 * expmed.c (extract_bit_field_1): Likewise.
311 * expr.c (expand_expr_real_2): Likewise.
312 * optabs-query.c (can_vec_perm_p): Likewise.
313 (can_vec_mask_load_store_p): Likewise.
314 * optabs.c (expand_vec_perm): Likewise.
315 * targhooks.c (default_get_mask_mode): Likewise.
316 * tree-vect-stmts.c (vectorizable_store): Likewise.
317 (vectorizable_load): Likewise.
318 (get_vectype_for_scalar_type_and_size): Likewise.
320 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
322 * machmode.h (mode_for_int_vector): New function.
323 * stor-layout.c (mode_for_int_vector): Likewise.
324 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
325 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
326 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
327 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
328 (s390_expand_vcond): Likewise.
330 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
332 * machmode.h (opt_machine_mode): New type.
333 (opt_mode<T>): Allow construction from anything that can be
335 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
336 (mode_for_size): Return an opt_machine_mode.
337 * stor-layout.c (mode_for_size): Likewise.
338 (mode_for_size_tree): Update call accordingly.
339 (bitwise_mode_for_mode): Likewise.
340 (make_fract_type): Likewise.
341 (make_accum_type): Likewise.
342 * caller-save.c (replace_reg_with_saved_mem): Update call
344 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
345 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
346 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
347 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
348 * expmed.c (extract_bit_field_1): Likewise.
349 * reload.c (get_secondary_mem): Likewise.
350 * varasm.c (assemble_integer): Likewise.
351 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
354 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
356 * machmode.h (decimal_float_mode_for_size): New function.
357 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
358 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
359 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
360 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
361 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
362 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
364 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
366 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
367 (get_builtin_sync_mode): Likewise.
368 (expand_ifn_atomic_compare_exchange): Likewise.
369 (expand_builtin_atomic_clear): Likewise.
370 (expand_builtin_atomic_test_and_set): Likewise.
371 (fold_builtin_atomic_always_lock_free): Likewise.
372 * calls.c (compute_argument_addresses): Likewise.
373 (emit_library_call_value_1): Likewise.
374 (store_one_arg): Likewise.
375 * combine.c (combine_instructions): Likewise.
376 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
377 * config/arm/arm.c (arm_function_value): Likewise.
378 (aapcs_allocate_return_reg): Likewise.
379 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
380 * config/i386/i386.c (construct_container): Likewise.
381 (ix86_gimplify_va_arg): Likewise.
382 (ix86_expand_sse_cmp): Likewise.
383 (emit_memmov): Likewise.
384 (emit_memset): Likewise.
385 (expand_small_movmem_or_setmem): Likewise.
386 (ix86_expand_pextr): Likewise.
387 (ix86_expand_pinsr): Likewise.
388 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
389 * config/microblaze/microblaze.c (microblaze_block_move_straight):
391 * config/mips/mips.c (mips_function_value_1) Likewise.
392 (mips_block_move_straight): Likewise.
393 (mips_expand_ins_as_unaligned_store): Likewise.
394 * config/powerpcspe/powerpcspe.c
395 (rs6000_darwin64_record_arg_advance_flush): Likewise.
396 (rs6000_darwin64_record_arg_flush): Likewise.
397 * config/rs6000/rs6000.c
398 (rs6000_darwin64_record_arg_advance_flush): Likewise.
399 (rs6000_darwin64_record_arg_flush): Likewise.
400 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
401 (sparc_function_value_1): Likewise.
402 * config/spu/spu.c (adjust_operand): Likewise.
403 (spu_emit_branch_or_set): Likewise.
404 (arith_immediate_p): Likewise.
405 * emit-rtl.c (gen_lowpart_common): Likewise.
406 * expr.c (expand_expr_real_1): Likewise.
407 * function.c (assign_parm_setup_block): Likewise.
408 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
409 * reload1.c (alter_reg): Likewise.
410 * stor-layout.c (mode_for_vector): Likewise.
411 (layout_type): Likewise.
413 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
415 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
416 (spu_convert_move): Likewise.
417 * lower-subreg.c (resolve_simple_move): Likewise.
419 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
422 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
423 define_insn to a define_expand.
424 (altivec_vsum2sws_direct): New define_insn.
425 (altivec_vsumsws): Convert from a define_insn to a define_expand.
427 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
429 * config/arm/arm.c (arm_option_params_internal): Improve setting of
432 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
437 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
438 realign stack if stack alignment needed is less than incoming
441 2017-09-05 Marek Polacek <polacek@redhat.com>
444 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
447 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
449 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
451 2017-09-05 Richard Biener <rguenther@suse.de>
453 PR tree-optimization/82084
454 * fold-const.c (can_native_encode_string_p): Handle wide characters.
456 2017-09-05 Richard Biener <rguenther@suse.de>
458 PR tree-optimization/82102
459 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
461 2017-09-05 Martin Liska <mliska@suse.cz>
463 PR tree-optimization/82032
464 * tree-cfg.c (generate_range_test): New function.
465 * tree-cfg.h (generate_range_test): Declared here.
466 * tree-cfgcleanup.c (convert_single_case_switch): New function.
467 (cleanup_control_expr_graph): Use it.
468 * tree-switch-conversion.c (try_switch_expansion): Remove
470 (emit_case_nodes): Use generate_range_test.
472 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
475 * config/i386/i386.md (*<btsc><mode>_mask): Add
476 TARGET_USE_BT to insn constraint.
477 (*btr<mode>_mask): Ditto.
479 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
481 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
482 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
484 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
487 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
488 TARGET_NEON and TARGET_IWMMXT.
489 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
490 TARGET_NEON and TARGET_IWMMXT.
491 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
493 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
495 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
496 (ix86_rewrite_tls_address): Ditto.
497 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
498 (ix86_rewrite_tls_address_1): Ditto.
499 (ix86_rewrite_tls_address): Ditto.
500 * config/i386/predicates.md (tls_address_pattern): New predicate.
501 * config/i386/i386.md (TLS address splitter): New splitter.
503 2017-09-04 Richard Biener <rguenther@suse.de>
505 PR tree-optimization/82084
506 * fold-const.h (can_native_encode_string_p): Declare.
507 * fold-const.c (can_native_encode_string_p): Factor out from ...
508 (native_encode_string): ... here.
509 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
510 vectorizing stores from constants we later cannot handle.
512 2017-09-04 Marek Polacek <polacek@redhat.com>
515 * doc/invoke.texi: Update -Wtautological-compare documentation.
517 2017-09-04 Jeff Law <law@redhat.com>
519 PR tree-optimization/64910
520 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
521 swap the first and last operand if the last is a constant.
523 2017-09-04 Marek Polacek <polacek@redhat.com>
526 * convert.c (do_narrow): When sanitizing signed integer overflows,
527 bail out for signed types.
528 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
530 2017-09-04 Richard Biener <rguenther@suse.de>
532 PR tree-optimization/82060
533 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
534 Move devirtualization after stmt folding and before EH/AB/noreturn
535 cleanup to get the stmt refs canonicalized. Use a bool instead
536 of gimple_modified_p since that doesn't work for NOPs. Schedule
537 NOPs generated by folding for removal.
539 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
540 Alan Hayward <alan.hayward@arm.com>
541 David Sherwood <david.sherwood@arm.com>
543 * coretypes.h (pad_direction): New enum.
544 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
545 (FUNCTION_ARG_PADDING): Likewise.
546 * target.def (function_arg_padding): New hook.
547 * targhooks.h (default_function_arg_padding): Declare.
548 * targhooks.c (default_function_arg_padding): New function.
549 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
550 (TARGET_FUNCTION_ARG_PADDING): ...this.
551 * doc/tm.texi: Regenerate.
552 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
553 instead of direction.
554 (compute_argument_addresses): Likewise.
555 (load_register_parameters): Likewise.
556 (emit_library_call_value_1): Likewise.
557 (store_one_arg): Use targetm.calls.function_arg_padding instead
558 of FUNCTION_ARG_PADDING.
559 (must_pass_in_stack_var_size_or_pad): Likewise.
560 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
561 (emit_group_store): Likewise.
562 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
563 instead of FUNCTION_ARG_PADDING.
564 (emit_push_insn): Likewise, and propagate enum change throughout
566 * function.h (direction): Delete.
567 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
569 * function.c (assign_parm_find_stack_rtl): Likewise.
570 (assign_parm_setup_block_p): Likewise.
571 (assign_parm_setup_block): Likewise.
572 (gimplify_parameters): Likewise.
573 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
574 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
576 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
577 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
578 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
579 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
580 (aarch64_function_arg_padding): ...this new function.
581 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
582 (TARGET_FUNCTION_ARG_PADDING): Redefine.
583 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
584 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
585 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
586 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
587 (arm_pad_arg_upward): Replace with...
588 (arm_function_arg_padding): ...this new function.
589 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
591 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
592 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
593 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
594 (ia64_hpux_function_arg_padding): Replace with...
595 (ia64_function_arg_padding): ...this new function. Use pad_direction
596 instead of direction. Check for TARGET_HPUX.
597 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
598 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
599 (iq2000_function_arg_padding): New function.
600 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
601 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
602 (mips_function_arg_padding): ...this new function.
603 (mips_pad_reg_upward): Update accordingly.
604 (TARGET_FUNCTION_ARG_PADDING): Redefine.
605 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
606 targetm.calls.function_arg_padding.
607 (FUNCTION_ARG_PADDING): Delete.
608 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
609 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
610 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
611 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
612 (nios2_block_reg_padding): Return pad_direction instead of direction.
613 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
614 instead of direction.
615 (nios2_function_arg_padding): Likewise. Make static.
616 (TARGET_FUNCTION_ARG_PADDING): Redefine.
617 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
618 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
619 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
620 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
621 (pa_function_arg_padding): Make static. Return pad_direction instead
623 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
624 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
625 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
626 instead of direction. Use targetm.calls.function_arg_padding.
627 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
628 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
629 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
630 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
631 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
633 (function_arg_padding): Rename to...
634 (rs6000_function_arg_padding): ...this. Make static. Return
635 pad_direction instead of direction.
636 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
637 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
638 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
639 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
640 instead of direction. Use targetm.calls.function_arg_padding.
641 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
642 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
643 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
644 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
645 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
646 (function_arg_padding): Rename to...
647 (rs6000_function_arg_padding): ...this. Make static. Return
648 pad_direction instead of direction.
649 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
650 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
651 * config/s390/s390.c (s390_function_arg_padding): New function.
652 (TARGET_FUNCTION_ARG_PADDING): Redefine.
653 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
654 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
655 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
656 (function_arg_padding): Rename to...
657 (sparc_function_arg_padding): ...this. Make static. Return
658 pad_direction instead of direction.
659 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
660 * config/spu/spu.c (spu_function_arg_padding): New function.
661 (TARGET_FUNCTION_ARG_PADDING): Redefine.
662 * system.h (FUNCTION_ARG_PADDING): Poison.
664 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
665 Alan Hayward <alan.hayward@arm.com>
666 David Sherwood <david.sherwood@arm.com>
668 * target.def (modes_tieable_p): New hook.
669 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
670 (TARGET_MODES_TIEABLE_P): ...this.
671 * doc/tm.texi.in: Regenerate.
672 * hooks.h (hook_bool_mode_mode_true): Declare.
673 * hooks.c (hook_bool_mode_mode_true): New function.
674 * combine.c (subst): Use targetm.modes_tieable_p instead of
676 * dse.c (find_shift_sequence): Likewise.
677 * expmed.c (extract_low_bits): Likewise.
678 * lower-subreg.c: Include target.h.
679 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
681 * rtlanal.c (rtx_cost): Likewise.
682 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
683 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
684 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
685 (TARGET_MODES_TIEABLE_P): Redefine.
686 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
687 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
688 (TARGET_MODES_TIEABLE_P): Redefine.
689 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
690 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
691 (arc_modes_tieable_p): New function.
692 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
693 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
694 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
695 (arm_modes_tieable_p): Make static.
696 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
697 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
698 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
699 (TARGET_MODES_TIEABLE_P): Redefine.
700 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
701 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
702 (TARGET_MODES_TIEABLE_P): Redefine.
703 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
704 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
705 (cr16_modes_tieable_p): New function.
706 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
707 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
708 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
709 (TRULY_NOOP_TRUNCATION): Update comment.
710 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
711 (TRULY_NOOP_TRUNCATION): Update comment.
712 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
713 (frv_modes_tieable_p): New function.
714 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
715 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
716 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
717 (TARGET_MODES_TIEABLE_P): Redefine.
718 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
719 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
720 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
721 (TARGET_MODES_TIEABLE_P): Redefine.
722 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
723 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
724 (ia64_modes_tieable_p): New function.
725 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
726 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
727 (iq2000_modes_tieable_p): New function.
728 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
729 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
730 (lm32_modes_tieable_p): New function.
731 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
732 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
733 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
734 (TARGET_MODES_TIEABLE_P): Redefine.
735 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
736 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
737 (m32r_modes_tieable_p): New function.
738 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
739 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
740 (m68k_modes_tieable_p): New function.
741 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
742 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
743 (mcore_modes_tieable_p): New function.
744 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
745 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
747 (TARGET_MODES_TIEABLE_P): Redefine.
748 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
749 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
750 * config/mips/mips.c (mips_modes_tieable_p): Make static.
751 (TARGET_MODES_TIEABLE_P): Redefine.
752 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
753 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
754 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
755 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
756 (mn10300_modes_tieable_p): ...this and make static.
757 (TARGET_MODES_TIEABLE_P): Redefine.
758 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
759 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
760 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
761 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
762 (msp430_modes_tieable_p): Make static.
763 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
764 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
765 (TARGET_MODES_TIEABLE_P): Redefine.
766 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
767 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
768 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
769 (TARGET_MODES_TIEABLE_P): Redefine.
770 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
771 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
772 * config/pa/pa.c (pa_modes_tieable_p): Make static.
773 (TARGET_MODES_TIEABLE_P): Redefine.
774 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
775 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
776 (pdp11_modes_tieable_p): New function.
777 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
778 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
779 (rs6000_modes_tieable_p): New function.
780 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
781 * config/powerpcspe/powerpcspe.md: Update comment.
782 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
783 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
784 (TARGET_MODES_TIEABLE_P): Redefine.
785 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
786 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
787 (rl78_modes_tieable_p): New function.
788 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
789 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
790 (rs6000_modes_tieable_p): New function.
791 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
792 * config/rs6000/rs6000.md: Update comment.
793 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
794 * config/rx/rx.c (rx_modes_tieable_p): New function.
795 (TARGET_MODES_TIEABLE_P): Redefine.
796 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
797 * config/s390/s390.c (s390_modes_tieable_p): New function.
798 (TARGET_MODES_TIEABLE_P): Redefine.
799 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
800 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
801 (sh_modes_tieable_p): New function.
802 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
803 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
804 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
805 (sparc_modes_tieable_p): Make static.
806 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
807 * config/spu/spu.c (spu_modes_tieable_p): New function.
808 (TARGET_MODES_TIEABLE_P): Redefine.
809 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
810 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
811 (TARGET_MODES_TIEABLE_P): Redefine.
812 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
813 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
814 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
815 * config/v850/v850.c (v850_modes_tieable_p): New function.
816 (TARGET_MODES_TIEABLE_P): Redefine.
817 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
818 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
819 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
820 (visium_modes_tieable_p): New function.
821 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
822 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
823 (xtensa_modes_tieable_p): New function.
824 * system.h (MODES_TIEABLE_P): Poison.
826 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
827 Alan Hayward <alan.hayward@arm.com>
828 David Sherwood <david.sherwood@arm.com>
830 * target.def (hard_regno_mode_ok): New hook.
831 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
832 (TARGET_HARD_REGNO_MODE_OK): ...this.
833 * doc/tm.texi.in: Regenerate.
834 * hooks.h (hook_bool_uint_mode_true): Declare.
835 * hooks.c (hook_bool_uint_mode_true): New function.
836 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
838 * genpreds.c (write_insn_preds_c): Add an include of target.h.
839 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
840 instead of HARD_REGNO_MODE_OK.
841 * caller-save.c: Include target.h.
842 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
844 * combine.c (can_combine_p): Likewise.
845 (combinable_i3pat): Likewise.
846 (can_change_dest_mode): Likewise.
847 * expr.c (init_expr_target): Likewise.
848 (convert_move): Likewise.
849 (convert_modes): Likewise.
850 * ira.c (setup_prohibited_class_mode_regs): Likewise.
851 (setup_prohibited_mode_move_regs): Likewise.
852 * ira.h (target_ira): Likewise.
853 * lra-assigns.c (find_hard_regno_for_1): Likewise.
854 * lra-constraints.c (process_alt_operands): Likewise.
855 (split_reg): Likewise.
856 * recog.c (peep2_find_free_register): Likewise.
857 * ree.c (combine_reaching_defs): Likewise.
858 * regcprop.c (maybe_mode_change): Likewise.
859 * reginfo.c (init_reg_sets_1): Likewise.
860 (choose_hard_reg_mode): Likewise.
861 (simplifiable_subregs): Likewise.
862 * regrename.c (check_new_reg_p): Likewise.
863 * reload.c (find_valid_class): Likewise.
864 (find_valid_class_1): Likewise.
865 (reload_inner_reg_of_subreg): Likewise.
866 (push_reload): Likewise.
867 (combine_reloads): Likewise.
868 (find_dummy_reload): Likewise.
869 (find_reloads): Likewise.
870 * reload1.c (find_reg): Likewise.
871 (set_reload_reg): Likewise.
872 (allocate_reload_reg): Likewise.
873 (choose_reload_regs): Likewise.
874 (reload_adjust_reg_for_temp): Likewise.
875 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
876 (simplify_subreg_regno): Likewise.
877 * sel-sched.c (init_regs_for_mode): Likewise.
878 * varasm.c (make_decl_rtl): Likewise.
879 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
880 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
882 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
883 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
884 (TARGET_HARD_REGNO_MODE_OK): Redefine.
885 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
886 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
887 (TARGET_HARD_REGNO_MODE_OK): Redefine.
888 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
889 (arc_mode_class): Delete.
890 (HARD_REGNO_MODE_OK): Delete.
891 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
892 (arc_hard_regno_mode_ok): Rename old array to...
893 (arc_hard_regno_mode_ok_modes): ...this.
894 (arc_conditional_register_usage): Update accordingly.
895 (arc_mode_class): Make static.
896 (arc_hard_regno_mode_ok): New function.
897 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
898 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
899 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
900 (arm_hard_regno_mode_ok): Make static.
901 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
903 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
904 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
905 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
907 (TARGET_HARD_REGNO_MODE_OK): Redefine.
908 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
909 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
910 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
911 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
912 (TARGET_HARD_REGNO_MODE_OK): Redefine.
913 * config/bfin/predicates.md (valid_reg_operand): Use
914 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
915 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
916 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
917 (TARGET_HARD_REGNO_MODE_OK): Redefine.
918 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
919 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
920 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
921 (cr16_hard_regno_mode_ok): Make static and return a bool.
922 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
923 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
924 (cris_hard_regno_mode_ok): New function.
925 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
926 (epiphany_mode_class): Delete.
927 (HARD_REGNO_MODE_OK): Delete.
928 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
929 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
930 (hard_regno_mode_ok): Rename to...
931 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
932 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
933 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
935 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
936 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
937 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
938 (frv_hard_regno_mode_ok): Make static and return a bool.
939 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
941 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
942 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
943 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
944 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
946 (TARGET_HARD_REGNO_MODE_OK): Redefine.
947 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
948 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
949 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
951 (TARGET_HARD_REGNO_MODE_OK): Redefine.
952 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
953 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
954 (ia64_hard_regno_mode_ok): New function.
955 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
956 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
957 (iq2000_hard_regno_mode_ok): New function.
958 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
959 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
960 (lm32_hard_regno_mode_ok): New function.
961 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
962 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
963 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
964 instead of HARD_REGNO_MODE_OK.
965 (m32c_hard_regno_ok): Rename to...
966 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
967 (m32c_cannot_change_mode_class): Update accordingly.
968 (TARGET_HARD_REGNO_MODE_OK): Redefine.
969 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
970 (m32r_mode_class): Delete.
971 (HARD_REGNO_MODE_OK): Delete.
972 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
973 (m32r_hard_regno_mode_ok): Rename to...
974 (m32r_hard_regno_modes): ...this.
975 (m32r_mode_class): Make static.
976 (m32r_hard_regno_mode_ok): New function.
977 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
978 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
979 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
980 (m68k_hard_regno_mode_ok): Make static.
981 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
982 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
983 (mcore_hard_regno_mode_ok): New function.
984 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
985 (HARD_REGNO_MODE_OK): Delete.
986 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
988 (microblaze_hard_regno_mode_ok_p): ...this and make static.
989 (microblaze_hard_regno_mode_ok): New function.
990 (TARGET_HARD_REGNO_MODE_OK): Redefine.
991 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
992 (mips_hard_regno_mode_ok): Delete.
993 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
994 (mips_hard_regno_mode_ok_p): ...this and make static.
995 (mips_hard_regno_mode_ok_p): Rename to...
996 (mips_hard_regno_mode_ok_uncached): ...this.
997 (mips_hard_regno_mode_ok): New function.
998 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
999 of HARD_REGNO_MODE_OK.
1000 (mips_option_override): Update after above name changes.
1001 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1002 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
1003 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
1004 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
1005 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
1006 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1007 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
1008 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
1009 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
1010 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1011 (msp430_hard_regno_mode_ok): Make static and return a bool.
1012 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
1013 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
1014 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
1016 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1017 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
1018 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
1019 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
1020 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
1021 (PA_HARD_REGNO_MODE_OK): ...this
1022 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
1023 (PA_HARD_REGNO_MODE_OK): ...this.
1024 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1025 (pa_hard_regno_mode_ok): New function.
1026 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
1027 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1028 (pdp11_hard_regno_mode_ok): New function.
1029 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
1030 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
1032 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
1034 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1035 (rs6000_hard_regno_mode_ok): Rename to...
1036 (rs6000_hard_regno_mode_ok_uncached): ...this.
1037 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1038 (rs6000_hard_regno_mode_ok): New function.
1039 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
1040 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
1041 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
1042 (riscv_hard_regno_mode_ok): ...this and make static.
1043 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1044 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
1045 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
1046 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1047 (rl78_hard_regno_mode_ok): Make static and return bool.
1048 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
1049 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
1051 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
1052 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1053 (rs6000_hard_regno_mode_ok): Rename to...
1054 (rs6000_hard_regno_mode_ok_uncached): ...this.
1055 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1056 (rs6000_hard_regno_mode_ok): New function.
1057 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
1058 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
1059 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1060 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
1061 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
1062 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
1063 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1064 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
1065 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
1066 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1067 (sh_hard_regno_mode_ok): Make static.
1068 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
1069 instead of HARD_REGNO_MODE_OK.
1070 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
1071 (sparc_mode_class): Delete.
1072 (HARD_REGNO_MODE_OK): Delete.
1073 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1074 (hard_regno_mode_classes): Make static.
1075 (sparc_mode_class): Likewise.
1076 (sparc_hard_regno_mode_ok): New function.
1077 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
1078 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
1079 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
1081 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1082 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
1083 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
1084 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
1085 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
1086 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1087 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
1088 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
1089 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1090 (visium_hard_regno_mode_ok): New function.
1091 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
1092 instead of HARD_REGNO_MODE_OK.
1093 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
1094 (HARD_REGNO_MODE_OK): Delete.
1095 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
1096 (xtensa_hard_regno_mode_ok_p): ...this and make static.
1097 (xtensa_option_override): Update accordingly.
1098 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1099 (xtensa_hard_regno_mode_ok): New function.
1100 * system.h (HARD_REGNO_MODE_OK): Poison.
1102 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1103 Alan Hayward <alan.hayward@arm.com>
1104 David Sherwood <david.sherwood@arm.com>
1106 * target.def (hard_regno_call_part_clobbered): New hook.
1107 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
1108 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
1109 * doc/tm.texi: Regenerate.
1110 * hooks.h (hook_bool_uint_mode_false): Declare.
1111 * hooks.c (hook_bool_uint_mode_false): New function.
1112 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1113 * cselib.c (cselib_process_insn): Use
1114 targetm.hard_regno_call_part_clobbered instead of
1115 HARD_REGNO_CALL_PART_CLOBBERED.
1116 * ira-conflicts.c (ira_build_conflicts): Likewise.
1117 * ira-costs.c (ira_tune_allocno_costs): Likewise.
1118 * lra-constraints.c (need_for_call_save_p): Likewise.
1119 * lra-lives.c: Include target.h.
1120 (check_pseudos_live_through_calls): Use
1121 targetm.hard_regno_call_part_clobbered instead of
1122 HARD_REGNO_CALL_PART_CLOBBERED.
1123 * regcprop.c: Include target.h.
1124 (copyprop_hardreg_forward_1): Use
1125 targetm.hard_regno_call_part_clobbered instead of
1126 HARD_REGNO_CALL_PART_CLOBBERED.
1127 * reginfo.c (choose_hard_reg_mode): Likewise.
1128 * regrename.c (check_new_reg_p): Likewise.
1129 * reload.c (find_equiv_reg): Likewise.
1130 * reload1.c (emit_reload_insns): Likewise.
1131 * sched-deps.c (deps_analyze_insn): Likewise.
1132 * sel-sched.c (init_regs_for_mode): Likewise.
1133 (mark_unavailable_hard_regs): Likewise.
1134 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
1135 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1136 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
1138 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1139 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1140 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
1142 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
1144 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1145 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1146 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
1148 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1149 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1150 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
1152 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1153 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
1155 * config/powerpcspe/powerpcspe.c
1156 (rs6000_hard_regno_call_part_clobbered): New function.
1157 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1158 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1159 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
1161 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1162 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1163 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
1165 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1166 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1167 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
1169 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1170 Alan Hayward <alan.hayward@arm.com>
1171 David Sherwood <david.sherwood@arm.com>
1173 * rtl.h (subreg_memory_offset): Declare.
1174 * emit-rtl.c (subreg_memory_offset): New function.
1175 * expmed.c (store_bit_field_1): Use it.
1176 * expr.c (undefined_operand_subword_p): Likewise.
1177 * simplify-rtx.c (simplify_subreg): Likewise.
1179 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
1181 PR rtl-optimization/57448
1184 * optabs.c (expand_atomic_load): Place compiler memory barriers if
1185 using atomic_load pattern.
1186 (expand_atomic_store): Likewise.
1188 2017-09-04 Jakub Jelinek <jakub@redhat.com>
1191 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
1192 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
1193 handling. Use replace_call_with_value with NULL instead of
1194 gsi_replace, unlink_stmt_vdef and release_defs.
1196 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
1199 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
1201 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1204 * rtl.h (emit_library_call_value_1): Declare.
1205 (emit_library_call): Replace declaration with a series of overloads.
1206 Remove the parameter count argument.
1207 (emit_library_call_value): Likewise.
1208 * calls.c (emit_library_call_value_1): Make global. Replace varargs
1209 with an "rtx_mode_t *".
1210 (emit_library_call_value): Delete.
1211 (emit_library_call): Likewise.
1212 * asan.c (asan_emit_stack_protection): Update calls accordingly.
1213 (asan_emit_allocas_unpoison): Likewise.
1214 * builtins.c (expand_builtin_powi): Likewise.
1215 (expand_asan_emit_allocas_unpoison): Likewise.
1216 * cfgexpand.c (expand_main_function): Likewise.
1217 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
1218 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
1219 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
1220 * config/arm/arm.c (arm_trampoline_init): Likewise.
1221 (arm_call_tls_get_addr): Likewise.
1222 (arm_expand_divmod_libfunc): Likewise.
1223 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
1224 (smulsi3_highpart): Likewise.
1225 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
1226 (c6x_expand_compare): Likewise.
1227 (c6x_expand_movmem): Likewise.
1228 * config/frv/frv.c (frv_trampoline_init): Likewise.
1229 * config/i386/i386.c (ix86_trampoline_init): Likewise.
1230 (ix86_expand_divmod_libfunc): Likewise.
1231 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
1232 (ia64_expand_compare): Likewise.
1233 (ia64_profile_hook): Likewise.
1234 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
1235 (nonlocal_goto): Likewise.
1236 (restore_stack_nonlocal): Likewise.
1237 * config/m32r/m32r.c (block_move_call): Likewise.
1238 (m32r_trampoline_init): Likewise.
1239 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
1240 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
1241 (m68k_call_m68k_read_tp): Likewise.
1242 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
1243 (microblaze_expand_divide): Likewise.
1244 * config/mips/mips.h (mips_args): Likewise.
1245 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
1246 (MIPS_ICACHE_SYNC): Likewise.
1247 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
1248 (nios2_trampoline_init): Likewise.
1249 * config/pa/pa.c (hppa_tls_call): Likewise.
1250 (pa_trampoline_init): Likewise.
1251 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
1252 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
1253 (expand_strn_compare): Likewise.
1254 (rs6000_generate_compare): Likewise.
1255 (rs6000_expand_float128_convert): Likewise.
1256 (output_profile_hook): Likewise.
1257 (rs6000_trampoline_init): Likewise.
1258 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
1259 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
1260 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
1261 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
1262 (rs6000_generate_compare): Likewise.
1263 (rs6000_expand_float128_convert): Likewise.
1264 (output_profile_hook): Likewise.
1265 (rs6000_trampoline_init): Likewise.
1266 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
1267 * config/sh/sh.c (sh_trampoline_init): Likewise.
1268 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
1269 (sparc_emit_float_lib_cmp): Likewise.
1270 (sparc32_initialize_trampoline): Likewise.
1271 (sparc64_initialize_trampoline): Likewise.
1272 (sparc_profile_hook): Likewise.
1273 * config/spu/spu.c (ea_load_store): Likewise.
1274 * config/spu/spu.md (floatunssidf2): Likewise.
1275 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
1276 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
1277 * config/visium/visium.c (expand_block_move_4): Likewise.
1278 (expand_block_move_2): Likewise.
1279 (expand_block_move_1): Likewise.
1280 (expand_block_set_4): Likewise.
1281 (expand_block_set_2): Likewise.
1282 (expand_block_set_1): Likewise.
1283 (visium_trampoline_init): Likewise.
1284 (visium_profile_hook): Likewise.
1285 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
1286 (xtensa_setup_frame_addresses): Likewise.
1287 (xtensa_trampoline_init): Likewise.
1288 * except.c (sjlj_emit_function_enter): Likewise.
1289 (sjlj_emit_function_exit): Likewise.
1290 * explow.c (allocate_dynamic_stack_space): Likewise.
1291 (probe_stack_range): Likewise.
1292 * expr.c (convert_mode_scalar): Likewise.
1293 * optabs.c (expand_binop): Likewise.
1294 (expand_twoval_binop_libfunc): Likewise.
1295 (expand_unop): Likewise.
1296 (prepare_cmp_insn): Likewise.
1297 (prepare_float_lib_cmp): Likewise.
1298 (expand_float): Likewise.
1299 (expand_fix): Likewise.
1300 (expand_fixed_convert): Likewise.
1301 (maybe_emit_sync_lock_test_and_set): Likewise.
1302 (expand_atomic_compare_and_swap): Likewise.
1303 (expand_mem_thread_fence): Likewise.
1304 (expand_atomic_fetch_op): Likewise.
1306 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
1308 * doc/generic.texi (OpenACC): Adjust URL.
1309 * doc/invoke.texi (C Dialect Options): Ditto.
1311 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
1313 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
1314 predicate for operand 1. Add (m,<S>) constraint.
1315 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
1316 Prevent memory operand 1 with register operand 2.
1318 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
1320 PR rtl-optimization/82024
1321 * combine.c (try_combine): If the combination result is a PARALLEL,
1322 and we only need to retain the SET in there that would be placed
1323 at I2, check that we can place that at I3 instead, before doing so.
1325 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1328 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
1330 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
1331 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
1334 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
1335 Jeff Law <law@redhat.com>
1337 * varasm.c (bss_initializer_p): Do not put constants into .bss
1338 (categorize_decl_for_section): Handle bss_initializer_p returning
1339 false when DECL_INITIAL is NULL.
1341 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1344 * config/s390/s390.c (s390_can_inline_p): New function.
1346 2017-09-01 Jeff Law <law@redhat.com>
1348 PR tree-optimization/82052
1349 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
1350 Always initialize the returned slot after a hash table miss
1351 when INSERT is true.
1353 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
1355 * config/s390/s390.md (mem_signal_fence): Remove.
1356 * doc/md.texi (mem_signal_fence): Remove.
1357 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
1359 * target-insns.def (mem_signal_fence): Remove.
1361 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1364 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
1367 * asan.c (create_odr_indicator): Strip name encoding from assembler
1368 name before appending it after __odr_asan_.
1370 2017-09-01 Martin Liska <mliska@suse.cz>
1372 PR tree-optimization/82059
1373 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
1374 frequency only when an edge is redirected.
1376 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1378 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
1379 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
1380 (arc_conditional_register_usage): Remove ARC600 lp_count
1382 (arc_file_start): Emit Tag_ARC_CPU_variation.
1383 (arc_can_use_doloop_p): New conditions to use ZOLs.
1384 (hwloop_fail): New function.
1385 (hwloop_optimize): Likewise.
1386 (hwloop_pattern_reg): Likewise.
1387 (arc_doloop_hooks): New struct, to be used with reorg_loops.
1388 (arc_reorg_loops): New function, calls reorg_loops.
1389 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
1390 (arc600_corereg_hazard): Remove ZOL checking, case handled by
1392 (arc_loop_hazard): Remove function, functionality moved into
1394 (arc_hazard): Remove arc_loop_hazard call.
1395 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
1396 into hwloop_optimize.
1397 (arc_label_align): Remove ZOL handling.
1398 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
1399 * config/arc/arc.md (doloop_begin): Remove pattern.
1400 (doloop_begin_i): Likewise.
1401 (doloop_end_i): Likewise.
1402 (doloop_fallback): Likewise.
1403 (doloop_fallback_m): Likewise.
1404 (doloop_end): Reimplement expand.
1405 (arc_lp): New pattern for LP instruction.
1406 (loop_end): New pattern.
1407 (loop_fail): Likewise.
1408 (decrement_and_branch_until_zero): Likewise.
1409 * config/arc/arc.opt (mlpc-width): New option.
1410 * doc/invoke.texi (mlpc-width): Document option.
1412 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1414 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
1415 (arc_ccfsm_advance): Fix checking for delay slots.
1416 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
1418 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1420 * config/arc/arc.md (movqi_insn): Add stores to save constant long
1422 (movhi_insn): Update store instruction constraint which are saving
1423 6-bit short immediates.
1424 (movsi_insn): Consider also short scaled load operations.
1425 (zero_extendhisi2_i): Use Usd constraint instead of T.
1426 (extendhisi2_i): Add q constraint.
1427 (arc_clzsi2): Add type and length attributes.
1428 (arc_ctzsi2): Likewise.
1429 * config/arc/constraints.md (Usc): Update constraint, the
1430 assembler can parse two relocations for a single instruction.
1432 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1434 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
1435 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
1437 2017-08-31 Olivier Hainque <hainque@adacore.com>
1439 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
1440 match as powerpc-wrs-vxworks*.
1442 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
1444 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
1445 register constraint for by-element operand.
1446 (aarch64_mls_elt_merge<mode>): Likewise.
1448 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1450 * config/arc/arc.c (arc_can_follow_jump): Check for short
1453 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1455 * config.gcc: Use g.opt for arc.
1456 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
1457 functionality moved to ...
1458 (legitimate_scaled_address_p): New function, ...here.
1459 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
1460 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
1461 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
1463 (arc_override_options): Handle G option.
1464 (arc_output_pic_addr_const): Correct function definition.
1465 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
1466 (arc_decl_anon_ns_mem_p): Delete.
1467 (arc_in_small_data_p): Overhaul this function to take into
1468 consideration the value given via G option.
1469 (arc_rewrite_small_data_1): Renamed and corrected old
1470 arc_rewrite_small_data function.
1471 (arc_rewrite_small_data): New function.
1472 (small_data_pattern): Don't use pic_offset_table_rtx.
1473 * config/arc/arc.h (CC1_SPEC): Recognize G option.
1474 * config/arc/simdext.md (movmisalignv2hi): Use
1475 prepare_move_operands function.
1477 (movmisalign*): Likewise.
1478 * doc/invoke.texi (ARC options): Document -G option.
1480 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1482 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
1484 * config/arc/arc.c (arc_print_operand): Output scalled address for
1485 sdata whenever is possible.
1486 (arc_in_small_data_p): Allow sdata for 64bit datum when double
1487 load/stores are available.
1488 (compact_sda_memory_operand): Check for the alignment required by
1489 code density instructions.
1490 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
1492 * config/arc/constraints.md (Usd): Update constraint.
1493 (Us0): New constraint.
1494 (Usc): Update constraint.
1496 2017-08-31 Richard Biener <rguenther@suse.de>
1499 * dwarf2out.c (dwarf2out_early_global_decl): Process each
1502 2017-08-31 Tamar Christina <tamar.christina@arm.com>
1504 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
1505 Resize type_signature.
1507 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1508 Alan Hayward <alan.hayward@arm.com>
1509 David Sherwood <david.sherwood@arm.com>
1511 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
1512 subregs whose inner modes can be stored in GPRs.
1513 (aarch64_classify_index): Likewise.
1515 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1516 Alan Hayward <alan.hayward@arm.com>
1517 David Sherwood <david.sherwood@arm.com>
1519 * config/aarch64/iterators.md (V_cmp_result): Rename to...
1520 (V_INT_EQUIV): ...this.
1521 (v_cmp_result): Rename to...
1522 (v_int_equiv): ...this.
1523 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
1524 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
1525 (copysign<mode>3): Likewise.
1526 (aarch64_simd_bsl<mode>_internal): Likewise.
1527 (aarch64_simd_bsl<mode>): Likewise.
1528 (vec_cmp<mode><mode>): Likewise.
1529 (vcond<mode><mode>): Likewise.
1530 (vcond<v_cmp_mixed><mode>): Likewise.
1531 (vcondu<mode><v_cmp_mixed>): Likewise.
1532 (aarch64_cm<optab><mode>): Likewise.
1533 (aarch64_cmtst<mode>): Likewise.
1534 (aarch64_fac<optab><mode>): Likewise.
1535 (vec_perm_const<mode>): Likewise.
1536 (vcond_mask_<mode><v_cmp_result>): Rename to...
1537 (vcond_mask_<mode><v_int_equiv>): ...this.
1538 (vec_cmp<mode><v_cmp_result>): Rename to...
1539 (vec_cmp<mode><v_int_equiv>): ...this.
1541 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1542 Alan Hayward <alan.hayward@arm.com>
1543 David Sherwood <david.sherwood@arm.com>
1545 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
1547 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
1548 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
1549 (UNSPEC_LD4_DREG): New unspecs.
1550 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
1551 (aarch64_ld2<mode>_dreg_be): Replace with...
1552 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
1554 (aarch64_ld3<mode>_dreg_le)
1555 (aarch64_ld3<mode>_dreg_be): Replace with...
1556 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
1558 (aarch64_ld4<mode>_dreg_le)
1559 (aarch64_ld4<mode>_dreg_be): Replace with...
1560 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
1563 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1565 PR tree-optimization/81987
1566 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
1567 insert an initializer in a location not dominated by the stride
1570 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1572 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
1573 on the entire header of the finally block in the fallthru case.
1575 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1577 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
1579 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
1581 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
1582 rs6000_emit_move_from_cr and call renamed function.
1583 (rs6000_emit_prologue): Call renamed functions.
1584 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
1585 movesi_from_cr, remove volatile CRs.
1587 2017-08-30 Jon Beniston <jon@beniston.com>
1588 Richard Biener <rguenther@suse.de>
1590 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
1591 of VECTOR_MODE_P check.
1592 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
1593 element vector types.
1595 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1597 * df.h (df_read_modify_subreg_p): Remove in favor of...
1598 * rtl.h (read_modify_subreg_p): ...this new function. Take a
1599 const_rtx instead of an rtx.
1600 * cprop.c (local_cprop_find_used_regs): Update accordingly.
1601 * df-problems.c (df_word_lr_mark_ref): Likewise.
1602 * ira-lives.c (mark_pseudo_reg_live): Likewise.
1603 (mark_pseudo_reg_dead): Likewise.
1604 (mark_ref_dead): Likewise.
1605 * reginfo.c (init_subregs_of_mode): Likewise.
1606 * sched-deps.c (sched_analyze_1): Likewise.
1607 * df-scan.c (df_def_record_1): Likewise.
1608 (df_uses_record): Likewise.
1609 (df_read_modify_subreg_p): Remove in favor of...
1610 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
1611 const_rtx instead of an rtx.
1613 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1614 Alan Hayward <alan.hayward@arm.com>
1615 David Sherwood <david.sherwood@arm.com>
1617 * rtl.h (partial_subreg_p): New function.
1618 * caller-save.c (save_call_clobbered_regs): Use it.
1619 * calls.c (expand_call): Likewise.
1620 * combine.c (combinable_i3pat): Likewise.
1621 (simplify_set): Likewise.
1622 (make_extraction): Likewise.
1623 (make_compound_operation_int): Likewise.
1624 (gen_lowpart_or_truncate): Likewise.
1625 (force_to_mode): Likewise.
1626 (make_field_assignment): Likewise.
1627 (reg_truncated_to_mode): Likewise.
1628 (record_truncated_value): Likewise.
1629 (move_deaths): Likewise.
1630 * cse.c (record_jump_cond): Likewise.
1631 (cse_insn): Likewise.
1632 * cselib.c (cselib_lookup_1): Likewise.
1633 * expmed.c (extract_bit_field_using_extv): Likewise.
1634 * function.c (assign_parm_setup_reg): Likewise.
1635 * ifcvt.c (noce_convert_multiple_sets): Likewise.
1636 * ira-build.c (create_insn_allocnos): Likewise.
1637 * lra-coalesce.c (merge_pseudos): Likewise.
1638 * lra-constraints.c (match_reload): Likewise.
1639 (simplify_operand_subreg): Likewise.
1640 (curr_insn_transform): Likewise.
1641 * lra-lives.c (process_bb_lives): Likewise.
1642 * lra.c (new_insn_reg): Likewise.
1643 (lra_substitute_pseudo): Likewise.
1644 * regcprop.c (mode_change_ok): Likewise.
1645 (maybe_mode_change): Likewise.
1646 (copyprop_hardreg_forward_1): Likewise.
1647 * reload.c (push_reload): Likewise.
1648 (find_reloads): Likewise.
1649 (find_reloads_subreg_address): Likewise.
1650 * reload1.c (alter_reg): Likewise.
1651 (eliminate_regs_1): Likewise.
1652 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1654 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
1656 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
1657 back to if statements, including unpack.
1659 2017-08-30 Martin Liska <mliska@suse.cz>
1662 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
1664 (func_checker::compare_asm_inputs_outputs): ... this function.
1665 (func_checker::compare_gimple_asm): Use the function to compare
1666 also ASM constrains.
1667 * ipa-icf-gimple.h: Rename the function.
1669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1670 Alan Hayward <alan.hayward@arm.com>
1671 David Sherwood <david.sherwood@arm.com>
1673 * coretypes.h (complex_mode): New type.
1674 * gdbhooks.py (build_pretty_printer): Handle it.
1675 * machmode.h (complex_mode): New class.
1676 (complex_mode::includes_p): New function.
1677 (is_complex_int_mode): Likewise.
1678 (is_complex_float_mode): Likewise.
1679 * genmodes.c (get_mode_class): Handle complex mode classes.
1680 * function.c (expand_function_end): Use is_complex_int_mode.
1682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1683 Alan Hayward <alan.hayward@arm.com>
1684 David Sherwood <david.sherwood@arm.com>
1686 * coretypes.h (scalar_mode_pod): New typedef.
1687 * gdbhooks.py (build_pretty_printer): Handle it.
1688 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
1689 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
1690 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
1691 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
1694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1695 Alan Hayward <alan.hayward@arm.com>
1696 David Sherwood <david.sherwood@arm.com>
1698 * machmode.h (mode_for_vector): Take a scalar_mode instead
1700 * stor-layout.c (mode_for_vector): Likewise.
1701 * explow.c (promote_mode): Use as_a <scalar_mode>.
1702 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
1704 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1705 Alan Hayward <alan.hayward@arm.com>
1706 David Sherwood <david.sherwood@arm.com>
1708 * target.def (preferred_simd_mode): Take a scalar_mode
1709 instead of a machine_mode.
1710 * targhooks.h (default_preferred_simd_mode): Likewise.
1711 * targhooks.c (default_preferred_simd_mode): Likewise.
1712 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
1713 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
1714 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
1715 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
1716 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
1717 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
1718 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
1719 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
1721 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
1722 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
1723 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
1724 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
1725 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
1726 * doc/tm.texi: Regenerate.
1727 * optabs-query.c (can_vec_mask_load_store_p): Return false for
1730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1731 Alan Hayward <alan.hayward@arm.com>
1732 David Sherwood <david.sherwood@arm.com>
1734 * target.def (scalar_mode_supported_p): Take a scalar_mode
1735 instead of a machine_mode.
1736 * targhooks.h (default_scalar_mode_supported_p): Likewise.
1737 * targhooks.c (default_scalar_mode_supported_p): Likewise.
1738 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
1739 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1740 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
1741 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
1742 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
1743 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
1744 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
1745 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
1746 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
1747 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
1748 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
1749 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
1751 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
1752 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
1753 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1754 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1755 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
1757 * doc/tm.texi: Regenerate.
1759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1760 Alan Hayward <alan.hayward@arm.com>
1761 David Sherwood <david.sherwood@arm.com>
1763 * coretypes.h (opt_scalar_mode): New typedef.
1764 * gdbhooks.py (build_pretty_printers): Handle it.
1765 * machmode.h (mode_iterator::get_2xwider): Add overload for
1767 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
1769 * expr.c (convert_mode_scalar): Likewise.
1770 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1771 * optabs.c (expand_float): Likewise.
1772 (expand_fix): Likewise.
1773 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1775 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1776 Alan Hayward <alan.hayward@arm.com>
1777 David Sherwood <david.sherwood@arm.com>
1779 * optabs.c (expand_float): Explicitly check for scalars before
1780 using a branching expansion.
1781 (expand_fix): Likewise.
1783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1784 Alan Hayward <alan.hayward@arm.com>
1785 David Sherwood <david.sherwood@arm.com>
1787 * expr.c (convert_mode): Split scalar handling out into...
1788 (convert_mode_scalar): ...this new function. Treat the modes
1791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1792 Alan Hayward <alan.hayward@arm.com>
1793 David Sherwood <david.sherwood@arm.com>
1795 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
1797 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
1799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1800 Alan Hayward <alan.hayward@arm.com>
1801 David Sherwood <david.sherwood@arm.com>
1803 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
1804 rather than a machine_mode.
1805 (fixed_from_string): Likewise.
1806 (fixed_convert): Likewise.
1807 (fixed_convert_from_int): Likewise.
1808 (fixed_convert_from_real): Likewise.
1809 (real_convert_from_fixed): Likewise.
1810 * fixed-value.c (fixed_from_double_int): Likewise.
1811 (fixed_from_string): Likewise.
1812 (fixed_convert): Likewise.
1813 (fixed_convert_from_int): Likewise.
1814 (fixed_convert_from_real): Likewise.
1815 (real_convert_from_fixed): Likewise.
1816 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
1818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1819 Alan Hayward <alan.hayward@arm.com>
1820 David Sherwood <david.sherwood@arm.com>
1822 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
1823 of separate mode class checks. Do not allow vector modes here.
1824 (immed_wide_int_const): Use as_a <scalar_mode>.
1825 * explow.c (trunc_int_for_mode): Likewise.
1826 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
1827 (wi::shwi): Likewise.
1828 (wi::min_value): Likewise.
1829 (wi::max_value): Likewise.
1830 * dwarf2out.c (loc_descriptor): Likewise.
1831 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
1834 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1835 Alan Hayward <alan.hayward@arm.com>
1836 David Sherwood <david.sherwood@arm.com>
1838 * tree.h (SCALAR_TYPE_MODE): New macro.
1839 * expr.c (expand_expr_addr_expr_1): Use it.
1840 (expand_expr_real_2): Likewise.
1841 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
1842 (fold_convert_const_fixed_from_int): Likewise.
1843 (fold_convert_const_fixed_from_real): Likewise.
1844 (native_encode_fixed): Likewise
1845 (native_encode_complex): Likewise
1846 (native_encode_vector): Likewise.
1847 (native_interpret_fixed): Likewise.
1848 (native_interpret_real): Likewise.
1849 (native_interpret_complex): Likewise.
1850 (native_interpret_vector): Likewise.
1851 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
1852 (simd_clone_adjust_argument_types): Likewise.
1853 (simd_clone_init_simd_arrays): Likewise.
1854 (simd_clone_adjust): Likewise.
1855 * stor-layout.c (layout_type): Likewise.
1856 * tree.c (build_minus_one_cst): Likewise.
1857 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
1858 * tree-inline.c (estimate_move_cost): Likewise.
1859 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
1860 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
1861 (vectorizable_reduction): Likewise.
1862 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
1863 (vect_recog_mixed_size_cond_pattern): Likewise.
1864 (check_bool_pattern): Likewise.
1865 (adjust_bool_pattern): Likewise.
1866 (search_type_for_mask_1): Likewise.
1867 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
1868 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1869 (vectorizable_load): Likewise.
1870 (vectorizable_store): Likewise.
1871 * ubsan.c (ubsan_encode_value): Likewise.
1872 * varasm.c (output_constant): Likewise.
1874 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1875 Alan Hayward <alan.hayward@arm.com>
1876 David Sherwood <david.sherwood@arm.com>
1878 * coretypes.h (scalar_mode): New class.
1879 * machmode.h (scalar_mode): Likewise.
1880 (scalar_mode::includes_p): New function.
1881 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
1882 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
1883 * genmodes.c (get_mode_class): Handle remaining scalar modes.
1884 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
1885 * expmed.c (store_bit_field_1): Likewise.
1886 (extract_bit_field_1): Likewise.
1887 * expr.c (write_complex_part): Likewise.
1888 (read_complex_part): Likewise.
1889 (emit_move_complex_push): Likewise.
1890 (expand_expr_real_2): Likewise.
1891 * function.c (assign_parm_setup_reg): Likewise.
1892 (assign_parms_unsplit_complex): Likewise.
1893 * optabs.c (expand_binop): Likewise.
1894 * rtlanal.c (subreg_get_info): Likewise.
1895 * simplify-rtx.c (simplify_immed_subreg): Likewise.
1896 * varasm.c (output_constant_pool_2): Likewise.
1898 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1899 Alan Hayward <alan.hayward@arm.com>
1900 David Sherwood <david.sherwood@arm.com>
1902 * expmed.c (extract_high_half): Use scalar_int_mode and remove
1904 (expmed_mult_highpart_optab): Likewise.
1905 (expmed_mult_highpart): Likewise.
1907 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1908 Alan Hayward <alan.hayward@arm.com>
1909 David Sherwood <david.sherwood@arm.com>
1911 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
1912 instead of a machine_mode.
1913 (builtin_memset_read_str): Likewise.
1914 * builtins.c (c_readstr): Likewise.
1915 (builtin_memcpy_read_str): Likewise.
1916 (builtin_strncpy_read_str): Likewise.
1917 (builtin_memset_read_str): Likewise.
1918 (builtin_memset_gen_str): Likewise.
1919 (expand_builtin_signbit): Use scalar_int_mode for local variables.
1920 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
1921 instead of a machine_mode.
1922 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
1924 (make_extraction): Likewise.
1925 (try_widen_shift_mode): Take and return scalar_int_modes instead
1927 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
1928 a scalar_int_mode instead of a machine_mode.
1929 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
1930 (avr_addr_space_pointer_mode): Likewise.
1931 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
1932 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
1933 (msp430_unwind_word_mode): Likewise.
1934 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
1935 (spu_addr_space_pointer_mode): Likewise.
1936 (spu_addr_space_address_mode): Likewise.
1937 (spu_libgcc_cmp_return_mode): Likewise.
1938 (spu_libgcc_shift_count_mode): Likewise.
1939 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
1940 (rl78_addr_space_pointer_mode): Likewise.
1941 (fl78_unwind_word_mode): Likewise.
1942 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
1944 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
1945 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
1946 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
1947 (mips_valid_pointer_mode): Likewise.
1948 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
1949 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
1950 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
1952 (ft32_addr_space_address_mode): Likewise.
1953 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
1954 scalar_int_mode instead of a machine_mode.
1955 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
1957 (m32c_addr_space_address_mode): Likewise.
1958 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
1959 (rs6000_eh_return_filter_mode): Likewise.
1960 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
1961 (rs6000_eh_return_filter_mode): Likewise.
1962 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
1963 (s390_libgcc_shift_count_mode): Likewise.
1964 (s390_unwind_word_mode): Likewise.
1965 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
1967 * target.def (mode_rep_extended): Likewise.
1968 (valid_pointer_mode): Likewise.
1969 (addr_space.valid_pointer_mode): Likewise.
1970 (eh_return_filter_mode): Return a scalar_int_mode rather than
1972 (libgcc_cmp_return_mode): Likewise.
1973 (libgcc_shift_count_mode): Likewise.
1974 (unwind_word_mode): Likewise.
1975 (addr_space.pointer_mode): Likewise.
1976 (addr_space.address_mode): Likewise.
1977 * doc/tm.texi: Regenerate.
1978 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
1980 (do_jump): Use scalar_int_mode for local variables.
1981 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
1982 rather than a machine_mode.
1983 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
1984 (scompare_loc_descriptor_wide): Likewise.
1985 (scompare_loc_descriptor_narrow): Likewise.
1986 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
1988 * except.c (sjlj_emit_dispatch_table): Likewise.
1989 (expand_builtin_eh_copy_values): Likewise.
1990 * explow.c (convert_memory_address_addr_space_1): Likewise.
1991 Take a scalar_int_mode rather than a machine_mode.
1992 (convert_memory_address_addr_space): Take a scalar_int_mode rather
1993 than a machine_mode.
1994 (memory_address_addr_space): Use scalar_int_mode for local variables.
1995 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
1996 rather than a machine_mode.
1997 * expmed.c (mask_rtx): Likewise.
1998 (init_expmed_one_conv): Likewise.
1999 (expand_mult_highpart_adjust): Likewise.
2000 (extract_high_half): Likewise.
2001 (expmed_mult_highpart_optab): Likewise.
2002 (expmed_mult_highpart): Likewise.
2003 (expand_smod_pow2): Likewise.
2004 (expand_sdiv_pow2): Likewise.
2005 (emit_store_flag_int): Likewise.
2006 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
2008 (extract_low_bits): Likewise.
2009 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
2011 * expr.c (pieces_addr::adjust): Likewise.
2012 (can_store_by_pieces): Likewise.
2013 (store_by_pieces): Likewise.
2014 (clear_by_pieces_1): Likewise.
2015 (expand_expr_addr_expr_1): Likewise.
2016 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
2017 (expand_expr_real_1): Likewise.
2018 (try_casesi): Likewise.
2019 * final.c (shorten_branches): Likewise.
2020 * fold-const.c (fold_convert_const_int_from_fixed): Change the
2021 type of "mode" to machine_mode.
2022 * internal-fn.c (expand_arith_overflow_result_store): Take a
2023 scalar_int_mode rather than a machine_mode.
2024 (expand_mul_overflow): Use scalar_int_mode for local variables.
2025 * loop-doloop.c (doloop_modify): Likewise.
2026 (doloop_optimize): Likewise.
2027 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
2028 than a machine_mode.
2029 (expand_doubleword_shift_condmove): Likewise.
2030 (expand_doubleword_shift): Likewise.
2031 (expand_doubleword_clz): Likewise.
2032 (expand_doubleword_popcount): Likewise.
2033 (expand_doubleword_parity): Likewise.
2034 (expand_absneg_bit): Use scalar_int_mode for local variables.
2035 (prepare_float_lib_cmp): Likewise.
2036 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
2037 rather than a machine_mode.
2038 (convert_memory_address_addr_space): Likewise.
2039 (get_mode_bounds): Likewise.
2040 (get_address_mode): Return a scalar_int_mode rather than a
2042 * rtlanal.c (get_address_mode): Likewise.
2043 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
2044 than a machine_mode.
2045 * targhooks.c (default_mode_rep_extended): Likewise.
2046 (default_valid_pointer_mode): Likewise.
2047 (default_addr_space_valid_pointer_mode): Likewise.
2048 (default_eh_return_filter_mode): Return a scalar_int_mode rather
2049 than a machine_mode.
2050 (default_libgcc_cmp_return_mode): Likewise.
2051 (default_libgcc_shift_count_mode): Likewise.
2052 (default_unwind_word_mode): Likewise.
2053 (default_addr_space_pointer_mode): Likewise.
2054 (default_addr_space_address_mode): Likewise.
2055 * targhooks.h (default_eh_return_filter_mode): Likewise.
2056 (default_libgcc_cmp_return_mode): Likewise.
2057 (default_libgcc_shift_count_mode): Likewise.
2058 (default_unwind_word_mode): Likewise.
2059 (default_addr_space_pointer_mode): Likewise.
2060 (default_addr_space_address_mode): Likewise.
2061 (default_mode_rep_extended): Take a scalar_int_mode rather than
2063 (default_valid_pointer_mode): Likewise.
2064 (default_addr_space_valid_pointer_mode): Likewise.
2065 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
2067 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
2068 rather than a machine_mode.
2069 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
2070 for local variables.
2071 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2072 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
2073 than a machine_mode.
2075 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2076 Alan Hayward <alan.hayward@arm.com>
2077 David Sherwood <david.sherwood@arm.com>
2079 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
2080 the mode argument to scalar_int_mode.
2081 (do_jump_by_parts_zero_rtx): Likewise.
2082 (do_jump_by_parts_equality_rtx): Likewise.
2083 (do_jump_by_parts_greater): Take a mode argument.
2084 (do_jump_by_parts_equality): Likewise.
2085 (do_jump_1): Update calls accordingly.
2087 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2088 Alan Hayward <alan.hayward@arm.com>
2089 David Sherwood <david.sherwood@arm.com>
2091 * is-a.h (safe_dyn_cast): New function.
2092 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
2093 (jump_table_for_label): Likewise.
2094 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
2095 instead of an rtx_insn *.
2096 (shorten_branches): Use dyn_cast instead of LABEL_P and
2097 JUMP_TABLE_DATA_P. Use jump_table_for_label and
2098 rtx_jump_table_data::get_data_mode.
2099 (final_scan_insn): Likewise.
2101 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2102 Alan Hayward <alan.hayward@arm.com>
2103 David Sherwood <david.sherwood@arm.com>
2105 * combine.c (try_combine): Use is_a <scalar_int_mode> when
2106 trying to combine a full-register integer set with a subreg
2109 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2110 Alan Hayward <alan.hayward@arm.com>
2111 David Sherwood <david.sherwood@arm.com>
2113 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
2114 that is always either address_mode or pointer_mode.
2116 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2117 Alan Hayward <alan.hayward@arm.com>
2118 David Sherwood <david.sherwood@arm.com>
2120 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
2121 when the two are known to be equal.
2123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2124 Alan Hayward <alan.hayward@arm.com>
2125 David Sherwood <david.sherwood@arm.com>
2127 * simplify-rtx.c (simplify_const_unary_operation): Use
2128 is_a <scalar_int_mode> instead of checking for a nonzero
2129 precision. Forcibly convert op_mode to a scalar_int_mode
2130 in that case. More clearly differentiate the operand and
2131 result modes and use the former when deciding what the value
2132 of a count-bits operation should be. Use is_int_mode instead
2133 of checking for a MODE_INT. Remove redundant check for whether
2134 this mode has a zero precision.
2136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2137 Alan Hayward <alan.hayward@arm.com>
2138 David Sherwood <david.sherwood@arm.com>
2140 * optabs.c (widen_leading): Change the type of the mode argument
2141 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
2142 (widen_bswap): Likewise.
2143 (expand_parity): Likewise.
2144 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
2145 (expand_ffs): Likewise.
2146 (epand_unop): Check for scalar integer modes before calling the
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 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
2155 (expand_expr_real_1): Update call accordingly.
2157 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2158 Alan Hayward <alan.hayward@arm.com>
2159 David Sherwood <david.sherwood@arm.com>
2161 * expmed.c (store_bit_field_using_insv): Add op0_mode and
2162 value_mode arguments. Use scalar_int_mode internally.
2163 (store_bit_field_1): Rename the new integer mode from imode
2164 to op0_mode and use it instead of GET_MODE (op0). Update calls
2165 to store_split_bit_field, store_bit_field_using_insv and
2166 store_fixed_bit_field.
2167 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
2168 Use scalar_int_mode internally. Use a bit count rather than a mode
2169 when calculating the largest bit size for get_best_mode.
2170 Update calls to store_split_bit_field and store_fixed_bit_field_1.
2171 (store_fixed_bit_field_1): Add mode and value_mode arguments.
2172 Remove assertion that OP0 has a scalar integer mode.
2173 (store_split_bit_field): Add op0_mode and value_mode arguments.
2174 Update calls to extract_fixed_bit_field.
2175 (extract_bit_field_using_extv): Add an op0_mode argument.
2176 Use scalar_int_mode internally.
2177 (extract_bit_field_1): Rename the new integer mode from imode to
2178 op0_mode and use it instead of GET_MODE (op0). Update calls to
2179 extract_split_bit_field, extract_bit_field_using_extv and
2180 extract_fixed_bit_field.
2181 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
2182 to extract_split_bit_field and extract_fixed_bit_field_1.
2183 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
2184 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
2186 (extract_split_bit_field): Add an op0_mode argument. Update call
2187 to extract_fixed_bit_field.
2189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2190 Alan Hayward <alan.hayward@arm.com>
2191 David Sherwood <david.sherwood@arm.com>
2193 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
2194 * explow.c (hard_function_value): Likewise.
2195 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
2196 convert_to_mode call outside the loop.
2197 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
2198 for the mode iterator. Require the mode specified by max_pieces
2200 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
2202 (copy_blkmode_to_reg): Likewise.
2203 (set_storage_via_setmem): Likewise.
2204 * optabs.c (prepare_cmp_insn): Likewise.
2205 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2206 * stor-layout.c (finish_bitfield_representative): Likewise.
2208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2209 Alan Hayward <alan.hayward@arm.com>
2210 David Sherwood <david.sherwood@arm.com>
2212 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
2213 * expr.c (convert_move): Use them.
2214 (convert_modes): Likewise.
2215 (store_expr_with_bounds): Likewise.
2217 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2218 Alan Hayward <alan.hayward@arm.com>
2219 David Sherwood <david.sherwood@arm.com>
2221 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
2222 parameter for the mode of "x". Remove the "known_x", "known_mode"
2223 and "known_ret" arguments. Change the type of the mode argument
2225 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
2226 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
2227 (reg_num_sign_bit_copies_for_combine): Likewise.
2228 * rtlanal.c (nonzero_bits1): Likewise.
2229 (num_sign_bit_copies1): Likewise.
2230 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
2231 (reg_num_sign_bit_copies_general): Likewise.
2232 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
2233 (reg_nonzero_bits_general): Likewise.
2235 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2236 Alan Hayward <alan.hayward@arm.com>
2237 David Sherwood <david.sherwood@arm.com>
2239 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
2240 than in subroutines. Return 1 for non-integer modes.
2241 (cached_num_sign_bit_copies): Change the type of the mode parameter
2243 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
2244 classes. Handle CONST_INT_P first and then check whether X also
2245 has a scalar integer mode. Check the same thing for inner registers
2246 of a SUBREG and for values that are being extended or truncated.
2248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2249 Alan Hayward <alan.hayward@arm.com>
2250 David Sherwood <david.sherwood@arm.com>
2252 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
2253 in subroutines. Return the mode mask for non-integer modes.
2254 (cached_nonzero_bits): Change the type of the mode parameter
2256 (nonzero_bits1): Likewise. Remove early exit for other mode
2257 classes. Handle CONST_INT_P first and then check whether X
2258 also has a scalar integer mode.
2260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2261 Alan Hayward <alan.hayward@arm.com>
2262 David Sherwood <david.sherwood@arm.com>
2264 * expr.c (widest_int_mode_for_size): Make the comment match the code.
2265 Return a scalar_int_mode and assert that the size is greater than
2267 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
2268 (op_by_pieces_d::op_by_pieces_d): Likewise.
2269 (op_by_pieces_d::run): Likewise.
2270 (can_store_by_pieces): Likewise.
2272 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2273 Alan Hayward <alan.hayward@arm.com>
2274 David Sherwood <david.sherwood@arm.com>
2276 * combine.c (extract_left_shift): Add a mode argument and update
2278 (make_compound_operation_int): Change the type of the mode parameter
2279 to scalar_int_mode and update the call to extract_left_shift.
2281 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2282 Alan Hayward <alan.hayward@arm.com>
2283 David Sherwood <david.sherwood@arm.com>
2285 * combine.c (simplify_and_const_int): Change the type of the mode
2286 parameter to scalar_int_mode.
2287 (simplify_and_const_int_1): Likewise. Update recursive call.
2289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2290 Alan Hayward <alan.hayward@arm.com>
2291 David Sherwood <david.sherwood@arm.com>
2293 * combine.c (simplify_compare_const): Check that the mode is a
2294 scalar_int_mode (rather than VOIDmode) before testing its
2296 (simplify_comparison): Move COMPARISON_P handling out of the
2297 loop and restrict the latter part of the loop to scalar_int_modes.
2298 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
2299 and when considering SUBREG_REGs. Use is_int_mode instead of
2300 checking GET_MODE_CLASS against MODE_INT.
2302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2303 Alan Hayward <alan.hayward@arm.com>
2304 David Sherwood <david.sherwood@arm.com>
2306 * combine.c (try_widen_shift_mode): Move check for equal modes to...
2307 (simplify_shift_const_1): ...here. Use scalar_int_mode for
2308 shift_unit_mode and for modes involved in scalar shifts.
2310 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2311 Alan Hayward <alan.hayward@arm.com>
2312 David Sherwood <david.sherwood@arm.com>
2314 * combine.c (force_int_to_mode): New function, split out from...
2315 (force_to_mode): ...here. Keep xmode up-to-date and use it
2316 instead of GET_MODE (x).
2318 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2319 Alan Hayward <alan.hayward@arm.com>
2320 David Sherwood <david.sherwood@arm.com>
2322 * optabs-query.h (extraction_insn::struct_mode): Change type to
2323 opt_scalar_int_mode and update comment.
2324 (extraction_insn::field_mode): Change type to scalar_int_mode.
2325 (extraction_insn::pos_mode): Likewise.
2326 * combine.c (make_extraction): Update accordingly.
2327 * optabs-query.c (get_traditional_extraction_insn): Likewise.
2328 (get_optab_extraction_insn): Likewise.
2329 * recog.c (simplify_while_replacing): Likewise.
2330 * expmed.c (narrow_bit_field_mem): Change the type of the mode
2331 parameter to opt_scalar_int_mode.
2333 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2334 Alan Hayward <alan.hayward@arm.com>
2335 David Sherwood <david.sherwood@arm.com>
2337 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
2338 to a scalar_int_mode instead of a machine_mode.
2339 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
2340 (get_best_mode): Return a boolean and use a pointer argument to store
2341 the selected mode. Replace the limit mode parameter with a bit limit.
2342 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
2343 for the values returned by bit_field_mode_iterator::next_mode.
2344 (store_bit_field): Update call to get_best_mode.
2345 (store_fixed_bit_field): Likewise.
2346 (extract_fixed_bit_field): Likewise.
2347 * expr.c (optimize_bitfield_assignment_op): Likewise.
2348 * fold-const.c (optimize_bit_field_compare): Likewise.
2349 (fold_truth_andor_1): Likewise.
2350 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
2351 Update for new type of m_mode.
2352 (get_best_mode): As above.
2354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2355 Alan Hayward <alan.hayward@arm.com>
2356 David Sherwood <david.sherwood@arm.com>
2358 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
2359 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
2360 (store_bit_field): Check is_a <scalar_int_mode> before calling
2361 strict_volatile_bitfield_p.
2362 (extract_bit_field): Likewise.
2364 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2365 Alan Hayward <alan.hayward@arm.com>
2366 David Sherwood <david.sherwood@arm.com>
2368 * target.def (cstore_mode): Return a scalar_int_mode.
2369 * doc/tm.texi: Regenerate.
2370 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
2371 * targhooks.h (default_cstore_mode): Likewise.
2372 * targhooks.c (default_cstore_mode): Likewise, using a forced
2374 * expmed.c (emit_cstore): Expect the target of the cstore to be
2377 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2378 Alan Hayward <alan.hayward@arm.com>
2379 David Sherwood <david.sherwood@arm.com>
2381 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
2383 (niter_desc): Likewise mode.
2384 (iv_analyze): Add a mode parameter.
2386 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
2387 and change its type to scalar_int_mode.
2388 * loop-iv.c: Update commentary at head of file.
2389 (iv_constant): Pass the mode paraeter before the rtx it describes
2390 and change its type to scalar_int_mode. Remove VOIDmode handling.
2391 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
2392 (iv_extend): Likewise.
2393 (shorten_into_mode): Likewise.
2394 (iv_add): Use scalar_int_mode.
2395 (iv_mult): Likewise.
2396 (iv_shift): Likewise.
2397 (canonicalize_iv_subregs): Likewise.
2398 (get_biv_step_1): Pass the outer_mode parameter before the rtx
2399 it describes and change its mode to scalar_int_mode. Also change
2400 the type of the returned inner_mode to scalar_int_mode.
2401 (get_biv_step): Likewise, turning outer_mode from a pointer
2402 into a direct parameter. Update call to get_biv_step_1.
2403 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
2404 iv_constant and get_biv_step.
2405 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
2406 and change its type to scalar_int_mode. Don't initialise iv->mode
2407 to VOIDmode and remove later checks for its still being VOIDmode.
2408 Update calls to iv_analyze_op and iv_analyze_expr. Check
2409 is_a <scalar_int_mode> when changing the mode under consideration.
2410 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
2411 Update call to iv_analyze_expr.
2412 (iv_analyze_op): Add a mode parameter. Reject subregs whose
2413 inner register is not also a scalar_int_mode. Update call to
2415 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
2416 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
2417 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
2418 separate mode class checks. Update calls to iv_analyze. Remove
2419 fix-up of VOIDmodes after iv_analyze_biv.
2420 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
2421 don't have a scalar_int_mode. Update call to biv_p.
2423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2424 Alan Hayward <alan.hayward@arm.com>
2425 David Sherwood <david.sherwood@arm.com>
2427 * cfgexpand.c (convert_debug_memory_address): Use
2428 as_a <scalar_int_mode>.
2429 * combine.c (expand_compound_operation): Likewise.
2430 (make_extraction): Likewise.
2431 (change_zero_ext): Likewise.
2432 (simplify_comparison): Likewise.
2433 * cse.c (cse_insn): Likewise.
2434 * dwarf2out.c (minmax_loc_descriptor): Likewise.
2435 (mem_loc_descriptor): Likewise.
2436 (loc_descriptor): Likewise.
2437 * expmed.c (init_expmed_one_mode): Likewise.
2438 (synth_mult): Likewise.
2439 (emit_store_flag_1): Likewise.
2440 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
2441 of a comparison with size.
2442 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
2443 (reduce_to_bit_field_precision): Likewise.
2444 * function.c (expand_function_end): Likewise.
2445 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
2446 * loop-doloop.c (doloop_modify): Likewise.
2447 * optabs.c (expand_binop): Likewise.
2448 (expand_unop): Likewise.
2449 (expand_copysign_absneg): Likewise.
2450 (prepare_cmp_insn): Likewise.
2451 (maybe_legitimize_operand): Likewise.
2452 * recog.c (const_scalar_int_operand): Likewise.
2453 * rtlanal.c (get_address_mode): Likewise.
2454 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2455 (simplify_cond_clz_ctz): Likewise.
2456 * tree-nested.c (get_nl_goto_field): Likewise.
2457 * tree.c (build_vector_type_for_mode): Likewise.
2458 * var-tracking.c (use_narrower_mode): Likewise.
2460 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2461 Alan Hayward <alan.hayward@arm.com>
2462 David Sherwood <david.sherwood@arm.com>
2464 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
2465 * builtins.c (expand_builtin_signbit): Use it.
2466 * cfgexpand.c (expand_debug_expr): Likewise.
2467 * dojump.c (do_jump): Likewise.
2468 (do_compare_and_jump): Likewise.
2469 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
2470 * expmed.c (make_tree): Likewise.
2471 * expr.c (expand_expr_real_2): Likewise.
2472 (expand_expr_real_1): Likewise.
2473 (try_casesi): Likewise.
2474 * fold-const-call.c (fold_const_call_ss): Likewise.
2475 * fold-const.c (unextend): Likewise.
2476 (extract_muldiv_1): Likewise.
2477 (fold_single_bit_test): Likewise.
2478 (native_encode_int): Likewise.
2479 (native_encode_string): Likewise.
2480 (native_interpret_int): Likewise.
2481 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
2482 * internal-fn.c (expand_addsub_overflow): Likewise.
2483 (expand_neg_overflow): Likewise.
2484 (expand_mul_overflow): Likewise.
2485 (expand_arith_overflow): Likewise.
2486 * match.pd: Likewise.
2487 * stor-layout.c (layout_type): Likewise.
2488 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2489 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
2490 * tree-ssanames.c (get_range_info): Likewise.
2491 * tree-switch-conversion.c (array_value_type) Likewise.
2492 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
2493 (vect_recog_divmod_pattern): Likewise.
2494 (vect_recog_mixed_size_cond_pattern): Likewise.
2495 * tree-vrp.c (extract_range_basic): Likewise.
2496 (simplify_float_conversion_using_ranges): Likewise.
2497 * tree.c (int_fits_type_p): Likewise.
2498 * ubsan.c (instrument_bool_enum_load): Likewise.
2499 * varasm.c (mergeable_string_section): Likewise.
2500 (narrowing_initializer_constant_valid_p): Likewise.
2501 (output_constant): Likewise.
2503 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2504 Alan Hayward <alan.hayward@arm.com>
2505 David Sherwood <david.sherwood@arm.com>
2507 * machmode.h (NARROWEST_INT_MODE): New macro.
2508 * expr.c (alignment_for_piecewise_move): Use it instead of
2509 GET_CLASS_NARROWEST_MODE (MODE_INT).
2510 (push_block): Likewise.
2511 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2513 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2516 Alan Hayward <alan.hayward@arm.com>
2517 David Sherwood <david.sherwood@arm.com>
2519 * postreload.c (move2add_valid_value_p): Change the type of the
2520 mode parameter to scalar_int_mode.
2521 (move2add_use_add2_insn): Add a mode parameter and use it instead
2523 (move2add_use_add3_insn): Likewise.
2524 (reload_cse_move2add): Update accordingly.
2526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2527 Alan Hayward <alan.hayward@arm.com>
2528 David Sherwood <david.sherwood@arm.com>
2530 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
2532 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
2533 * optabs.c (expand_unop): 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 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
2540 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
2541 (popcount_loc_descriptor): Likewise.
2542 (bswap_loc_descriptor): Likewise.
2543 (rotate_loc_descriptor): Likewise.
2544 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
2545 calling the functions above.
2547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2548 Alan Hayward <alan.hayward@arm.com>
2549 David Sherwood <david.sherwood@arm.com>
2551 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
2553 (try_combine): Likewise.
2554 (simplify_if_then_else): Likewise.
2555 * cse.c (cse_insn): Likewise.
2556 * dwarf2out.c (mem_loc_descriptor): Likewise.
2557 * emit-rtl.c (gen_lowpart_common): Likewise.
2558 * simplify-rtx.c (simplify_truncation): Likewise.
2559 (simplify_binary_operation_1): Likewise.
2560 (simplify_const_relational_operation): Likewise.
2561 (simplify_ternary_operation): Likewise.
2562 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
2564 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2565 Alan Hayward <alan.hayward@arm.com>
2566 David Sherwood <david.sherwood@arm.com>
2568 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
2569 * reload.c (push_reload): Likewise.
2570 (find_reloads): Likewise.
2572 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2573 Alan Hayward <alan.hayward@arm.com>
2574 David Sherwood <david.sherwood@arm.com>
2576 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
2577 (make_compound_operation_int): Likewise.
2578 (change_zero_ext): Likewise.
2579 * expr.c (convert_move): Likewise.
2580 (convert_modes): Likewise.
2581 * fwprop.c (forward_propagate_subreg): Likewise.
2582 * loop-iv.c (get_biv_step_1): Likewise.
2583 * optabs.c (widen_operand): Likewise.
2584 * postreload.c (move2add_valid_value_p): Likewise.
2585 * recog.c (simplify_while_replacing): Likewise.
2586 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2587 (simplify_binary_operation_1): Likewise. Remove redundant
2588 mode equality check.
2590 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2591 Alan Hayward <alan.hayward@arm.com>
2592 David Sherwood <david.sherwood@arm.com>
2594 * combine.c (combine_simplify_rtx): Add checks for
2595 is_a <scalar_int_mode>.
2596 (simplify_if_then_else): Likewise.
2597 (make_field_assignment): Likewise.
2598 (simplify_comparison): Likewise.
2599 * ifcvt.c (noce_try_bitop): Likewise.
2600 * loop-invariant.c (canonicalize_address_mult): Likewise.
2601 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2603 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2604 Alan Hayward <alan.hayward@arm.com>
2605 David Sherwood <david.sherwood@arm.com>
2607 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
2608 is_a <scalar_int_mode> instead of != BLKmode.
2610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2611 Alan Hayward <alan.hayward@arm.com>
2612 David Sherwood <david.sherwood@arm.com>
2614 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
2615 instead of != VOIDmode.
2616 * combine.c (if_then_else_cond): Likewise.
2617 (change_zero_ext): Likewise.
2618 * dwarf2out.c (mem_loc_descriptor): Likewise.
2619 (loc_descriptor): Likewise.
2620 * rtlanal.c (canonicalize_condition): Likewise.
2621 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
2623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2624 Alan Hayward <alan.hayward@arm.com>
2625 David Sherwood <david.sherwood@arm.com>
2627 * simplify-rtx.c (simplify_binary_operation_1): Use
2628 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
2630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2631 Alan Hayward <alan.hayward@arm.com>
2632 David Sherwood <david.sherwood@arm.com>
2634 * wide-int.h (int_traits<unsigned char>) New class.
2635 (int_traits<unsigned short>) Likewise.
2636 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
2637 Use GET_MODE_UNIT_PRECISION and remove redundant test for
2639 * combine.c (set_nonzero_bits_and_sign_copies): Use
2640 is_a <scalar_int_mode>.
2641 (find_split_point): Likewise.
2642 (combine_simplify_rtx): Likewise.
2643 (simplify_logical): Likewise.
2644 (expand_compound_operation): Likewise.
2645 (expand_field_assignment): Likewise.
2646 (make_compound_operation): Likewise.
2647 (extended_count): Likewise.
2648 (change_zero_ext): Likewise.
2649 (simplify_comparison): Likewise.
2650 * dwarf2out.c (scompare_loc_descriptor): Likewise.
2651 (ucompare_loc_descriptor): Likewise.
2652 (minmax_loc_descriptor): Likewise.
2653 (mem_loc_descriptor): Likewise.
2654 (loc_descriptor): Likewise.
2655 * expmed.c (init_expmed_one_mode): Likewise.
2656 * lra-constraints.c (lra_constraint_offset): Likewise.
2657 * optabs.c (prepare_libcall_arg): Likewise.
2658 * postreload.c (move2add_note_store): Likewise.
2659 * reload.c (operands_match_p): Likewise.
2660 * rtl.h (load_extend_op): Likewise.
2661 * rtlhooks.c (gen_lowpart_general): Likewise.
2662 * simplify-rtx.c (simplify_truncation): Likewise.
2663 (simplify_unary_operation_1): Likewise.
2664 (simplify_binary_operation_1): Likewise.
2665 (simplify_const_binary_operation): Likewise.
2666 (simplify_const_relational_operation): Likewise.
2667 (simplify_subreg): Likewise.
2668 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2669 * var-tracking.c (adjust_mems): Likewise.
2670 (prepare_call_arguments): Likewise.
2672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2673 Alan Hayward <alan.hayward@arm.com>
2674 David Sherwood <david.sherwood@arm.com>
2676 * machmode.h (is_int_mode): New fuction.
2677 * combine.c (find_split_point): Use it.
2678 (combine_simplify_rtx): Likewise.
2679 (simplify_if_then_else): Likewise.
2680 (simplify_set): Likewise.
2681 (simplify_shift_const_1): Likewise.
2682 (simplify_comparison): Likewise.
2683 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
2684 * cse.c (notreg_cost): Likewise.
2685 (cse_insn): Likewise.
2686 * cselib.c (cselib_lookup_1): Likewise.
2687 * dojump.c (do_jump_1): Likewise.
2688 (do_compare_rtx_and_jump): Likewise.
2689 * dse.c (get_call_args): Likewise.
2690 * dwarf2out.c (rtl_for_decl_init): Likewise.
2691 (native_encode_initializer): Likewise.
2692 * expmed.c (emit_store_flag_1): Likewise.
2693 (emit_store_flag): Likewise.
2694 * expr.c (convert_modes): Likewise.
2695 (store_field): Likewise.
2696 (expand_expr_real_1): Likewise.
2697 * fold-const.c (fold_read_from_constant_string): Likewise.
2698 * gimple-ssa-sprintf.c (get_format_string): Likewise.
2699 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
2700 * optabs.c (expand_binop): Likewise.
2701 (expand_unop): Likewise.
2702 (expand_abs_nojump): Likewise.
2703 (expand_one_cmpl_abs_nojump): Likewise.
2704 * simplify-rtx.c (mode_signbit_p): Likewise.
2705 (val_signbit_p): Likewise.
2706 (val_signbit_known_set_p): Likewise.
2707 (val_signbit_known_clear_p): Likewise.
2708 (simplify_relational_operation_1): Likewise.
2709 * tree.c (vector_type_mode): Likewise.
2711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2712 Alan Hayward <alan.hayward@arm.com>
2713 David Sherwood <david.sherwood@arm.com>
2715 * machmode.h (smallest_mode_for_size): Fix formatting.
2716 (smallest_int_mode_for_size): New function.
2717 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
2718 instead of smallest_mode_for_size.
2719 * combine.c (make_extraction): Likewise.
2720 * config/arc/arc.c (arc_expand_movmem): Likewise.
2721 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
2722 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2723 * config/s390/s390.c (s390_expand_insv): Likewise.
2724 * config/sparc/sparc.c (assign_int_registers): Likewise.
2725 * config/spu/spu.c (spu_function_value): Likewise.
2726 (spu_function_arg): Likewise.
2727 * coverage.c (get_gcov_type): Likewise.
2728 (get_gcov_unsigned_t): Likewise.
2729 * dse.c (find_shift_sequence): Likewise.
2730 * expmed.c (store_bit_field_1): Likewise.
2731 * expr.c (convert_move): Likewise.
2732 (store_field): Likewise.
2733 * internal-fn.c (expand_arith_overflow): Likewise.
2734 * optabs-query.c (get_best_extraction_insn): Likewise.
2735 * optabs.c (expand_twoval_binop_libfunc): Likewise.
2736 * stor-layout.c (layout_type): Likewise.
2737 (initialize_sizetypes): Likewise.
2738 * targhooks.c (default_get_mask_mode): Likewise.
2739 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
2741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2742 Alan Hayward <alan.hayward@arm.com>
2743 David Sherwood <david.sherwood@arm.com>
2745 * machmode.h (opt_mode::else_blk): New function.
2746 (int_mode_for_mode): Declare.
2747 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
2748 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
2750 * cfgexpand.c (expand_debug_expr): Likewise.
2751 * combine.c (gen_lowpart_or_truncate): Likewise.
2752 (gen_lowpart_for_combine): Likewise.
2753 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
2754 * config/avr/avr.c (avr_to_int_mode): Likewise.
2755 (avr_out_plus_1): Likewise.
2756 (avr_out_plus): Likewise.
2757 (avr_out_round): Likewise.
2758 * config/i386/i386.c (ix86_split_to_parts): Likewise.
2759 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2760 (s390_expand_vcond): Likewise.
2761 * config/spu/spu.c (spu_split_immediate): Likewise.
2762 (spu_expand_mov): Likewise.
2763 * dse.c (get_stored_val): Likewise.
2764 * expmed.c (store_bit_field_1): Likewise.
2765 (convert_extracted_bit_field): Use int_mode_for_mode instead of
2767 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
2768 (extract_low_bits): Likewise.
2769 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
2770 handling rather than repeating the check.
2771 (emit_group_store): Likewise.
2772 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
2773 * optabs.c (expand_absneg_bit): Likewise.
2774 (expand_copysign_absneg): Likewise.
2775 (expand_copysign_bit): Likewise.
2776 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
2777 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
2778 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
2779 * var-tracking.c (prepare_call_arguments): Likewise.
2780 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
2781 int_mode_for_mode instead of mode_for_size.
2782 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2784 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2785 Alan Hayward <alan.hayward@arm.com>
2786 David Sherwood <david.sherwood@arm.com>
2788 * machmode.h (int_mode_for_size): New function.
2789 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
2790 instead of mode_for_size.
2791 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
2793 * combine.c (expand_field_assignment): Use int_mode_for_size
2794 instead of mode_for_size.
2795 (make_extraction): Likewise.
2796 (simplify_shift_const_1): Likewise.
2797 (simplify_comparison): Likewise.
2798 * dojump.c (do_jump): Likewise.
2799 * dwarf2out.c (mem_loc_descriptor): Likewise.
2800 * emit-rtl.c (init_derived_machine_modes): Likewise.
2801 * expmed.c (flip_storage_order): Likewise.
2802 (convert_extracted_bit_field): Likewise.
2803 * expr.c (copy_blkmode_from_reg): Likewise.
2804 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
2805 * internal-fn.c (expand_mul_overflow): Likewise.
2806 * lower-subreg.c (simple_move): Likewise.
2807 * optabs-libfuncs.c (init_optabs): Likewise.
2808 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2809 * tree.c (vector_type_mode): Likewise.
2810 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
2811 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
2812 * tree-vect-generic.c (expand_vector_parallel): Likewise.
2813 * tree-vect-stmts.c (vectorizable_load): Likewise.
2814 (vectorizable_store): Likewise.
2816 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2817 Alan Hayward <alan.hayward@arm.com>
2818 David Sherwood <david.sherwood@arm.com>
2820 * coretypes.h (pod_mode): New type.
2821 (scalar_int_mode_pod): New typedef.
2822 * machmode.h (pod_mode): New class.
2823 (int_n_data_t::m): Change type to scalar_int_mode_pod.
2824 * genmodes.c (emit_mode_int_n): Update accordingly.
2825 * lower-subreg.h (target_lower_subreg): Change type to
2826 scalar_int_mode_pod.
2827 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
2828 scalar_int_mode_pod.
2830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2831 Alan Hayward <alan.hayward@arm.com>
2832 David Sherwood <david.sherwood@arm.com>
2834 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
2835 machine_mode to scalar_int_mode.
2836 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
2837 (rs6000_option_override_internal): Remove cast to int.
2838 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
2839 machine_mode to scalar_int_mode.
2840 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
2841 (rs6000_option_override_internal): Remove cast to int.
2842 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
2843 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
2845 * config/s390/s390.c (s390_expand_builtin): Likewise.
2846 * coretypes.h (scalar_int_mode): New type.
2847 (opt_scalar_int_mode): New typedef.
2848 * machmode.h (scalar_int_mode): New class.
2849 (scalar_int_mode::includes_p): New function.
2850 (byte_mode): Change type to scalar_int_mode.
2851 (word_mode): Likewise.
2852 (ptr_mode): Likewise.
2853 * emit-rtl.c (byte_mode): Likewise.
2854 (word_mode): Likewise.
2855 (ptr_mode): Likewise.
2856 (init_derived_machine_modes): Update accordingly.
2857 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
2858 and MODE_PARTIAL_INT.
2859 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
2860 opt_scalar_int_mode.
2862 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2863 Alan Hayward <alan.hayward@arm.com>
2864 David Sherwood <david.sherwood@arm.com>
2866 * target.def (libgcc_floating_mode_supported_p): Take a
2868 * doc/tm.texi: Regenerate.
2869 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
2871 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
2872 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
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 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
2880 * doc/tm.texi: Regenerate.
2881 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
2882 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
2883 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
2884 * targhooks.h (default_floatn_mode): Likewise.
2885 * targhooks.c (default_floatn_mode): Likewise.
2886 * tree.c (build_common_tree_nodes): Update accordingly.
2888 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2889 Alan Hayward <alan.hayward@arm.com>
2890 David Sherwood <david.sherwood@arm.com>
2892 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
2893 (mode_iterator::iterate_p): Likewise.
2894 (mode_iterator::get_wider): Likewise.
2895 * expr.c (init_expr_target): Use opt_scalar_float_mode.
2897 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2898 Alan Hayward <alan.hayward@arm.com>
2899 David Sherwood <david.sherwood@arm.com>
2901 * coretypes.h (opt_scalar_float_mode): New typedef.
2902 * machmode.h (float_mode_for_size): New function.
2903 * emit-rtl.c (double_mode): Delete.
2904 (init_emit_once): Use float_mode_for_size.
2905 * stor-layout.c (layout_type): Likewise.
2906 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
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 * output.h (assemble_real): Take a scalar_float_mode.
2913 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
2914 * config/arm/arm.md (consttable_4): Likewise.
2915 (consttable_8): Likewise.
2916 (consttable_16): Likewise.
2917 * config/mips/mips.md (consttable_float): Likewise.
2918 * config/s390/s390.c (s390_output_pool_entry): Likewise.
2919 * varasm.c (assemble_real): Take a scalar_float_mode.
2920 (output_constant_pool_2): Update accordingly.
2921 (output_constant): Likewise.
2923 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2924 Alan Hayward <alan.hayward@arm.com>
2925 David Sherwood <david.sherwood@arm.com>
2927 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
2928 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
2929 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
2930 (native_encode_real): Likewise.
2931 (native_interpret_real): Likewise.
2932 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
2933 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2935 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2936 Alan Hayward <alan.hayward@arm.com>
2937 David Sherwood <david.sherwood@arm.com>
2939 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
2940 <scalar_float_mode>. Simplify.
2941 (gen_extend_conv_libfunc): Likewise.
2943 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2944 Alan Hayward <alan.hayward@arm.com>
2945 David Sherwood <david.sherwood@arm.com>
2947 * coretypes.h (scalar_float_mode): New type.
2948 * machmode.h (mode_traits::from_int): Use machine_mode if
2949 USE_ENUM_MODES is defined.
2950 (is_a): New function.
2952 (dyn_cast): Likewise.
2953 (scalar_float_mode): New class.
2954 (scalar_float_mode::includes_p): New function.
2955 (is_float_mode): Likewise.
2956 * gdbhooks.py (MachineModePrinter): New class.
2957 (build_pretty_printer): Use it for scalar_float_mode.
2958 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
2959 (format_helper::format_helper): Turn into a template.
2960 * genmodes.c (get_mode_class): New function.
2961 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
2962 or machine_mode if none.
2963 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
2964 as_a <scalar_float_mode>.
2965 * dwarf2out.c (mem_loc_descriptor): Likewise.
2966 (insert_float): Likewise.
2967 (add_const_value_attribute): Likewise.
2968 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2969 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
2970 (expand_unop): Update accordingly.
2971 (expand_abs_nojump): Likewise.
2972 (expand_copysign_absneg): Take a scalar_float_mode.
2973 (expand_copysign_bit): Likewise.
2974 (expand_copysign): Update accordingly.
2976 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2977 Alan Hayward <alan.hayward@arm.com>
2978 David Sherwood <david.sherwood@arm.com>
2980 * coretypes.h (opt_mode): New class.
2981 * machmode.h (opt_mode): Likewise.
2982 (opt_mode::else_void): New function.
2983 (opt_mode::require): Likewise.
2984 (opt_mode::exists): Likewise.
2985 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
2986 (GET_MODE_2XWIDER_MODE): Likewise.
2987 (mode_iterator::get_wider): Update accordingly.
2988 (mode_iterator::get_2xwider): Likewise.
2989 (mode_iterator::get_known_wider): Likewise, turning into a template.
2990 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
2991 forcing a wider mode to exist.
2992 * config/cr16/cr16.h (LONG_REG_P): Likewise.
2993 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2994 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
2995 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
2996 * lower-subreg.c (init_lower_subreg): Likewise.
2997 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
2998 on the final iteration.
2999 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
3000 a wider mode exists before asking for a move pattern.
3001 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
3002 forcing a wider mode to exist.
3003 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
3004 returning false if no such mode exists.
3005 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
3006 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
3007 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
3008 Avoid checking for a MODE_INT if we already know the mode is not a
3010 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
3011 forcing a wider mode to exist.
3012 (expmed_mult_highpart_optab): Likewise.
3013 (expmed_mult_highpart): Likewise.
3014 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
3016 * lto-streamer-in.c (lto_input_mode_table): Likewise.
3017 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
3018 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3019 * internal-fn.c (expand_mul_overflow): Update use of
3020 GET_MODE_2XWIDER_MODE.
3021 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3022 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
3023 GET_MODE_WIDER_MODE.
3024 (convert_plusminus_to_widen): Likewise.
3025 * tree-switch-conversion.c (array_value_type): Likewise.
3026 * var-tracking.c (emit_note_insn_var_location): Likewise.
3027 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3028 Return false inside rather than outside the loop if no wider mode
3030 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
3031 and GET_MODE_2XWIDER_MODE
3032 (can_compare_p): Use else_void.
3033 * gdbhooks.py (OptMachineModePrinter): New class.
3034 (build_pretty_printer): Use it for opt_mode.
3036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3037 Alan Hayward <alan.hayward@arm.com>
3038 David Sherwood <david.sherwood@arm.com>
3040 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
3041 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
3043 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3044 Alan Hayward <alan.hayward@arm.com>
3045 David Sherwood <david.sherwood@arm.com>
3047 * machmode.h (mode_traits): New structure.
3048 (get_narrowest_mode): New function.
3049 (mode_iterator::start): Likewise.
3050 (mode_iterator::iterate_p): Likewise.
3051 (mode_iterator::get_wider): Likewise.
3052 (mode_iterator::get_known_wider): Likewise.
3053 (mode_iterator::get_2xwider): Likewise.
3054 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
3055 (FOR_EACH_MODE): Likewise.
3056 (FOR_EACH_MODE_FROM): Likewise.
3057 (FOR_EACH_MODE_UNTIL): Likewise.
3058 (FOR_EACH_WIDER_MODE): Likewise.
3059 (FOR_EACH_2XWIDER_MODE): Likewise.
3060 * builtins.c (expand_builtin_strlen): Use new mode iterators.
3061 * combine.c (simplify_comparison): Likewise
3062 * config/i386/i386.c (type_natural_mode): Likewise.
3063 * cse.c (cse_insn): Likewise.
3064 * dse.c (find_shift_sequence): Likewise.
3065 * emit-rtl.c (init_derived_machine_modes): Likewise.
3066 (init_emit_once): Likewise.
3067 * explow.c (hard_function_value): Likewise.
3068 * expmed.c (extract_fixed_bit_field_1): Likewise.
3069 (extract_bit_field_1): Likewise.
3070 (expand_divmod): Likewise.
3071 (emit_store_flag_1): Likewise.
3072 * expr.c (init_expr_target): Likewise.
3073 (convert_move): Likewise.
3074 (alignment_for_piecewise_move): Likewise.
3075 (widest_int_mode_for_size): Likewise.
3076 (emit_block_move_via_movmem): Likewise.
3077 (copy_blkmode_to_reg): Likewise.
3078 (set_storage_via_setmem): Likewise.
3079 (compress_float_constant): Likewise.
3080 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3081 * optabs-query.c (get_best_extraction_insn): Likewise.
3082 * optabs.c (expand_binop): Likewise.
3083 (expand_twoval_unop): Likewise.
3084 (expand_twoval_binop): Likewise.
3085 (widen_leading): Likewise.
3086 (widen_bswap): Likewise.
3087 (expand_parity): Likewise.
3088 (expand_unop): Likewise.
3089 (prepare_cmp_insn): Likewise.
3090 (prepare_float_lib_cmp): Likewise.
3091 (expand_float): Likewise.
3092 (expand_fix): Likewise.
3093 (expand_sfix_optab): Likewise.
3094 * postreload.c (move2add_use_add2_insn): Likewise.
3095 * reg-stack.c (reg_to_stack): Likewise.
3096 * reginfo.c (choose_hard_reg_mode): Likewise.
3097 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3098 * stor-layout.c (mode_for_size): Likewise.
3099 (smallest_mode_for_size): Likewise.
3100 (mode_for_vector): Likewise.
3101 (finish_bitfield_representative): Likewise.
3102 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
3103 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
3104 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3105 * var-tracking.c (prepare_call_arguments): Likewise.
3107 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3108 Alan Hayward <alan.hayward@arm.com>
3109 David Sherwood <david.sherwood@arm.com>
3111 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
3112 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
3113 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
3114 * machmode.h (mode_size): Move earlier in file.
3115 (mode_precision): Likewise.
3116 (mode_inner): Likewise.
3117 (mode_nunits): Likewise.
3118 (mode_unit_size): Likewise.
3119 (unit_unit_precision): Likewise.
3120 (mode_wider): Likewise.
3121 (mode_2xwider): Likewise.
3122 (machine_mode): New class.
3123 (mode_to_bytes): New function.
3124 (mode_to_bits): Likewise.
3125 (mode_to_precision): Likewise.
3126 (mode_to_inner): Likewise.
3127 (mode_to_unit_size): Likewise.
3128 (mode_to_unit_precision): Likewise.
3129 (mode_to_nunits): Likewise.
3130 (GET_MODE_SIZE): Use mode_to_bytes.
3131 (GET_MODE_BITSIZE): Use mode_to_bits.
3132 (GET_MODE_PRECISION): Use mode_to_precision.
3133 (GET_MODE_INNER): Use mode_to_inner.
3134 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
3135 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
3136 (GET_MODE_NUNITS): Use mode_to_nunits.
3137 * system.h (ALWAYS_INLINE): New macro.
3138 * config/powerpcspe/powerpcspe-c.c
3139 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
3140 int for arg1_mode and arg2_mode.
3142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3143 Alan Hayward <alan.hayward@arm.com>
3144 David Sherwood <david.sherwood@arm.com>
3146 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
3147 Prefix mode names with E_ in case statements.
3148 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3149 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
3150 (aarch64_split_simd_move): Likewise.
3151 (aarch64_gen_storewb_pair): Likewise.
3152 (aarch64_gen_loadwb_pair): Likewise.
3153 (aarch64_gen_store_pair): Likewise.
3154 (aarch64_gen_load_pair): Likewise.
3155 (aarch64_get_condition_code_1): Likewise.
3156 (aarch64_constant_pool_reload_icode): Likewise.
3157 (get_rsqrte_type): Likewise.
3158 (get_rsqrts_type): Likewise.
3159 (get_recpe_type): Likewise.
3160 (get_recps_type): Likewise.
3161 (aarch64_gimplify_va_arg_expr): Likewise.
3162 (aarch64_simd_container_mode): Likewise.
3163 (aarch64_emit_load_exclusive): Likewise.
3164 (aarch64_emit_store_exclusive): Likewise.
3165 (aarch64_expand_compare_and_swap): Likewise.
3166 (aarch64_gen_atomic_cas): Likewise.
3167 (aarch64_emit_bic): Likewise.
3168 (aarch64_emit_atomic_swap): Likewise.
3169 (aarch64_emit_atomic_load_op): Likewise.
3170 (aarch64_evpc_trn): Likewise.
3171 (aarch64_evpc_uzp): Likewise.
3172 (aarch64_evpc_zip): Likewise.
3173 (aarch64_evpc_ext): Likewise.
3174 (aarch64_evpc_rev): Likewise.
3175 (aarch64_evpc_dup): Likewise.
3176 (aarch64_gen_ccmp_first): Likewise.
3177 (aarch64_gen_ccmp_next): Likewise.
3178 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3179 (alpha_emit_xfloating_libcall): Likewise.
3180 (emit_insxl): Likewise.
3181 (alpha_arg_type): Likewise.
3182 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
3183 (arc_preferred_simd_mode): Likewise.
3184 (arc_secondary_reload): Likewise.
3185 (get_arc_condition_code): Likewise.
3186 (arc_print_operand): Likewise.
3187 (arc_legitimate_constant_p): Likewise.
3188 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3189 * config/arc/arc.md (casesi_load): Likewise.
3190 (casesi_compact_jump): Likewise.
3191 * config/arc/predicates.md (proper_comparison_operator): Likewise.
3192 (cc_use_register): Likewise.
3193 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3194 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
3195 (arm_init_iwmmxt_builtins): Likewise.
3196 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
3197 (neon_expand_vector_init): Likewise.
3198 (arm_attr_length_move_neon): Likewise.
3199 (maybe_get_arm_condition_code): Likewise.
3200 (arm_emit_vector_const): Likewise.
3201 (arm_preferred_simd_mode): Likewise.
3202 (arm_output_iwmmxt_tinsr): Likewise.
3203 (thumb1_output_casesi): Likewise.
3204 (thumb2_output_casesi): Likewise.
3205 (arm_emit_load_exclusive): Likewise.
3206 (arm_emit_store_exclusive): Likewise.
3207 (arm_expand_compare_and_swap): Likewise.
3208 (arm_evpc_neon_vuzp): Likewise.
3209 (arm_evpc_neon_vzip): Likewise.
3210 (arm_evpc_neon_vrev): Likewise.
3211 (arm_evpc_neon_vtrn): Likewise.
3212 (arm_evpc_neon_vext): Likewise.
3213 (arm_validize_comparison): Likewise.
3214 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
3215 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
3216 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
3217 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
3218 (c6x_preferred_simd_mode): Likewise.
3219 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
3220 (epiphany_rtx_costs): Likewise.
3221 * config/epiphany/predicates.md (proper_comparison_operator):
3223 * config/frv/frv.c (condexec_memory_operand): Likewise.
3224 (frv_emit_move): Likewise.
3225 (output_move_single): Likewise.
3226 (output_condmove_single): Likewise.
3227 (frv_hard_regno_mode_ok): Likewise.
3228 (frv_matching_accg_mode): Likewise.
3229 * config/h8300/h8300.c (split_adds_subs): Likewise.
3230 (h8300_rtx_costs): Likewise.
3231 (h8300_print_operand): Likewise.
3232 (compute_mov_length): Likewise.
3233 (output_logical_op): Likewise.
3234 (compute_logical_op_length): Likewise.
3235 (compute_logical_op_cc): Likewise.
3236 (h8300_shift_needs_scratch_p): Likewise.
3237 (output_a_shift): Likewise.
3238 (compute_a_shift_length): Likewise.
3239 (compute_a_shift_cc): Likewise.
3240 (expand_a_rotate): Likewise.
3241 (output_a_rotate): Likewise.
3242 * config/i386/i386.c (classify_argument): Likewise.
3243 (function_arg_advance_32): Likewise.
3244 (function_arg_32): Likewise.
3245 (function_arg_64): Likewise.
3246 (function_value_64): Likewise.
3247 (ix86_gimplify_va_arg): Likewise.
3248 (ix86_legitimate_constant_p): Likewise.
3249 (put_condition_code): Likewise.
3250 (split_double_mode): Likewise.
3251 (ix86_avx256_split_vector_move_misalign): Likewise.
3252 (ix86_expand_vector_logical_operator): Likewise.
3253 (ix86_split_idivmod): Likewise.
3254 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
3255 (ix86_build_const_vector): Likewise.
3256 (ix86_build_signbit_mask): Likewise.
3257 (ix86_match_ccmode): Likewise.
3258 (ix86_cc_modes_compatible): Likewise.
3259 (ix86_expand_branch): Likewise.
3260 (ix86_expand_sse_cmp): Likewise.
3261 (ix86_expand_sse_movcc): Likewise.
3262 (ix86_expand_int_sse_cmp): Likewise.
3263 (ix86_expand_vec_perm_vpermi2): Likewise.
3264 (ix86_expand_vec_perm): Likewise.
3265 (ix86_expand_sse_unpack): Likewise.
3266 (ix86_expand_int_addcc): Likewise.
3267 (ix86_split_to_parts): Likewise.
3268 (ix86_vectorize_builtin_gather): Likewise.
3269 (ix86_vectorize_builtin_scatter): Likewise.
3270 (avx_vpermilp_parallel): Likewise.
3271 (inline_memory_move_cost): Likewise.
3272 (ix86_tieable_integer_mode_p): Likewise.
3273 (x86_maybe_negate_const_int): Likewise.
3274 (ix86_expand_vector_init_duplicate): Likewise.
3275 (ix86_expand_vector_init_one_nonzero): Likewise.
3276 (ix86_expand_vector_init_one_var): Likewise.
3277 (ix86_expand_vector_init_concat): Likewise.
3278 (ix86_expand_vector_init_interleave): Likewise.
3279 (ix86_expand_vector_init_general): Likewise.
3280 (ix86_expand_vector_set): Likewise.
3281 (ix86_expand_vector_extract): Likewise.
3282 (emit_reduc_half): Likewise.
3283 (ix86_emit_i387_round): Likewise.
3284 (ix86_mangle_type): Likewise.
3285 (ix86_expand_round_sse4): Likewise.
3286 (expand_vec_perm_blend): Likewise.
3287 (canonicalize_vector_int_perm): Likewise.
3288 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
3289 (expand_vec_perm_1): Likewise.
3290 (expand_vec_perm_interleave3): Likewise.
3291 (expand_vec_perm_even_odd_pack): Likewise.
3292 (expand_vec_perm_even_odd_1): Likewise.
3293 (expand_vec_perm_broadcast_1): Likewise.
3294 (ix86_vectorize_vec_perm_const_ok): Likewise.
3295 (ix86_expand_vecop_qihi): Likewise.
3296 (ix86_expand_mul_widen_hilo): Likewise.
3297 (ix86_expand_sse2_abs): Likewise.
3298 (ix86_expand_pextr): Likewise.
3299 (ix86_expand_pinsr): Likewise.
3300 (ix86_preferred_simd_mode): Likewise.
3301 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
3302 * config/i386/sse.md (*andnot<mode>3): Likewise.
3303 (<mask_codefor><code><mode>3<mask_name>): Likewise.
3304 (*<code><mode>3): Likewise.
3305 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
3306 (ia64_expand_atomic_op): Likewise.
3307 (ia64_arg_type): Likewise.
3308 (ia64_mode_to_int): Likewise.
3309 (ia64_scalar_mode_supported_p): Likewise.
3310 (ia64_vector_mode_supported_p): Likewise.
3311 (expand_vec_perm_broadcast): Likewise.
3312 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
3313 (iq2000_function_arg_advance): Likewise.
3314 (iq2000_function_arg): Likewise.
3315 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
3316 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
3317 (m68k_libcall_value): Likewise.
3318 (m68k_function_value): Likewise.
3319 (sched_attr_op_type): Likewise.
3320 * config/mcore/mcore.c (mcore_output_move): Likewise.
3321 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
3323 (microblaze_function_arg): Likewise.
3324 * config/mips/mips.c (mips16_build_call_stub): Likewise.
3325 (mips_print_operand): Likewise.
3326 (mips_mode_ok_for_mov_fmt_p): Likewise.
3327 (mips_vector_mode_supported_p): Likewise.
3328 (mips_preferred_simd_mode): Likewise.
3329 (mips_expand_vpc_loongson_even_odd): Likewise.
3330 (mips_expand_vec_unpack): Likewise.
3331 (mips_expand_vi_broadcast): Likewise.
3332 (mips_expand_vector_init): Likewise.
3333 (mips_expand_vec_reduc): Likewise.
3334 (mips_expand_msa_cmp): Likewise.
3335 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
3336 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
3337 (cc_flags_for_mode): Likewise.
3338 * config/msp430/msp430.c (msp430_print_operand): Likewise.
3339 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
3340 (nds32_output_casesi_pc_relative): Likewise.
3341 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3342 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
3343 (nvptx_gen_unpack): Likewise.
3344 (nvptx_gen_pack): Likewise.
3345 (nvptx_gen_shuffle): Likewise.
3346 (nvptx_gen_wcast): Likewise.
3347 (nvptx_preferred_simd_mode): Likewise.
3348 * config/pa/pa.c (pa_secondary_reload): Likewise.
3349 * config/pa/predicates.md (base14_operand): Likewise.
3350 * config/powerpcspe/powerpcspe-c.c
3351 (altivec_resolve_overloaded_builtin): Likewise.
3352 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
3354 (rs6000_preferred_simd_mode): Likewise.
3355 (output_vec_const_move): Likewise.
3356 (rs6000_expand_vector_extract): Likewise.
3357 (rs6000_split_vec_extract_var): Likewise.
3358 (reg_offset_addressing_ok_p): Likewise.
3359 (rs6000_legitimate_offset_address_p): Likewise.
3360 (rs6000_legitimize_address): Likewise.
3361 (rs6000_emit_set_const): Likewise.
3362 (rs6000_const_vec): Likewise.
3363 (rs6000_emit_move): Likewise.
3364 (spe_build_register_parallel): Likewise.
3365 (rs6000_darwin64_record_arg_recurse): Likewise.
3366 (swap_selector_for_mode): Likewise.
3367 (spe_init_builtins): Likewise.
3368 (paired_init_builtins): Likewise.
3369 (altivec_init_builtins): Likewise.
3370 (do_load_for_compare): Likewise.
3371 (rs6000_generate_compare): Likewise.
3372 (rs6000_expand_float128_convert): Likewise.
3373 (emit_load_locked): Likewise.
3374 (emit_store_conditional): Likewise.
3375 (rs6000_output_function_epilogue): Likewise.
3376 (rs6000_handle_altivec_attribute): Likewise.
3377 (rs6000_function_value): Likewise.
3378 (emit_fusion_gpr_load): Likewise.
3379 (emit_fusion_p9_load): Likewise.
3380 (emit_fusion_p9_store): Likewise.
3381 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
3382 (fusion_gpr_mem_load): Likewise.
3383 (fusion_addis_mem_combo_load): Likewise.
3384 (fusion_addis_mem_combo_store): Likewise.
3385 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
3386 (fusion_gpr_mem_load): Likewise.
3387 (fusion_addis_mem_combo_load): Likewise.
3388 (fusion_addis_mem_combo_store): Likewise.
3389 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3391 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
3392 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
3393 (rs6000_preferred_simd_mode): Likewise.
3394 (output_vec_const_move): Likewise.
3395 (rs6000_expand_vector_extract): Likewise.
3396 (rs6000_split_vec_extract_var): Likewise.
3397 (reg_offset_addressing_ok_p): Likewise.
3398 (rs6000_legitimate_offset_address_p): Likewise.
3399 (rs6000_legitimize_address): Likewise.
3400 (rs6000_emit_set_const): Likewise.
3401 (rs6000_const_vec): Likewise.
3402 (rs6000_emit_move): Likewise.
3403 (rs6000_darwin64_record_arg_recurse): Likewise.
3404 (swap_selector_for_mode): Likewise.
3405 (paired_init_builtins): Likewise.
3406 (altivec_init_builtins): Likewise.
3407 (rs6000_expand_float128_convert): Likewise.
3408 (emit_load_locked): Likewise.
3409 (emit_store_conditional): Likewise.
3410 (rs6000_output_function_epilogue): Likewise.
3411 (rs6000_handle_altivec_attribute): Likewise.
3412 (rs6000_function_value): Likewise.
3413 (emit_fusion_gpr_load): Likewise.
3414 (emit_fusion_p9_load): Likewise.
3415 (emit_fusion_p9_store): Likewise.
3416 * config/rx/rx.c (rx_gen_move_template): Likewise.
3417 (flags_from_mode): Likewise.
3418 * config/s390/predicates.md (s390_alc_comparison): Likewise.
3419 (s390_slb_comparison): Likewise.
3420 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
3421 (s390_vector_mode_supported_p): Likewise.
3422 (s390_cc_modes_compatible): Likewise.
3423 (s390_match_ccmode_set): Likewise.
3424 (s390_canonicalize_comparison): Likewise.
3425 (s390_emit_compare_and_swap): Likewise.
3426 (s390_branch_condition_mask): Likewise.
3427 (s390_rtx_costs): Likewise.
3428 (s390_secondary_reload): Likewise.
3429 (__SECONDARY_RELOAD_CASE): Likewise.
3430 (s390_expand_cs): Likewise.
3431 (s390_preferred_simd_mode): Likewise.
3432 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
3433 * config/sh/sh.c (sh_print_operand): Likewise.
3434 (dump_table): Likewise.
3435 (sh_secondary_reload): Likewise.
3436 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3437 * config/sh/sh.md (casesi_worker_1): Likewise.
3438 (casesi_worker_2): Likewise.
3439 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
3440 (fcc_comparison_operator): Likewise.
3441 * config/sparc/sparc.c (sparc_expand_move): Likewise.
3442 (emit_soft_tfmode_cvt): Likewise.
3443 (sparc_preferred_simd_mode): Likewise.
3444 (output_cbranch): Likewise.
3445 (sparc_print_operand): Likewise.
3446 (sparc_expand_vec_perm_bmask): Likewise.
3447 (vector_init_bshuffle): Likewise.
3448 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3449 (spu_vector_mode_supported_p): Likewise.
3450 (spu_expand_insv): Likewise.
3451 (spu_emit_branch_or_set): Likewise.
3452 (spu_handle_vector_attribute): Likewise.
3453 (spu_builtin_splats): Likewise.
3454 (spu_builtin_extract): Likewise.
3455 (spu_builtin_promote): Likewise.
3456 (spu_expand_sign_extend): Likewise.
3457 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3458 (tilegx_simd_int): Likewise.
3459 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
3460 (tilepro_simd_int): Likewise.
3461 * config/v850/v850.c (const_double_split): Likewise.
3462 (v850_print_operand): Likewise.
3463 (ep_memory_offset): Likewise.
3464 * config/vax/vax.c (vax_rtx_costs): Likewise.
3465 (vax_output_int_move): Likewise.
3466 (vax_output_int_add): Likewise.
3467 (vax_output_int_subtract): Likewise.
3468 * config/visium/predicates.md (visium_branch_operator): Likewise.
3469 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
3470 (visium_print_operand_address): Likewise.
3471 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3472 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
3473 (xtensa_expand_conditional_branch): Likewise.
3474 (xtensa_copy_incoming_a7): Likewise.
3475 (xtensa_output_literal): Likewise.
3476 * dfp.c (decimal_real_maxval): Likewise.
3477 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3480 Alan Hayward <alan.hayward@arm.com>
3481 David Sherwood <david.sherwood@arm.com>
3483 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
3484 (mode_nunits_inline): Likewise.
3485 (mode_inner_inline): Likewise.
3486 (mode_unit_size_inline): Likewise.
3487 (mode_unit_precision_inline): Likewise.
3488 (emit_insn_modes_h): Likewise. Also emit a #define of the
3490 (emit_mode_wider): Add an E_ prefix to mode names.
3491 (emit_mode_complex): Likewise.
3492 (emit_mode_inner): Likewise.
3493 (emit_mode_adjustments): Likewise.
3494 (emit_mode_int_n): Likewise.
3495 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
3496 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
3497 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
3498 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
3499 (CRC32_BUILTIN, ENTRY): Likewise.
3500 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
3501 (aarch64_pop_regs): Likewise.
3502 (aarch64_process_components): Likewise.
3503 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
3504 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
3505 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
3506 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
3507 * config/arm/arm.c (arm_init_libfuncs): Likewise.
3508 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
3510 * config/i386/i386-builtin.def (pcmpestr): Likewise.
3511 (pcmpistr): Likewise.
3512 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
3513 * config/mmix/mmix.c (mmix_output_condition): Likewise.
3514 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
3516 * config/rl78/rl78.c (mduc_regs): Likewise.
3517 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3518 (htm_expand_builtin): Likewise.
3519 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
3520 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
3521 * config/xtensa/xtensa.c (print_operand): Likewise.
3522 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
3523 (NUM_MODE_VECTOR_INT): Likewise.
3524 * genoutput.c (null_operand): Likewise.
3525 (output_operand_data): Likewise.
3526 * genrecog.c (print_parameter_value): Likewise.
3527 * lra.c (debug_operand_data): Likewise.
3529 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3531 * dwarf2out.c (scompare_loc_descriptor_wide)
3532 (scompare_loc_descriptor_narrow): New functions, split out from...
3533 (scompare_loc_descriptor): ...here.
3534 * expmed.c (emit_store_flag_int): New function, split out from...
3535 (emit_store_flag): ...here.
3537 2017-08-30 Richard Biener <rguenther@suse.de>
3539 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
3540 (dwarf2out_early_finish): Move setting of AT_pubnames from
3541 early debug output to early finish.
3543 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3545 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
3546 and -mdata-region to the assembler.
3548 2017-08-30 Richard Biener <rguenther@suse.de>
3550 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
3552 (gen_subprogram_die): Add DW_AT_object_pointer only early.
3553 (dwarf2out_early_global_decl): Only generate a DIE for the
3554 abstract origin if it doesn't already exist or is a declaration DIE.
3555 (resolve_addr): Do not add the linkage name twice when
3556 generating a stub DIE for the DW_TAG_GNU_call_site target.
3558 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3560 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3561 Use machine_mode rather than int for arg1_mode.
3563 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
3566 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
3567 that the second argument of the built-in functions to unpack
3568 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
3569 switch statement instead a lot of if statements.
3570 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
3571 Allow 64-bit values to be in Altivec registers as well as
3572 traditional floating point registers.
3573 (pack<mode>, FMOVE128_VSX iterator): Likewise.
3575 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
3577 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
3578 MAX_REGS_PER_ADDRESS == 1.
3580 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
3582 * config/i386/i386.opt (flag_fentry): Do not init to -1.
3583 * config/i386/i386.c (ix86_option_override_internal): Simplify
3584 setting of opts->x_flag_entry.
3586 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3587 Jakub Jelinek <jakub@redhat.com>
3588 Richard Biener <rguenther@suse.de>
3590 PR tree-optimization/81503
3591 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
3592 folded constant fits in the target type; reorder tests for clarity.
3594 2017-08-29 Martin Liska <mliska@suse.cz>
3596 * passes.def: Include pass_lower_switch.
3597 * stmt.c (dump_case_nodes): Remove and move to
3598 tree-switch-conversion.
3599 (case_values_threshold): Likewise.
3600 (expand_switch_as_decision_tree_p): Likewise.
3601 (emit_case_decision_tree): Likewise.
3602 (expand_case): Likewise.
3603 (balance_case_nodes): Likewise.
3604 (node_has_low_bound): Likewise.
3605 (node_has_high_bound): Likewise.
3606 (node_is_bounded): Likewise.
3607 (emit_case_nodes): Likewise.
3608 (struct simple_case_node): New struct.
3609 (add_case_node): Remove.
3610 (emit_case_dispatch_table): Use vector instead of case_list.
3611 (reset_out_edges_aux): Remove.
3612 (compute_cases_per_edge): Likewise.
3613 (expand_case): Build list of simple_case_node.
3614 (expand_sjlj_dispatch_table): Use it.
3615 * tree-switch-conversion.c (struct case_node): Moved from
3616 stmt.c and adjusted.
3617 (emit_case_nodes): Likewise.
3618 (node_has_low_bound): Likewise.
3619 (node_has_high_bound): Likewise.
3620 (node_is_bounded): Likewise.
3621 (case_values_threshold): Likewise.
3622 (reset_out_edges_aux): Likewise.
3623 (compute_cases_per_edge): Likewise.
3624 (add_case_node): Likewise.
3625 (dump_case_nodes): Likewise.
3626 (balance_case_nodes): Likewise.
3627 (expand_switch_as_decision_tree_p): Likewise.
3628 (emit_jump): Likewise.
3629 (emit_case_decision_tree): Likewise.
3630 (try_switch_expansion): Likewise.
3631 (do_jump_if_equal): Likewise.
3632 (emit_cmp_and_jump_insns): Likewise.
3633 (fix_phi_operands_for_edge): New function.
3634 (record_phi_operand_mapping): Likewise.
3635 (class pass_lower_switch): New pass.
3636 (pass_lower_switch::execute): New function.
3637 (make_pass_lower_switch): Likewise.
3638 (conditional_probability):
3639 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
3640 * tree-pass.h: Add make_pass_lower_switch.
3642 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3645 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
3648 2017-08-29 Richard Biener <rguenther@suse.de>
3650 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
3651 we do not add a DW_AT_inline attribute twice.
3652 (gen_subprogram_die): Remove code setting DW_AT_inline on
3653 DECL_ABSTRACT_P nodes.
3655 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
3657 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
3658 calls to internal functions.
3659 (gimplify_modify_expr): Likewise.
3660 * tree-call-cdce.c (use_internal_fn): Likewise.
3661 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
3662 (convert_to_divmod): Set the nothrow flag.
3663 * tree-if-conv.c (predicate_mem_writes): Likewise.
3664 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3665 (vectorizable_call): Likewise.
3666 (vectorizable_store): Likewise.
3667 (vectorizable_load): Likewise.
3668 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
3669 (vect_recog_mask_conversion_pattern): Likewise.
3671 2017-08-29 Martin Liska <mliska@suse.cz>
3674 * gcc.c (driver_handle_option): Add new argument.
3675 * opts-common.c (handle_option): Pass
3676 target_option_override_hook.
3677 * opts-global.c (lang_handle_option): Add new option.
3678 (set_default_handlers): Add new argument.
3679 (decode_options): Likewise.
3680 * opts.c (target_handle_option): Likewise.
3681 (common_handle_option): Call target_option_override_hook.
3682 * opts.h (struct cl_option_handler_func): Add hook for
3683 target option override.
3684 (struct cl_option_handlers): Likewise.
3685 (set_default_handlers): Add new argument.
3686 (decode_options): Likewise.
3687 (common_handle_option): Likewise.
3688 (target_handle_option): Likewise.
3689 * toplev.c (toplev::main): Pass targetm.target_option.override
3692 2017-08-29 Richard Biener <rguenther@suse.de>
3693 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3695 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
3696 life to the active subtree.
3698 2017-08-28 Jeff Law <law@redhat.com>
3700 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
3701 derive_equivalences.
3702 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
3704 (edge_info::derive_equivalences): New private member function
3706 * tree-ssa-dom.c (class edge_info): Changed from a struct
3707 to a class. Add ctor/dtor, methods and data members.
3708 (edge_info::edge_info): Renamed from allocate_edge_info.
3709 Initialize additional members.
3710 (edge_info::~edge_info): New.
3711 (free_dom_edge_info): Delete the edge info.
3712 (record_edge_info): Use new class & associated member functions.
3713 Tighten forms for testing for edge equivalences.
3714 (record_temporary_equivalences): Iterate over the simple
3715 equivalences rather than assuming there's only one per edge.
3716 (cprop_into_successor_phis): Iterate over the simple
3717 equivalences rather than assuming there's only one per edge.
3718 (optimize_stmt): Use operand_equal_p rather than pointer
3719 equality for mini-DSE code.
3721 2017-08-28 Nathan Sidwell <nathan@acm.org>
3723 * gcc.c (execute): Fold SIGPIPE handling into switch
3724 statement. Adjust internal error message.
3726 2017-08-28 Richard Biener <rguenther@suse.de>
3729 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
3730 Do nothing for removed DIEs.
3732 2017-08-28 Richard Biener <rguenther@suse.de>
3734 PR tree-optimization/81977
3735 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
3738 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
3741 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
3742 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
3743 using targetm.gen_mem_thread_fence.
3745 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
3748 * config/i386/i386.md (*<btsc><mode>): Change operand 2
3749 predicate to register_operand. Reorder operands.
3750 (*btr<mode>): Ditto.
3751 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
3752 (*btr<mode>_mask): Ditto.
3754 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
3756 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
3757 * config/rs6000/xmmintrin.h: New file.
3758 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
3760 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3763 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
3764 parameter and_insn and return it.
3765 (recombine_lvx_pattern): Insert a copy to ensure availability of
3766 the base register of the copied masking operation at the point of
3767 the instruction replacement.
3768 (recombine_stvx_pattern): Likewise.
3770 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3772 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
3773 undocumented switches.
3774 (-mpower9-dform-vector): Likewise.
3775 (-mpower9-dform): Likewise.
3776 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
3777 comments to delete references to -mpower9-dform* switches.
3778 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
3779 Delete reference to -mpower9-dform* switches, test for
3780 -mpower9-vector instead.
3781 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
3782 (OTHER_P9_VECTOR_MASKS): Likewise.
3783 (POWERPC_MASKS): Likewise.
3784 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
3785 tests against -mpower9-dform* to -mpower9-vector. Delete code
3786 that checked for -mpower9-dform* consistancy with other options.
3787 Add test for -mpower9-misc to enable other power9 switches.
3788 (rs6000_init_hard_regno_mode_ok): Likewise.
3789 (rs6000_option_override_internal): Likewise.
3790 (rs6000_emit_prologue): Likewise.
3791 (rs6000_emit_epilogue): Likewise.
3792 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
3793 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
3794 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
3796 (emit_fusion_p9_store): Likewise.
3797 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
3798 resetting these macros if the assembler does not support ISA 3.0
3800 (TARGET_P9_DFORM_VECTOR): Likewise.
3801 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
3802 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
3804 2017-08-25 Alan Modra <amodra@gmail.com>
3806 PR rtl-optimization/81747
3807 * cse.c (cse_extended_basic_block): Don't attempt to record
3808 equivalences for degenerate conditional branches that branch
3809 to their fall-through.
3811 2017-08-24 Martin Sebor <msebor@redhat.com>
3814 * gimple-fold.c (size_must_be_zero_p): New function.
3815 (gimple_fold_builtin_memory_op): Call it.
3817 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
3819 * config/rs6000/mm_malloc.h: New file.
3821 2017-08-24 Bin Cheng <bin.cheng@arm.com>
3823 PR tree-optimization/81913
3824 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
3825 analysis when either IVs in condition can wrap.
3827 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
3829 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
3830 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
3832 2017-08-24 Richard Biener <rguenther@suse.de>
3835 * targhooks.c (default_target_can_inline_p): Properly
3836 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
3837 is present and always compare.
3838 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
3839 infer -mfpmath=sse from TARGET_SSE_P.
3840 (ix86_can_inline_p): Properly use target_option_default_node when
3841 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
3843 2017-08-24 Richard Biener <rguenther@suse.de>
3846 * dwarf2out.c (output_die): Handle flag_generate_offload like
3848 (output_comp_unit): Likewise.
3849 (gen_array_type_die): Likewise.
3850 (dwarf2out_early_finish): Likewise.
3851 (note_variable_value_in_expr): Likewise.
3852 (dwarf2out_finish): Likewise. Adjust assert.
3853 * cgraphunit.c (symbol_table::compile): Move setting of
3854 flag_generate_offload earlier ...
3855 (symbol_table::finalize_compilation_unit): ... here, before
3856 early debug finalization.
3858 2017-08-24 Richard Biener <rguenther@suse.de>
3860 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
3861 and ipa-fnsummary.h.
3862 (ix86_can_inline_p): When ix86_fpmath flags do not match
3863 check whether the callee uses FP math at all.
3865 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
3868 * tree-ssanames.c (get_nonzero_bits): Use element_precision
3869 instead of TYPE_PRECISION.
3871 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
3872 Alan Hayward <alan.hayward@arm.com>
3873 David Sherwood <david.sherwood@arm.com>
3875 * combine.c (make_extraction): Use subreg_offset_from_lsb.
3877 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
3879 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
3881 (ix86_frame::stack_realign_allocate): New field.
3882 (struct machine_frame_state): Modify comments.
3883 (machine_frame_state::sp_realigned_fp_end): New field.
3884 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
3886 (sp_valid_at): Add assertion to assure no attempt to access invalid
3887 offset of a realigned stack.
3888 (fp_valid_at): Likewise.
3889 (choose_baseaddr): Modify comments.
3890 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
3891 ix86_expand_prologue.
3892 (ix86_expand_prologue): Modify stack realignment and allocation.
3893 (ix86_expand_epilogue): Modify comments.
3894 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
3895 avx2_runtime, avx512f, and avx512f_runtime.
3897 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
3899 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
3900 (mstackrealign): Do not init to -1.
3901 * config/i386/i386.c (ix86_option_override_internal):
3902 Check opts_set, not opts when setting default value of
3903 opts->x_ix86_force_align_arg_pointer.
3905 2017-08-23 Richard Biener <rguenther@suse.de>
3907 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
3908 lang_hooks.decl_printable_name.
3909 * print-rtl-function.c (print_rtx_function): Likewise.
3910 * tree-pretty-print.c (dump_function_header): Likewise.
3912 2017-08-23 Richard Biener <rguenther@suse.de>
3915 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
3916 -g0 at compile-time.
3918 2017-08-23 Tamar Christina <tamar.christina@arm.com>
3921 * doc/sourcebuild.texi (Other hardware attributes):
3924 2017-08-23 Tamar Christina <tamar.christina@arm.com>
3927 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
3928 Fix single-use check.
3930 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
3932 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
3934 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
3936 * doc/install.texi: Modify to add more details on running selected
3939 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
3941 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
3942 is combined with -mabi=ms.
3943 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
3946 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3948 PR tree-optimization/81488
3949 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
3950 and cached_basis fields.
3951 (MAX_SPREAD): New constant.
3952 (alloc_cand_and_find_basis): Initialize new fields.
3953 (clear_visited): New function.
3954 (create_phi_basis_1): Rename from create_phi_basis, set visited
3955 and cached_basis fields.
3956 (create_phi_basis): New wrapper function.
3957 (phi_add_costs_1): Rename from phi_add_costs, add spread
3958 parameter, set visited field, short-circuit when limits reached.
3959 (phi_add_costs): New wrapper function.
3960 (record_phi_increments_1): Rename from record_phi_increments, set
3962 (record_phi_increments): New wrapper function.
3963 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
3964 (phi_incr_cost): New wrapper function.
3965 (all_phi_incrs_profitable_1): Rename from
3966 all_phi_incrs_profitable, set visited field.
3967 (all_phi_incrs_profitable): New wrapper function.
3969 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
3970 Alan Hayward <alan.hayward@arm.com>
3971 David Sherwood <david.sherwood@arm.com>
3973 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
3974 that takes the outer and inner modes.
3975 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
3976 comparison as the canonical test for a paradoxical subreg.
3977 * combine.c (simplify_set): Use paradoxical_subreg_p.
3978 (make_extraction): Likewise.
3979 (force_to_mode): Likewise.
3980 (rtx_equal_for_field_assignment_p): Likewise.
3981 (gen_lowpart_for_combine): Likewise.
3982 (simplify_comparison): Likewise.
3983 * cse.c (equiv_constant): Likewise.
3984 * expmed.c (store_bit_field_1): Likewise.
3985 * final.c (alter_subreg): Likewise.
3986 * fwprop.c (propagate_rtx): Likewise.
3987 (forward_propagate_subreg): Likewise.
3988 * ira-conflicts.c (ira_build_conflicts): Likewise.
3989 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
3990 * lra-constraints.c (curr_insn_transform): Likewise.
3991 (split_reg): Likewise.
3992 * lra-eliminations.c (move_plus_up): Likewise.
3993 (lra_eliminate_regs_1): Likewise.
3994 * recog.c (general_operand): Likewise.
3995 * ree.c (combine_reaching_defs): Likewise.
3996 * reload.c (push_reload): Likewise.
3997 (find_reloads): Likewise.
3998 * reload1.c (elimination_effects): Likewise.
3999 (compute_reload_subreg_offset): Likewise.
4000 (choose_reload_regs): Likewise.
4001 * rtlanal.c (subreg_lsb_1): Likewise.
4002 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4003 (simplify_subreg): Likewise.
4004 * var-tracking.c (track_loc_p): Likewise.
4005 * emit-rtl.c (byte_lowpart_offset): Likewise.
4006 (paradoxical_subreg_p): Delete out-of-line definition.
4008 2017-08-22 Jeff Law <law@redhat.com>
4010 PR tree-optimization/81741
4011 PR tree-optimization/71947
4012 * tree-ssa-dom.c: Include tree-inline.h.
4013 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
4014 equivalences if one is more expensive to compute than the other.
4015 * tree-ssa-scopedtables.h (class const_or_copies): Make
4016 record_const_or_copy_raw method private.
4017 (class avail_exprs_stack): New method simplify_binary_operation.
4018 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
4019 avail_exprs_stack::simplify_binary_operation as needed.
4020 (avail_exprs_stack::simplify_binary_operation): New function.
4022 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
4024 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
4025 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
4026 (DOT_SYMBOLS): Likewise.
4027 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
4028 (RELOCATABLE_NEEDS_FIXUP): Likewise.
4029 (RS6000_ABI_NAME): Likewise.
4030 (TARGET_CMODEL): Likewise.
4031 (TOC_SECTION_ASM_OP): Likewise.
4032 (SET_CMODEL): New macro.
4033 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
4035 2017-08-22 Richard Biener <rguenther@suse.de>
4037 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
4038 to allow for free-lang-data replacements similar to verify_type_variant.
4040 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4041 Alan Hayward <alan.hayward@arm.com>
4042 David Sherwood <david.sherwood@arm.com>
4044 * config/aarch64/aarch64.md (casesi): Use DImode rather than
4045 VOIDmode for the LABEL_REF.
4047 2017-08-22 Richard Biener <rguenther@suse.de>
4049 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
4051 2017-08-22 Richard Biener <rguenther@suse.de>
4053 * common.opt (feliminate-dwarf2-dups): Ignore.
4054 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
4055 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
4056 same_die_p_wrap, compute_section_prefix,
4057 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
4058 (comdat_symbol_id, comdat_symbol_number): Likewise.
4059 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
4061 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
4062 (output_die): Mark unreachable path unreachable.
4063 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
4064 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
4065 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
4066 (dwarf2out_early_finish): Likewise.
4068 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
4070 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
4072 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
4075 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
4076 not VAR_P. Filter attribute warnings with OPT_Wattributes.
4077 (avr_attribute_table) <io, io_low, address>: Initialize
4078 .decl_required with true.
4080 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
4082 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
4083 undocumented debugging options.
4084 (-mvsx-scalar-double): Likewise.
4085 (-mallow-df-permute): Likewise.
4086 (-mvectorize-builtins): Likewise.
4087 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4088 (rs6000_builtin_vectorized_function): Likewise.
4089 (rs6000_builtin_md_vectorized_function): Likewise.
4090 (rs6000_opt_vars): Likewise.
4092 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
4095 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
4096 (*btrq_imm): Rename from *btrq.
4097 (*btcq_imm): Rename from *btcq.
4098 (btsc): New code attribute.
4099 (*<btsc><mode>): New insn pattern.
4100 (*btr<mode>): Ditto.
4101 (*<btsc><mode>_mask): New insn_and_split pattern.
4102 (*btr<mode>_mask): Ditto.
4104 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4105 Alan Hayward <alan.hayward@arm.com>
4106 David Sherwood <david.sherwood@arm.com>
4108 * function.c (pad_below): Simplify padding calculation.
4110 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4111 Alan Hayward <alan.hayward@arm.com>
4112 David Sherwood <david.sherwood@arm.com>
4114 * target.def (function_prologue): Remove frame size argument.
4115 (function_epilogue): Likewise.
4116 * doc/tm.texi: Regenerate.
4117 * final.c (final_start_function): Update call to function_prologue.
4118 (final_end_function): Update call to function_epilogue.
4119 (default_function_pro_epilogue): Remove frame size argument.
4120 * output.h (default_function_pro_epilogue): Likewise.
4121 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
4122 (arm_output_function_prologue): Likewise.
4123 * config/frv/frv.c (frv_function_prologue): Likewise.
4124 (frv_function_epilogue): Likewise.
4125 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
4126 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
4127 (ia64_output_function_epilogue): Likewise.
4128 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
4129 (m32r_output_function_epilogue): Likewise.
4130 * config/microblaze/microblaze.c (microblaze_function_prologue)
4131 (microblaze_function_epilogue): Likewise.
4132 * config/mips/mips.c (mips_output_function_prologue): Likewise.
4133 (mips_output_function_epilogue): Likewise.
4134 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
4135 (mmix_target_asm_function_epilogue): Likewise.
4136 * config/msp430/msp430.c (msp430_start_function): Likewise.
4137 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
4138 (nds32_asm_function_epilogue): Likewise.
4139 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
4140 * config/pa/pa.c (pa_output_function_prologue): Likewise.
4141 (pa_output_function_epilogue): Likewise.
4142 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
4143 (rs6000_output_function_epilogue): Likewise.
4144 * config/rl78/rl78.c (rl78_start_function): Likewise.
4145 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
4146 (rs6000_output_function_epilogue): Likewise.
4147 * config/rx/rx.c (rx_output_function_prologue): Likewise.
4148 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
4149 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
4150 (sparc_asm_function_epilogue): Likewise.
4152 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4154 * tree.h (type_has_mode_precision_p): New function.
4155 * convert.c (convert_to_integer_1): Use it.
4156 * expr.c (expand_expr_real_2): Likewise.
4157 (expand_expr_real_1): Likewise.
4158 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
4159 * match.pd: Likewise.
4160 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
4161 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
4162 * tree-tailcall.c (process_assignment): Likewise.
4163 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4164 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
4165 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
4166 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4167 (vectorizable_assignment): Likewise.
4168 (vectorizable_shift): Likewise.
4169 (vectorizable_operation): Likewise.
4170 * tree-vrp.c (register_edge_assert_for_2): Likewise.
4172 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
4174 * match.pd: Add pow (C, x) simplification.
4176 2017-08-21 Richard Biener <rguenther@suse.de>
4178 PR tree-optimization/81900
4179 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
4180 for blocks with abnormal predecessors.
4181 (compute_antic): Do not set visited flag prematurely.
4183 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
4186 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
4188 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4190 * stor-layout.h (vector_type_mode): Move to...
4191 * tree.h (vector_type_mode): ...here.
4192 * stor-layout.c (vector_type_mode): Move to...
4193 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
4195 2017-08-21 Richard Biener <rguenther@suse.de>
4197 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
4198 register_external_die hooks.
4199 (debug_false_tree_charstarstar_uhwistar): Declare.
4200 (debug_nothing_tree_charstar_uhwi): Likewise.
4201 * debug.c (do_nothing_debug_hooks): Adjust.
4202 (debug_false_tree_charstarstar_uhwistar): New do nothing.
4203 (debug_nothing_tree_charstar_uhwi): Likewise.
4204 * dbxout.c (dbx_debug_hooks): Adjust.
4205 (xcoff_debug_hooks): Likewise.
4206 * sdbout.c (sdb_debug_hooks): Likewise.
4207 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4208 * dwarf2out.c (macinfo_label_base): New global.
4209 (dwarf2out_register_external_die): New function for the
4210 register_external_die hook.
4211 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
4212 (dwarf2_debug_hooks): Use them.
4213 (dwarf2_lineno_debug_hooks): Adjust.
4214 (struct die_struct): Add with_offset flag.
4215 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
4216 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
4217 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
4218 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
4219 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
4220 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
4221 defining section names for the early LTO debug variants.
4222 (reset_indirect_string): New helper.
4223 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
4224 (print_dw_val): Add support for offsetted symbol references.
4225 (get_ultimate_context): Split out from is_cxx.
4226 (is_cxx): Use get_ultimate_context.
4227 (is_fortran): Add decl overload.
4228 (compute_comp_unit_symbol): Split out worker from
4229 compute_section_prefix.
4230 (compute_section_prefix): Call compute_comp_unit_symbol and
4231 set comdat_type_p here.
4232 (output_die): Skip DIE symbol output for the LTO added one.
4233 Handle DIE symbol references with offset.
4234 (output_comp_unit): Guard section name mangling properly.
4235 For LTO debug sections emit a symbol at the section beginning
4236 which we use to refer to its DIEs.
4237 (add_abstract_origin_attribute): For DIEs registered via
4238 dwarf2out_register_external_die directly refer to the early
4239 DIE rather than indirectly through the shadow one we created.
4240 Remove obsolete call to dwarf2out_abstract_function for
4241 non-function/block origins.
4242 (gen_array_type_die): When generating early LTO debug do
4243 not emit DW_AT_string_length.
4244 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
4245 late when in LTO. As suggested place a gcc_unreachable for
4246 the DECL_ABSTRACT_P case.
4247 (gen_subprogram_die): Avoid another specification DIE
4248 for early built declarations/definitions for the late LTO case.
4249 (gen_variable_die): Add type references for late duplicated VLA dies
4251 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
4252 we have the abstract instance already.
4253 (process_scope_var): Adjust decl DIE contexts in LTO which
4254 first puts them in limbo.
4255 (gen_decl_die): Do not generate type DIEs late apart from
4256 types for VLAs or for decls we do not yet have a DIE. Do not
4257 call dwarf2out_abstract_function late.
4258 (dwarf2out_early_global_decl): Make sure to create DIEs
4259 for abstract instances of a decl first.
4260 (dwarf2out_late_global_decl): Adjust comment.
4261 (output_macinfo_op): With multiple macro sections use
4262 macinfo_label_base to distinguish labels.
4263 (output_macinfo): Likewise. Update macinfo_label_base.
4264 Pass in the line info label.
4265 (note_variable_value_in_expr): When generating LTO resolve
4266 all variable values here by generating DIEs as needed.
4267 (init_sections_and_labels): Add early LTO debug flag parameter
4268 and generate different sections and names if set. Add generation
4269 counter for the labels so we can have multiple of them.
4270 (reset_dies): Helper to allow DIEs to be output multiple times.
4271 (dwarf2out_finish): When outputting DIEs to the fat part of an
4272 LTO object first reset DIEs.
4273 (dwarf2out_early_finish): Output early DIEs when generating LTO.
4274 (modified_type_die): Check for decl_ultimate_origin being self
4276 (gen_type_die_with_usage): Likewise.
4277 (gen_typedef_die): Allow decl_ultimate_origin being self.
4278 (set_decl_abstract_flags): Remove.
4279 (set_block_abstract_flags): Likewise.
4280 (dwarf2out_abstract_function): Treat the early generated DIEs
4281 as the abstract copy and only add DW_AT_inline and
4282 DW_AT_artificial here and call set_decl_origin_self.
4283 If the DIE has an abstract origin don't do anything.
4284 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
4285 if we have none yet (Go fails to build one, PR78628).
4286 (variably_modified_type_p): Prevent endless recursion for Ada
4287 cyclic pointer types.
4288 * lto-streamer-in.c: Include debug.h.
4289 (dref_queue): New global.
4290 (lto_read_tree_1): Stream in DIE references.
4291 (lto_input_tree): Register DIE references.
4292 (input_function): Stream DECL_DEBUG_ARGS.
4293 * lto-streamer-out.c: Include debug.h.
4294 (lto_write_tree_1): Output DIE references.
4295 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
4296 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
4297 (output_function): Stream DECL_DEBUG_ARGS.
4298 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
4299 Stream DECL_ABSTRACT_ORIGIN.
4300 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
4301 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
4302 DECL_CONTEXT for file-scope decls.
4303 * lto-streamer.h (struct dref_entry): Declare.
4304 (dref_queue): Likewise.
4305 * cfgexpand.c (pass_expand::execute): Do not call the
4306 outlining_inline_function hook here.
4307 * lto-wrapper.c (debug_obj): New global.
4308 (tool_cleanup): Unlink it if required.
4309 (debug_objcopy): New function.
4310 (run_gcc): Handle early debug sections in the IL files by
4311 extracting them to separate files, partially linkin them and
4312 feeding the result back as result to the linker.
4313 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
4314 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
4315 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
4316 sections into a separate segment.
4317 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
4319 (darwin_asm_dwarf_section): Likewise.
4320 (darwin_asm_output_dwarf_offset): Likewise.
4321 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
4323 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4324 Alan Hayward <alan.hayward@arm.com>
4325 David Sherwood <david.sherwood@arm.com>
4327 * read-md.h (md_reader::record_potential_iterator_use): Replace
4328 pointer argument with an rtx and an index.
4329 * read-rtl.c (iterator_group::apply_iterator): Likewise.
4330 (apply_mode_iterator): Likewise.
4331 (apply_code_iterator): Likewise.
4332 (apply_int_iterator): Likewise.
4333 (apply_subst_iterator): Likewise.
4334 (record_iterator_use): Likewise.
4335 (record_attribute_use): Likewise.
4336 (md_reader::record_potential_iterator_use): Likewise. Update calls
4337 to record_iterator_use and apply_iterator.
4338 (iterator_use): Replace ptr with x and index.
4339 (attribute_use): Likewise.
4340 (apply_attribute_uses): Update calls to apply_iterator.
4341 (apply_iterators): Likewise. Update initialization of iterator_use.
4342 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
4343 and record_potential_iterator_use.
4344 (rtx_reader::read_rtx_operand): Likewise.
4346 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4347 Alan Hayward <alan.hayward@arm.com>
4348 David Sherwood <david.sherwood@arm.com>
4350 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
4353 2017-08-21 Richard Biener <rguenther@suse.de>
4356 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
4357 at struct end conservatively when comparing common bases.
4359 2017-08-21 Richard Biener <rguenther@suse.de>
4361 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
4362 (mem_ref_in_stmt): Remove.
4363 (determine_max_movement): Use ref index to get at the reference.
4364 (invariantness_dom_walker::before_dom_children): Deal with
4365 lim data already initialized.
4366 (gather_mem_refs_stmt): Initialize lim data and record ref index.
4368 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4370 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
4371 (TARGET_ISA_ROUND): Ditto.
4372 (TARGET_ROUND): Ditto.
4373 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
4374 * config/i386/i386.md: Ditto.
4375 * config/i386/sse.md: Ditto.
4376 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
4377 with OPTION_MASK_ISA_SSE4_1.
4379 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4382 * doc/extend.texi (x86 Built-in Functions): Correct the name of
4383 __builtin_ia32_lzcnt_u16.
4385 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4388 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
4389 (rs6000_set_current_function): Rewrite function to use it.
4391 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
4394 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
4395 and TYPE_WARN_IF_NOT_ALIGN.
4396 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
4397 (handle_warn_if_not_align): New.
4398 (place_union_field): Call handle_warn_if_not_align.
4399 (place_field): Call handle_warn_if_not_align.
4400 Copy TYPE_WARN_IF_NOT_ALIGN.
4401 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
4402 (layout_type): Likewise.
4403 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
4405 (tree_decl_common): Add warn_if_not_align.
4406 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
4407 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
4408 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
4409 (DECL_WARN_IF_NOT_ALIGN): Likewise.
4410 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
4411 * doc/extend.texi: Document warn_if_not_aligned attribute.
4412 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
4414 2017-08-17 Martin Liska <mliska@suse.cz>
4417 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
4418 (get_data_dependence): Use it as pointer type.
4419 (distribute_loop): Likewise.
4421 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4423 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
4424 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
4425 (p8_vmrgow_<mode>_direct): New define_insn.
4426 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
4427 handle endianness for vmrgew and vmrgow permute patterns.
4429 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4431 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
4432 * config/rs6000/rs6000-cpus.def: Remove comment.
4433 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
4434 (POWERPC_MASKS): Likewise.
4435 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
4436 use of TARGET_VSX_TIMODE.
4437 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
4438 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
4439 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
4440 (rs6000_option_override_internal): Remove dead code.
4441 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
4442 (rs6000_legitimize_reload_address): Likewise.
4443 (rs6000_legitimate_address_p): Likewise.
4444 (rs6000_opt_masks): Delete "vsx-timode".
4445 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
4446 from function comment.
4447 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
4448 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
4449 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
4451 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
4452 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
4453 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
4456 (define_peephole2): Likewise.
4458 2017-08-17 Martin Sebor <msebor@redhat.com>
4461 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
4462 past the end of an array.
4463 (test_pp_format): Add test cases.
4465 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
4467 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
4468 missing ECF_NOTHROW flags.
4470 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4473 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
4474 operands residing in integer registers.
4475 (*vsx_le_perm_load_<mode>): Likewise.
4476 (*vsx_le_perm_store_<mode>): Likewise.
4477 (define_peephole2): Add peepholes to optimize the above.
4479 2017-08-17 Marek Polacek <polacek@redhat.com>
4482 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
4483 to mimic what shorten_compare did. Change the return type to bool.
4484 (fold_cond_expr_with_comparison): Update call to
4485 operand_equal_for_comparison_p.
4486 (fold_ternary_loc): Likewise.
4488 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
4490 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
4492 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
4494 2017-08-17 Richard Biener <rguenther@suse.de>
4496 * tree-ssa-structalias.c (solve_graph): When propagating
4497 to successors update the graphs succ edges and avoid duplicate work.
4499 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
4502 * config/i386/i386.c (ix86_option_override_internal): Save target
4503 specific options after ix86_stack_protector_guard_reg was changed.
4505 2017-08-17 Richard Biener <rguenther@suse.de>
4507 PR tree-optimization/81827
4508 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
4509 (new_var_info): Initialize it conservatively.
4510 (get_call_vi): Mark register vars.
4511 (new_scalar_tmp_constraint_exp): Likewise.
4512 (handle_rhs_call): Likewise.
4513 (handle_const_call): Likewise.
4514 (create_function_info_for): Likewise.
4515 (solve_constraints): Sort varinfos to separate register from
4516 non-register vars to pack points-to solution bitmaps during
4519 2017-08-17 Marek Polacek <polacek@redhat.com>
4521 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
4523 2017-08-17 Richard Biener <rguenther@suse.de>
4525 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
4526 to true when overflow is undefined and we saturated the result.
4528 2017-08-17 Alan Modra <amodra@gmail.com>
4531 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
4532 Don't use store multiple if only one reg needs saving.
4533 (interesting_frame_related_regno): New function.
4534 (rs6000_frame_related): Don't emit frame info for regs that
4536 (rs6000_emit_epilogue): Likewise.
4538 2017-08-16 Nathan Sidwell <nathan@acm.org>
4540 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
4541 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
4542 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
4543 * tree.c (free_lang_data_in_type): Use else-if chain. Always
4544 clear TYPE_LANG_1. Remove obsolete member-function stripping.
4545 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
4546 (verify_type): Adjust for TYPE_BINFO move.
4547 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
4548 process TYPE_BINFO directly.
4549 (hash_tree): Likewise.
4550 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
4552 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
4555 2017-08-16 David Malcolm <dmalcolm@redhat.com>
4557 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
4559 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
4562 * config/i386/i386.md (*anddi_1_btr): Change predicates of
4563 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
4564 Add ix86_binary_operator_ok to insn constraint.
4565 (*iordi_1_bts): Ditto.
4566 (*xordi_1_btc): Ditto.
4567 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
4568 Update corresponding peephole2 pattern.
4572 2017-08-16 Bin Cheng <bin.cheng@arm.com>
4574 PR tree-optimization/81832
4575 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
4576 copy loop header which has IFN_LOOP_DIST_ALIAS call.
4578 2017-08-16 Marek Polacek <polacek@redhat.com>
4581 * fold-const.c (fold_indirect_ref_1): Restore original behavior
4582 regarding size_zero_node.
4584 2017-08-16 Martin Liska <mliska@suse.cz>
4587 * config.gcc: Respect previously set extra_objs in case
4590 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
4592 PR tree-optimization/81835
4593 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
4594 the phi SSA_NAME. Check that the condition in a COND_EXPR does
4595 not depend on the phi.
4597 2017-08-16 Alan Modra <amodra@gmail.com>
4599 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
4602 2017-08-16 Alan Modra <amodra@gmail.com>
4604 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
4605 (save_reg_p): ..into this. Update all callers.
4606 (first_reg_to_save): Simplify.
4608 2017-08-16 Alan Modra <amodra@gmail.com>
4610 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
4613 2017-08-15 Joseph Myers <joseph@codesourcery.com>
4617 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
4618 constant count if that count is less than 32.
4620 2017-08-15 Nathan Sidwell <nathan@acm.org>
4622 * gcc.c (execute): Emit friendlier message if inferior is killed
4623 by an external cause.
4625 2017-08-15 Richard Biener <rguenther@suse.de>
4627 PR tree-optimization/81790
4628 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
4629 CONSTRUCTORs from simplifying and VN.
4631 2017-08-14 Martin Sebor <msebor@redhat.com>
4633 * builtin-attrs.def: Add comments.
4635 2017-08-14 Martin Sebor <msebor@redhat.com>
4638 * doc/extend.texi (attribute nonstring): Document new attribute.
4640 2017-08-14 Martin Sebor <msebor@redhat.com>
4643 * tree-diagnostic.c (default_tree_printer): Handle %G.
4644 * gimple-pretty-print.h (percent_G_format): Declare new function.
4645 * gimple-pretty-print.c (percent_G_format): Define.
4646 * tree-pretty-print.c (percent_K_format): Add argument.
4648 2017-08-14 Martin Sebor <msebor@redhat.com>
4650 PR translation/79998
4651 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
4652 Remove a stray space.
4654 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
4657 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
4658 (*iordi_1_bts): Ditto.
4659 (*xordi_1_btc): Ditto.
4661 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4664 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
4665 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4667 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
4668 quoted strings, and make more translator-friendly.
4669 (darwin_rs6000_override_options): Likewise.
4670 (rs6000_option_override_internal): Likewise.
4671 (rs6000_return_in_memory): Fix overlong line.
4672 (init_cmulative_args): Use quoted strings, and make more
4673 translator-friendly.
4674 (rs6000_pass_by_reference): Fix overlong line.
4675 (def_builtin): Use quoted strings.
4676 (altivec_expand_predicate_builtin): Use quoted strings, and make
4677 more translator-friendly.
4678 (htm_expand_builtin): Use quoted strings.
4679 (cpu_expand_builtin): Use quoted strings, and make more
4680 translator-friendly.
4681 (altivec_expand_builtin): Likewise.
4682 (paired_expand_predicate_builtin): Likewise.
4683 (rs6000_invalid_builtin): Likewise.
4684 (builtin_function_type): Use quoted strings.
4685 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
4686 more translator-friendly.
4687 (rs6000_trampoline_init): Likewise.
4688 (rs6000_handle_altivec_attribute): Likewise.
4689 (rs6000_inner_target_options): Use quoted strings.
4690 (rs6000_disable_incompatible_switches): Likewise.
4691 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
4692 strings, and make more translator-friendly.
4693 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
4695 2017-08-14 Bin Cheng <bin.cheng@arm.com>
4697 PR tree-optimization/81799
4698 * tree-loop-distribution.c (version_loop_by_alias_check): Force
4699 cond_expr to simple gimple operand.
4701 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
4704 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
4706 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
4710 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
4711 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
4712 TARGET_GASISR_PROLOGUES.
4713 * config/avr/avr.c (avr_option_override): Same.
4714 (avr_pass_pre_proep::execute): Same.
4716 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
4719 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
4720 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
4722 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
4724 * config/i386/i386.md (*load_tp_<mode>): Redefine as
4725 define_insn_and_split. Split to a memory load from 0 in
4726 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
4727 using PTR mode iterator.
4728 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
4729 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
4730 (*add_tp_<mode>): Redefine as define_insn_and_split.
4731 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
4732 address space. Merge with *add_tp_x32 using PTR mode iterator.
4733 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
4734 Split to an add with a memory load from 0 in
4735 DEFAULT_TLS_SEG_REG address space.
4737 2017-08-12 Andrew Pinski <apinski@cavium.com>
4739 * config/aarch64/aarch64-option-extensions.def (rdma):
4740 Fix feature string to what Linux prints out in /proc/cpuinfo.
4742 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
4745 * dwarf2out.c (modified_type_die): For C typedef types that have
4746 an ultimate origin, process the ultimate origin instead of the
4748 (gen_typedef_die): Assert that input DECLs have no ultimate
4750 (gen_type_die_with_usage): For typedef variants that have an
4751 ultimate origin, just call gen_decl_die on the original DECL.
4752 (process_scope_var): Avoid creating DIEs for local typedefs and
4753 concrete static variables.
4755 2017-08-12 Alan Modra <amodra@gmail.com>
4759 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
4760 match gnu-user.h startfile.
4761 (ENDFILE_LINUX_SPEC): Similarly.
4763 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
4766 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
4768 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
4770 2017-08-11 Tamar Christina <tamar.christina@arm.com>
4771 * config/aarch64/aarch64.md (mov<mode>): Change.
4772 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4773 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
4774 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
4776 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
4778 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
4779 for storage order barriers.
4781 2017-08-11 Martin Liska <mliska@suse.cz>
4784 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
4785 variables of void type.
4787 2017-08-11 Martin Liska <mliska@suse.cz>
4789 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
4790 TARGET_SUPPORTS_ALIASES.
4791 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
4792 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
4793 (optimize_weakref): Likewise.
4794 * symtab.c (symtab_node::noninterposable_alias): Likewise.
4795 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
4796 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
4798 2017-08-11 Martin Liska <mliska@suse.cz>
4801 * config/i386/i386.c (make_resolver_func): Do complete
4802 refactoring of the function.
4804 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4807 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
4808 * config/i386/i386.c (ix86_stack_protect_guard): Use
4809 ix86_stack_protect_guard_symbol_str to generate varible declaration.
4810 * doc/invoke.texi (x86 Options): Document
4811 -mstack-protector-guard-symbol= option.
4813 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4815 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
4816 * config/i386/i386.c (ix86_split_stack_guard): New function.
4817 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
4818 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
4819 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
4820 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
4821 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
4822 (split_stack_space_check): Call ix86_split_stack_guard.
4824 2017-08-10 Martin Sebor <msebor@redhat.com>
4826 * print-tree.c (print_node): Print location using the established
4828 Replace spaces with colons.
4829 (debug_raw, debug): Ditto.
4831 2017-08-10 Martin Sebor <msebor@redhat.com>
4834 * pretty-print.c (pp_format): Correct the handling of %s precision.
4836 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
4839 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
4841 (ix86_finalize_stack_frame_flags): This. Also clear
4842 frame_pointer_needed if -fno-omit-frame-pointer is used without
4844 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
4845 with ix86_finalize_stack_frame_flags.
4846 (ix86_expand_epilogue): Likewise.
4847 (ix86_expand_split_stack_prologue): Likewise.
4848 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
4850 2017-08-10 Martin Liska <mliska@suse.cz>
4853 * c-attribs.c (handle_target_attribute):
4854 Report warning for an empty string argument of target attribute.
4856 2017-08-09 Jakub Jelinek <jakub@redhat.com>
4859 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
4861 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
4862 or DECL_NONLOCAL labels.
4863 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
4864 or DECL_NONLOCAL labels here.
4866 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
4868 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
4869 to indicate when early gimple folding has been disabled.
4870 (rs6000_gimple_fold_builtin): Add debug content.
4871 (rs6000_invalid_builtin): Fix whitespace.
4872 (rs6000_expand_builtin): Fix whitespace.
4873 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
4875 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
4878 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
4879 SAVE_MULTIPLE if not all the registers that saves, should be saved.
4881 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
4883 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
4884 (qdf24xx): Likewise.
4885 * config/aarch64/aarch64.md: Include falkor.md.
4886 * config/aarch64/falkor.md: New.
4888 2017-08-09 Marek Polacek <polacek@redhat.com>
4891 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
4892 * diagnostic.c (emit_diagnostic): Add a comment.
4893 (emit_diagnostic_valist): New function.
4895 2017-08-09 Marek Polacek <polacek@redhat.com>
4898 * input.c (make_location): New overload.
4899 * input.h (make_location): Declare.
4901 2017-08-08 Alan Modra <amodra@gmail.com>
4902 H.J. Lu <hongjiu.lu@intel.com>
4905 * gcc.c (NO_PIE_SPEC): Delete.
4906 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
4908 (LINK_PIE_SPEC): ..to here.
4909 (LINK_COMMAND_SPEC): Support -no-pie.
4910 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
4911 chain of crtbegin*.o selection, update for PIE_SPEC changes and
4913 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
4914 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
4915 (ENDFILE_CRTEND_SPEC): Similarly.
4917 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
4920 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
4921 (mstack-protector-guard-offset=): Ditto.
4922 * config/i386/i386.c (ix86_option_override): Handle
4923 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
4925 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
4926 ix86_stack_protect_guard_offset variables.
4927 (TARGET_STACK_PROTECT_GUARD): Always define.
4928 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
4929 and -mstack-protector-guard-offset= options.
4931 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4933 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
4934 boundary case for the last candidate.
4936 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4938 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
4939 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
4941 2017-08-08 Tamar Christina <tamar.christina@arm.com>
4944 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
4945 * config/aarch64/aarch64-builtins.c
4946 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
4947 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
4948 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
4950 2017-08-08 Tamar Christina <tamar.christina@arm.com>
4951 Andrew Pinski <pinskia@gmail.com>
4954 * internal-fn.def (XORSIGN): New.
4955 * optabs.def (xorsign_optab): New.
4956 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
4957 (convert_expand_mult_copysign): New.
4958 (pass_optimize_widening_mul::execute): Call
4959 convert_expand_mult_copysign.
4961 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4963 PR tree-optimization/81354
4964 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
4965 Insert on edges rather than explicitly creating landing pads.
4966 (analyze_candidates_and_replace): Commit edge inserts.
4968 2017-08-08 Richard Biener <rguenther@suse.de>
4971 * tree-ssa-loop-niter.c: Include tree-dfa.h.
4972 (expand_simple_operations): Also look through ADDR_EXPRs with
4973 MEM_REF bases treating them as POINTER_PLUS_EXPR.
4975 2017-08-08 Richard Biener <rguenther@suse.de>
4977 PR tree-optimization/81723
4978 * tree-vect-slp.c (struct bst_traits): New hash traits.
4979 (bst_fail): New global.
4980 (vect_build_slp_tree_2): New worker, split out from ...
4981 (vect_build_slp_tree): ... this now wrapping it with using
4982 bst_fail set to cache SLP tree build fails. Properly handle
4984 (vect_analyze_slp_instance): Allocate and free bst_fail.
4986 2017-08-08 Martin Liska <mliska@suse.cz>
4989 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
4990 LABEL_DECLs that can be from a different function.
4992 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4994 PR tree-optimization/81744
4995 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
4996 loop's number of iterations.
4998 2017-08-08 Martin Liska <mliska@suse.cz>
5000 * asan.c: Include header files.
5001 * attribs.c (build_decl_attribute_variant): New function moved
5003 (build_type_attribute_qual_variant): Likewise.
5004 (cmp_attrib_identifiers): Likewise.
5005 (simple_cst_list_equal): Likewise.
5006 (omp_declare_simd_clauses_equal): Likewise.
5007 (attribute_value_equal): Likewise.
5008 (comp_type_attributes): Likewise.
5009 (build_type_attribute_variant): Likewise.
5010 (lookup_ident_attribute): Likewise.
5011 (remove_attribute): Likewise.
5012 (merge_attributes): Likewise.
5013 (merge_type_attributes): Likewise.
5014 (merge_decl_attributes): Likewise.
5015 (merge_dllimport_decl_attributes): Likewise.
5016 (handle_dll_attribute): Likewise.
5017 (attribute_list_equal): Likewise.
5018 (attribute_list_contained): Likewise.
5019 * attribs.h (lookup_attribute): New function moved from tree.[ch].
5020 (lookup_attribute_by_prefix): Likewise.
5021 * bb-reorder.c: Include header files.
5022 * builtins.c: Likewise.
5023 * calls.c: Likewise.
5024 * cfgexpand.c: Likewise.
5025 * cgraph.c: Likewise.
5026 * cgraphunit.c: Likewise.
5027 * convert.c: Likewise.
5028 * dwarf2out.c: Likewise.
5029 * final.c: Likewise.
5030 * fold-const.c: Likewise.
5031 * function.c: Likewise.
5032 * gimple-expr.c: Likewise.
5033 * gimple-fold.c: Likewise.
5034 * gimple-pretty-print.c: Likewise.
5035 * gimple.c: Likewise.
5036 * gimplify.c: Likewise.
5037 * hsa-common.c: Likewise.
5038 * hsa-gen.c: Likewise.
5039 * internal-fn.c: Likewise.
5040 * ipa-chkp.c: Likewise.
5041 * ipa-cp.c: Likewise.
5042 * ipa-devirt.c: Likewise.
5043 * ipa-fnsummary.c: Likewise.
5044 * ipa-inline.c: Likewise.
5045 * ipa-visibility.c: Likewise.
5047 * lto-cgraph.c: Likewise.
5048 * omp-expand.c: Likewise.
5049 * omp-general.c: Likewise.
5050 * omp-low.c: Likewise.
5051 * omp-offload.c: Likewise.
5052 * omp-simd-clone.c: Likewise.
5053 * opts-global.c: Likewise.
5054 * passes.c: Likewise.
5055 * predict.c: Likewise.
5056 * sancov.c: Likewise.
5057 * sanopt.c: Likewise.
5058 * symtab.c: Likewise.
5059 * toplev.c: Likewise.
5060 * trans-mem.c: Likewise.
5061 * tree-chkp.c: Likewise.
5062 * tree-eh.c: Likewise.
5063 * tree-into-ssa.c: Likewise.
5064 * tree-object-size.c: Likewise.
5065 * tree-parloops.c: Likewise.
5066 * tree-profile.c: Likewise.
5067 * tree-ssa-ccp.c: Likewise.
5068 * tree-ssa-live.c: Likewise.
5069 * tree-ssa-loop.c: Likewise.
5070 * tree-ssa-sccvn.c: Likewise.
5071 * tree-ssa-structalias.c: Likewise.
5072 * tree-ssa.c: Likewise.
5073 * tree-streamer-in.c: Likewise.
5074 * tree-vectorizer.c: Likewise.
5075 * tree-vrp.c: Likewise.
5077 * ubsan.c: Likewise.
5078 * varasm.c: Likewise.
5079 * varpool.c: Likewise.
5080 * tree.c: Remove functions moved to attribs.[ch].
5082 * config/aarch64/aarch64.c: Add attrs.h header file.
5083 * config/alpha/alpha.c: Likewise.
5084 * config/arc/arc.c: Likewise.
5085 * config/arm/arm.c: Likewise.
5086 * config/avr/avr.c: Likewise.
5087 * config/bfin/bfin.c: Likewise.
5088 * config/c6x/c6x.c: Likewise.
5089 * config/cr16/cr16.c: Likewise.
5090 * config/cris/cris.c: Likewise.
5091 * config/darwin.c: Likewise.
5092 * config/epiphany/epiphany.c: Likewise.
5093 * config/fr30/fr30.c: Likewise.
5094 * config/frv/frv.c: Likewise.
5095 * config/ft32/ft32.c: Likewise.
5096 * config/h8300/h8300.c: Likewise.
5097 * config/i386/winnt.c: Likewise.
5098 * config/ia64/ia64.c: Likewise.
5099 * config/iq2000/iq2000.c: Likewise.
5100 * config/lm32/lm32.c: Likewise.
5101 * config/m32c/m32c.c: Likewise.
5102 * config/m32r/m32r.c: Likewise.
5103 * config/m68k/m68k.c: Likewise.
5104 * config/mcore/mcore.c: Likewise.
5105 * config/microblaze/microblaze.c: Likewise.
5106 * config/mips/mips.c: Likewise.
5107 * config/mmix/mmix.c: Likewise.
5108 * config/mn10300/mn10300.c: Likewise.
5109 * config/moxie/moxie.c: Likewise.
5110 * config/msp430/msp430.c: Likewise.
5111 * config/nds32/nds32-isr.c: Likewise.
5112 * config/nds32/nds32.c: Likewise.
5113 * config/nios2/nios2.c: Likewise.
5114 * config/nvptx/nvptx.c: Likewise.
5115 * config/pa/pa.c: Likewise.
5116 * config/pdp11/pdp11.c: Likewise.
5117 * config/powerpcspe/powerpcspe.c: Likewise.
5118 * config/riscv/riscv.c: Likewise.
5119 * config/rl78/rl78.c: Likewise.
5120 * config/rx/rx.c: Likewise.
5121 * config/s390/s390.c: Likewise.
5122 * config/sh/sh.c: Likewise.
5123 * config/sol2.c: Likewise.
5124 * config/sparc/sparc.c: Likewise.
5125 * config/spu/spu.c: Likewise.
5126 * config/stormy16/stormy16.c: Likewise.
5127 * config/tilegx/tilegx.c: Likewise.
5128 * config/tilepro/tilepro.c: Likewise.
5129 * config/v850/v850.c: Likewise.
5130 * config/vax/vax.c: Likewise.
5131 * config/visium/visium.c: Likewise.
5132 * config/xtensa/xtensa.c: Likewise.
5134 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5137 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
5138 constraints since the -mupper-regs-* switches have been
5140 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
5141 into a vector from a double word element that was extracted from
5142 another vector, and eliminate extra XXPERMDI instructions.
5143 (vsx_concat_<mode>_2): Likewise.
5144 (vsx_concat_<mode>_3): Likewise.
5145 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
5146 concat to allow optimizing inserts from previous extracts.
5148 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
5150 * config/i386/i386.c (ix86_stack_protect_guard): Generate
5151 memory reference to a SSP offset in TLS address space.
5152 (ix86_print_operand) <case '@'>: Remove.
5153 (ix86_print_operand_punct_valid_p): Remove '@' code.
5154 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
5156 (stack_tls_protect_set_<mode>): Remove.
5157 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
5158 (stack_tls_protect_test_<mode>): Remove.
5159 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
5161 2017-08-07 Olivier Hainque <hainque@adacore.com>
5164 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
5166 2017-08-07 Douglas Rupp <rupp@adacore.com>
5168 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
5169 variable was referenced as multidir in command.
5171 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5174 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
5177 2017-08-07 Martin Liska <mliska@suse.cz>
5179 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
5180 * config/rl78/rl78.c: Add include of attribs.h.
5181 * config/sh/sh.c: Likewise.
5182 * config/v850/v850.c: Likewise.
5184 2017-08-07 Tom de Vries <tom@codesourcery.com>
5187 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
5189 2017-08-07 Martin Liska <mliska@suse.cz>
5191 * config/mips/mips.c: Include attribs.h.
5193 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
5196 * doc/invoke.texi: Document change in behvaior for -Ofast for
5199 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
5201 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
5203 (aarch64_pop_regs): Likewise.
5204 (aarch64_gen_load_pair): Likewise.
5205 (aarch64_save_callee_saves): Likewise.
5206 (aarch64_restore_callee_saves): Likewise.
5208 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5210 * config/i386/i386.c: Revert the last change.
5212 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5215 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5217 (ix86_finalize_stack_frame_flags): This. Also clear
5218 frame_pointer_needed if -fno-omit-frame-pointer is used without
5220 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5221 with ix86_finalize_stack_frame_flags.
5222 (ix86_expand_epilogue): Likewise.
5223 (ix86_expand_split_stack_prologue): Likewise.
5225 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5228 * config/i386/i386.c (get_builtin_code_for_version): Set priority
5229 to P_AES for Westmere.
5231 2017-08-07 Jonathan Yong <10walls@gmail.com>
5233 * config/i386/mingw.opt (fset-stack-executable): Removed.
5234 * config/i386/cygming.opt (fset-stack-executable): Moved
5236 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
5238 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
5240 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
5242 2017-08-07 Marek Polacek <polacek@redhat.com>
5245 * fold-const.c (fold_indirect_ref_1): Check type_domain.
5247 2017-08-07 Martin Liska <mliska@suse.cz>
5249 * attribs.h (canonicalize_attr_name): New function.
5250 (cmp_attribs): Move from c-format.c and adjusted.
5251 (is_attribute_p): Moved from tree.h.
5252 * tree-inline.c: Add new includes.
5253 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
5254 (private_is_attribute_p): Remove.
5255 (private_lookup_attribute): Likewise.
5256 (private_lookup_attribute_by_prefix): Simplify.
5257 (remove_attribute): Use is_attribute_p.
5258 * tree.h: Remove removed declarations.
5260 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5263 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
5264 instead of computing it in the function. Formatting fix.
5265 (expand_case): Don't rely on default_edge being the first edge,
5266 clear it if removing it, pass default_edge to
5267 emit_case_dispatch_table.
5268 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
5271 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
5273 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
5274 insn in the function, emit NOP after the insn.
5276 2017-08-06 Tom de Vries <tom@codesourcery.com>
5278 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
5281 2017-08-06 Tom de Vries <tom@codesourcery.com>
5283 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
5286 2017-08-04 Yury Gribov <tetra2005@gmail.com>
5288 PR tree-optimization/57371
5289 * match.pd: New pattern.
5291 2017-08-04 Marek Polacek <polacek@redhat.com>
5294 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
5295 perform the computation in offset_int.
5297 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5299 PR tree-optimization/81136
5300 * tree-vectorizer.h: Include tree-hash-traits.h.
5301 (vec_base_alignments): New typedef.
5302 (vec_info): Add a base_alignments field.
5303 (vect_record_base_alignments): Declare.
5304 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
5306 (DR_IS_CONDITIONAL_IN_STMT): New macro.
5307 (create_data_ref): Add an is_conditional_in_stmt argument.
5308 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
5309 the is_conditional_in_stmt field.
5310 (data_ref_loc): Add an is_conditional_in_stmt field.
5311 (get_references_in_stmt): Set the is_conditional_in_stmt field.
5312 (find_data_references_in_stmt): Update call to create_data_ref.
5313 (graphite_find_data_references_in_stmt): Likewise.
5314 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
5315 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
5316 (vect_record_base_alignment): New function.
5317 (vect_record_base_alignments): Likewise.
5318 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
5319 for nested statements even if we fail to compute a misalignment.
5320 Use pooled base alignments for unconditional references.
5321 (vect_find_same_alignment_drs): Compare base addresses instead
5323 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
5324 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
5326 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5328 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
5329 Add an explicit name for the enum. Use auto_vec for slp_instances
5331 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
5333 (_bb_vec_info): Add a constructor and destructor.
5334 (vinfo_for_stmt): Return NULL for uids of -1 as well.
5335 (destroy_loop_vec_info): Delete.
5336 (vect_destroy_datarefs): Likewise.
5337 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
5338 (vec_info::vec_info): New function.
5339 (vec_info::~vec_info): Likewise.
5340 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
5341 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
5342 destroy_loop_vec_info.
5343 * tree-vect-loop.c (new_loop_vec_info): Replace with...
5344 (_loop_vec_info::_loop_vec_info): ...this.
5345 (destroy_loop_vec_info): Replace with...
5346 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
5347 the stmt_vec_infos. Leave handling of vec_info information to its
5348 destructor. Remove explicit vector releases.
5349 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
5350 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
5351 * tree-vect-slp.c (new_bb_vec_info): Replace with...
5352 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
5353 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
5354 (destroy_bb_vec_info): Replace with...
5355 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
5356 information to its destructor.
5357 (vect_slp_analyze_bb_1): Use new and delete instead of
5358 new_bb_vec_info and destroy_bb_vec_info.
5359 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
5362 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5364 * tree-data-ref.h (subscript): Add access_fn field.
5365 (data_dependence_relation): Add could_be_independent_p.
5366 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
5367 (same_access_functions): Move to tree-data-ref.c.
5368 * tree-data-ref.c (ref_contains_union_access_p): New function.
5369 (access_fn_component_p): Likewise.
5370 (access_fn_components_comparable_p): Likewise.
5371 (dr_analyze_indices): Add a reference to access_fn_component_p.
5372 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
5374 (constant_access_functions): Likewise.
5375 (add_other_self_distances): Likewise.
5376 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
5377 (initialize_data_dependence_relation): Use XCNEW and remove
5378 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
5379 of access functions that have the same type. Allow the
5380 subsequence to end with different bases in some circumstances.
5381 Record the chosen access functions in SUB_ACCESS_FN.
5382 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
5383 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
5384 (subscript_dependence_tester_1): Likewise dra and drb.
5385 (build_classic_dist_vector): Update calls accordingly.
5386 (subscript_dependence_tester): Likewise.
5387 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
5388 DDR_COULD_BE_INDEPENDENT_P.
5389 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
5390 comp_alias_ddrs instead of may_alias_ddrs.
5391 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
5393 (vect_analyze_data_ref_dependence): Use it if
5394 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
5395 distance vectors if that fails.
5396 (dependence_distance_ge_vf): New function.
5397 (vect_prune_runtime_alias_test_list): Use it. Don't clear
5398 LOOP_VINFO_MAY_ALIAS_DDRS.
5400 2017-08-04 Richard Biener <rguenther@suse.de>
5403 * fold-const.c (fold_binary_loc): Properly restrict
5404 minus_var0 && minus_var1 case when associating undefined overflow
5407 2017-08-04 Tom de Vries <tom@codesourcery.com>
5409 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
5411 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5413 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5414 Don't start diagnostic messages with a capital letter.
5415 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5417 (rs6000_invalid_builtin): Likewise.
5418 (rs6000_trampoline_init): Likewise.
5420 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5423 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
5424 after setting changeable df flags.
5426 2017-08-03 Richard Biener <rguenther@suse.de>
5428 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
5429 up if the use is in USE - X.
5431 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
5433 * toplev.c (dumpfile.h): New include.
5434 (internal_error_reentered): New static function. Use it...
5435 (internal_error_function): ...here to handle reentered internal_error.
5437 2017-08-03 Richard Biener <rguenther@suse.de>
5440 * fold-const.c (split_tree): Add minus_var and minus_con
5441 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
5442 here but always use minus_*.
5443 (associate_trees): Assert we never associate with MINUS_EXPR
5444 and NULL first operand. Do not recurse for PLUS_EXPR operands
5445 when associating as MINUS_EXPR either.
5446 (fold_binary_loc): Track minus_var and minus_con.
5448 2017-08-03 Tom de Vries <tom@codesourcery.com>
5451 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
5452 ACCEL_COMPILER, apply finish_options on
5453 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
5455 2017-08-03 Tom de Vries <tom@codesourcery.com>
5458 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
5459 function_entry_patch_area_size > 0.
5461 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5464 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
5465 instead of 10??LU, perform unit multiplication in wide_int,
5466 don't change alloc_object_size_limit if the limit is larger
5469 PR tree-optimization/81655
5470 PR tree-optimization/81588
5471 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
5472 the case when ranges[i].low and high are 1 for unsigned type with
5476 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
5477 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
5479 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5481 * tree-vrp.h: Add include guard.
5483 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
5486 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
5487 (ud2): New insn pattern.
5488 * config/i386/i386.c (ix86_expand_epilogue):
5489 For naked functions, generate ud2 instead of trap insn.
5491 2017-08-02 Marek Polacek <polacek@redhat.com>
5494 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
5496 2017-08-02 Tom de Vries <tom@codesourcery.com>
5497 Cesar Philippidis <cesar@codesourcery.com>
5499 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
5500 Add missing edge probabilities.
5502 2017-08-02 Tamar Christina <tamar.christina@arm.com>
5504 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
5507 2017-08-02 Jakub Jelinek <jakub@redhat.com>
5510 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
5511 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
5512 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
5513 prologue_seq sequences - if any.
5515 2017-08-02 Richard Biener <rguenther@suse.de>
5517 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
5518 via vectors if supported, integer extracts via punning if supported
5519 or otherwise vector extracts.
5521 2017-08-02 Richard Biener <rguenther@suse.de>
5523 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
5525 (bitmap_insert_into_set): ... this.
5527 2017-08-02 Richard Biener <rguenther@suse.de>
5529 PR tree-optimization/81633
5531 2015-08-17 Alan Hayward <alan.hayward@arm.com>
5533 PR tree-optimization/71752
5534 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
5536 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
5538 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
5539 (machine_function::call_ms2sysv_pad_out): Remove field.
5540 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
5541 (ix86_compute_frame_layout): Likewise.
5543 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
5546 * config/i386/i386.c (ix86_set_func_type): Disallow naked
5547 attribute with interrupt attribute.
5549 2017-08-01 Andrew Pinski <apinski@cavium.com>
5551 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
5552 BIT_INSERT_EXPR's operand 1
5553 to see if the types precision matches.
5555 2017-08-01 Martin Liska <mliska@suse.cz>
5558 * builtins.c (expand_builtin_memcpy_args): Remove.
5559 (expand_builtin_memcpy): Call newly added function
5560 expand_builtin_memory_copy_args.
5561 (expand_builtin_memcpy_with_bounds): Likewise.
5562 (expand_builtin_mempcpy): Remove last argument.
5563 (expand_builtin_mempcpy_with_bounds): Likewise.
5564 (expand_builtin_memory_copy_args): New function created from
5565 expand_builtin_mempcpy_args with small modifications.
5566 (expand_builtin_mempcpy_args): Remove.
5567 (expand_builtin_stpcpy): Remove unused argument.
5568 (expand_builtin): Likewise.
5569 (expand_builtin_with_bounds): Likewise.
5571 2017-08-01 Martin Liska <mliska@suse.cz>
5574 Make mempcpy more optimal (PR middle-end/70140).
5576 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5579 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5580 __builtin_vec_cmpne verify both arguments are compatible vectors
5581 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
5582 verify arg1_type is a pointer or array type. For __builtin_vec_st,
5583 move computation of aligned to after checking the argument types.
5587 * config/rs6000/vsx.md (vextract_fp_from_shorth,
5588 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
5591 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
5592 Doug Rupp <rupp@adacore.com>
5593 Olivier Hainque <hainque@adacore.com>
5595 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
5596 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
5598 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
5599 for TARGET_OS_CPP_BUILTIN.
5600 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
5601 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
5603 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
5604 passing required abi options to the assembler for EABI configurations.
5605 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
5606 of .text.hot and .text.unlikely sections for kernel modules when
5607 using ARM style exceptions.
5608 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
5609 options. Add EXTRA_CC1_SPEC.
5610 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
5612 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
5614 (ARM_TARGET2_DWARF_FORMAT): Define.
5615 * config/arm/t-vxworks: Adjust multilib control to removal of the
5616 Diab command line options.
5618 2017-08-01 Martin Liska <mliska@suse.cz>
5620 PR gcov-profile/81561
5621 * gcov.c (unblock): Make unblocking safe as we need to preserve
5622 index correspondence of blocks and block_lists.
5624 2017-08-01 Richard Biener <rguenther@suse.de>
5626 PR tree-optimization/81181
5627 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5628 (compute_antic): ... end of iteration here.
5630 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
5632 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
5633 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
5634 (ftree-slp-vectorize): Likewise.
5635 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
5636 can no longer be set independent of flag_tree_loop_vectorize.
5637 * omp-general.c (emp_max_vf): Likewise.
5638 * opts.c (enable_fdo_optimizations): Remove references to
5639 flag_tree_vectorize, these are now implicit.
5640 (common_handle_option): Remove handling for OPT_ftree_vectorize,
5641 and leave it for the options machinery.
5643 2017-08-01 Martin Liska <mliska@suse.cz>
5646 * builtins.c (expand_builtin_memcpy_args): Remove.
5647 (expand_builtin_memcpy): Call newly added function
5648 expand_builtin_memory_copy_args.
5649 (expand_builtin_memcpy_with_bounds): Likewise.
5650 (expand_builtin_mempcpy): Remove last argument.
5651 (expand_builtin_mempcpy_with_bounds): Likewise.
5652 (expand_builtin_memory_copy_args): New function created from
5653 expand_builtin_mempcpy_args with small modifications.
5654 (expand_builtin_mempcpy_args): Remove.
5655 (expand_builtin_stpcpy): Remove unused argument.
5656 (expand_builtin): Likewise.
5657 (expand_builtin_with_bounds): Likewise.
5659 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5662 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
5663 print "ds:" only for immediates in generic address space.
5665 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5668 * config/i386/i386.c (ix86_funciton_naked): New prototype.
5669 (ix86_function_ok_for_sibcall): Return false for naked functions.
5671 2017-08-01 Richard Biener <rguenther@suse.de>
5673 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
5674 (compute_antic): Seed worklist with exit block predecessors.
5675 * cfganal.c (dfs_find_deadend): For a cycle return the source
5676 of the edge closing it.
5678 2017-08-01 Tamar Christina <tamar.christina@arm.com>
5680 * config/aarch64/aarch64.c
5681 (aarch64_can_const_movi_rtx_p): Move 0 check.
5683 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5685 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
5686 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
5688 * match.pd: Ditto in address comparison pattern.
5690 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5692 PR tree-optimization/81627
5693 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
5694 closed ssa form for store-store chain.
5696 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5698 PR tree-optimization/81620
5699 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
5700 for store-store chain.
5702 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5704 PR tree-optimization/81588
5705 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
5706 ranges[i].in_p, invert comparison code ccode. For >/>=,
5707 swap rhs1 and rhs2 and comparison code unconditionally,
5708 for </<= don't do that. Don't swap rhs1/rhs2 again if
5709 ranges[i].in_p, instead invert comparison code ccode if
5710 opcode or oe->rank is BIT_IOR_EXPR.
5713 * optabs.def (vec_extract_optab, vec_init_optab): Change from
5714 a direct optab to conversion optab.
5715 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
5716 with GET_MODE_INNER as last argument instead of optab_handler.
5717 * expmed.c (extract_bit_field_1): Likewise. Use vector from
5718 vector extraction if possible and optab is available.
5719 * expr.c (store_constructor): Use convert_optab_handler instead
5720 of optab_handler. Use vector initialization from smaller
5721 vectors if possible and optab is available.
5722 * tree-vect-stmts.c (vectorizable_load): Likewise.
5723 * doc/md.texi (vec_extract, vec_init): Document that the optabs
5725 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
5726 of vec_init from half-sized vectors with the same element mode.
5727 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
5728 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
5729 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
5730 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
5731 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
5732 after mode in gen_vec_extract* calls.
5733 (vec_extract<mode>): Renamed to ...
5734 (vec_extract<mode><ssescalarmodelower>): ... this.
5735 (vec_extract<mode><ssehalfvecmodelower>): New expander.
5736 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
5737 element mode after mode in gen_vec_init* calls.
5738 (VEC_INIT_HALF_MODE): New mode iterator.
5739 (vec_init<mode>): Renamed to ...
5740 (vec_init<mode><ssescalarmodelower>): ... this.
5741 (vec_init<mode><ssehalfvecmodelower>): New expander.
5742 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
5743 (vec_extractv2sfsf): ... this.
5744 (vec_initv2sf): Renamed to ...
5745 (vec_initv2sfsf): ... this.
5746 (vec_extractv2si): Renamed to ...
5747 (vec_extractv2sisi): ... this.
5748 (vec_initv2si): Renamed to ...
5749 (vec_initv2sisi): ... this.
5750 (vec_extractv4hi): Renamed to ...
5751 (vec_extractv4hihi): ... this.
5752 (vec_initv4hi): Renamed to ...
5753 (vec_initv4hihi): ... this.
5754 (vec_extractv8qi): Renamed to ...
5755 (vec_extractv8qiqi): ... this.
5756 (vec_initv8qi): Renamed to ...
5757 (vec_initv8qiqi): ... this.
5758 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
5759 (vec_init<mode>): Renamed to ...
5760 (vec_init<mode><VEC_base_l>): ... this.
5761 (vec_extract<mode>): Renamed to ...
5762 (vec_extract<mode><VEC_base_l>): ... this.
5763 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
5764 (vec_initv2sfsf): ... this.
5765 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
5766 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5767 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
5768 element mode after mode in gen_vec_init* calls.
5769 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
5770 (vec_init<mode><Vel>): ... this.
5771 (vec_extract<mode>): Renamed to ...
5772 (vec_extract<mode><Vel>): ... this.
5773 * config/aarch64/iterators.md (Vel): New mode attribute.
5774 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
5775 Add element mode after mode in gen_vec_extract* calls.
5776 * config/s390/vector.md (non_vec_l): New mode attribute.
5777 (vec_extract<mode>): Renamed to ...
5778 (vec_extract<mode><non_vec_l>): ... this.
5779 (vec_init<mode>): Renamed to ...
5780 (vec_init<mode><non_vec_l>): ... this.
5781 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
5782 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
5784 * config/arm/iterators.md (V_elem_l): New mode attribute.
5785 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
5786 (vec_extract<mode><V_elem_l>): ... this.
5787 (vec_extractv2di): Renamed to ...
5788 (vec_extractv2didi): ... this.
5789 (vec_init<mode>): Renamed to ...
5790 (vec_init<mode><V_elem_l>): ... this.
5791 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
5792 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
5793 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
5794 Add element mode after gen_vec_extract* calls.
5795 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
5796 (vec_init<mode><unitmode>): ... this.
5797 (vec_extract<mode>): Renamed to ...
5798 (vec_extract<mode><unitmode>): ... this.
5799 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
5800 (vec_init<mode><unitmode>): ... this.
5801 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
5802 (vec_initv2sfsf): ... this.
5803 (vec_extractv2sf): Renamed to ...
5804 (vec_extractv2sfsf): ... this.
5805 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
5806 Add element mode after gen_vec_extract* calls.
5807 * config/mips/mips.md (unitmode): New mode iterator.
5808 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
5809 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
5810 * config/spu/spu.md (inner_l): New mode attribute.
5811 (vec_init<mode>): Renamed to ...
5812 (vec_init<mode><inner_l>): ... this.
5813 (vec_extract<mode>): Renamed to ...
5814 (vec_extract<mode><inner_l>): ... this.
5815 * config/sparc/sparc.md (veltmode): New mode iterator.
5816 (vec_init<VMALL:mode>): Renamed to ...
5817 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
5818 * config/ia64/vect.md (vec_initv2si): Renamed to ...
5819 (vec_initv2sisi): ... this.
5820 (vec_initv2sf): Renamed to ...
5821 (vec_initv2sfsf): ... this.
5822 (vec_extractv2sf): Renamed to ...
5823 (vec_extractv2sfsf): ... this.
5824 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
5825 (vec_init<mode>): Renamed to ...
5826 (vec_init<mode><VEC_base_l>): ... this.
5827 (vec_extract<mode>): Renamed to ...
5828 (vec_extract<mode><VEC_base_l>): ... this.
5829 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
5830 (vec_initv2sfsf): ... this.
5831 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
5832 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5833 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
5834 gen_vec_init* calls.
5836 2017-08-01 Richard Biener <rguenther@suse.de>
5838 PR tree-optimization/81297
5839 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
5840 TREE_OVERFLOW from INTEGER_CSTs.
5842 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
5844 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
5846 2017-07-31 Carl Love <cel@us.ibm.com>
5848 * config/rs6000/rs6000-c: Add support for built-in functions
5849 vector signed char vec_xl_be (signed long long, signed char *);
5850 vector unsigned char vec_xl_be (signed long long, unsigned char *);
5851 vector signed int vec_xl_be (signed long long, signed int *);
5852 vector unsigned int vec_xl_be (signed long long, unsigned int *);
5853 vector signed long long vec_xl_be (signed long long, signed long long *);
5854 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
5855 vector signed short vec_xl_be (signed long long, signed short *);
5856 vector unsigned short vec_xl_be (signed long long, unsigned short *);
5857 vector double vec_xl_be (signed long long, double *);
5858 vector float vec_xl_be (signed long long, float *);
5859 * config/rs6000/altivec.h (vec_xl_be): Add #define.
5860 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
5861 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
5863 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
5864 (altivec_expand_builtin): Add switch statement to call
5865 altivec_expand_xl_be for each builtin.
5866 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
5867 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
5868 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
5869 __builtin_vsx_le_be_v16qi.
5870 * doc/extend.texi: Update the built-in documentation file for the
5871 new built-in functions.
5873 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5876 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
5878 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5880 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5882 * config.gcc: Add z14.
5883 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
5884 CPU model numbers for z13s and z14.
5885 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
5887 * config/s390/s390-opts.h (enum processor_type): Rename
5888 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5889 * config/s390/s390.c (processor_table): Add field for CPU name to
5890 be passed to Binutils.
5891 (s390_asm_output_machine_for_arch): Use the new field in
5892 processor_table for Binutils.
5893 (s390_expand_builtin): Replace arch12 with z14.
5894 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5895 (s390_get_sched_attrmask): Likewise.
5896 (s390_get_unit_mask): Likewise.
5897 * config/s390/s390.opt: Add z14 to processor_type enum.
5899 2017-07-31 Martin Jambor <mjambor@suse.cz>
5902 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
5903 regardless of optimization level.
5905 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
5906 Martin Liska <mliska@suse.cz>
5908 * predict.def: Remove old comment and adjust probability.
5909 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
5912 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5915 * config/i386/i386.c (ix86_function_naked): New function.
5916 (ix86_can_use_return_insn_p): Return false for naked functions.
5917 (ix86_expand_prologue): Skip prologue for naked functions.
5918 (ix86_expand_epilogue): Skip epilogue for naked functions
5919 and emit trap instruction.
5920 (ix86_warn_func_return): New function.
5921 (ix86_attribute_table): Add "naked" attribute specification.
5922 (TARGET_WARN_FUNC_RETURN): Define.
5923 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
5925 2017-07-31 Martin Liska <mliska@suse.cz>
5927 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
5928 (dump_gimple_bb_header): Always dump BB info.
5929 (pp_cfg_jump): Do not append info about BB when dumping a jump.
5931 2017-07-31 Martin Liska <mliska@suse.cz>
5934 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
5935 also with current_function_decl non-null equality.
5937 2017-07-31 Jakub Jelinek <jakub@redhat.com>
5940 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
5941 change type to the element type, instead add eltype variable and
5942 use it where we are interested in the element type.
5944 PR tree-optimization/81603
5945 * ipa-polymorphic-call.c
5946 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
5947 offset arithmetic in offset_int, bail out if the resulting bit offset
5948 doesn't fit into shwi.
5950 2017-07-31 Martin Liska <mliska@suse.cz>
5952 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
5953 (gimplify_save_expr): Fix comment.
5955 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
5958 * config/i386/i386.c (ix86_function_arg): Update arguments for
5960 (ix86_compute_frame_layout): Set the initial stack offset to
5961 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
5962 INCOMING_FRAME_SP_OFFSET.
5963 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
5964 stack before exception handler returns.
5965 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
5966 the 'ERROR_CODE' for exception handler.
5968 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
5970 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
5971 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
5972 (ASM_OUTPUT_REG_POP): Ditto.
5973 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
5974 instead of asm_fprintf to output pure string.
5976 2017-07-29 Jakub Jelinek <jakub@redhat.com>
5978 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
5979 to imported_module_or_decl hook.
5980 (debug_nothing_tree_tree_tree_bool): Remove.
5981 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
5982 * debug.c (do_nothing_debug_hooks): Use
5983 debug_nothing_tree_tree_tree_bool_bool instead of
5984 debug_nothing_tree_tree_tree_bool.
5985 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5986 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
5987 * sdbout.c (sdb_debug_hooks): Likewise.
5988 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
5989 (gen_namespace_die): Add DW_AT_export_symbols attribute if
5991 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
5992 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
5993 attribute, don't add anything.
5995 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5997 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
5998 (fold_build2_stat_loc): Likewise.
5999 (fold_build3_stat_loc): Likewise.
6000 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
6001 (fold_build1_loc): Remove macro.
6002 (fold_build2_loc): Likewise.
6003 (fold_build3_loc): Likewise.
6005 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6007 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
6008 (gimple_build_debug_bind_source_stat): Likewise.
6009 * gimple.h (gimple_build_debug_bind): Remove macro.
6010 (gimple_build_debug_bind_source): Likewise.
6012 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6014 * bitmap.c (bitmap_alloc): Adjust.
6015 (bitmap_gc_alloc): Likewise.
6016 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
6018 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6020 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
6021 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
6022 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
6023 (bitmap_gc_alloc_stat): Likewise.
6024 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
6026 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6028 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
6029 * rtl.h (shallow_copy_rtx): Remove macro.
6031 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6033 * emit-rtl.c (gen_raw_REG): Adjust.
6034 * gengenrtl.c (gendef): Likewise.
6035 * rtl.c (rtx_alloc_stat): Remove _stat from name.
6036 * rtl.h (rtx_alloc): Remove macro.
6038 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6040 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
6041 (build_tree_list_stat): Likewise.
6042 * tree.h (build_tree_list): Remove macro.
6043 (build_tree_list_vec): Likewise.
6045 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6047 * tree.c (make_vector_stat): Remove _stat from name.
6048 (build_vector_stat): Likewise.
6049 * tree.h (make_vector_stat): Remove macro.
6050 (build_vector_stat): Likewise.
6052 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6054 * tree.h (build_var_debug_value): Remove prototype.
6056 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6058 * tree.c (tree_cons_stat): Remove _stat from name.
6059 * tree.h (tree_cons): Remove macro.
6061 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6063 * tree.c (build_vl_exp_stat): Remove _stat from name.
6064 * tree.h (build_vl_exp): Remove macro.
6066 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6068 * tree.c (build_decl_stat): Remove _stat from name.
6069 * tree.h (build_decl): Remove macro.
6071 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6073 * gimple.c (gimple_build_with_ops_stat): Adjust.
6074 (gimple_alloc_stat): Remove _stat from name.
6075 * gimple.h (gimple_alloc): Remove macro.
6077 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6079 * tree.c (make_tree_vec_stat): Remove _stat from name.
6080 (grow_tree_vec_stat): Likewise.
6081 * tree.h (make_tree_vec_stat): Adjust prototype.
6082 (grow_tree_vec_stat): Likewise.
6083 (make_tree_vec): Remove macro.
6084 (grow_tree_vec): Likewise.
6086 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6088 * fold-const.c (fold_build1_stat_loc): Adjust.
6089 (fold_build2_stat_loc): Likewise.
6090 (fold_build3_stat_loc): Likewise.
6091 * tree.c (build0_stat): Remove _stat from name.
6092 (build1_stat): Likewise.
6093 (build2_stat): Likewise.
6094 (build3_stat): Likewise.
6095 (build4_stat): Likewise.
6096 (build5_stat): Likewise.
6097 * tree.h (build1_loc): Remove macro, and rename _stat function
6099 (build2_loc): Likewise.
6100 (build3_loc): Likewise.
6101 (build4_loc): Likewise.
6102 (build5_loc): Likewise.
6104 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6106 * tree.c (make_int_cst_stat): Remove _stat from name.
6107 * tree.h (make_int_cst_stat): Adjust prototype.
6108 (make_int_cst): Remove macro.
6110 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6112 * tree.c (make_tre_binfo_stat): Remove _stat from name.
6113 * tree.h (make_tree_binfo_stat): Adjust prototype.
6114 (make_tree_binfo): Remove.
6116 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6118 * tree.c (copy_node_stat): Rename to copy_node.
6119 (build_distinct_type_copy): Adjust.
6120 * tree.h (copy_node_stat): Adjust prototype.
6121 (copy_node): Remove macro.
6123 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6125 * tree.c (make_node_stat): rename to make_node.
6126 (build_tree_list_stat): Adjust.
6127 (build0_stat): Likewise.
6128 (build2_stat): Likewise.
6129 (build3_stat): Likewise.
6130 (build4_stat): Likewise.
6131 (build5_stat): Likewise.
6132 (build_decl_stat): Likewise.
6133 * tree.h (make_node_stat): Adjust prototype.
6134 (make_node): remove macro.
6136 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
6138 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
6139 (PPC_FEATURE2_SCV): Likewise.
6140 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
6142 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6144 * config/aarch64/aarch64.c
6145 (aarch64_internal_mov_immediate): Add new special pattern.
6146 * config/aarch64/aarch64.md (*movdi_aarch64):
6147 Add reg/32bit const mov case.
6149 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6150 Richard Sandiford <richard.sandiford@linaro.org>
6152 * config/aarch64/aarch64.md (mov<mode>): Generalize.
6153 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6154 Add integer and movi cases.
6155 (movi-split-hf-df-sf split, fp16): New.
6156 (enabled): Added TARGET_FP_F16INST.
6157 * config/aarch64/iterators.md (GPF_HF): New.
6158 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
6160 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6162 * config/aarch64/aarch64.c
6163 (aarch64_simd_container_mode): Add prototype.
6164 (aarch64_expand_mov_immediate): Add HI support.
6165 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
6166 (aarch64_can_const_movi_rtx_p): New.
6167 (aarch64_preferred_reload_class):
6168 Remove restrictions of using FP registers for certain SIMD operations.
6169 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
6170 (aarch64_valid_floating_const): Add integer move validation.
6171 (aarch64_simd_imm_scalar_p): Remove.
6172 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
6173 (aarch64_legitimate_constant_p): Expand list of supported cases.
6174 * config/aarch64/aarch64-protos.h
6175 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
6176 (aarch64_reinterpret_float_as_int): New.
6177 (aarch64_simd_imm_scalar_p): Remove.
6178 * config/aarch64/constraints.md (Uvi): New.
6179 (Dd): Split into Ds and new Dd.
6180 * config/aarch64/aarch64.md (*movsi_aarch64):
6182 (*movdi_aarch64): Add SIMD mov case.
6184 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6186 * tree-predcom.c: (struct chain): Handle store-store chain in which
6187 stores for elimination only store loop invariant values.
6188 (execute_pred_commoning_chain): Ditto.
6189 (prepare_initializers_chain_store_elim): Ditto.
6190 (prepare_finalizers): Ditto.
6191 (is_inv_store_elimination_chain): New function.
6192 (initialize_root_vars_store_elim_1): New function.
6194 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6196 * tree-predcom.c: Revise general description of the pass.
6197 (enum chain_type): New enum type for store elimination.
6198 (struct chain): New field supporting store elimination.
6199 (struct component): Ditto.
6200 (dump_chain): Dump store-stores chain.
6201 (release_chain): Release resources.
6202 (split_data_refs_to_components): Compute and create component
6203 contains only stores for elimination.
6204 (get_chain_last_ref_at): New function.
6205 (make_invariant_chain): Initialization.
6206 (make_rooted_chain): Specify chain type in parameter and record it.
6207 (add_looparound_copies): Skip for store-stores chain.
6208 (determine_roots_comp): Compute type of chain and pass it to
6210 (initialize_root_vars_store_elim_2): New function.
6211 (finalize_eliminated_stores): New function.
6212 (remove_stmt): Handle store for elimination.
6213 (execute_pred_commoning_chain): Execute predictive commoning on
6215 (determine_unroll_factor): Skip unroll for store-stores chain.
6216 (prepare_initializers_chain_store_elim): New function.
6217 (prepare_initializers_chain): Hanlde store-store chain.
6218 (prepare_finalizers_chain, prepare_finalizers): New function.
6219 (tree_predictive_commoning_loop): Return integer value indicating
6220 if loop is unrolled or lcssa form is corrupted.
6221 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
6223 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6225 * tree-predcom.c (initialize_root): Delete.
6226 (execute_pred_commoning_chain): Initialize root vars and replace
6227 reference of non-combined chain directly, rather than call above
6230 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6232 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
6233 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
6235 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6237 * tree-predcom.c (struct chain): New field init_seq.
6238 (release_chain): Release init_seq.
6239 (prepare_initializers_chain): Record intialization stmts in above
6241 (insert_init_seqs): New function.
6242 (tree_predictive_commoning_loop): Call insert_init_seqs.
6244 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6246 * tree-predcom.c (determine_roots_comp): Skip trivial components.
6248 2017-07-28 Richard Biener <rguenther@suse.de>
6250 * match.pd: Remove superfluous :c.
6251 * genmatch.c (simplify::id): Add member.
6252 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
6254 (current_id): New global.
6255 (dt_node::parent): Move from ...
6256 (dt_operand::parent): ... here. Add for_id member.
6257 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
6258 (decision_tree::find_node): Relax order requirement when
6259 merging DT_TRUE nodes to ones inbetween the current simplify
6260 and the one we try to merge with. Add diagnostic whenever
6261 we need to enforce pattern order by not merging.
6262 (decision_tree::insert): Set current_id.
6263 (decision_tree::print_node): Dump parent node and for_id.
6264 (parser::last_id): Add member.
6265 (parser::push_simplify): Assign unique id.
6266 (parser::parser): Initialize last_id.
6268 2017-07-28 Martin Liska <mliska@suse.cz>
6271 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
6272 gimple_build_debug_bind.
6274 2017-07-28 Richard Biener <rguenther@suse.de>
6276 PR tree-optimization/81502
6277 * match.pd: Add pattern combining BIT_INSERT_EXPR with
6279 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
6281 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
6282 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
6283 for BIT_FIELD_REF args.
6284 * fold-const.c (make_bit_field_ref): Likewise.
6285 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6287 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6290 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
6291 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
6292 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
6293 Or it into SANITIZER_UNDEFINED.
6294 * ubsan.c: Include gimple-fold.h and varasm.h.
6295 (ubsan_expand_ptr_ifn): New function.
6296 (instrument_pointer_overflow): New function.
6297 (maybe_instrument_pointer_overflow): New function.
6298 (instrument_object_size): Formatting fix.
6299 (pass_ubsan::execute): Call instrument_pointer_overflow
6300 and maybe_instrument_pointer_overflow.
6301 * internal-fn.c (expand_UBSAN_PTR): New function.
6302 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
6303 * sanitizer.def (__ubsan_handle_pointer_overflow,
6304 __ubsan_handle_pointer_overflow_abort): New builtins.
6305 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
6306 * internal-fn.def (UBSAN_PTR): New internal function.
6307 * opts.c (sanitizer_opts): Add pointer-overflow.
6308 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
6309 * fold-const.c (build_range_check): Compute pointer range check in
6310 integral type if pointer arithmetics would be needed. Formatting
6313 2017-07-28 Martin Liska <mliska@suse.cz>
6316 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
6317 parameters that are of a variable-length.
6319 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
6321 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
6323 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
6324 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
6325 (CRT_CALL_STATIC_FUNCTION): Likewise.
6326 (ASM_DEFAULT_SPEC): New define.
6327 (ASM_SPEC32): Likewise.
6328 (ASM_SPEC64): Likewise.
6329 (ASM_SPEC_COMMON): Likewise.
6330 (ASM_SPEC): Likewise.
6331 (INVALID_64BIT): Likewise.
6332 (LINK_OS_DEFAULT_SPEC): Likewise.
6333 (LINK_OS_SPEC32): Likewise.
6334 (LINK_OS_SPEC64): Likewise.
6335 (POWERPC_LINUX): Likewise.
6336 (PTRDIFF_TYPE): Likewise.
6337 (RESTORE_FP_PREFIX): Likewise.
6338 (RESTORE_FP_SUFFIX): Likewise.
6339 (SAVE_FP_PREFIX): Likewise.
6340 (SAVE_FP_SUFFIX): Likewise.
6341 (SIZE_TYPE): Likewise.
6342 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
6343 (TARGET_64BIT): Likewise.
6344 (TARGET_64BIT): Likewise.
6345 (TARGET_AIX): Likewise.
6346 (WCHAR_TYPE_SIZE): Likewise.
6347 (WCHAR_TYPE): Undefine.
6348 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
6349 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
6350 (CPP_OS_RTEMS_SPEC): Delete.
6351 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
6352 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
6354 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
6356 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6358 PR tree-optimization/81578
6359 * tree-parloops.c (build_new_reduction): Bail out if
6360 reduction_code isn't one of the standard OpenMP reductions.
6361 Move the details printing after that decision.
6363 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6365 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
6366 related to reload_in_progress.
6367 (splat_input_operand): Likewise.
6368 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
6370 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
6372 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
6373 (TARGET_INSTANTIATE_DECLS): Likewise.
6374 (legitimate_indexed_address_p): Delete reload_in_progress code.
6375 (rs6000_debug_legitimate_address_p): Likewise.
6376 (rs6000_eliminate_indexed_memrefs): Likewise.
6377 (rs6000_emit_le_vsx_store): Likewise.
6378 (rs6000_emit_move_si_sf_subreg): Likewise.
6379 (rs6000_emit_move): Likewise.
6380 (register_to_reg_type): Likewise.
6381 (rs6000_pre_atomic_barrier): Likewise.
6382 (rs6000_machopic_legitimize_pic_address): Likewise.
6383 (rs6000_allocate_stack_temp): Likewise.
6384 (rs6000_address_for_fpconvert): Likewise.
6385 (rs6000_address_for_altivec): Likewise.
6386 (rs6000_secondary_memory_needed_rtx): Delete function.
6387 (rs6000_check_sdmode): Likewise.
6388 (rs6000_alloc_sdmode_stack_slot): Likewise.
6389 (rs6000_instantiate_decls): Likewise.
6390 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
6391 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
6392 Delete reload_in_progress.
6393 (*vec_reload_and_plus_<mptrsize>): Likewise.
6394 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
6395 (vsx_div_v2di): Likewise.
6396 (vsx_udiv_v2di): Likewise.
6398 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6400 * config/rs6000/rs6000.opt (mlra): Replace with stub.
6401 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
6402 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
6403 (rs6000_debug_reg_global): Delete print of LRA status.
6404 (rs6000_option_override_internal): Delete dead LRA related code.
6405 (rs6000_lra_p): Delete function.
6406 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
6408 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
6410 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
6411 * config/riscv/rtems.h: New file.
6413 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6414 Sudakshina Das <sudi.das@arm.com>
6416 * config/aarch64/aarch64.md
6417 (define_split for and<mode>3nr_compare): Move
6418 non aarch64_logical_operand to a register.
6419 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
6420 register immediate operand to a register.
6421 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
6423 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6426 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
6428 2017-07-27 Richard Biener <rguenther@suse.de>
6430 PR tree-optimization/81573
6431 PR tree-optimization/81494
6432 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
6433 multi defuse cycle case.
6435 2017-07-27 Richard Biener <rguenther@suse.de>
6437 PR tree-optimization/81571
6438 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
6441 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
6443 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
6444 earlier and only if MASK_FPU is set. Adjust formatting.
6446 2017-07-27 Martin Liska <mliska@suse.cz>
6448 * opt-functions.awk: Add validation of value of Init.
6449 * optc-gen.awk: Pass new argument.
6451 2017-07-27 Martin Liska <mliska@suse.cz>
6453 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
6454 Fix wrong condition.
6456 2017-07-27 Martin Liska <mliska@suse.cz>
6458 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
6459 BBs and edges seen by autoFDO.
6461 2017-07-27 Richard Biener <rguenther@suse.de>
6463 PR tree-optimization/81502
6464 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
6465 with incompatible but same sized type.
6466 (execute_update_addresses_taken): Likewise.
6468 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
6470 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
6471 flag_tree_loop_vectorize rather than flag_tree_vectorize.
6473 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6476 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
6477 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
6478 Change s_operand to memory_operand.
6480 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
6482 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
6483 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
6484 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
6485 Emit instructions rather than returning an expression. Handle TFmode
6486 and KFmode by casting to TImode.
6487 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
6488 (rs6000_emit_le_vsx_store): Likewise.
6489 * config/rs6000/vsx.md (VSX_TI): New iterator.
6490 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
6491 (*vsx_le_undo_permute_<mode>): Likewise.
6492 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
6493 emit the split sequence.
6494 (*vsx_le_perm_store_<mode>): Likewise.
6496 2017-07-27 Jakub Jelinek <jakub@redhat.com>
6498 PR tree-optimization/81555
6499 PR tree-optimization/81556
6500 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
6501 if true, force CHANGED for the recursive invocation.
6502 (reassociate_bb): Remember original length of ops array, pass
6503 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
6505 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
6506 attributes for noipa attribute. For naked attribute use
6507 lookup_attribute first before lookup_attribute_spec.
6508 * final.c (rest_of_handle_final): Disable IPA RA for functions with
6510 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
6511 for functions with noipa attribute.
6512 (cgraph_externally_visible_p): Return true for functions with noipa
6514 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
6515 for functions with noipa attribute.
6516 * doc/extend.texi: Document noipa function attribute.
6517 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
6518 also for functions with noipa attribute.
6519 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
6521 2017-07-26 Andrew Pinski <apinski@cavium.com>
6523 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
6524 vec_unalign_load_cost and vec_unalign_store_cost.
6526 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6528 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
6529 -mvsx-small-integer option.
6530 (ISA_3_0_MASKS_IEEE): Likewise.
6531 (OTHER_VSX_VECTOR_MASKS): Likewise.
6532 (POWERPC_MASKS): Likewise.
6533 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
6534 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
6535 code, only testing for DImode being allowed in non-VSX floating
6537 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
6538 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
6540 (rs6000_option_override_internal): Delete -mvsx-small-integer.
6541 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
6542 TARGET_P8_VECTOR test.
6543 (rs6000_secondary_reload_simple_move): Likewise.
6544 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
6545 since TARGET_P9_VECTOR was already tested.
6546 (rs6000_opt_masks): Remove -mvsx-small-integer.
6547 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
6548 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6550 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
6551 test for TARGET_VEXTRACTUB was used, and that uses
6553 (p9 extract splitter): Likewise.
6554 (vsx_extract_<mode>_di_p9): Likewise.
6555 (vsx_extract_<mode>_store_p9): Likewise.
6556 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
6557 for TARGET_P9_VECTOR was used. Delete code that is now dead with
6558 the elimination of TARGET_VSX_SMALL_INTEGER.
6559 (vsx_extract_<mode>_p8): Likewise.
6560 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
6561 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
6562 (vsx_set_<mode>_p9): Likewise.
6563 (vsx_set_v4sf_p9): Likewise.
6564 (vsx_set_v4sf_p9_zero): Likewise.
6565 (vsx_insert_extract_v4sf_p9): Likewise.
6566 (vsx_insert_extract_v4sf_p9_2): Likewise.
6567 * config/rs6000/rs6000.md (sign extend splitter): Change
6568 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
6569 (floatsi<mode>2_lfiwax_mem): Likewise.
6570 (floatunssi<mode>2_lfiwzx_mem): Likewise.
6571 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
6572 since a test for TARGET_P9_VECTOR was used.
6573 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6574 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
6575 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6576 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6577 TARGET_P8_VECTOR test.
6578 (fix_trunc<mode>si2_stfiwx): Likewise.
6579 (fix_trunc<mode>si2_internal): Likewise.
6580 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
6581 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6583 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6584 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6585 TARGET_P8_VECTOR test.
6586 (fixuns_trunc<mode>si2_stfiwx): Likewise.
6587 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
6588 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6590 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6591 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
6592 since a test for TARGET_P9_VECTOR was used.
6593 (splitter for loading small constants): Likewise.
6595 2017-07-26 Andrew Pinski <apinski@cavium.com>
6597 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
6600 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
6603 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
6604 (fp_valid_at): Likewise.
6606 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6608 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
6609 (qdf24xx_addrcost_table): Likewise.
6610 (cortexa57_tunings): Update to use generic_branch_cost.
6611 (cortexa72_tunings): Likewise.
6612 (cortexa73_tunings): Likewise.
6613 (qdf24xx_tunings): Likewise.
6615 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6617 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
6618 (thunderx2t99_branch_cost): Likewise.
6619 (cortexa35_tunings): Update to use generic_branch_cost.
6620 (cortexa53_tunings): Likewise.
6621 (cortexa57_tunings): Likewise.
6622 (cortexa72_tunings): Likewise.
6623 (cortexa73_tunings): Likewise.
6624 (thunderx2t99_tunings): Likewise.
6626 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6628 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
6629 (sparc_option_override): Honour MASK_FSMULD.
6630 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
6631 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
6632 * config/sparc/sparc.opt (mfsmuld): New option.
6633 * doc/invoke.texi (mfsmuld): Document option.
6635 2017-07-26 Marek Polacek <polacek@redhat.com>
6638 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
6640 2017-07-26 Richard Biener <rguenther@suse.de>
6642 * gimple-match-head.c (do_valueize): Return OP if valueize
6644 (get_def): New helper to get at the def stmt of a SSA name
6646 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
6647 do_valueize to get at the def stmt.
6648 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
6650 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
6653 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
6655 2017-07-26 Martin Liska <mliska@suse.cz>
6658 * function.c (expand_function_start): Make expansion of
6659 nonlocal_goto_save_area after parm_birth_insn.
6661 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6663 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
6664 from all CPU target flags enable members.
6666 2017-07-26 Richard Biener <rguenther@suse.de>
6668 * genmatch.c (dt_simplify::gen): Make iterator vars const.
6669 (decision_tree::gen): Make 'type' const.
6670 (write_predicate): Likewise.
6672 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6674 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
6675 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
6676 (rs6000_option_override_internal): Likewise.
6677 (rs6000_expand_vector_set): Likewise.
6678 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
6679 (TARGET_UPPER_REGS_SF): Likewise.
6680 (TARGET_UPPER_REGS_DI): Likewise.
6681 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
6682 (TARGET_DIRECT_MOVE_64BIT): Likewise.
6683 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
6684 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6685 (Splitters for DI constants in Altivec registers): Likewise.
6686 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
6687 (vsx_set_v4sf_p9): Likewise.
6688 (vsx_set_v4sf_p9_zero): Likewise.
6689 (vsx_insert_extract_v4sf_p9): Likewise.
6690 (vsx_insert_extract_v4sf_p9_2): Likewise.
6692 2017-07-25 Carl Love <cel@us.ibm.com>
6694 * doc/extend.texi: Update the built-in documentation file for the
6695 existing built-in functions
6696 vector signed char vec_cnttz (vector signed char);
6697 vector unsigned char vec_cnttz (vector unsigned char);
6698 vector signed short vec_cnttz (vector signed short);
6699 vector unsigned short vec_cnttz (vector unsigned short);
6700 vector signed int vec_cnttz (vector signed int);
6701 vector unsigned int vec_cnttz (vector unsigned int);
6702 vector signed long long vec_cnttz (vector signed long long);
6703 vector unsigned long long vec_cnttz (vector unsigned long long);
6705 2017-07-25 Andrew Pinski <apinski@cavium.com>
6707 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
6708 accesses where the use is for the first operand of a BIT_INSERT.
6710 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6713 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6714 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6716 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6718 * config/i386/gstabs.h: Delete.
6719 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
6721 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
6723 * config/i386/i386.c (ix86_decompose_address): Do not check for
6724 register RTX when looking at index_reg or base_reg.
6725 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
6727 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
6729 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
6730 to update EH info here.
6732 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6734 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
6736 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6738 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
6740 2017-07-25 Torsten Duwe <duwe@suse.de>
6742 * common.opt: Introduce -fpatchable-function-entry
6743 command line option, and its variables function_entry_patch_area_size
6744 and function_entry_patch_area_start.
6745 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
6746 including a two-value parser.
6747 * target.def (print_patchable_function_entry): New target hook.
6748 * targhooks.h (default_print_patchable_function_entry): New function.
6749 * targhooks.c (default_print_patchable_function_entry): Likewise.
6750 * toplev.c (process_options): Switch off IPA-RA if
6751 patchable function entries are being generated.
6752 * varasm.c (assemble_start_function): Look at the
6753 patchable-function-entry command line switch and current
6754 function attributes and maybe generate NOP instructions by
6755 calling the print_patchable_function_entry hook.
6756 * doc/extend.texi: Document patchable_function_entry attribute.
6757 * doc/invoke.texi: Document -fpatchable_function_entry
6758 command line option.
6759 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
6761 * doc/tm.texi: Re-generate.
6763 2017-07-25 Jakub Jelinek <jakub@redhat.com>
6766 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
6767 TARGET_AVX512DQ rather than TARGET_AVX512BW.
6769 2017-07-25 Tamar Christina <tamar.christina@arm.com>
6771 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
6773 2017-07-25 Richard Biener <rguenther@suse.de>
6775 PR tree-optimization/81455
6776 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
6777 not walk in cycles when looking for guards.
6779 2017-07-25 Richard Biener <rguenther@suse.de>
6781 PR tree-optimization/81529
6782 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
6783 when optimizing backedge uses.
6785 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
6787 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
6789 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
6790 to dl_section_ref. On AIX, append an expression to subtract
6791 the size of the section length to dl_section_ref.
6793 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
6795 * configure.ac: If any of the config.* scripts fail, exit 1.
6796 * configure: Regenerate.
6798 2017-07-25 Richard Biener <rguenther@suse.de>
6801 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
6802 of immediate uses, be more verbose on errors.
6804 2017-07-25 Richard Biener <rguenther@suse.de>
6806 PR tree-optimization/81510
6807 * tree-vect-loop.c (vect_is_simple_reduction): When the
6808 reduction stmt is not inside the loop bail out.
6810 2017-07-25 Richard Biener <rguenther@suse.de>
6812 PR tree-optimization/81303
6813 * tree-vect-loop-manip.c (vect_loop_versioning): Build
6814 profitability check against LOOP_VINFO_NITERSM1.
6816 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6818 * domwalk.c (cmp_bb_postorder): Simplify.
6819 (sort_bbs_postorder): New function. Use it...
6820 (dom_walker::walk): ...here to optimize common cases.
6822 2017-07-25 Martin Liska <mliska@suse.cz>
6825 * ipa-visibility.c (function_and_variable_visibility): Make the
6826 redirection just on target that supports aliasing.
6827 Fix GNU coding style.
6829 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
6832 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
6834 * config/arm/rtems.h: Likewise.
6835 * config/bfin/rtems.h: Likewise.
6836 * config/i386/rtemself.h: Likewise.
6837 * config/lm32/rtems.h: Likewise.
6838 * config/m32c/rtems.h: Likewise.
6839 * config/m68k/rtemself.h: Likewise.
6840 * config/microblaze/rtems.h: Likewise.
6841 * config/mips/rtems.h: Likewise.
6842 * config/moxie/rtems.h: Likewise.
6843 * config/nios2/rtems.h: Likewise.
6844 * config/powerpcspe/rtems.h: Likewise.
6845 * config/rs6000/rtems.h: Likewise.
6846 * config/rtems.h: Likewise.
6847 * config/sh/rtems.h: Likewise.
6848 * config/sh/rtemself.h: Likewise.
6849 * config/sparc/rtemself.h: Likewise.
6851 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6854 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
6855 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
6856 * tree-ssa-structalias.c (alias_get_name): Same.
6858 2017-07-25 Bin Cheng <bin.cheng@arm.com>
6861 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
6862 instructions if no du chain is found.
6864 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6866 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
6868 2017-07-25 Richard Biener <rguenther@suse.de>
6871 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
6874 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6876 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
6878 (ISA_2_7_MASKS_SERVER): Likewise.
6879 (ISA_3_0_MASKS_IEEE): Likewise.
6880 (OTHER_P8_VECTOR_MASKS): Likewise.
6881 (OTHER_VSX_VECTOR_MASKS): Likewise.
6882 (POWERPC_MASKS): Likewise.
6883 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
6884 duplicate list of options.
6885 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
6886 explicit -mupper-regs options.
6887 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
6888 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
6889 alias for -mupper-regs-df.
6890 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6891 (rs6000_init_hard_regno_mode_ok): Likewise.
6892 (rs6000_option_override_internal): Likewise.
6893 (rs6000_opt_masks): Likewise.
6894 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
6895 options in terms of whether -mvsx or -mpower8-vector was used.
6896 (TARGET_UPPER_REGS_DI): Likewise.
6897 (TARGET_UPPER_REGS_SF): Likewise.
6898 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
6899 -mupper-regs-* options.
6901 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6903 * passes.c (emergency_dump_function): Print some empty lines and a
6904 header before the RTL dump.
6906 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6908 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
6910 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
6913 * config/aarch64/aarch64.c (aarch64_classify_symbol):
6914 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
6916 2017-07-24 Carl Love <cel@us.ibm.com>
6918 * config/rs6000/rs6000-c.c: Add support for built-in functions
6919 vector float vec_extract_fp32_from_shorth (vector unsigned short);
6920 vector float vec_extract_fp32_from_shortl (vector unsigned short);
6921 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
6922 vec_extract_fp_from_shortl): Add defines for the two builtins.
6923 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
6924 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
6926 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
6927 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
6928 * doc/extend.texi: Update the built-in documentation file for the
6929 new built-in function.
6931 2017-07-24 Jakub Jelinek <jakub@redhat.com>
6934 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
6936 * doc/generic.texi: Likewise.
6937 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6938 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6940 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
6942 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
6943 (aarch64_mls_elt_merge<mode>): Likewise.
6945 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
6947 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
6948 having __cxa_atexit.
6950 2017-07-23 Michael Collison <michael.collison@arm.com>
6952 * config/arm/arm.c (arm_option_override): Deprecate
6953 use of -mstructure-size-boundary.
6954 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
6955 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
6957 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6960 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
6961 Reduce cost estimate for direct moves.
6963 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
6966 * config/i386/i386.c (ix86_option_override_internal): Disable
6967 BMI, BMI2 and TBM instructions for -m16.
6969 2017-07-21 Carl Love <cel@us.ibm.com>
6971 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6972 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6973 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
6974 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
6975 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
6976 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
6977 VMULOSW): New enum "unspec" values.
6978 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
6979 altivec_vmulosw): New patterns.
6980 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
6981 VMULOSW): Add definitions.
6983 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
6985 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
6986 (qdf24xx): Likewise.
6987 * config/aarch64/aarch64-options-extensions.def (rdma); New.
6988 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
6989 (AARCH64_FL_V8_1): Renumber.
6990 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
6991 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
6992 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
6993 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
6994 rdma to feature modifiers list.
6996 2017-07-21 Yury Gribov <tetra2005@gmail.com>
6999 * ipa-visibility (function_and_variable_visibility): Convert
7000 recursive PLT call to direct call if appropriate.
7002 2017-07-21 Andrew Pinski <apinski@cavium.com>
7004 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
7005 operand 1 to see if the types precision matches.
7006 * fold-const.c (operand_equal_p): Likewise.
7008 2017-07-21 Richard Biener <rguenther@suse.de>
7010 PR tree-optimization/81303
7011 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
7012 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
7013 (vect_peeling_hash_get_lowest_cost): Adjust.
7014 (vect_enhance_data_refs_alignment): Likewise. Use
7015 vect_get_peeling_costs_all_drs to compute the penalty for no
7016 peeling to match up costs.
7018 2017-07-21 Richard Biener <rguenther@suse.de>
7020 PR tree-optimization/81500
7021 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
7022 we didn't identify a reduction path.
7024 2017-07-21 Tom de Vries <tom@codesourcery.com>
7025 Cesar Philippidis <cesar@codesourcery.com>
7027 PR gcov-profile/81442
7028 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
7031 2017-07-21 Tom de Vries <tom@codesourcery.com>
7034 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
7036 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
7037 nvptx_override_options_after_change.
7039 2017-07-21 Ulrich Drepper <drepper@redhat.com>
7041 * dwarf2out.c (output_file_names): Avoid double testing for
7044 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
7046 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
7048 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7050 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
7052 (try_crossjump_to_edge): Do not punt on partitioned functions.
7054 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7056 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
7057 Put all BBs reachable only via paths crossing cold region to cold
7059 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
7061 2016-07-21 Richard Biener <rguenther@suse.de>
7063 PR tree-optimization/81303
7064 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
7065 into account prologue and epilogue iterations when raising
7066 min_profitable_iters to sth at least covering one vector iteration.
7068 2017-07-21 Tamar Christina <tamar.christina@arm.com>
7070 * config/arm/arm.c (arm_test_cpu_arch_dat):
7073 2017-07-20 Nathan Sidwell <nathan@acm.org>
7075 Remove TYPE_METHODS.
7076 * tree.h (TYPE_METHODS): Delete.
7077 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
7078 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
7079 (dbxout_type_methods): Scan TYPE_FIELDS.
7080 (dbxout_type): Don't check TYPE_METHODS here.
7081 * function.c (use_register_for_decl): Always ignore register for
7082 class types when not optimizing.
7083 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
7084 * tree.c (free_lang_data_in_type): Stitch out member functions and
7085 templates from TYPE_FIELDS.
7086 (build_distinct_type_copy, verify_type_variant,
7087 verify_type): Member fns are on TYPE_FIELDS.
7088 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
7089 * tree-pretty-print.c (dump_generic_node): Likewise.
7091 2017-07-20 Jakub Jelinek <jakub@redhat.com>
7094 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
7095 V2TImode and V4TImode.
7096 (ix86_expand_vector_extract): Likewise.
7097 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
7098 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
7099 (ssescalarmode): Handle V4TImode and V2TImode.
7100 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
7101 (*vec_extractv2ti, *vec_extractv4ti): New insns.
7102 (VEXTRACTI128_MODE): New mode iterator.
7103 (splitter for *vec_extractv?ti first element): New.
7104 (VEC_INIT_MODE): New mode iterator.
7105 (vec_init<mode>): Consolidate 3 expanders into one using
7106 VEC_INIT_MODE mode iterator.
7108 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7110 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
7111 non_spilled_static_chain_regno_p.
7113 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7115 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
7117 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
7119 * bb-reorder.c (connect_traces): Allow copying of blocks within
7122 2017-07-20 Richard Biener <rguenther@suse.de>
7124 * gimple.h (gimple_phi_result): Add gphi * overload.
7125 (gimple_phi_result_ptr): Likewise.
7126 (gimple_phi_arg): Likewise. Adjust index assert to only
7127 allow actual argument accesses rather than all slots available
7129 (gimple_phi_arg_def): Add gphi * overload.
7130 * tree-phinodes.c (make_phi_node): Initialize only actual
7132 (resize_phi_node): Clear memory not covered by old node,
7133 do not initialize excess argument slots.
7134 (reserve_phi_args_for_new_edge): Initialize new argument slot
7137 2017-07-20 Bin Cheng <bin.cheng@arm.com>
7139 PR tree-optimization/81388
7141 2016-07-21 Bin Cheng <bin.cheng@arm.com>
7143 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
7144 by removing computation of may_be_zero.
7146 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7147 Tom de Vries <tom@codesourcery.com>
7150 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
7151 when gimple level profile disagrees with what RTL expander did.
7153 2017-07-20 Richard Biener <rguenther@suse.de>
7155 PR tree-optimization/61171
7156 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
7157 (vect_analyze_stmt): Add slp instance parameter.
7158 (vectorizable_reduction): Likewise.
7159 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
7160 (vect_is_simple_reduction): Deal with chains not detected
7161 as SLP reduction chain, specifically not properly associated
7162 chains containing a mix of plus/minus.
7163 (get_reduction_op): Remove.
7164 (get_initial_defs_for_reduction): Simplify, pass in whether
7165 this is a reduction chain, pass in the SLP node for the PHIs.
7166 (vect_create_epilog_for_reduction): Get the SLP instance as
7168 (vectorizable_reduction): Get the SLP instance as arg.
7169 During analysis remember the SLP node with the PHIs in the
7170 instance. Simplify getting at the vectorized reduction PHIs.
7171 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
7172 through SLP instance.
7173 (vect_slp_analyze_operations): Likewise.
7174 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
7175 (vect_transform_stmt): Likewise.
7177 2017-07-20 Tom de Vries <tom@codesourcery.com>
7179 PR tree-optimization/81489
7180 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
7181 read of phi arg location to before loop that modifies phi.
7183 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7185 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
7188 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7191 * except.c (execute): Fix ordering issue.
7193 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7195 PR rtl-optimization/81423
7196 * combine.c (make_compound_operation_int): Don't try to optimize
7197 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
7199 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7201 PR rtl-optimization/81423
7202 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
7203 with a constant that is -1 in the truncated to mode.
7205 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7207 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
7208 (determine_unlikely_bbs): ... here.
7209 * predict.h (propagate_unlikely_bbs_forward): Declare.
7210 * cfgexpand.c (pass_expand::execute): Use it.
7211 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
7213 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
7214 propagate_unlikely_bbs_forward.
7216 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7219 * except.c (maybe_add_nop_after_section_switch): New function.
7222 2017-07-19 Tom de Vries <tom@codesourcery.com>
7224 * gimple.h (gimple_phi_set_arg): Make assert more strict.
7226 2017-07-19 Tom de Vries <tom@codesourcery.com>
7228 * gimple.h (gimple_phi_arg): Make assert more strict.
7230 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
7232 * config.gcc (powerpc*-*-*): Add mmintrin.h.
7233 * config/rs6000/mmintrin.h: New file.
7234 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
7236 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7238 PR tree-optimization/81346
7239 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
7241 2017-07-19 Tom de Vries <tom@codesourcery.com>
7243 * config/nvptx/nvptx.md (VECIM): Add V2DI.
7245 2017-07-19 Tom de Vries <tom@codesourcery.com>
7247 * config/nvptx/nvptx-modes.def: Add V2DImode.
7248 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
7249 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
7250 (nvptx_output_mov_insn): Handle lack of mov.b128.
7251 (nvptx_print_operand): Handle 'H' and 'L' codes.
7252 (nvptx_vector_mode_supported): Allow V2DImode.
7253 (nvptx_preferred_simd_mode): New function.
7254 (nvptx_data_alignment): New function.
7255 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
7256 nvptx_preferred_simd_mode.
7257 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
7259 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
7261 2017-07-19 Tom de Vries <tom@codesourcery.com>
7263 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
7264 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
7265 (nvptx_vector_mode_supported): New function. Allow V2SImode.
7266 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
7267 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
7268 (mov<VECIM>_insn): New define_insn.
7269 (define_expand "mov<VECIM>): New define_expand.
7271 2017-07-19 Tom de Vries <tom@codesourcery.com>
7273 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
7275 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7277 PR tree-optimization/81346
7278 * fold-const.h (fold_div_compare, range_check_type): Declare.
7279 * fold-const.c (range_check_type): New function.
7280 (build_range_check): Use range_check_type.
7281 (fold_div_compare): No longer static, rewritten into
7282 a match.pd helper function.
7283 (fold_comparison): Don't call fold_div_compare here.
7284 * match.pd (X / C1 op C2): New optimization using fold_div_compare
7287 2017-07-19 Nathan Sidwell <nathan@acm.org>
7289 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
7290 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
7291 * tree.c (find_decls_types_r, verify_type): Use
7292 TYPE_{MIN,MAX}_VALUE_RAW.
7293 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
7294 (hash_tree): Likewise.
7295 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7297 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7300 2017-07-18 Tom de Vries <tom@codesourcery.com>
7303 * omp-expand.c (expand_omp_for_static_chunk): Handle
7304 equal-argument loop exit phi.
7306 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
7309 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
7310 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
7311 operand 2 predicate.
7312 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
7313 operand 2 predicate.
7314 (ror,rol -> rorx splitters): Use const_int_operand as
7315 operand 2 predicate.
7317 2017-06-18 Richard Biener <rguenther@suse.de>
7319 PR tree-optimization/81410
7320 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
7321 the gap in the ! slp_perm SLP case after each group.
7323 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7326 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
7329 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7332 * predict.c (set_even_probabilities): Cleanup; do not affect
7333 probabilities that are already known.
7334 (combine_predictions_for_bb): Call even when count is set.
7336 2017-07-18 Nathan Sidwell <nathan@acm.org>
7338 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
7341 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7344 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
7345 optimization for loop niter analysis.
7347 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
7350 * config/avr/avr.c (avr_optimize_casesi): Don't use
7351 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
7353 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7355 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
7356 body_cost_vec from _vect_peel_extended_info.
7357 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
7358 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
7361 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7363 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
7365 2017-07-18 Richard Biener <rguenther@suse.de>
7367 PR tree-optimization/80620
7368 PR tree-optimization/81403
7369 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
7370 info when re-using a VN table entry.
7372 2017-07-18 Richard Biener <rguenther@suse.de>
7374 PR tree-optimization/81418
7375 * tree-vect-loop.c (vectorizable_reduction): Properly compute
7376 vectype_in. Verify that with lane-reducing reduction operations
7377 we have a single def-use cycle.
7379 2017-07-17 Carl Love <cel@us.ibm.com>
7381 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
7383 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7384 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7385 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7386 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7387 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7388 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7389 VMULOSW): New enum "unspec" values.
7390 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
7391 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
7392 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7393 altivec_vmulosw): New patterns.
7394 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7395 VMULOSW): Add definitions.
7396 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
7398 * config/alpha/alpha.c: Include predict.h.
7400 2017-07-17 Yury Gribov <tetra2005@gmail.com>
7402 * tree-vrp.c (compare_assert_loc): Fix comparison function
7403 to return predictable results.
7405 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7407 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
7410 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
7411 * doc/invoke.texi (mexpand-adddi): Update text.
7413 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7415 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
7416 that also clobbers the CC register. The old expand code is moved
7418 (*arc_clzsi2): ... here.
7419 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
7420 the CC register. The old expand code is moved to ...
7421 (arc_ctzsi2): ... here.
7423 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7425 * config/arc/arc.opt (mindexed-loads): Use initial value
7426 TARGET_INDEXED_LOADS_DEFAULT.
7427 (mauto-modify-reg): Use initial value
7428 TARGET_AUTO_MODIFY_REG_DEFAULT.
7429 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7430 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7431 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7432 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7434 2017-07-17 Martin Liska <mliska@suse.cz>
7437 * opts.c (finish_options): Do not allow -fgnu-tm
7438 w/ -fsanitize={kernel-,}address. Say sorry.
7440 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7443 * tree-loop-distribution.c (classify_partition): Only assert on
7444 numer of iterations.
7445 (merge_dep_scc_partitions): Delete prameter. Update function call.
7446 (distribute_loop): Remove code handling loop with unknown niters.
7447 (pass_loop_distribution::execute): Skip loop with unknown niters.
7449 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7452 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
7453 function sort_partitions_by_post_order.
7455 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7457 PR tree-optimization/81374
7458 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
7459 the max index of basic blocks, rather than number of basic blocks.
7461 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7463 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
7465 (arc_legitimate_pic_operand_p): Likewise.
7466 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
7468 (arc_needs_pcl_p): Likewise.
7469 (arc_legitimate_pc_offset_p): Likewise.
7470 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
7471 function is also used in constrains.md.
7472 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
7473 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
7474 PLUS. Only return true/false in known cases, otherwise assert.
7475 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
7476 is already called in arc_legitimate_constant_p.
7477 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
7479 (LEGITIMATE_PIC_OPERAND_P): Use
7480 arc_raw_symbolic_reference_mentioned_p function.
7481 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
7486 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7487 Andrew Burgess <andrew.burgess@embecosm.com>
7489 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
7490 (arc_return_address_register): New function.
7491 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
7492 (arc_handle_fndecl_attribute): Add naked attribute.
7493 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7494 (TARGET_WARN_FUNC_RETURN): Likewise.
7495 (arc_allocate_stack_slots_for_args): New function.
7496 (arc_warn_func_return): Likewise.
7497 (machine_function): Change type fn_type.
7498 (arc_compute_function_type): Consider new naked function type,
7499 change function return type.
7500 (arc_must_save_register): Adapt to handle new
7501 arc_compute_function_type's return type.
7502 (arc_expand_prologue): Likewise.
7503 (arc_expand_epilogue): Likewise.
7504 (arc_return_address_regs): Delete.
7505 (arc_return_address_register): New function.
7506 (arc_epilogue_uses): Use above function.
7507 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
7508 (arc_function_type): Change encoding, add naked type.
7509 (ARC_INTERRUPT_P): Change to handle the new encoding.
7510 (ARC_FAST_INTERRUPT_P): Likewise.
7511 (ARC_NORMAL_P): Define.
7512 (ARC_NAKED_P): Likewise.
7513 (arc_compute_function_type): Delete prototype.
7514 * config/arc/arc.md (in_ret_delay_slot): Use
7515 arc_return_address_register function.
7516 (simple_return): Likewise.
7517 (p_return_i): Likewise.
7519 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7521 PR tree-optimization/81428
7522 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
7523 can't be built for those types.
7525 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7527 Remove stuff dead since r239246.
7529 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
7530 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
7531 (avr_inform_devices): Remove dead stuff.
7533 2017-07-17 Tamar Christina <tamar.christina@arm.com>
7535 * config/arm/arm_neon.h: Fix softp typo.
7537 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7539 PR tree-optimization/81365
7540 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
7541 aggregate moves onto bb predecessor edges, make sure there are no
7542 loads that could alias the lhs in between the start of bb and the
7545 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7548 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
7549 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
7550 [LSHIFTRT, outer_code = TRUNCATE]: Same.
7552 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7554 PR tree-optimization/81396
7555 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
7556 (init_symbolic_number): Initialize it to 1.
7557 (perform_symbolic_merge): Add n_ops from both operands into the new
7559 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
7560 without base_addr as useless if they need more than one operation.
7561 (bswap_replace): Handle !bswap case for NULL base_addr.
7563 2017-07-17 Tom de Vries <tom@codesourcery.com>
7566 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
7569 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
7571 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
7572 conditional builtin define __FIX_LEON3FT_B2BST.
7574 2017-07-17 Daniel Cederman <cederman@gaisler.com>
7576 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
7577 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
7580 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7582 PR rtl-optimization/81424
7583 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
7584 to remove potential trapping from operands if -fnon-call-exceptions.
7586 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7588 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
7589 profile_proability for scalling.
7590 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
7592 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7594 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
7596 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7598 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
7599 fixpoint arithmetics.
7601 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7603 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
7604 fixpoint arithmetics.
7606 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7608 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
7609 fixpoint arithmetics.
7611 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7613 * profile-count.h (profile_probability::from_reg_br_prob_note,
7614 profile_probability::to_reg_br_prob_note): New functions.
7615 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
7616 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
7617 * predict.c (probability_reliable_p): Update.
7618 (edge_probability_reliable_p): Update.
7619 (br_prob_note_reliable_p): Update.
7620 (invert_br_probabilities): Update.
7621 (add_reg_br_prob_note): New function.
7622 (combine_predictions_for_insn): Update.
7623 * asan.c (asan_clear_shadow): Update.
7624 * cfgbuild.c (compute_outgoing_frequencies): Update.
7625 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
7626 (update_br_prob_note): Update.
7627 (rtl_verify_edges): Update.
7628 (purge_dead_edges): Update.
7629 (fixup_reorder_chain): Update.
7630 * emit-rtl.c (try_split): Update.
7631 * ifcvt.c (cond_exec_process_insns): Update.
7632 (cond_exec_process_if_block): Update.
7633 (dead_or_predicable): Update.
7634 * internal-fn.c (expand_addsub_overflow): Update.
7635 (expand_neg_overflow): Update.
7636 (expand_mul_overflow): Update.
7637 * loop-doloop.c (doloop_modify): Update.
7638 * loop-unroll.c (compare_and_jump_seq): Update.
7639 * optabs.c (emit_cmp_and_jump_insn_1): Update.
7640 * predict.h: Update.
7641 * reorg.c (mostly_true_jump): Update.
7643 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
7644 * config/alpha/alpha.c (emit_unlikely_jump): Update.
7645 * config/arc/arc.c: (emit_unlikely_jump): Update.
7646 * config/arm/arm.c: (emit_unlikely_jump): Update.
7647 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
7648 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
7649 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
7650 (ix86_print_operand): Update.
7651 (ix86_split_fp_branch): Update.
7652 (predict_jump): Update.
7653 * config/ia64/ia64.c (ia64_print_operand): Update.
7654 * config/mmix/mmix.c (mmix_print_operand): Update.
7655 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
7656 (rs6000_expand_split_stack_prologue): Update.
7657 * config/rs6000/rs6000.c: Update.
7658 * config/s390/s390.c (s390_expand_vec_strlen): Update.
7659 (s390_expand_vec_movstr): Update.
7660 (s390_expand_cs_tdsi): Update.
7661 (s390_expand_split_stack_prologue): Update.
7662 * config/sh/sh.c (sh_print_operand): Update.
7663 (expand_cbranchsi4): Update.
7664 (expand_cbranchdi4): Update.
7665 * config/sparc/sparc.c (output_v9branch): Update.
7666 * config/spu/spu.c (get_branch_target): Update.
7667 (ea_load_store_inline): Update.
7668 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
7669 * config/tilepro/tilepro.c: Update.
7671 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7673 * gimplify.c (mostly_copy_tree_r): Revert latest change.
7674 (gimplify_save_expr): Likewise.
7676 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7678 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
7680 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7682 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
7684 * timevar.def (TV_IPA_FNSUMMARY): Define.
7686 2017-07-16 Daniel Cederman <cederman@gaisler.com>
7688 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
7689 to back store errata sensitive sequence from being generated.
7690 (sqrtdf2_fix): Likewise.
7692 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7694 * tree-ssa-threadupdate.c (compute_path_counts,
7695 update_joiner_offpath_counts): Use profile_probability.
7697 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
7700 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7702 * config/arm/arm-c.c (arm_cpu_builtins): Define
7703 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7705 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
7707 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7708 array entries to represent __ieee128 versions of the
7709 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
7710 scalar_extract_sig, and scalar_insert_exp built-in functions.
7711 (altivec_resolve_overloaded_builtin): Add special case handling
7712 for the __builtin_scalar_insert_exp function, as represented by
7713 the P9V_BUILTIN_VEC_VSIEDP constant.
7714 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
7715 exponent support for __ieee128 argument.
7716 (VSESQP): Add scalar extract signature support for __ieee128
7718 (VSTDCNQP): Add scalar test negative support for __ieee128
7720 (VSIEQP): Add scalar insert exponent support for __int128 argument
7721 with __ieee128 result.
7722 (VSIEQPF): Add scalar insert exponent support for __ieee128
7723 argument with __ieee128 result.
7724 (VSTDCQP): Add scalar test data class support for __ieee128
7726 (VSTDCNQP): Add overload support for scalar test negative with
7728 (VSTDCQP): Add overload support for scalar test data class
7730 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
7732 (UNSPEC_VSX_SIEXPQP): New constant.
7733 (xsxexpqp): New insn for VSX scalar extract exponent quad
7735 (xsxsigqp): New insn for VSX scalar extract significand quad
7737 (xsiexpqpf): New insn for VSX scalar insert exponent quad
7738 precision with floating point argument.
7739 (xststdcqp): New expand for VSX scalar test data class quad
7741 (xststdcnegqp): New expand for VSX scalar test negative quad
7743 (xststdcqp): New insn to match expansions for VSX scalar test data
7744 class quad precision and VSX scalar test negative quad precision.
7745 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7746 special case operand checking to enforce that second operand of
7747 VSX scalar test data class with quad precision argument is a 7-bit
7749 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
7750 prototypes and descriptions of __ieee128 versions of
7751 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
7752 scalar_test_data_class, and scalar_test_neg built-in functions.
7754 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7756 PR tree-optimization/81162
7757 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
7758 replace a negate with an add.
7760 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
7762 * doc/invoke.texi (arm/-mcpu): Document +crypto.
7764 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7766 * config/arm/arm-c.c (arm_cpu_builtins): Define
7767 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7769 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7771 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
7772 (armv8-r): Set ARM Cortex-R52 as default CPU.
7773 * config/arm/arm-tables.opt: Regenerate.
7774 * config/arm/arm-tune.md: Regenerate.
7775 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7777 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
7778 extension for -mcpu=cortex-r52.
7780 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7782 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
7783 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
7784 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
7785 (fp-armv8): Define it as FP_ARMv8 only.
7786 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
7787 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
7789 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
7790 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
7791 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
7792 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
7793 than TARGET_FPU_ARMV8.
7794 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
7795 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
7796 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
7798 * config/arm/neon.md (neon_vrint): Likewise.
7799 (neon_vcvt): Likewise.
7800 (neon_<fmaxmin_op><mode>): Likewise.
7801 (<fmaxmin><mode>3): Likewise.
7802 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
7803 * config/arm/predicates.md (arm_cond_move_operator): Check against
7804 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
7806 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
7808 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
7811 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7813 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
7814 loop in comment with memset.
7816 2017-07-14 Martin Liska <mliska@suse.cz>
7818 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
7819 * dwarf2out.c (is_java): Remove the function.
7820 (output_pubname): Remove usage of the function.
7821 (lower_bound_default): Remove usage of DW_LANG_Java.
7822 (gen_compile_unit_die): Likewise.
7823 * gcc.c: Remove compiler defaults for .java and .zip files.
7824 * gimple-expr.c (remove_suffix): Change as there's no longer
7825 extension than 4-letter one.
7826 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
7827 (gimplify_save_expr): Likewise.
7828 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
7829 as it's possible even for other languages than Java.
7830 * langhooks.h (struct lang_hooks): Remove Java from a comment.
7831 * lto-opts.c (lto_write_options): Remove reference to Java.
7832 * opts.c (strip_off_ending): Update file extension handling.
7833 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
7834 * tree-eh.c (lower_resx): Likewise.
7835 * tree.c (free_lang_data_in_type): Remove dead code.
7836 (find_decls_types_r): Likewise.
7837 (build_common_builtin_nodes): Remove Java from a comment.
7838 (verify_type): Remove dead code.
7839 * varasm.c (assemble_external): Remove Java from a comment.
7841 2017-07-14 Martin Liska <mliska@suse.cz>
7843 * opts.c (finish_options): Add quotes.
7844 (common_handle_option): Likewise.
7846 2017-07-14 Martin Liska <mliska@suse.cz>
7848 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
7849 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
7851 * dwarf2out.c (lower_bound_default): Do not handle
7853 (gen_compile_unit_die): Likewise.
7854 * gcc.c: Remove default extension binding for GNU Pascal.
7855 * stmt.c: Remove Pascal language from a comment.
7856 * xcoffout.c: Likewise.
7858 2017-07-13 David Malcolm <dmalcolm@redhat.com>
7861 * diagnostic-show-locus.c (fixit_cmp): New function.
7862 (layout::layout): Sort m_fixit_hints.
7863 (column_range::column_range): Assert that the values are valid.
7864 (struct char_span): New struct.
7865 (correction::overwrite): New method.
7866 (struct source_line): New struct.
7867 (line_corrections::add_hint): Add assertions. Reimplement memcpy
7868 calls in terms of classes source_line and char_span, and
7869 correction::overwrite.
7870 (selftest::test_overlapped_fixit_printing_2): New function.
7871 (selftest::diagnostic_show_locus_c_tests): Call it.
7873 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7875 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
7876 early if there is no lhs.
7878 2017-07-13 Martin Liska <mliska@suse.cz>
7880 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
7881 (gen_reference_type_die): Likewise.
7882 * stor-layout.c: Remove Pascal-related comment.
7884 2017-07-13 Martin Liska <mliska@suse.cz>
7886 * opts.c (finish_options): Add quotes to error messages.
7887 (parse_sanitizer_options): Likewise.
7889 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
7891 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
7893 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
7895 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
7897 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
7899 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
7901 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
7903 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7906 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
7907 provides the hardware capability bits, define the macro
7908 __BUILTIN_CPU_SUPPORTS__.
7909 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
7910 if GLIBC does not provide the hardware capability bits. Add a
7911 gcc_unreachable call if the built-in cpu function is neither
7912 __builtin_cpu_is nor __builtin_cpu_supports.
7913 (rs6000_get_function_versions_dispatcher): Change the warning
7914 that an old GLIBC is used which does not export the capability
7915 bits to be an error.
7916 * doc/extend.texi (target_clones attribute): Document the
7917 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
7918 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
7919 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
7920 the macros defined by GCC if the newer GLIBC is available.
7922 2017-07-12 Jeff Law <law@redhat.com>
7924 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
7925 remaining includes slightly.
7926 * config/riscv/riscv-builtins.c: Include profile-count.h.
7928 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
7931 * config/avr/avr.c (avr_set_current_function): In diagnostic
7932 messages: Quote keywords and (parts of) identifiers.
7933 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
7936 2017-07-12 Carl Love <cel@us.ibm.com>
7938 * config/rs6000/rs6000-c.c: Add support for built-in functions
7939 vector bool char vec_revb (vector bool char);
7940 vector bool short vec_revb (vector short char);
7941 vector bool int vec_revb (vector bool int);
7942 vector bool long long vec_revb (vector bool long long);
7943 * doc/extend.texi: Update the built-in documentation file for the
7944 new built-in functions.
7946 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7948 * config/s390/s390.md: Remove movcc splitter.
7950 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7952 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
7953 load/store on condition.
7955 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
7958 * config/avr/avr.c (avr_encode_section_info)
7959 [progmem && !TREE_READONLY]: Error if progmem object needs
7962 2017-07-11 Michael Collison <michael.collison@arm.com>
7964 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
7967 2017-07-11 Carl Love <cel@us.ibm.com>
7969 * config/rs6000/rs6000-c.c: Add support for builtins
7970 vector unsigned int vec_parity_lsbb (vector signed int);
7971 vector unsigned int vec_parity_lsbb (vector unsigned int);
7972 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
7973 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
7974 vector unsigned long long vec_parity_lsbb (vector signed long long);
7975 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
7976 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
7977 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
7978 * doc/extend.texi: Update the built-in documentation file for the
7979 new built-in functions.
7981 2017-07-11 David Malcolm <dmalcolm@redhat.com>
7983 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
7984 (layout::m_primary_loc): New field.
7985 (layout::layout): Initialize new field. Move location filtering
7986 logic from here to...
7987 (layout::maybe_add_location_range): ...this new method. Add
7988 support for filtering to just the lines already specified by other
7990 (layout::will_show_line_p): New method.
7991 (gcc_rich_location::add_location_if_nearby): New method.
7992 (selftest::test_add_location_if_nearby): New test function.
7993 (selftest::diagnostic_show_locus_c_tests): Call it.
7994 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
7997 2017-07-11 Tom de Vries <tom@codesourcery.com>
7999 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
8000 (bb_first_real_insn): New function.
8001 (nvptx_single): Add extra initialization of broadcasted condition
8004 2017-07-11 Nathan Sidwell <nathan@acm.org>
8006 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
8008 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
8010 * doc/extend.texi (AVR Function Attributes): Remove weblink to
8011 Binutils doc as TEXI will mess them up.
8012 * doc/invoke.texi (AVR Options): Same here.
8014 2017-07-11 Daniel Cederman <cederman@gaisler.com>
8016 * config/sparc/sparc.opt (mfix-ut700): New option.
8017 (mfix-gr712rc): Likewise.
8018 (sparc_fix_b2bst): New variable.
8019 * doc/invoke.texi (SPARC options): Document them.
8020 (ARM options): Fix warnings.
8021 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
8022 instructions to prevent sequences that can trigger the store-store
8023 errata for certain LEON3FT processors.
8024 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
8025 (sparc_option_override): Set sparc_fix_b2bst appropriately.
8026 * config/sparc/sparc.md (fix_b2bst): New attribute.
8027 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
8029 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
8032 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
8034 (*rsqrtsf2_sse): Ditto.
8036 (div<mode>3): Macroize insn from divdf3 and divsf3
8037 using MODEF mode iterator.
8039 2017-07-10 Martin Sebor <msebor@redhat.com>
8041 PR tree-optimization/80397
8042 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
8043 instead of testing for equality to INTEGER_TYPE.
8045 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
8047 * config.gcc: Remove uclibc from arc target spec.
8049 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
8051 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
8053 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8056 * lto-wrapper.c (remove_option): New function.
8057 (merge_and_complain): Merge PIC/PIE options more realistically.
8059 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8061 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
8065 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
8066 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
8067 * config.in: Regenerate.
8068 * configure: Regenerate.
8069 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
8070 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
8071 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
8072 (TARGET_GASISR_PROLOGUES): ...target mask.
8073 * common/config/avr/avr-common.c
8074 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
8075 Set -mgas-isr-prologues.
8076 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
8077 INSERT_PASS_BEFORE for it.
8078 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
8079 * config/avr/avr.c (avr_option_override)
8080 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
8081 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
8082 (avr_attribute_table) <no_gccisr>: Add new function attribute.
8083 (avr_set_current_function) <is_no_gccisr>: Init machine field.
8084 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
8086 (make_avr_pass_pre_proep): New function.
8087 (emit_push_sfr) <treg>: Add argument to function and use it
8089 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
8090 and set machine->gasisr.yes.
8091 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
8092 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
8093 __gcc_isr.n_pushed to .L__stack_usage.
8094 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
8095 (avr_asm_final_postscan_insn): ...this new static function.
8096 * config/avr/avr.h (machine_function)
8097 <is_no_gccisr, use_L__stack_usage>: New fields.
8098 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
8099 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
8100 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
8101 (gasisr, *gasisr): New expander and insn.
8102 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
8103 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
8104 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
8106 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
8108 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
8111 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8113 Move jump-tables out of .text again.
8116 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
8117 (ASM_OUTPUT_ADDR_VEC): New function.
8118 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
8119 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
8120 INSN_ADDRESSes as asm comment.
8121 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
8122 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
8123 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
8124 * config/avr/avr.md (*tablejump): Adjust comment.
8125 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
8126 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
8128 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
8129 (avr_output_addr_vec): New proto.
8130 (avr_log_t) <insn_addresses>: New field.
8132 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
8135 * config/i386/i386.c (ix86_function_arg_advance): Set
8136 outgoing_args_on_stack to true if there are outgoing arguments
8138 (ix86_function_arg): Likewise.
8139 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
8140 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
8141 * config/i386/i386.h (machine_function): Add
8142 outgoing_args_on_stack.
8144 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
8146 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8148 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
8150 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
8152 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
8153 (REAL_H): Remove $(MACHMODE_H).
8154 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
8156 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
8157 $(MACHMODE_H) and double-int.h.
8158 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
8160 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
8163 2017-07-07 Andrew Pinski <apinski@cavium.com>
8165 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
8166 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
8168 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8170 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
8171 Add warning if GCC was not configured to link against a GLIBC that
8172 exports the hardware capability bits.
8173 (make_resolver_func): Make resolver function private and not a
8174 COMDAT function. Create the name with clone_function_name instead
8175 of make_unique_name.
8178 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
8179 correct operand in doing the split.
8181 2017-07-07 Carl Love <cel@us.ibm.com>
8183 * config/rs6000/rs6000-c: Add support for built-in function
8184 vector unsigned short vec_pack_to_short_fp32 (vector float,
8186 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
8187 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
8188 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
8189 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
8190 (convert_4f32_8i16): Add define_expand.
8191 * doc/extend.texi: Update the built-in documentation file for the
8192 new built-in function.
8194 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8196 * config/sparc/m8.md: New file.
8197 * config/sparc/sparc.md: Include m8.md.
8199 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8201 * config/sparc/sparc.opt: New option -mvis4b.
8202 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
8203 (sparc_option_override): Handle VIS4B.
8204 (enum sparc_builtins): Define
8205 SPARC_BUILTIN_DICTUNPACK{8,16,32},
8206 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
8207 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
8208 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
8209 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
8210 (check_constant_argument): New function.
8211 (sparc_vis_init_builtins): Define builtins
8212 __builtin_vis_dictunpack{8,16,32},
8213 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
8214 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
8215 __builtin_vis_fpcmpde{8,16,32}shl and
8216 __builtin_vis_fpcmpur{8,16,32}shl.
8217 (sparc_expand_builtin): Check that the constant operands to
8218 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
8219 constant and in range.
8220 * config/sparc/sparc-c.c (sparc_target_macros): Handle
8222 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
8223 (SPARC_IMM5_P): Likewise.
8224 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
8225 (enabled): Handle vis4b.
8226 (UNSPEC_DICTUNPACK): New unspec.
8227 (UNSPEC_FPCMPSHL): Likewise.
8228 (UNSPEC_FPUCMPSHL): Likewise.
8229 (UNSPEC_FPCMPDESHL): Likewise.
8230 (UNSPEC_FPCMPURSHL): Likewise.
8231 (cpu_feature): New CPU feature `vis4b'.
8232 (dictunpack{8,16,32}): New insns.
8233 (FPCSMODE): New mode iterator.
8234 (fpcscond): New code iterator.
8235 (fpcsucond): Likewise.
8236 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
8237 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
8238 (fpcmpde{8,16,32}{si,di}shl): Likewise.
8239 (fpcmpur{8,16,32}{si,di}shl): Likewise.
8240 * config/sparc/constraints.md: Define constraints `q' for unsigned
8241 2-bit integer constants and `t' for unsigned 5-bit integer
8243 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
8245 (imm5_operand_dictunpack16): Likewise.
8246 (imm5_operand_dictunpack32): Likewise.
8247 (imm2_operand): Likewise.
8248 * doc/invoke.texi (SPARC Options): Document -mvis4b.
8249 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
8250 ditunpack* and fpcmp*shl builtins.
8252 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8254 * config.gcc: Handle m8 in --with-{cpu,tune} options.
8255 * config.in: Add HAVE_AS_SPARC6 define.
8256 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
8258 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
8260 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
8261 (CPP_CPU_SPEC): Handle m8.
8262 (ASM_CPU_SPEC): Likewise.
8263 * config/sparc/sparc-opts.h (enum processor_type): Add
8265 * config/sparc/sparc.c (m8_costs): New struct.
8266 (sparc_option_override): Handle TARGET_CPU_m8.
8267 (sparc32_initialize_trampoline): Likewise.
8268 (sparc64_initialize_trampoline): Likewise.
8269 (sparc_issue_rate): Likewise.
8270 (sparc_register_move_cost): Likewise.
8271 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
8272 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
8273 (ASM_CPU64_DEFAULT_SPEC): Likewise.
8274 (CPP_CPU_SPEC): Handle M8.
8275 (ASM_CPU_SPEC): Likewise.
8276 (AS_M8_FLAG): Define.
8277 * config/sparc/sparc.md: Add m8 to the cpu attribute.
8278 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
8279 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
8281 * configure: Regenerate.
8282 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
8285 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8287 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
8289 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
8290 ("*movdi_insn_sp32"): Do not set v3pipe.
8291 ("*movsi_insn"): Likewise.
8292 ("*movdi_insn_sp64"): Likewise.
8293 ("*movsf_insn"): Likewise.
8294 ("*movdf_insn_sp32"): Likewise.
8295 ("*movdf_insn_sp64"): Likewise.
8296 ("*zero_extendsidi2_insn_sp64"): Likewise.
8297 ("*sign_extendsidi2_insn"): Likewise.
8298 ("*mov<VM32:mode>_insn"): Likewise.
8299 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8300 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8301 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8302 ("<vlop:code><VL:mode>3"): Likewise.
8303 ("*not_<vlop:code><VL:mode>3"): Likewise.
8304 ("*nand<VL:mode>_vis"): Likewise.
8305 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
8306 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
8307 ("one_cmpl<VL:mode>2"): Likewise.
8308 ("faligndata<VM64:mode>_vis"): Likewise.
8309 ("alignaddrsi_vis"): Likewise.
8310 ("alignaddrdi_vis"): Likweise.
8311 ("alignaddrlsi_vis"): Likewise.
8312 ("alignaddrldi_vis"): Likewise.
8313 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8314 ("bmaskdi_vis"): Likewise.
8315 ("bmasksi_vis"): Likewise.
8316 ("bshuffle<VM64:mode>_vis"): Likewise.
8317 ("cmask8<P:mode>_vis"): Likewise.
8318 ("cmask16<P:mode>_vis"): Likewise.
8319 ("cmask32<P:mode>_vis"): Likewise.
8320 ("pdistn<P:mode>_vis"): Likewise.
8321 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8323 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8325 * config/sparc/sparc.md ("subtype"): New insn attribute.
8326 ("*wrgsr_sp64"): Set insn subtype.
8327 ("*rdgsr_sp64"): Likewise.
8328 ("alignaddrsi_vis"): Likewise.
8329 ("alignaddrdi_vis"): Likewise.
8330 ("alignaddrlsi_vis"): Likewise.
8331 ("alignaddrldi_vis"): Likewise.
8332 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8333 ("fexpand_vis"): Likewise.
8334 ("fpmerge_vis"): Likewise.
8335 ("faligndata<VM64:mode>_vis"): Likewise.
8336 ("bshuffle<VM64:mode>_vis"): Likewise.
8337 ("cmask8<P:mode>_vis"): Likewise.
8338 ("cmask16<P:mode>_vis"): Likewise.
8339 ("cmask32<P:mode>_vis"): Likewise.
8340 ("fchksm16_vis"): Likewise.
8341 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
8342 ("fmean16_vis"): Likewise.
8343 ("fp<plusminus_insn>64_vis"): Likewise.
8344 ("<plusminus_insn>v8qi3"): Likewise.
8345 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8346 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
8347 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
8348 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
8349 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
8350 ("*movqi_insn"): Likewise.
8351 ("*movhi_insn"): Likewise.
8352 ("*movsi_insn"): Likewise.
8353 ("movsi_pic_gotdata_op"): Likewise.
8354 ("*movdi_insn_sp32"): Likewise.
8355 ("*movdi_insn_sp64"): Likewise.
8356 ("movdi_pic_gotdata_op"): Likewise.
8357 ("*movsf_insn"): Likewise.
8358 ("*movdf_insn_sp32"): Likewise.
8359 ("*movdf_insn_sp64"): Likewise.
8360 ("*zero_extendhisi2_insn"): Likewise.
8361 ("*zero_extendqihi2_insn"): Likewise.
8362 ("*zero_extendqisi2_insn"): Likewise.
8363 ("*zero_extendqidi2_insn"): Likewise.
8364 ("*zero_extendhidi2_insn"): Likewise.
8365 ("*zero_extendsidi2_insn_sp64"): Likewise.
8366 ("ldfsr"): Likewise.
8367 ("prefetch_64"): Likewise.
8368 ("prefetch_32"): Likewise.
8369 ("tie_ld32"): Likewise.
8370 ("tie_ld64"): Likewise.
8371 ("*tldo_ldub_sp32"): Likewise.
8372 ("*tldo_ldub1_sp32"): Likewise.
8373 ("*tldo_ldub2_sp32"): Likewise.
8374 ("*tldo_ldub_sp64"): Likewise.
8375 ("*tldo_ldub1_sp64"): Likewise.
8376 ("*tldo_ldub2_sp64"): Likewise.
8377 ("*tldo_ldub3_sp64"): Likewise.
8378 ("*tldo_lduh_sp32"): Likewise.
8379 ("*tldo_lduh1_sp32"): Likewise.
8380 ("*tldo_lduh_sp64"): Likewise.
8381 ("*tldo_lduh1_sp64"): Likewise.
8382 ("*tldo_lduh2_sp64"): Likewise.
8383 ("*tldo_lduw_sp32"): Likewise.
8384 ("*tldo_lduw_sp64"): Likewise.
8385 ("*tldo_lduw1_sp64"): Likewise.
8386 ("*tldo_ldx_sp64"): Likewise.
8387 ("*mov<VM32:mode>_insn"): Likewise.
8388 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8389 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8391 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8393 * config/sparc/sparc.md ("type"): New insn type viscmp.
8394 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
8396 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
8397 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
8398 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8399 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
8401 ("n7_vis_logical_11cycle"): Likewise.
8402 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
8403 * config/sparc/niagara2.md ("niag3_vis": Likewise.
8404 * config/sparc/niagara.md ("niag_vis"): Likewise.
8405 * config/sparc/ultra3.md ("us3_fga"): Likewise.
8406 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
8408 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8410 * config/sparc/sparc.md: New instruction type `bmask'.
8411 (bmaskdi_vis): Use the `bmask' type.
8412 (bmasksi_vis): Likewise.
8413 * config/sparc/ultra3.md (us3_array): Likewise.
8414 * config/sparc/niagara7.md (n7_array): Likewise.
8415 * config/sparc/niagara4.md (n4_array): Likewise.
8416 * config/sparc/niagara2.md (niag2_vis): Likewise.
8417 (niag3_vis): Likewise.
8418 * config/sparc/niagara.md (niag_vis): Likewise.
8420 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8422 * ipa-comdats.c: Remove optimize check from gate.
8423 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
8424 for functions not optimized.
8425 (ipa_fn_summary_read): Skip optimize check.
8426 (ipa_fn_summary_write): Likewise.
8427 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
8429 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
8431 (can_inline_edge_p): Check flag_pcc_struct_return for match.
8432 (check_callers): Give up on caller which is not optimized.
8433 (inline_small_functions): Likewise.
8434 (ipa_inline): Do not give up when not optimizing.
8435 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
8436 away unoptimizes cdtors.
8437 (whole_program_function_and_variable_visibility): Do
8438 ipa_discover_readonly_nonaddressable_vars in LTO mode.
8439 * ipa.c (process_references): Do not check optimize.
8440 (symbol_table::remove_unreachable_nodes): Update optimize check.
8441 (set_writeonly_bit): Update optimize check.
8442 (pass_ipa_cdtor_merge::gate): Do not check optimize.
8443 (pass_ipa_single_use::gate): Remove.
8445 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8447 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
8448 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
8449 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
8450 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
8451 permute_load, permute_store, adjust_extract, adjust_splat,
8452 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
8453 replace_swap_with_copy, dump_swap_insn_table,
8454 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
8455 recombine_lvx_pattern, recombine_stvx_pattern,
8456 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
8457 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
8458 to file rs6000-p8swap.c.
8459 * config/rs6000/rs6000-p8swap.c: New file.
8460 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
8461 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
8462 and rs6000*-*-* targets.
8464 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8466 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
8468 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8470 * lto-wrapper.c (merge_and_complain): Do not merge
8471 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
8472 fsigned_zeros, ftrapping_math, fwrapv.
8473 (append_compiler_options): Do not track these options.
8474 (append_linker_options): Likewie
8476 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8478 * cgraphunit.c (cgraph_node::finalize_function): When
8479 !flag_toplevel_reorde set no_reorder flag.
8480 (varpool_node::finalize_decl): Likewise.
8481 (symbol_table::compile): Drop no toplevel reorder path.
8483 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8485 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
8486 edges; zero probability is not better than uninitialized.
8488 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8490 * asan.h (asan_sanitize_allocas_p): Declare.
8491 * asan.c (asan_sanitize_allocas_p): New function.
8492 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
8493 (handle_builtin_alloca): Likewise.
8494 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
8495 if !asan_sanitize_allocas_p.
8496 * params.def (asan-instrument-allocas): Add new option.
8497 * params.h (ASAN_PROTECT_ALLOCAS): Define.
8498 * opts.c (common_handle_option): Disable allocas sanitization for
8501 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8503 * asan.c: Include gimple-fold.h.
8504 (get_last_alloca_addr): New function.
8505 (handle_builtin_stackrestore): Likewise.
8506 (handle_builtin_alloca): Likewise.
8507 (asan_emit_allocas_unpoison): Likewise.
8508 (get_mem_refs_of_builtin_call): Add new parameter, remove const
8509 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
8510 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
8511 (instrument_builtin_call): Pass gimple iterator to
8512 get_mem_refs_of_builtin_call.
8513 (last_alloca_addr): New global.
8514 * asan.h (asan_emit_allocas_unpoison): Declare.
8515 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
8516 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
8517 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
8518 if function calls alloca.
8519 * gimple-fold.c (replace_call_with_value): Remove static keyword.
8520 * gimple-fold.h (replace_call_with_value): Declare.
8521 * internal-fn.c: Include asan.h.
8522 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
8523 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
8525 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8527 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
8528 (C_SELFTEST_FLAGS): New.
8529 (CPP_SELFTEST_FLAGS): New.
8530 (SELFTEST_DEPS): New, from deps of s-selftest.
8531 (C_SELFTEST_DEPS): New, from deps of s-selftest.
8532 (CPP_SELFTEST_DEPS): New.
8533 (selftest): Add dependency on s-selftest-c++.
8534 (s-selftest): Rename to...
8535 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
8536 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
8537 than SELFTEST_FLAGS.
8538 (selftest-gdb): Rename to...
8539 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
8541 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
8542 (selftest-valgrind): Rename to...
8543 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
8545 (selftest-valgrind): Reintroduce as an alias for
8546 selftest-c-valgrind.
8547 (s-selftest-c++): New.
8548 (selftest-c++-gdb): New.
8549 (selftest-c++-valgrind): New.
8551 2017-07-06 Olivier Hainque <hainque@adacore.com>
8553 * gcc.c (process_command): When deciding if undefined variables
8554 should be ignored when processing specs, accept "gcc -v" as well.
8556 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8558 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
8559 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
8561 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
8563 * config/arm/arm-cpus.in (armv8-r): Add new entry.
8564 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
8565 * config/arm/arm-tables.opt: Regenerate.
8566 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
8568 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
8570 2017-07-06 Carl Love <cel@us.ibm.com>
8572 * ChangeLog: Clean up from mid air collision
8574 2017-07-06 Carl Love <cel@us.ibm.com>
8576 * config/rs6000/rs6000-c.c: Add support for built-in functions
8577 vector signed int vec_subc (vector signed int, vector signed int);
8578 vector signed __int128 vec_subc (vector signed __int128,
8579 vector signed __int128);
8580 vector unsigned __int128 vec_subc (vector unsigned __int128,
8581 vector unsigned __int128);
8582 vector signed int vec_sube (vector signed int, vector signed int,
8584 vector unsigned int vec_sube (vector unsigned int,
8585 vector unsigned int,
8586 vector unsigned int);
8587 vector signed __int128 vec_sube (vector signed __int128,
8588 vector signed __int128,
8589 vector signed__int128);
8590 vector unsigned __int128 vec_sube (vector unsigned __int128,
8591 vector unsigned __int128,
8592 vector unsigned __int128);
8593 vector signed int vec_subec (vector signed int, vector signed int,
8595 vector unsigned int vec_subec (vector unsigned int,
8596 vector unsigned int,
8597 vector unsigned int);
8598 vector signed __int128 vec_subec (vector signed __int128,
8599 vector signed __int128,
8600 vector signed__int128);
8601 vector unsigned __int128 vec_subec (vector unsigned __int128,
8602 vector unsigned __int128,
8603 vector unsigned __int128);
8604 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
8605 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
8606 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
8607 BU_ALTIVEC_OVERLOAD_X definitions.
8608 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
8609 * doc/extend.texi: Update the built-in documentation file for the new
8612 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8615 * diagnostic-show-locus.c (layout::layout): Use start and finish
8616 spelling location for the start and finish of each range.
8617 * genmatch.c (linemap_client_expand_location_to_spelling_point):
8618 Add unused aspect param.
8619 * input.c (expand_location_1): Add "aspect" param, and use it
8620 to access the correct part of the location.
8621 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
8623 (expand_location_to_spelling_point): Likewise.
8624 (linemap_client_expand_location_to_spelling_point): Add "aspect"
8625 param, and pass it to expand_location_1.
8627 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
8629 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
8630 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
8631 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
8632 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
8633 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
8634 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
8635 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
8636 _mm_maskz_getmant_ss): New intrinsics.
8637 (__builtin_ia32_getexpss128_mask): Changed to ...
8638 __builtin_ia32_getexpss128_round ... this.
8639 (__builtin_ia32_getexpsd128_mask): Changed to ...
8640 __builtin_ia32_getexpsd128_round ... this.
8641 * config/i386/i386-builtin-types.def
8642 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
8643 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
8644 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
8645 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
8646 __builtin_ia32_getmantss_mask_round): New builtins.
8647 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
8648 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
8649 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
8650 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
8651 * config/i386/sse.md
8652 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
8653 avx512f_sgetexp<mode><mask_scalar_name>
8654 <round_saeonly_scalar_name> ... this.
8655 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
8656 %0, %1, %2<round_saeonly_op3>}): Changed to ...
8657 vgetexp<ssescalarmodesuffix>
8658 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8659 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
8660 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
8661 avx512f_vgetmant<mode><mask_scalar_name>
8662 <round_saeonly_scalar_name> ... this.
8663 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
8664 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
8665 vgetmant<ssescalarmodesuffix>
8666 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
8667 %0<mask_scalar_operand4>, %1, %2
8668 <round_saeonly_scalar_mask_op4>, %3} ... this.
8669 * config/i386/subst.md (mask_scalar_operand4,
8670 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
8671 round_saeonly_scalar_nimm_predicate): New subst attributes.
8673 2017-07-06 Julia Koval <julia.koval@intel.com>
8675 * config/i386/i386.c (ix86_erase_embedded_rounding):
8676 Remove code for old rounding pattern.
8678 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
8680 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
8682 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
8684 * doc/sourcebuild.texi (Test Directives, Variants of
8685 dg-require-support): Add documentation for dg-require-stack-check.
8687 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
8689 * config/i386/subst.md (mask_scalar, round_scalar,
8690 round_saeonly_scalar): New meta-templates.
8691 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
8692 round_scalar_mask_operand3, round_scalar_mask_op3,
8693 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
8694 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
8695 round_saeonly_scalar_constraint,
8696 round_saeonly_scalar_prefix): New subst attribute.
8697 * config/i386/sse.md
8698 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
8699 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
8700 <round_scalar_name> ... this.
8701 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
8702 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
8703 <round_scalar_name> ... this.
8704 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
8705 <sse>_vm<code><mode>3<mask_scalar_name>
8706 <round_saeonly_scalar_name> ... this.
8707 (v<plusminus_mnemonic><ssescalarmodesuffix>
8708 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8709 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8710 v<plusminus_mnemonic><ssescalarmodesuffix>
8711 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8712 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8713 (v<multdiv_mnemonic><ssescalarmodesuffix>
8714 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8715 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8716 v<multdiv_mnemonic><ssescalarmodesuffix>
8717 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8718 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8719 (v<maxmin_float><ssescalarmodesuffix>
8720 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
8721 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
8722 v<maxmin_float><ssescalarmodesuffix>
8723 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8724 %0<mask_scalar_operand3>, %1, %<iptr>2
8725 <round_saeonly_scalar_mask_op3>} ... this.
8727 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
8729 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
8730 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
8732 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
8733 Alan Hayward <alan.hayward@arm.com>
8734 David Sherwood <david.sherwood@arm.com>
8736 * combine.c (simplify_if_then_else): Remove "enum" before
8738 * compare-elim.c (can_eliminate_compare): Likewise.
8739 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8741 (aarch64_lookup_simd_builtin_type): Likewise.
8742 (aarch64_simd_builtin_type): Likewise.
8743 (aarch64_init_simd_builtin_types): Likewise.
8744 (aarch64_simd_expand_args): Likewise.
8745 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
8747 (aarch64_reverse_mask): Likewise.
8748 (aarch64_simd_emit_reg_reg_move): Likewise.
8749 (aarch64_gen_adjusted_ldpstp): Likewise.
8750 (aarch64_ccmp_mode_to_code): Likewise.
8751 (aarch64_operands_ok_for_ldpstp): Likewise.
8752 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8753 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
8755 (aarch64_min_divisions_for_recip_mul): Likewise.
8756 (aarch64_reassociation_width): Likewise.
8757 (aarch64_get_condition_code_1): Likewise.
8758 (aarch64_simd_emit_reg_reg_move): Likewise.
8759 (aarch64_simd_attr_length_rglist): Likewise.
8760 (aarch64_reverse_mask): Likewise.
8761 (aarch64_operands_ok_for_ldpstp): Likewise.
8762 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8763 (aarch64_gen_adjusted_ldpstp): Likewise.
8764 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
8766 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
8767 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8768 (arm_lookup_simd_builtin_type): Likewise.
8769 (arm_simd_builtin_type): Likewise.
8770 (arm_init_simd_builtin_types): Likewise.
8771 (arm_expand_builtin_args): Likewise.
8772 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
8773 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
8774 (ft32_setup_incoming_varargs): Likewise.
8775 (ft32_function_arg): Likewise.
8776 (ft32_function_arg_advance): Likewise.
8777 (ft32_pass_by_reference): Likewise.
8778 (ft32_arg_partial_bytes): Likewise.
8779 (ft32_valid_pointer_mode): Likewise.
8780 (ft32_addr_space_pointer_mode): Likewise.
8781 (ft32_addr_space_legitimate_address_p): Likewise.
8782 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
8784 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
8785 (ix86_emit_outlined_ms2sysv_restore): Likewise.
8786 (iamcu_alignment): Likewise.
8787 (canonicalize_vector_int_perm): Likewise.
8788 (ix86_noce_conversion_profitable_p): Likewise.
8789 (ix86_mpx_bound_mode): Likewise.
8790 (ix86_operands_ok_for_move_multiple): Likewise.
8791 * config/microblaze/microblaze-protos.h
8792 (microblaze_expand_conditional_branch_reg): Likewise.
8793 * config/microblaze/microblaze.c
8794 (microblaze_expand_conditional_branch_reg): Likewise.
8795 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8797 (rs6000_reassociation_width): Likewise.
8798 (rs6000_invalid_binary_op): Likewise.
8799 (fusion_p9_p): Likewise.
8800 (emit_fusion_p9_load): Likewise.
8801 (emit_fusion_p9_store): Likewise.
8802 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
8804 (riscv_hard_regno_mode_ok_p): Likewise.
8805 (riscv_address_insns): Likewise.
8806 (riscv_split_symbol): Likewise.
8807 (riscv_legitimize_move): Likewise.
8808 (riscv_function_value): Likewise.
8809 (riscv_hard_regno_nregs): Likewise.
8810 (riscv_expand_builtin): Likewise.
8811 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
8812 (riscv_build_integer): Likewise.
8813 (riscv_split_integer): Likewise.
8814 (riscv_legitimate_constant_p): Likewise.
8815 (riscv_cannot_force_const_mem): Likewise.
8816 (riscv_regno_mode_ok_for_base_p): Likewise.
8817 (riscv_valid_base_register_p): Likewise.
8818 (riscv_valid_offset_p): Likewise.
8819 (riscv_valid_lo_sum_p): Likewise.
8820 (riscv_classify_address): Likewise.
8821 (riscv_legitimate_address_p): Likewise.
8822 (riscv_address_insns): Likewise.
8823 (riscv_load_store_insns): Likewise.
8824 (riscv_force_binary): Likewise.
8825 (riscv_split_symbol): Likewise.
8826 (riscv_force_address): Likewise.
8827 (riscv_legitimize_address): Likewise.
8828 (riscv_move_integer): Likewise.
8829 (riscv_legitimize_const_move): Likewise.
8830 (riscv_legitimize_move): Likewise.
8831 (riscv_address_cost): Likewise.
8832 (riscv_subword): Likewise.
8833 (riscv_output_move): Likewise.
8834 (riscv_canonicalize_int_order_test): Likewise.
8835 (riscv_emit_int_order_test): Likewise.
8836 (riscv_function_arg_boundary): Likewise.
8837 (riscv_pass_mode_in_fpr_p): Likewise.
8838 (riscv_pass_fpr_single): Likewise.
8839 (riscv_pass_fpr_pair): Likewise.
8840 (riscv_get_arg_info): Likewise.
8841 (riscv_function_arg): Likewise.
8842 (riscv_function_arg_advance): Likewise.
8843 (riscv_arg_partial_bytes): Likewise.
8844 (riscv_function_value): Likewise.
8845 (riscv_pass_by_reference): Likewise.
8846 (riscv_setup_incoming_varargs): Likewise.
8847 (riscv_print_operand): Likewise.
8848 (riscv_elf_select_rtx_section): Likewise.
8849 (riscv_save_restore_reg): Likewise.
8850 (riscv_for_each_saved_reg): Likewise.
8851 (riscv_register_move_cost): Likewise.
8852 (riscv_hard_regno_mode_ok_p): Likewise.
8853 (riscv_hard_regno_nregs): Likewise.
8854 (riscv_class_max_nregs): Likewise.
8855 (riscv_memory_move_cost): Likewise.
8856 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
8857 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
8858 (rl78_addr_space_address_mode): Likewise.
8859 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8861 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8862 (rs6000_reassociation_width): Likewise.
8863 (rs6000_invalid_binary_op): Likewise.
8864 (fusion_p9_p): Likewise.
8865 (emit_fusion_p9_load): Likewise.
8866 (emit_fusion_p9_store): Likewise.
8867 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
8868 (ok_for_simple_move_operands): Likewise.
8869 (ok_for_simple_move_strict_operands): Likewise.
8870 (ok_for_simple_arith_logic_operands): Likewise.
8871 (visium_legitimize_reload_address): Likewise.
8872 (visium_select_cc_mode): Likewise.
8873 (output_cbranch): Likewise.
8874 (visium_split_double_move): Likewise.
8875 (visium_expand_copysign): Likewise.
8876 (visium_expand_int_cstore): Likewise.
8877 (visium_expand_fp_cstore): Likewise.
8878 * config/visium/visium.c (visium_pass_by_reference): Likewise.
8879 (visium_function_arg): Likewise.
8880 (visium_function_arg_advance): Likewise.
8881 (visium_libcall_value): Likewise.
8882 (visium_setup_incoming_varargs): Likewise.
8883 (visium_legitimate_constant_p): Likewise.
8884 (visium_legitimate_address_p): Likewise.
8885 (visium_legitimize_address): Likewise.
8886 (visium_secondary_reload): Likewise.
8887 (visium_register_move_cost): Likewise.
8888 (visium_memory_move_cost): Likewise.
8889 (prepare_move_operands): Likewise.
8890 (ok_for_simple_move_operands): Likewise.
8891 (ok_for_simple_move_strict_operands): Likewise.
8892 (ok_for_simple_arith_logic_operands): Likewise.
8893 (visium_function_value_1): Likewise.
8894 (rtx_ok_for_offset_p): Likewise.
8895 (visium_legitimize_reload_address): Likewise.
8896 (visium_split_double_move): Likewise.
8897 (visium_expand_copysign): Likewise.
8898 (visium_expand_int_cstore): Likewise.
8899 (visium_expand_fp_cstore): Likewise.
8900 (visium_split_cstore): Likewise.
8901 (visium_select_cc_mode): Likewise.
8902 (visium_split_cbranch): Likewise.
8903 (output_cbranch): Likewise.
8904 (visium_print_operand_address): Likewise.
8905 * expmed.c (flip_storage_order): Likewise.
8906 * expmed.h (emit_cstore): Likewise.
8907 (flip_storage_order): Likewise.
8908 * genrecog.c (validate_pattern): Likewise.
8909 * hsa-gen.c (gen_hsa_addr): Likewise.
8910 * internal-fn.c (expand_arith_overflow): Likewise.
8911 * ira-color.c (allocno_copy_cost_saving): Likewise.
8912 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8913 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
8914 (process_invariant_for_inheritance): Likewise.
8915 * lra-eliminations.c (move_plus_up): Likewise.
8916 * omp-low.c (lower_oacc_reductions): Likewise.
8917 * simplify-rtx.c (simplify_subreg): Likewise.
8918 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
8919 (TARGET_CHKP_BOUND_MODE): Likewise..
8920 * targhooks.c (default_chkp_bound_mode): Likewise.
8921 (default_setup_incoming_vararg_bounds): Likewise.
8922 * targhooks.h (default_chkp_bound_mode): Likewise.
8923 (default_setup_incoming_vararg_bounds): Likewise.
8924 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
8925 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
8926 (have_whole_vector_shift): Likewise.
8927 * tree-vect-stmts.c (vectorizable_load): Likewise.
8928 * doc/tm.texi: Regenerate.
8930 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
8932 Graceful degrade if Binutils PR21472 is not available.
8935 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
8936 .rodata in flash test fails.
8937 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
8938 * confgure: Regenerate.
8939 * config.in: Regenerate.
8940 * config/avr/avr.c (avr_asm_named_section)
8941 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
8942 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
8943 (avr_asm_init_sections): Same.
8945 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8947 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
8948 (fma<VH:mode>4_intrinsic): Likewise.
8949 (*fmsub<VCVTF:mode>4): Likewise.
8950 (*fmsub<VH:mode>4_intrinsic): Likewise.
8952 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
8955 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
8956 Don't depend on "optimize > 0".
8957 (out_movhi_r_mr, out_movqi_mr_r): Same.
8958 (out_movhi_mr_r, out_movqi_r_mr): Same.
8959 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
8960 io_address_operand on "optimize > 0".
8962 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8964 * tree-loop-distribution.c: Add general explanantion on the pass.
8965 (generate_loops_for_partition): Mark distributed loop.
8966 (pg_add_dependence_edges): New parameter. Handle alias data
8967 dependence specially and record it in the parameter if asked.
8968 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
8969 (init_partition_graph_vertices, add_partition_graph_edge): New.
8970 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
8971 (free_partition_graph_vdata, build_partition_graph): New.
8972 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
8973 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
8974 (data_ref_segment_size, latch_dominated_by_data_ref): New.
8975 (compute_alias_check_pairs, version_loop_by_alias_check): New.
8976 (version_for_distribution_p, finalize_partitions): New.
8977 (distribute_loop): Handle alias data dependence specially. Factor
8978 out loop fusion code as functions and call these functions.
8980 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8982 * tree-loop-distribution.c (classify_partition): New parameter and
8983 better handle reduction statement.
8984 (rdg_build_partitions): Revise comment.
8985 (distribute_loop): Compute statements in all partitions and pass it
8986 to classify_partition.
8988 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8990 * tree-loop-distribution.c (enum partition_type): New.
8991 (struct partition): New field type.
8992 (partition_merge_into): Add parameter. Update partition type.
8993 (data_dep_in_cycle_p, update_type_for_merge): New functions.
8994 (build_rdg_partition_for_vertex): Compute partition type.
8995 (rdg_build_partitions): Dump partition type.
8996 (distribute_loop): Update calls to partition_merge_into.
8998 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9000 * tree-loop-distribution.c (struct ddr_hasher): New.
9001 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
9003 (classify_partition): Call get_data_dependence.
9004 (pg_add_dependence_edges): Ditto.
9005 (distribute_loop): Release data dependence hash table.
9007 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9009 * tree-loop-distribution.c (ref_base_address): Delete.
9010 (similar_memory_accesses): Rename ...
9011 (share_memory_accesses): ... to this. Check if partitions access
9012 the same memory reference.
9013 (distribute_loop): Call share_memory_accesses.
9015 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9017 * tree-loop-distribution.c (struct partition): New field recording
9019 (partition_alloc, partition_free): Init and release data refs.
9020 (partition_merge_into): Merge data refs.
9021 (build_rdg_partition_for_vertex): Collect data refs for partition.
9022 (pg_add_dependence_edges): Change parameters from vector to bitmap.
9024 (distribute_loop): Remve data refs from vertice data of partition
9027 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9029 * tree-loop-distribution.c (params.h): Include header file.
9030 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
9031 (datarefs_vec): New global var.
9032 (create_rdg_vertices): Use datarefs_vec directly.
9033 (free_rdg): Don't free data references.
9034 (build_rdg): Update use. Don't free data references.
9035 (distribute_loop): Compute global variable for data references.
9036 Bail out if there are too many data references.
9038 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9040 * tree-loop-distribution.c (loop_nest): New global var.
9041 (build_rdg): Use loop directly, rather than loop nest.
9042 (pg_add_dependence_edges): Remove loop nest parameter. Use global
9044 (distribute_loop): Compute global variable loop nest. Update use.
9046 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9048 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
9049 (partition_merge_into): New parameter. Dump reason for fusion.
9050 (distribute_loop): Update use of partition_merge_into.
9052 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9054 * tree-loop-distribution.c (bb_top_order_index): New.
9055 (bb_top_order_index_size, bb_top_order_cmp): New.
9056 (stmts_from_loop): Use topological order.
9057 (pass_loop_distribution::execute): Compute and release topological
9058 order for basic blocks.
9060 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9062 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
9065 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9067 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
9068 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
9069 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
9070 * internal-fn.def (LOOP_DIST_ALIAS): New.
9071 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
9072 (fold_loop_internal_call): ... this.
9073 (vect_loop_dist_alias_call): New function.
9074 (set_uid_loop_bbs): Call fold_loop_internal_call.
9075 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
9078 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9081 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
9082 Require dead FLAGS_REG at the beginning of a peephole.
9084 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9087 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
9088 arguments in the call to __builtin_ia32_sbb_u32.
9089 (_subborrow_u64): Swap _X and _Y arguments in the call to
9090 __builtin_ia32_sbb_u64.
9092 2017-07-04 Jakub Jelinek <jakub@redhat.com>
9095 * tree-vrp.c (compare_assert_loc): Turn into a function template
9096 with stable template parameter. Only test if a->e is NULL,
9097 !a->e == !b->e has been verified already. Use e == NULL or
9098 e != NULL instead of e or ! e tests. If stable is true, don't use
9099 iterative_hash_expr, on the other side allow a or b or both NULL
9100 and sort the NULLs last.
9101 (process_assert_insertions): Sort using compare_assert_loc<false>
9102 instead of compare_assert_loc, later sort using
9103 compare_assert_loc<true> before calling process_assert_insertions_for
9104 in a loop. Use break instead of continue once seen NULL pointer.
9106 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9108 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9109 Cortex-R7 and Cortex-R8 processors.
9111 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9113 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
9114 uninitialized while src is not.
9116 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
9118 * common/config/arm/arm-common.c: Adjust include path for
9120 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
9121 (arm-cpu.h): Create in build directory. Adjust dependency rules.
9122 (arm-cpu-data.h): Likewise.
9123 (arm-cpu-cdata.h): Likewise.
9124 * config/arm/arm-cpu.h: Delete.
9125 * config/arm/arm-cpu-cdata.h: Delete.
9126 * config/arm/arm-cpu-data.h: Delete.
9128 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
9130 * config/arm/arm-cpus.in (cortex-a55): New.
9131 (cortex-a75): Likewise.
9132 (cortex-a75.cortex-a55): Likewise.
9133 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
9135 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
9136 * config/arm/arm-cpu-cdata.h: Regenerate.
9137 * config/arm/arm-cpu-data.h: Regenerate.
9138 * config/arm/arm-cpu.h: Regenerate.
9139 * config/arm/arm-tables.opt: Regenerate.
9140 * config/arm/arm-tune.md: Regenerate.
9142 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9144 * haifa-sched.c (sched_create_recovery_edges): Update profile.
9146 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9148 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
9151 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
9153 PR tree-optimization/81292
9154 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
9155 full_string_p, also call adjust_related_strinfos if the adjustment
9156 is simple, otherwise invalidate related strinfos.
9158 2017-07-04 Martin Liska <mliska@suse.cz>
9161 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
9162 newly created variable as DECL_IGNORED_P.
9164 2017-07-04 Martin Liska <mliska@suse.cz>
9167 * ipa-inline.c (inline_small_functions):
9168 Use xstrdup_for_dump.
9170 2017-07-04 Tom de Vries <tom@codesourcery.com>
9172 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
9174 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9177 * config/darwin.c (darwin_function_switched_text_sections):
9180 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
9182 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
9184 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9186 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
9188 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9190 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
9191 min_profitable_iters, and th as inclusive lower bounds.
9192 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
9193 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
9194 for min_profitable_iters and min_profitable_estimate.
9195 (vect_transform_loop): Treat th as an inclusive lower bound.
9196 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
9198 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9201 * config/darwin.c (darwin_function_switched_text_sections):
9202 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
9203 in two pieces, and suppress the use of buf.
9205 2017-07-03 Nathan Sidwell <nathan@acm.org>
9207 * hash-table.h (hash_table_mod1): Fix indentation.
9209 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9212 * predict.c (force_edge_cold): Be more careful about propagation
9214 * profile-count.h (profile_probability::guessed,
9215 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
9217 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
9219 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
9221 * doc/invoke.texi (rcpc architecture extension): Document it.
9223 2017-07-03 Richard Biener <rguenther@suse.de>
9225 PR tree-optimization/60510
9226 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
9227 the scalar reduction PHI and use it.
9228 (vectorizable_reduction): Properly guard the single_defuse_cycle
9229 path for non-SLP reduction chains where we cannot use it.
9230 Rework reduc_def/index and vector type deduction. Rework
9231 vector operand gathering during reduction op code-gen.
9232 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
9233 chains dissolve the chain and leave it to non-SLP reduction
9236 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9238 * tree-data-ref.h (dr_alignment): Declare.
9239 * tree-data-ref.c (dr_alignment): New function.
9240 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
9241 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
9243 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
9245 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9247 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
9248 and base_misalignment fields.
9249 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
9250 * tree-data-ref.c: Include builtins.h.
9251 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
9252 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
9253 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
9254 * tree-vect-data-refs.c: Include tree-cfg.h.
9255 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
9256 fields instead of calculating an alignment here.
9257 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
9258 innermost_loop_behavior fields.
9260 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9262 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
9264 (DR_STEP_ALIGNMENT): New macro.
9265 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
9266 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
9267 (create_data_ref): Print it.
9268 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
9269 to tell whether the step preserves vector (mis)alignment.
9270 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9271 Move the check for an integer step and generalise to all INTEGER_CST.
9272 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
9273 Print the outer step alignment.
9275 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9277 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
9278 with offset_alignment.
9279 (DR_ALIGNED_TO): Delete.
9280 (DR_OFFSET_ALIGNMENT): New macro.
9281 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
9282 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
9283 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
9284 (create_data_ref): Likewise.
9285 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9286 (vect_analyze_data_refs): Likewise.
9287 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
9288 creating dummy innermost behavior.
9290 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9292 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
9293 with a "innermost_loop_behavior *" and refeence tree.
9294 * tree-data-ref.c (dr_analyze_innermost): Likewise.
9295 (create_data_ref): Update call accordingly.
9296 * tree-predcom.c (find_looparound_phi): Likewise.
9298 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9300 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
9301 fields with dr_wrt_vec_loop.
9302 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
9303 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
9304 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
9305 (vect_dr_behavior): New function.
9306 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9307 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
9308 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
9309 track whether the step preserves the misalignment.
9310 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9311 Use vect_dr_behavior.
9312 (vect_setup_realignment): Update call accordingly.
9313 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
9314 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
9315 call to vect_create_addr_base_for_vector_ref.
9316 (vect_create_cond_for_align_checks): Likewise.
9317 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
9318 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
9319 (vect_recog_mask_conversion_pattern): Likewise.
9320 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
9321 (new_stmt_vec_info): Remove redundant zeroing.
9323 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9325 * common/config/arm/arm-common.c (arm_be8_option): New function.
9326 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
9327 (ISA_ARMv6): Add isa_bit_be8.
9328 * config/arm/arm.h (arm_be8_option): Add prototype.
9329 (BE8_SPEC_FUNCTION): New define.
9330 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
9331 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
9332 (mlittle-endian): Similarly.
9333 (mbe8, mbe32): New options.
9334 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
9335 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
9337 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9339 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
9341 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9343 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
9344 (cleanup_tree_cfg_bb): Use it.
9345 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
9347 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
9349 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9352 * loop-doloop.c (add_test): Update profile.
9354 2017-07-03 Martin Liska <mliska@suse.cz>
9357 * sanopt.c (rewrite_usage_of_param): New function.
9358 (sanitize_rewrite_addressable_params): Likewise.
9359 (pass_sanopt::execute): Call rewrite_usage_of_param.
9361 2017-07-03 Richard Biener <rguenther@suse.de>
9363 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
9364 back to using VIEW_CONVERT_EXPR.
9366 2017-07-03 Martin Liska <mliska@suse.cz>
9369 * doc/extend.texi: Document when a resolver function is
9370 generated for target_clones.
9372 2017-07-03 Martin Liska <mliska@suse.cz>
9374 * asan.c (asan_emit_stack_protection): Unpoison just red zones
9375 and shadow memory of auto variables which are subject of
9376 use-after-scope sanitization.
9377 (asan_expand_mark_ifn): Add do set only when is_poison.
9379 2016-07-03 Richard Biener <rguenther@suse.de>
9381 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
9383 (vect_force_simple_reduction): Record reduction def -> phi mapping.
9384 (vectorizable_reduction): Perform reduction PHI creation when
9385 visiting a reduction PHI and adjust and simplify code generation
9386 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
9387 (vect_transform_loop): Visit reduction PHIs.
9388 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
9389 defs into the SLP tree.
9390 (vect_build_slp_tree): Reduction defs terminate the recursion.
9391 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
9393 (vect_get_vec_defs_for_stmt_copy): Export.
9394 (vect_get_vec_defs): Likewise.
9395 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
9397 (vect_get_vec_defs_for_stmt_copy): Declare.
9398 (vect_get_vec_defs): Likewise.
9400 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9402 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
9403 parameter with a "loop" parameter and use it instead of the
9404 loop containing DR_STMT. Don't check simple_iv when doing
9405 BB analysis. Describe the two analysis modes in the comment.
9407 2017-07-03 Tom de Vries <tom@codesourcery.com>
9409 PR tree-optimization/69468
9410 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
9411 (find_same_succ_bb): Handle ignore_edge_flags.
9413 2017-07-03 Tom de Vries <tom@codesourcery.com>
9415 PR tree-optimization/81192
9416 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
9418 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
9420 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
9422 2017-07-03 Tom de Vries <tom@codesourcery.com>
9424 PR tree-optimization/81192
9425 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
9426 BB_SAME_SUCC (bb) == NULL.
9428 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9430 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
9433 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9435 * dumpfile.c: Include profile-count.h
9436 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
9438 (insert_cond_bb): Update profile.
9439 * tree-cfg.h (insert_cond_bb): Update prototype.
9440 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
9441 * tree-dump.c: Do not include tree-cfg.
9443 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9445 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
9447 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9449 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
9452 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9454 * tree-complex.c (expand_complex_div_wide): update profile.
9456 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9457 Alan Hayward <alan.hayward@arm.com>
9458 David Sherwood <david.sherwood@arm.com>
9460 * Makefile.in (MACHMODE_H): Remove insn-modes.h
9461 (CORETYPES_H): New define.
9462 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
9463 (insn-modes-inline.h, s-modes-inline-h): New rules.
9464 (generated_files): Add insn-modes-inline.h.
9465 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
9466 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
9467 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
9468 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
9469 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
9470 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
9471 (build/gencodes.o, build/genconditions.o): Likewise.
9472 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
9473 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
9474 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
9475 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
9476 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
9477 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
9478 * coretypes.h: Include everything up to real.h for generators.
9479 Include insn-modes.h first. Include wide-int-print.h after
9480 wide-int.h. Include insn-modes-inline.h and then machmode.h.
9481 * machmode.h: Don't include insn-modes.h here.
9482 * function-tests.c: Remove includes of signop.h, machmode.h,
9483 double-int.h and wide-int.h.
9485 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
9487 * optc-save-gen.awk: Likewise.
9488 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
9489 * godump.c: Remove include of wide-int-print.h.
9490 * pretty-print.h: Likewise.
9491 * wide-int-print.cc: Likewise.
9492 * wide-int.cc: Likewise.
9493 * hash-map-tests.c: Remove include of signop.h.
9494 * hash-set-tests.c: Likewise.
9495 * rtl-tests.c: Likewise.
9496 * mkconfig.sh: Remove include of machmode.h.
9497 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
9499 (emit_insn_modes_inline_h): ...this new function. Emit the code
9500 into an insn-modes-inline.h header file, adding appropriate
9501 include guards and end comments.
9502 (emit_insn_modes_c_header): Remove include of machmode.h.
9503 (emit_min_insn_modes_c_header): Include coretypes.h rather than
9505 (main): Handle -i flag and call emit_insn_modes_inline_h when
9508 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9510 * tree-ssa-strlen.c (strinfo): Rename the length field to
9511 nonzero_chars. Add a full_string_p field.
9512 (compare_nonzero_chars, zero_length_string_p): New functions.
9513 (get_addr_stridx): Add an offset_out parameter.
9514 Use compare_nonzero_chars.
9515 (get_stridx): Update accordingly. Use compare_nonzero_chars.
9516 (new_strinfo): Update after above changes to strinfo.
9517 (set_endptr_and_length): Set full_string_p.
9518 (get_string_length): Update after above changes to strinfo.
9519 (unshare_strinfo): Update call to new_strinfo.
9520 (maybe_invalidate): Likewise.
9521 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
9522 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
9523 as a uhwi instead of an shwi. Update after above changes to
9524 strinfo and new_strinfo.
9525 (zero_length_string): Assert that chainsi contains full strings.
9526 Use zero_length_string_p. Update call to new_strinfo.
9527 (adjust_related_strinfos): Update after above changes to strinfo.
9528 Copy full_string_p from origsi.
9529 (adjust_last_stmt): Use zero_length_string_p.
9530 (handle_builtin_strlen): Update after above changes to strinfo and
9531 new_strinfo. Install the lhs as the string length if the previous
9532 entry didn't describe a full string.
9533 (handle_builtin_strchr): Update after above changes to strinfo
9535 (handle_builtin_strcpy): Likewise.
9536 (handle_builtin_strcat): Likewise.
9537 (handle_builtin_malloc): Likewise.
9538 (handle_pointer_plus): Likewise.
9539 (handle_builtin_memcpy): Likewise. Track nonzero characters
9540 that aren't necessarily followed by a nul terminator.
9541 (handle_char_store): Likewise.
9543 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9545 PR tree-optimization/80769
9546 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
9547 for malloc and calloc. Document the new invariant that all related
9548 strinfos have delayed lengths or none do.
9549 (verify_related_strinfos): Move earlier in file.
9550 (set_endptr_and_length): New function, split out from...
9551 (get_string_length): ...here. Also set the lengths of related
9553 (zero_length_string): Assert that chainsi has known (rather than
9555 (adjust_related_strinfos): Likewise.
9557 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9559 PR tree-optimization/81136
9560 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
9561 assert that two references with the same misalignment have the same
9562 compile-time misalignment if those compile-time misalignments
9565 2017-07-01 Andi Kleen <ak@linux.intel.com>
9567 * print-tree.c (print_node): Print all attributes.
9569 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9571 * cfg.c (scale_bbs_frequencies): New function.
9572 * cfg.h (scale_bbs_frequencies): Declare it.
9573 * cfgloopanal.c (single_likely_exit): Cleanup.
9574 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
9576 (scale_loop_profile): Likewise.
9577 (loop_version): Likewise.
9578 (create_empty_loop_on_edge): Update.
9579 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
9580 scale_loop_frequencies, scale_loop_profile, loopify,
9581 loop_version): Update prototypes.
9582 * modulo-sched.c (sms_schedule): Update.
9583 * predict.c (unlikely_executed_edge_p): Also check probability.
9584 (probably_never_executed_edge_p): Fix typo.
9585 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9586 * tree-parloops.c (gen_parallel_loop): Update.
9587 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
9588 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9589 * tree-ssa-loop-split.c (split_loop): Update.
9590 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9591 * tree-vect-loop-manip.c (vect_do_peeling): Update.
9592 (vect_loop_versioning): Update.
9593 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9595 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9597 * trans-mem.c (split_bb_make_tm_edge): Update profile.
9599 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9601 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
9602 to keep profile consistent.
9604 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9606 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
9607 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
9608 * profile-count.h (max_safe_multiplier): Make unsigned.
9609 (profile_count::guessed_zero): New.
9611 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9613 * bb-reorder.c (fix_up_crossing_landing_pad,
9614 fix_crossing_conditional_branches): Use make_single_succ_edge
9615 to keep profile consistent.
9617 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9619 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
9622 2017-07-01 Jakub Jelinek <jakub@redhat.com>
9625 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
9626 the right scopes, make sure cond_jump isn't preserved between multiple
9627 iterations. Search for fallthru edge whenever there are 3+ edges and
9628 use find_fallthru_edge for it.
9630 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9632 Patch by Alexander Monakov <amonakov@ispras.ru>
9633 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
9634 probabilities consistently.
9636 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9638 * pa.c (pa_expand_compare_and_swap_loop): Update call of
9639 emit_cmp_and_jump_insns.
9641 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9644 * tree-inline.c (expand_call_inline): Combine profile statuses.
9646 2017-06-30 Andrew Pinski <apinski@cavium.com>
9648 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
9649 fold_stmt returned true.
9651 2017-06-30 Nathan Sidwell <nathan@acm.org>
9653 * ggc.h (empty_string): Delete.
9654 * cfgexpand.c (expand_asm_stmt): Use plain "".
9655 * optabs.c (expand_asm_memory_barrier): Likewise.
9656 * stringpool.c (empty_string): Delete.
9657 (digit_vector, digit_string): Delete.
9658 (ggc_alloc_string): Use plain "", don't optimize single digit
9659 strings. Use ggc_alloc_atomic.
9661 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
9663 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
9664 comparison set and one other set, use the cost of the non-comparison
9667 2017-06-30 Nathan Sidwell <nathan@acm.org>
9669 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
9672 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
9674 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
9675 loops. Remove now unneeded calls to gimple_switch_set_label() that
9676 just set removed labels to NULL_TREE.
9678 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
9680 * tree-ssanames.c (set_range_info_raw): Abstract from ...
9681 (set_range_info): ...here. Only call set_range_info_raw if domain
9683 (set_nonzero_bits): Call set_range_info_raw.
9684 * tree-ssanames.h (set_range_info_raw): New.
9686 2017-06-30 Jakub Jelinek <jakub@redhat.com>
9689 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
9690 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
9691 of nonimmediate_operand and <store_mask_constraint> instead of m
9692 for the input operand. For V8FI iterator, always split if input
9693 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
9694 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
9695 <store_mask_predicate> instead of register_operand and
9696 <store_mask_constraint> instead of v for the input operand. Make
9697 sure both operands aren't MEMs for if not <mask_applied>.
9699 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
9701 * lto-wrapper.c (copy_file) Close both file descriptors before
9704 2017-06-30 Martin Liska <mliska@suse.cz>
9707 * multiple_target.c (create_dispatcher_calls): Make ifunc
9708 also for function that don't have calls or are not referenced.
9710 2017-06-30 Richard Biener <rguenther@suse.de>
9712 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
9713 analyze the first scalar stmt. Move vector type computation
9714 for the BB case here from ...
9715 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
9716 live operation processing in the SLP case properly.
9718 2017-06-30 Richard Biener <rguenther@suse.de>
9720 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
9722 2017-06-30 Martin Liska <mliska@suse.cz>
9725 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
9726 before BUILT_IN_UNWIND_RESUME when ASAN is used.
9728 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
9730 * doc/invoke.texi (AArch64): Add missing options and remove redundant
9733 2017-06-30 Richard Biener <rguenther@suse.de>
9735 PR tree-optimization/81249
9736 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
9737 condition reduction result to original scalar type.
9739 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9741 * profile-count.h (enum profile_quality): Fix typos and whitespace
9744 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9746 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
9747 type for branch probabilities.
9749 2017-06-29 Julian Brown <julian@codesourcery.com>
9750 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9752 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
9753 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
9754 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
9755 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
9757 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9759 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
9760 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
9761 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
9762 CC usage from generic code to here.
9763 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
9764 CC usage into the target macros.
9766 2017-06-29 Maya Rashish <coypu@sdf.org>
9768 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
9771 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9773 * arm/arm-builtins.c: Include profile-count.h
9774 * except.c (sjlj_emit_function_enter): Use
9775 profile_probability::unlikely.
9777 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9779 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
9780 and tocrel_offset be pointer args rather than implicitly using
9782 (legitimate_constant_pool_address_p, rs6000_emit_move,
9783 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
9784 tocrel_offset and use in toc_relative_expr_p call.
9785 (print_operand, print_operand_address): Use static tocrel_base_oac
9786 and tocrel_offset_oac.
9787 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
9790 2017-06-29 Maya Rashish <coypu@sdf.org>
9792 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
9794 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
9796 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
9797 objects, take into account only the alignment of 'op0' and 'mode1' if
9800 2017-06-29 Steve Ellcey <sellcey@cavium.com>
9802 * ccmp.c (ccmp_tree_comparison_p): New function.
9803 (ccmp_candidate_p): Update to use above function.
9804 (get_compare_parts): New function.
9805 (expand_ccmp_next): Update to use new functions.
9806 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
9808 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
9809 take mode as argument.
9810 * ccmp.h (expand_ccmp_expr): Add mode as argument.
9811 * expr.c (expand_expr_real_1): Pass mode as argument.
9813 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
9815 * combine.c (combine_instructions): Print insns to dump_file, together
9818 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9820 * asan.c (asan_emit_stack_protection): Update.
9821 (create_cond_insert_point): Update.
9822 * auto-profile.c (afdo_propagate_circuit): Update.
9823 * basic-block.h (struct edge_def): Turn probability to
9824 profile_probability.
9825 (EDGE_FREQUENCY): Update.
9826 * bb-reorder.c (find_traces_1_round): Update.
9827 (better_edge_p): Update.
9828 (sanitize_hot_paths): Update.
9829 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
9830 (make_single_succ_edge): Update.
9831 (check_bb_profile): Update.
9832 (dump_edge_info): Update.
9833 (update_bb_profile_for_threading): Update.
9834 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
9835 probabilitycount to 0.
9836 * cfgbuild.c (compute_outgoing_frequencies): Update.
9837 * cfgcleanup.c (try_forward_edges): Update.
9838 (outgoing_edges_match): Update.
9839 (try_crossjump_to_edge): Update.
9840 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
9841 (expand_gimple_tailcall): Update.
9842 (construct_init_block): Use make_single_succ_edge.
9843 (construct_exit_block): Use make_single_succ_edge.
9844 * cfghooks.c (verify_flow_info): Update.
9845 (redirect_edge_succ_nodup): Update.
9846 (split_edge): Update.
9847 (account_profile_record): Update.
9848 * cfgloopanal.c (single_likely_exit): Update.
9849 * cfgloopmanip.c (scale_loop_profile): Update.
9850 (set_zero_probability): Remove.
9851 (duplicate_loop_to_header_edge): Update.
9852 * cfgloopmanip.h (loop_version): Update prototype.
9853 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9854 (force_nonfallthru_and_redirect): Update.
9855 (update_br_prob_note): Update.
9856 (rtl_verify_edges): Update.
9857 (purge_dead_edges): Update.
9858 (rtl_lv_add_condition_to_bb): Update.
9859 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
9860 * cgraphunit.c (init_lowered_empty_function): Update.
9861 (cgraph_node::expand_thunk): Update.
9862 * cilk-common.c: Include profile-count.h
9863 * dojump.c (inv): Remove.
9864 (jumpifnot): Update.
9865 (jumpifnot_1): Update.
9866 (do_jump_1): Update.
9868 (do_jump_by_parts_greater_rtx): Update.
9869 (do_compare_rtx_and_jump): Update.
9870 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
9871 do_jump_1. do_compare_rtx_and_jump): Update prototype.
9872 * dwarf2cfi.c: Include profile-count.h
9873 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
9874 (sjlj_emit_dispatch_table): Likewise.
9875 * explow.c: Include profile-count.h
9876 * expmed.c (emit_store_flag_force): Update.
9877 (do_cmp_and_jump): Update.
9878 * expr.c (compare_by_pieces_d::generate): Update.
9879 (compare_by_pieces_d::finish_mode): Update.
9880 (emit_block_move_via_loop): Update.
9881 (store_expr_with_bounds): Update.
9882 (store_constructor): Update.
9883 (expand_expr_real_2): Update.
9884 (expand_expr_real_1): Update.
9885 * expr.h (try_casesi, try_tablejump): Update prototypes.
9886 * gimple-pretty-print.c (dump_probability): Update.
9887 (dump_profile): New.
9888 (dump_gimple_label): Update.
9889 (dump_gimple_bb_header): Update.
9890 * graph.c (draw_cfg_node_succ_edges): Update.
9891 * hsa-gen.c (convert_switch_statements): Update.
9892 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
9893 (find_if_case_1): Update.
9894 (find_if_case_2): Update.
9895 * internal-fn.c (expand_arith_overflow_result_store): Update.
9896 (expand_addsub_overflow): Update.
9897 (expand_neg_overflow): Update.
9898 (expand_mul_overflow): Update.
9899 (expand_vector_ubsan_overflow): Update.
9900 * ipa-cp.c (good_cloning_opportunity_p): Update.
9901 * ipa-split.c (split_function): Use make_single_succ_edge.
9902 * ipa-utils.c (ipa_merge_profiles): Update.
9903 * loop-doloop.c (add_test): Update.
9904 (doloop_modify): Update.
9905 * loop-unroll.c (compare_and_jump_seq): Update.
9906 (unroll_loop_runtime_iterations): Update.
9907 * lra-constraints.c (lra_inheritance): Update.
9908 * lto-streamer-in.c (input_cfg): Update.
9909 * lto-streamer-out.c (output_cfg): Update.
9910 * mcf.c (adjust_cfg_counts): Update.
9911 * modulo-sched.c (sms_schedule): Update.
9912 * omp-expand.c (expand_omp_for_init_counts): Update.
9913 (extract_omp_for_update_vars): Update.
9914 (expand_omp_ordered_sink): Update.
9915 (expand_omp_for_ordered_loops): Update.
9916 (expand_omp_for_generic): Update.
9917 (expand_omp_for_static_nochunk): Update.
9918 (expand_omp_for_static_chunk): Update.
9919 (expand_cilk_for): Update.
9920 (expand_omp_simd): Update.
9921 (expand_omp_taskloop_for_outer): Update.
9922 (expand_omp_taskloop_for_inner): Update.
9923 * omp-simd-clone.c (simd_clone_adjust): Update.
9924 * optabs.c (expand_doubleword_shift): Update.
9925 (expand_abs): Update.
9926 (emit_cmp_and_jump_insn_1): Update.
9927 (expand_compare_and_swap_loop): Update.
9928 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
9929 * predict.c (predictable_edge_p): Update.
9930 (edge_probability_reliable_p): Update.
9931 (set_even_probabilities): Update.
9932 (combine_predictions_for_insn): Update.
9933 (combine_predictions_for_bb): Update.
9934 (propagate_freq): Update.
9935 (estimate_bb_frequencies): Update.
9936 (force_edge_cold): Update.
9937 * profile-count.c (profile_count::dump): Add missing space into dump.
9938 (profile_count::debug): Add newline.
9939 (profile_count::differs_from_p): Explicitly convert to unsigned.
9940 (profile_count::stream_in): Update.
9941 (profile_probability::dump): New member function.
9942 (profile_probability::debug): New member function.
9943 (profile_probability::differs_from_p): New member function.
9944 (profile_probability::differs_lot_from_p): New member function.
9945 (profile_probability::stream_in): New member function.
9946 (profile_probability::stream_out): New member function.
9947 * profile-count.h (profile_count_quality): Rename to ...
9948 (profile_quality): ... this one.
9949 (profile_probability): New.
9950 (profile_count): Update.
9951 * profile.c (compute_branch_probabilities): Update.
9952 * recog.c (peep2_attempt): Update.
9953 * sched-ebb.c (schedule_ebbs): Update.
9954 * sched-rgn.c (find_single_block_region): Update.
9955 (compute_dom_prob_ps): Update.
9956 (schedule_region): Update.
9957 * sel-sched-ir.c (compute_succs_info): Update.
9958 * stmt.c (struct case_node): Update.
9959 (do_jump_if_equal): Update.
9960 (get_outgoing_edge_probs): Update.
9961 (conditional_probability): Update.
9962 (emit_case_dispatch_table): Update.
9963 (expand_case): Update.
9964 (expand_sjlj_dispatch_table): Update.
9965 (emit_case_nodes): Update.
9966 * targhooks.c: Update.
9967 * tracer.c (better_p): Update.
9968 (find_best_successor): Update.
9969 * trans-mem.c (expand_transaction): Update.
9970 * tree-call-cdce.c: Update.
9971 * tree-cfg.c (gimple_split_edge): Upate.
9972 (move_sese_region_to_fn): Upate.
9973 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
9974 * tree-eh.c (lower_resx): Upate.
9975 (cleanup_empty_eh_move_lp): Upate.
9976 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9977 * tree-inline.c (copy_edges_for_bb): Update.
9978 (copy_cfg_body): Update.
9979 * tree-parloops.c (gen_parallel_loop): Update.
9980 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
9981 (gimple_gen_time_profiler): Update.
9982 * tree-ssa-dce.c (remove_dead_stmt): Update.
9983 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
9984 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
9985 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
9986 (unloop_loops): Update.
9987 (try_peel_loop): Update.
9988 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9989 * tree-ssa-loop-split.c (connect_loops): Update.
9990 (split_loop): Update.
9991 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9992 (hoist_guard): Update.
9993 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
9994 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
9995 (value_replacement): Update.
9996 * tree-ssa-reassoc.c (branch_fixup): Update.
9997 * tree-ssa-tail-merge.c (replace_block_by): Update.
9998 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
9999 (create_edge_and_update_destination_phis): Update.
10000 (compute_path_counts): Update.
10001 (recompute_probabilities): Update.
10002 (update_joiner_offpath_counts): Update.
10003 (freqs_to_counts_path): Update.
10004 (duplicate_thread_path): Update.
10005 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10006 (struct switch_conv_info): Update.
10007 (gen_inbound_check): Update.
10008 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10009 (vect_do_peeling): Update.
10010 (vect_loop_versioning): Update.
10011 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10012 (optimize_mask_stores): Update.
10013 * ubsan.c (ubsan_expand_null_ifn): Update.
10014 * value-prof.c (gimple_divmod_fixed_value): Update.
10015 (gimple_divmod_fixed_value_transform): Update.
10016 (gimple_mod_pow2): Update.
10017 (gimple_mod_pow2_value_transform): Update.
10018 (gimple_mod_subtract): Update.
10019 (gimple_mod_subtract_transform): Update.
10020 (gimple_ic): Update.
10021 (gimple_stringop_fixed_value): Update.
10022 (gimple_stringops_transform): Update.
10023 * value-prof.h: Update.
10025 2017-06-29 Carl Love <cel@us.ibm.com>
10027 * config/rs6000/rs6000-c.c: Add support for built-in functions
10028 vector signed int vec_signed (vector float);
10029 vector signed long long vec_signed (vector double);
10030 vector signed int vec_signed2 (vector double, vector double);
10031 vector signed int vec_signede (vector double);
10032 vector signed int vec_signedo (vector double);
10033 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
10034 instruction generator.
10035 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10036 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
10037 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
10039 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
10040 vunsignede_v2df): Add define_expands.
10041 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
10042 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
10043 VEC_UNSIGNEDO): Add definitions.
10044 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10045 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
10046 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
10047 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
10048 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
10049 * config/rs6000/altivec.h (vec_signed, vec_signed2,
10050 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
10051 vec_unsignede, vec_unsignedo): Add builtin defines.
10052 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
10054 * doc/extend.texi: Update the built-in documentation file for the
10055 new built-in functions.
10057 2017-06-29 Richard Biener <rguenther@suse.de>
10059 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
10060 reduction chains to LOOP_VINFO_REDUCTIONS.
10061 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
10062 SLP reductions after processing reduction chains.
10064 2017-06-29 Nathan Sidwell <nathan@acm.org>
10066 * builtins.c (fold_builtin_FUNCTION): Use
10067 lang_hooks.decl_printable_name.
10069 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
10071 PR middle-end/81194
10072 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
10073 with only one label.
10074 * stmt.c (expand_case): Assert NCASES is greater than one.
10076 2017-06-29 Richard Biener <rguenther@suse.de>
10078 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
10080 (group_case_labels): Likewise.
10081 (find_taken_edge): Push sanity checking on val to workers...
10082 (find_taken_edge_cond_expr): ... here
10083 (find_taken_edge_switch_expr): ... and here, handle cases
10084 with just a default label.
10085 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
10086 (group_case_labels): Likewise.
10087 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
10088 group_case_labels does anything cleanup the CFG again.
10090 2017-06-29 Bin Cheng <bin.cheng@arm.com>
10092 PR tree-optimization/81196
10093 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
10094 exit condition comparing two IVs.
10096 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
10098 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
10099 profile to the dummy entry at the end of the list of architectures.
10100 * config/arm/arm-cpu-cdata.h: Regenerated.
10102 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10103 Michael Collison <michael.collison@arm.com>
10106 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
10108 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
10109 (*aarch64_reg_<mode>3_minus_mask): New pattern.
10110 (*aarch64_<optab>_reg_di3_mask2): New pattern.
10111 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
10112 of shift when the shift amount is masked with constant equal to
10113 the size of the mode.
10114 * config/aarch64/predicates.md (subreg_lowpart_operator): New
10117 2017-06-29 Martin Liska <mliska@suse.cz>
10119 * config/i386/i386.opt: Change range from [1,5] to [0,5].
10121 2017-06-29 Yury Gribov <tetra2005@gmail.com>
10124 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
10127 (edge_growth_cache_entry::edge_growth_cache_entry): New
10129 (reset_edge_growth_cache): Update to use constructor.
10131 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10133 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
10134 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10135 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10137 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
10139 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
10140 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
10142 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
10144 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
10145 (*-linux-uclibc*): Add t-uclibc tmake_file.
10146 * config/t-musl: New.
10147 * config/t-uclibc: New.
10149 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
10151 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
10153 (gen_comm_data): Emit architectural setting of arch_prof.
10154 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
10156 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
10157 (armv8-m.base, armv8-m.main): Likewise.
10158 * arm-protos.h (arm_build_target): Add profile field.
10159 (arch_option): Likewise.
10160 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
10162 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
10163 arm_active_target.profile.
10165 2017-06-28 Richard Biener <rguenther@suse.de>
10167 PR middle-end/81227
10168 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
10169 TYPE_OVERFLOW_WRAPS.
10170 * match.pd (negate_expr_p): Likewise.
10171 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
10172 fold_build2, not fold_binary.
10174 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10176 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
10177 Convert memory address to Pmode.
10178 (aarch64_print_operand): Assert MEM operands are always Pmode.
10180 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10183 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
10184 Remove redundant if.
10185 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
10186 * config/arm/aarch-common-protos.h
10187 (aarch_forward_to_shift_is_not_shifted_re): Remove.
10188 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
10190 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
10193 * multiple_target.c (create_dispatcher_calls): Set the default
10194 clone to be static, not public.
10196 2017-06-28 Richard Biener <rguenther@suse.de>
10198 * tree-vect-loop.c (vectorizable_reduction): Move special
10199 cond reduction IV var creation ...
10200 (vect_create_epilog_for_reduction): ... here. Remove induction_index
10201 parameter. Use STMT_VINFO_VECTYPE.
10202 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
10205 2017-06-28 Martin Liska <mliska@suse.cz>
10208 * ipa-visibility.c (non_local_p): Handle visibility.
10210 2017-06-28 Martin Liska <mliska@suse.cz>
10213 * common.opt: Add IntegerRange to various options.
10214 * opt-functions.awk (integer_range_info): New function.
10215 * optc-gen.awk: Add integer_range_info to cl_options struct.
10216 * opts-common.c (decode_cmdline_option): Handle
10217 CL_ERR_INT_RANGE_ARG.
10218 (cmdline_handle_error): Likewise.
10219 * opts.c (print_filtered_help): Show valid interval in
10220 when --help is provided.
10221 * opts.h (struct cl_option): Add range_min and range_max fields.
10222 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
10224 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
10226 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
10227 (x * C EQ/NE y * C): New transformation.
10229 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
10231 * genmultilib (combination_space): Accept '+' in option names.
10233 2017-06-28 Martin Liska <mliska@suse.cz>
10236 * asan.c (instrument_derefs): Bail out inner references
10237 that are hard register variables.
10239 2017-06-28 Jakub Jelinek <jakub@redhat.com>
10242 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
10243 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
10245 2017-06-28 Richard Biener <rguenther@suse.de>
10247 * tree-vectorizer.h (vect_get_vec_defs): Remove.
10248 (vect_get_slp_defs): Adjust.
10249 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
10251 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
10253 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10254 get_initial_defs_for_reduction instead of vect_get_vec_defs.
10255 (vectorizable_reduction): Adjust.
10256 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
10258 (vect_get_slp_defs): Likewise.
10259 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
10260 (vectorizable_bswap): Adjust.
10261 (vectorizable_call): Likewise.
10262 (vectorizable_conversion): Likewise.
10263 (vectorizable_assignment): Likewise.
10264 (vectorizable_shift): Likewise.
10265 (vectorizable_operation): Likewise.
10266 (vectorizable_store): Likewise.
10267 (vectorizable_condition): Likewise.
10268 (vectorizable_comparison): Likewise.
10270 2017-06-28 Michael Collison <michael.collison@arm.com>
10273 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
10275 (arm_gen_movmemqi): Removed unused variable 'i'.
10276 Convert 'for' loop into 'while' loop.
10277 (arm_expand_prologue): Remove last unnecessary set of insn.
10278 (thumb_pop): Remove unused variable 'pushed_words'.
10279 (thumb_exit): Remove last unnecessary set of regs_to_pop.
10281 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10283 * config/s390/predicates.md: Use s390_rel_address_ok_p.
10284 * config/s390/s390-protos.h: Add prototype of
10285 s390_rel_address_ok_p.
10286 * config/s390/s390.c (s390_got_symbol): New function.
10287 (s390_rel_address_ok_p): New function.
10288 (legitimize_pic_address): Use s390_rel_address_ok_p.
10289 (s390_load_got): Use s390_got_symbol.
10290 (s390_option_override): Issue error if
10291 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
10292 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
10294 * config/s390/s390.opt: New option mpic-data-is-text-relative.
10296 2017-06-27 Andrew Pinski <apinski@cavium.com>
10298 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
10299 (X * copysign (1.0, X)): New pattern.
10300 (X * copysign (1.0, -X)): New pattern.
10301 (copysign (-1.0, CST)): New pattern.
10303 2017-06-27 Joseph Myers <joseph@codesourcery.com>
10305 * genmultilib (combination_space): Remove variable.
10306 Validate reuse rules against regular expression for any sequence
10307 of multilib options in any order.
10309 2017-06-27 Michael Collison <michael.collison@arm.com>
10311 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
10312 call aarch64_split_simd_combine.
10313 * (aarch64_combine_internal<mode>): Delete pattern.
10314 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
10315 Allow register and subreg operands.
10317 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10319 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
10320 specific need, just fallback on defaults.
10321 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
10323 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10324 Olivier Hainque <hainque@adacore.com>
10326 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
10328 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
10329 targets. Pick a default if no particular attempt applied.
10330 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
10333 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10335 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
10336 (x86_64-wrs-vxworks7): Likewise.
10338 2017-06-27 Marek Polacek <polacek@redhat.com>
10341 * ubsan.c (instrument_null): Check get_base_address's result for null.
10343 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10345 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
10347 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10349 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
10350 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
10351 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
10352 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
10353 New function types.
10354 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
10355 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
10356 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
10357 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
10358 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
10359 BUILT_IN_FEUPDATEENV): New builtins.
10360 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
10361 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
10362 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
10363 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
10365 (builtin_structptr_types): Adjust size.
10366 * tree.c (builtin_structptr_types): Add four entries.
10368 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10369 Olivier Hainque <hainque@adacore.com>
10371 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
10372 (TLS_SYM): New local macro, forcing reference to __tls__ on
10373 link command lines for VxWorks 7 RTPs, triggering initialization
10375 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
10376 OS features TLS support, true for RTPs on VxWorks 7.
10377 * config/vxworks.c (vxworks_override_options): Setup emutls
10380 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10382 * predict.c (test_prediction_value_range): Use -1U instead of -1
10383 to avoid narrowing conversion warning.
10384 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
10385 to avoid narrowing conversion warning.
10386 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
10388 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
10390 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10392 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
10393 64bit configurations.
10394 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
10395 (SIZE_TYPE): Likewise.
10396 * config/vxworks.c (vxworks_emutls_var_fields): Use
10397 long_unsigned_type_node instead of unsigned_type_node as the offset
10398 field type, which is "pointer" mode in emutls.c.
10400 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10403 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
10405 PR middle-end/81207
10406 * gimple-fold.c (replace_call_with_call_and_fold): Handle
10407 gimple_vuse copying separately from gimple_vdef copying.
10409 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10411 * value-prof.c (free_hist): Remove call to memset and the enclosing if
10414 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10415 Olivier Hainque <hainque@adacore.com>
10417 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
10418 for all vxworks7 targets.
10419 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
10420 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
10421 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
10422 variations for VX6/VX7 and 32/64bits later on in ...
10423 (VXWORKS_LIB_SPEC): Leverage new macros.
10424 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
10425 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
10427 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10429 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
10430 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
10432 2017-06-26 Carl Love <cel@us.ibm.com>
10434 * config/rs6000/rs6000-c.c: Add support for built-in functions
10435 vector bool char vec_reve (vector bool char);
10436 vector signed char vec_reve (vector signed char);
10437 vector unsigned char vec_reve (vector unsigned char);
10438 vector bool int vec_reve (vector bool int);
10439 vector signed int vec_reve (vector signed int);
10440 vector unsigned int vec_reve (vector unsigned int);
10441 vector bool long long vec_reve (vector bool long long);
10442 vector signed long long vec_reve (vector signed long long);
10443 vector unsigned long long vec_reve (vector unsigned long long);
10444 vector bool short vec_reve (vector bool short);
10445 vector signed short vec_reve (vector signed short);
10446 vector double vec_reve (vector double);
10447 vector float vec_reve (vector float);
10448 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
10449 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
10450 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
10451 (altivec_vreve): New pattern.
10452 * config/rs6000/altivec.h (vec_reve): New define.
10453 * doc/extend.texi (vec_rev): Update the built-in documentation file
10454 for the new built-in functions.
10456 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10458 PR tree-optimization/71815
10459 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
10461 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
10463 (slsr_process_phi): Likewise.
10464 (replace_uncond_cands_and_profitable_phis): Don't replace a
10465 multiply candidate with a stride of 1 (copy or cast).
10466 (phi_incr_cost): Call uses_consumed_by_stmt rather than
10468 (lowest_cost_path): Likewise.
10469 (total_savings): Likewise.
10471 2017-06-26 Richard Biener <rguenther@suse.de>
10474 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
10475 Use def_builtin_pure for all gather builtins.
10477 2017-06-26 Richard Biener <rguenther@suse.de>
10479 PR tree-optimization/81203
10480 * tree-tailcall.c (find_tail_calls): Do not move stmts into
10481 non-dominating BBs.
10483 2017-06-26 Marek Polacek <polacek@redhat.com>
10486 * doc/invoke.texi: Document -Wmultistatement-macros.
10488 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
10490 * doc/sourcebuild.texi (ARM-specific attributes): Document new
10491 arm_neon_ok_no_float_abi effective target.
10493 2017-06-26 Richard Biener <rguenther@suse.de>
10495 PR tree-optimization/80928
10496 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
10497 (copy_bbs): Set BB_DUPLICATED flag early.
10498 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
10500 (execute_on_shrinking_pred): Likewise.
10501 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
10502 BB_DUPLICATED blocks.
10503 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
10504 iterate over all PHIs considering removal of *gsi.
10506 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
10508 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
10511 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10513 * config/rs6000/rs6000-string.c: (expand_block_clear,
10514 do_load_for_compare, select_block_compare_mode,
10515 compute_current_alignment, expand_block_compare,
10516 expand_strncmp_align_check, expand_strn_compare,
10517 expand_block_move, rs6000_output_load_multiple)
10518 Move functions related to string/block move/compare
10519 to a separate file.
10520 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
10521 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
10522 for this function which is now used in two files.
10523 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
10524 * config.gcc: Add rs6000-string.o to extra_objs for
10525 targets powerpc*-*-* and rs6000*-*-*.
10527 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
10530 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
10531 32-bit, since indexed is not valid for DImode.
10532 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
10533 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
10534 (define_peephole2 for Altivec d-form load): Add 32-bit support.
10535 (define_peephole2 for Altivec d-form store): Likewise.
10538 * multiple_target.c (create_dispatcher_calls): Only create the
10539 dispatcher call if the function is the default clone of a
10540 versioned function.
10542 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
10544 PR middle-end/80902
10545 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
10546 a call, force the call to not be a tail call.
10548 2017-06-23 Jeff Law <law@redhat.com>
10550 * doc/contrib.texi: Add entry for Steven Pemberton's work on
10553 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
10555 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
10556 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
10557 handling for early expansion of vector shifts (sl,sr,sra,rl).
10558 (builtin_function_type): Add vector shift right instructions
10559 to the unsigned argument list.
10561 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
10563 rtl-optimizatoin/79286
10564 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
10565 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
10566 trap. PIC register plus a const unspec without offset can never trap.
10568 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
10570 * tree.h (builtin_structptr_type): New type.
10571 (builtin_structptr_types): Declare new array.
10572 * tree.c (builtin_structptr_types): New array.
10573 (free_lang_data, build_common_tree_nodes): Use it.
10575 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
10578 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
10581 2017-06-22 Matt Turner <mattst88@gmail.com>
10583 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
10584 Lake models to skylake case. Assume skylake for unknown
10585 models with clflushopt.
10587 2017-06-22 Jeff Law <law@redhat.com>
10589 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
10590 frame sizes that do not satisfy aarch64_uimm12_shift.
10592 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
10594 * profile-count.h (apply_probability,
10595 apply_scale, probability_in): Fix checks for zero.
10597 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10599 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
10600 * doc/cppdiropts.texi (-I @var{dir}): Document it.
10602 2016-06-22 Richard Biener <rguenther@suse.de>
10604 * tree-vect-loop.c (vect_model_reduction_cost): Handle
10605 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
10606 REDUC_MAX_EXPR support.
10607 (vectorizable_reduction): Likewise.
10608 (vect_create_epilog_for_reduction): Likewise.
10610 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
10612 * match.pd (A / (1 << B) -> A >> B): New.
10613 * generic-match-head.c: Include optabs-tree.h.
10614 * gimple-match-head.c: Likewise.
10615 * optabs-tree.h (target_supports_op_p): New.
10616 * optabs-tree.c (target_supports_op_p): New.
10618 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10620 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
10621 $gcc_cv_ld --help output.
10622 (gcc_cv_ld_demangle): Likewise.
10623 (gcc_cv_ld_eh_frame_hdr): Likewise.
10624 (gcc_cv_ld_pie): Likewise.
10625 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
10626 (gcc_cv_ld_buildid): Likewise.
10627 (gcc_cv_ld_sysroot): Likewise.
10628 (ld_bndplt_support): Likewise.
10629 (ld_pushpopstate_support): Likewise.
10630 * configure: Regenerate.
10631 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
10633 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10636 * config/i386/sse.md (round<mode>2): Renumber match_dup and
10637 operands indexes to avoid gap between operands and match_dups.
10639 2017-06-21 Andrew Pinski <apinski@cavium.com>
10641 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
10642 Increment Arith_shift and Arith_shift_reg by 1.
10643 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
10645 * config/aarch64/aarch64.c (thunderx_tunings): Enable
10646 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
10647 (aarch64_strip_extend): Add new argument and test for it.
10648 (aarch64_cheap_mult_shift_p): New function.
10649 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
10650 add a cost if it is true.
10651 Update calls to aarch64_strip_extend.
10652 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
10654 2017-06-21 Andrew Pinski <apinski@cavium.com>
10656 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
10658 (thunderxt88): Likewise.
10659 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
10660 (thunderx_prefetch_tune): New variable.
10661 (thunderx2t99_prefetch_tune): Update for the correct values.
10662 (thunderxt88_tunings): New variable.
10663 (thunderx_tunings): Use thunderx_prefetch_tune instead of
10664 generic_prefetch_tune.
10665 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
10667 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10669 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
10670 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
10671 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
10672 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
10673 (aarch64_atomic_cas<mode>, GPI): Likewise.
10675 2017-06-21 Martin Liska <mliska@suse.cz>
10677 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
10678 statements on cold and hot labels.
10679 * predict.c (tree_estimate_probability_bb): Remove the
10680 prediction from this place.
10682 2017-06-21 Martin Liska <mliska@suse.cz>
10684 PR tree-optimization/79489
10685 * gimplify.c (maybe_add_early_return_predict_stmt): New
10687 (gimplify_return_expr): Call the function.
10688 * predict.c (tree_estimate_probability_bb): Remove handling
10690 * predict.def: Update comment about early return predictor.
10691 * gimple-predict.h (is_gimple_predict): New function.
10692 * predict.def: Change default value of early return to 66.
10693 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
10695 * passes.def: Put pass_strip_predict_hints to the beginning of
10698 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
10700 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
10701 FUNCTION_DECL declarations.
10702 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
10704 (dwaf2out_decl): Likewise.
10705 * godump.c (go_early_global_decl): Skip call to the real debug hook
10706 for FUNCTION_DECL declarations.
10707 * passes.c (rest_of_decl_compilation): Skip call to the
10708 early_global_decl debug hook for FUNCTION_DECL declarations, unless
10709 -fdump-go-spec is passed.
10711 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10713 * config/i386/i386.c (struct builtin_isa): New field pure_p.
10714 Reorder for compactness.
10715 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
10716 (def_builtin_pure, def_builtin_pure2): New functions.
10717 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
10719 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10721 * match.pd (nop_convert): New predicate.
10722 ((A +- CST1) +- CST2): Allow some NOP conversions.
10724 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10727 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
10728 with ctors/dtors if GOVD_SHARED is set.
10730 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10732 * config/aarch64/aarch64.md (movti_aarch64):
10733 Emit mov rather than orr.
10734 (movtf_aarch64): Likewise.
10735 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
10736 Emit mov rather than orr.
10738 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10740 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
10741 Swap alternatives, make integer dup more expensive.
10743 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10745 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
10746 Return true for non-tls symbols.
10748 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
10750 * config/aarch64/aarch64-cores.def (cortex-a55): New.
10751 (cortex-a75): Likewise.
10752 (cortex-a75.cortex-a55): Likewise.
10753 * config/aarch64/aarch64-tune.md: Regenerate.
10754 * doc/invoke.texi (-mtune): Document new values for -mtune.
10756 2017-06-21 Tom de Vries <tom@codesourcery.com>
10758 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
10759 stack_size feature.
10760 (Effective-Target Keywords, Other attributes): Suggest using
10761 dg-add-options stack_size feature to get stack limit in stack_size
10762 effective target documentation.
10764 2017-06-21 Julian Brown <julian@codesourcery.com>
10765 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10767 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
10768 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
10769 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
10771 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
10772 attribute type list for neon_multiply.
10773 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
10774 attribute type list for neon_multiply.
10775 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
10776 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
10777 attribute type list for neon_multiply.
10778 * config/arm/types.md (crypto_pmull): Add.
10779 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
10780 attribute type list.
10782 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
10784 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
10787 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10789 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
10790 to make sure not to dereference a NULL cost_classes_ptr pointer.
10792 2017-06-20 Carl Love <cel@us.ibm.com>
10794 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10795 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10796 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10797 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10798 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10799 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10800 VMULOSW): New enum "unspec" values.
10801 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10802 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10803 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10804 altivec_vmulosw): New patterns.
10805 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10806 VMULOSW): Add definitions.
10808 2017-06-20 Julia Koval <julia.koval@intel.com>
10810 * config/i386/i386.c: Fix rounding expand for new pattern.
10811 * config/i386/subst.md: Fix pattern (parallel -> unspec).
10813 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10815 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
10816 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
10818 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10820 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
10823 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10825 * config/aarch64/aarch64-cores.def: Rearrange to sort by
10826 architecture, then by implementer ID.
10827 * config/aarch64/aarch64-tune.md: Regenerate.
10829 2017-06-20 Richard Biener <rguenther@suse.de>
10831 PR middle-end/81097
10832 * fold-const.c (split_tree): Fold to type before negating.
10834 2017-06-20 David Malcolm <dmalcolm@redhat.com>
10836 * diagnostic-show-locus.c
10837 (selftest::test_fixit_deletion_affecting_newline): New function.
10838 (selftest::diagnostic_show_locus_c_tests): Call it.
10840 2017-06-20 Andreas Schwab <schwab@suse.de>
10843 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
10846 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10848 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
10849 __ARM_FEATURE_COPROC according to support.
10851 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10853 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10854 Rewritten to avoid overflow for > 32-bit pointers.
10857 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
10858 by removing enum keyword.
10859 (ubsan_type_descriptor): Likewise. Formatting fix.
10862 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
10863 splitter): Require TARGET_SSE2 in the condition.
10865 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
10868 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
10869 for doing vector set of SFmode on ISA 3.0.
10870 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
10871 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
10873 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
10874 SFmode value into a V4SF variable that was extracted from another
10875 V4SF variable without converting the element to double precision
10876 and back to single precision vector format.
10877 (vsx_insert_extract_v4sf_p9_2): Likewise.
10879 2017-06-19 Jakub Jelinek <jakub@redhat.com>
10881 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
10882 in UWHI to avoid undefined overflow.
10885 * ubsan.h (enum ubsan_encode_value_phase): New.
10886 (ubsan_encode_value): Change second argument to
10887 enum ubsan_encode_value_phase with default value of
10888 UBSAN_ENCODE_VALUE_GENERIC.
10889 * ubsan.c (ubsan_encode_value): Change second argument to
10890 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
10891 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
10892 create_tmp_var_raw instead of create_tmp_var and use a
10894 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
10895 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
10896 ubsan_encode_value callers.
10899 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
10900 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
10901 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
10903 2017-06-19 Richard Biener <rguenther@suse.de>
10905 PR middle-end/81118
10906 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
10907 estimates if we changed anything.
10909 2017-06-19 Richard Biener <rguenther@suse.de>
10911 PR tree-optimization/80887
10912 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
10913 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
10914 simplified lookups, then reset mprts_hook.
10915 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
10917 (try_to_simplify): Likewise.
10919 2017-06-19 Martin Liska <mliska@suse.cz>
10922 * gimplify.c (gimplify_switch_expr):
10923 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
10925 2017-06-19 Martin Liska <mliska@suse.cz>
10927 * doc/install.texi: Document that PGO runs in 4 stages.
10929 2017-06-19 Martin Liska <mliska@suse.cz>
10932 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
10933 to dispatcher function name.
10934 * multiple_target.c (replace_function_decl): New function.
10935 (create_dispatcher_calls): Redirect both edges and references.
10937 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
10939 * profile-count.c (profile_count::dump): Dump quality.
10940 (profile_count::differs_from_p): Update for unsigned val.
10941 * profile-count.h (profile_count_quality): New enum.
10942 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
10944 2017-06-19 Richard Biener <rguenther@suse.de>
10946 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
10947 struct function as arg.
10948 (estimate_numbers_of_iterations): Export overload with loop arg.
10949 (free_numbers_of_iterations_estimates_loop): Use an overload of
10950 free_numbers_of_iterations_estimates instead.
10951 * tree-cfg.c (remove_bb): Adjust.
10952 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
10953 * tree-parloops.c (gen_parallel_loop): Likewise.
10954 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
10956 (tree_unroll_loops_completely): Likewise.
10957 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
10958 Use an overload instead and export.
10959 (estimated_loop_iterations): Adjust.
10960 (max_loop_iterations): Likewise.
10961 (likely_max_loop_iterations): Likewise.
10962 (estimate_numbers_of_iterations): Take struct function as arg
10964 (loop_exits_before_overflow): Adjust.
10965 (free_numbers_of_iterations_estimates_loop): Use an overload.
10966 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
10967 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
10969 2017-06-19 Richard Biener <rguenther@suse.de>
10972 * ipa-prop.c (find_constructor_constant_at_offset): Handle
10973 RANGE_EXPR conservatively.
10975 2017-06-16 Carl Love <cel@us.ibm.com>
10977 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10978 definitions for vec_float, vec_float2, vec_floato,
10979 vec_floate built-ins.
10980 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
10981 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
10983 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
10984 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
10985 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
10986 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
10987 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
10988 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
10989 vec_floato): Add builtin defines.
10990 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
10991 Update the built-in documentation file for the new built-in
10994 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10996 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
10997 (mthumb): Mark as the negative of -marm.
10999 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11001 * doc/invoke.texi (ARM Options, -mcpu): Document supported
11003 (ARM Options, -mtune): Document that this accepts the same
11004 extension options as -mcpu.
11005 (ARM Options, -mfpu): Document addition of -mfpu=auto.
11007 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11009 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
11010 permitted extensions.
11012 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11014 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
11015 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
11016 (armv8-m.main): Add option +nodsp.
11017 * config/arm/arm-cpu-cdata.h: Regenerated.
11019 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11021 * config/arm/t-fuchsia: New file.
11022 * config.gcc (arm*-*-fuchsia*): Use it.
11024 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11026 * config/arm/t-symbian: Rewrite for new option infrastructure.
11028 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11030 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
11031 (MULTILIB_REQUIRED): Likewise.
11033 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11035 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
11036 (MULTILIB_RESUE): Likewise.
11037 (MULTILIB_MATCHES): Likewise.
11038 (MULTLIB_REQUIRED): Likewise.
11040 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11042 * config/arm/t-rtems: Rewrite for new option framework.
11044 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11046 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
11047 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
11048 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
11049 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
11050 * config/arm/t-multilib: ... here.
11051 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
11052 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
11053 armv7-a and armv8*-a when A-profile libraries have not been built.
11054 * config/arm/t-rmprofile: Rewrite.
11056 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11058 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
11059 with a backslash. Remove the backslash after substituting unescaped
11061 * doc/fragments.texi (MULTILIB_REUSE): Document it.
11063 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11065 * config.gcc: (arm*-*-*): When building a-profile libraries, force
11066 the driver to pass through the default setting of -mfloat-abi.
11067 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
11069 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
11070 (all_feat_combs): New rule.
11071 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
11073 * config/arm/t-aprofile: Rewrite.
11075 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11077 * config/arm/arm.h (FPUTYPE_AUTO): Define.
11078 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
11079 fpu is not specified by the user/command-line.
11080 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
11081 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
11082 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
11083 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
11084 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
11085 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
11087 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11089 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
11090 * config/arm/t-arm-elf: Rewritten.
11092 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11094 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
11095 have some floating-point instructions.
11096 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
11097 (TARGET_MAYBE_HARD_FLOAT): New macro.
11098 * config/arm/arm-builtins.c (arm_init_builtins): Use
11099 TARGET_MAYBE_HARD_FLOAT.
11100 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
11102 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11104 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
11105 (configargs.h): Include it.
11106 (arm_print_hint_for_fpu_option): New function.
11107 (arm_parse_fpu_option): New function.
11108 (candidate_extension): New class.
11109 (arm_canon_for_multilib): New function.
11110 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
11111 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
11112 (ARCH_CANONICAL_SPECS): New macro.
11113 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
11115 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11117 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
11118 are set after handling multilib fragments. Set target_cpu_default2
11121 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11123 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
11125 (arm*-*-*): Set target_cpu_default2 to a quoted string.
11126 * config/arm/parsecpu.awk (check_cpu): Validate any extension
11128 (check_arch): Likewise.
11129 * config/arm/arm.c (arm_configure_build_target): Handle
11130 TARGET_CPU_DEFAULT being a string constant. Scan any feature
11131 options in the default.
11133 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11135 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
11136 when an option is an alias of another.
11137 * config/arm/parsecpu.awk (optalias): New parser token.
11138 (gen_comm_data): Mark non-alias options as such. Emit entries
11139 for extension aliases.
11140 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
11141 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
11142 (armv6kz, armv6zk, armv6t2): Likewise.
11143 (armv7): Make vfpv3-d16 an alias.
11144 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
11146 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
11147 Sort in canonical order.
11148 (armv8-a): Sort in canonical order.
11149 (armv8.1-a, armv8.2-a): Likewise.
11150 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
11152 (cortex-a9): Sort in canonical order.
11153 * config/arm/arm.c (selftests.h): Include it.
11154 (arm_test_cpu_arch_data): New function.
11155 (arm_run_self_tests): New function.
11156 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
11157 (targetm): Move declaration to the end of the file.
11158 * arm-cpu-cdata.h: Regenerated.
11160 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11162 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
11163 call to target_mode_check describing the type of option passed.
11164 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
11165 (arm_target_thumb_only): Use arm_parse_arch_option_name or
11166 arm_parse_cpu_option_name to match parameters against list of
11168 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
11169 arm_arch_core_flags data structure.
11170 * config/arm/arm-cpu_cdata.h: Regenerated.
11172 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11174 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
11176 (arm_print_hint_for_cpu_option): Likewise.
11177 (arm_print_hint_for_arch_option): Likewise.
11178 (arm_parse_cpu_option_name): Likewise.
11179 (arm_parse_arch_option_name): Likewise.
11180 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
11181 of entries in the all_fpus list.
11182 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
11183 (arm_parse_cpu_option_name): Declare.
11184 (arm_parse_arch_option_name): Declare.
11185 (arm_parse_option_features): Declare.
11186 (arm_intialize_isa): Declare.
11187 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
11189 (gen_comm_data): ... here. Make definitions non-static.
11190 * config/arm/arm-cpu-data.h: Regenerated.
11191 * config/arm/arm-cpu-cdata.h: Regenerated.
11193 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11195 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
11196 (cpu_arch_extension): New structure.
11197 (cpu_arch_option, arch_option, cpu_option): New structures.
11198 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
11199 architecture types.
11200 (gen_data): Generate new format data tables.
11201 * config/arm/arm.c (cpu_tune): New structure.
11202 (cpu_option, processors): Delete.
11203 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
11204 (arm_print_hint_for_cpu_option): ... this and ...
11205 (arm_print_hint_for_arch_option): ... this.
11206 (arm_parse_arch_cpu_name): Delete. Replace with ...
11207 (arm_parse_cpu_option_name): ... this and ...
11208 (arm_parse_arch_option_name): ... this.
11209 (arm_unrecognized_feature): Change type of target parameter to
11211 (arm_parse_arch_cpu_features): Delete. Replace with ...
11212 (arm_parse_option_features): ... this.
11213 (arm_configure_build_target): Rework to use new configuration data
11215 (arm_print_tune_info): Rework for new configuration data tables.
11216 * config/arm/arm-cpu-data.h: Regenerated.
11217 * config/arm/arm-cpu.h: Regenerated.
11219 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11221 * Makefile.in (OBJS): Move sbitmap.o from here ...
11222 (OBJS-libcommon): ... to here.
11224 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11226 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
11227 (ISA_ALL_CRYPTO): New macro.
11228 (ISA_ALL_SIMD): New macro
11229 (ISA_ALL_FP): New macro.
11230 * config/arm/arm.c (fpu_bitlist): Update initializer.
11231 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
11233 (arm9e): Add fpu. Add option for nofp
11234 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
11235 (arm926ej-s, arm1026ej-s): Likewise.
11236 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
11237 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
11238 neon-fp16, neon-vfpv4, nofp and nosimd.
11239 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
11240 (cortex-a8): Add fpu. Add option for nofp.
11241 (cortex-a9): Add fpu. Add options for nosimd and nofp.
11242 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
11243 (cortex-r4f): Add fpu.
11244 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
11245 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
11247 (cortex-r8): Likewise.
11248 (cortex-m4): Add fpu. Add option for nofp.
11249 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
11250 (cortex-a17.cortex-a7): Likewise.
11251 (cortex-a32): Add fpu. Add options for crypto and nofp.
11252 (cortex-a35, cortex-a53): Likewise.
11253 (cortex-a57): Add fpu. Add option for crypto.
11254 (cortex-a72, cortex-a73): Likewise.
11255 (exynos-m1): Likewise.
11256 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
11257 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
11258 (cortex-m33): Add fpu. Add option for nofp.
11259 * config/arm/arm-cpu-cdata.h: Regenerated
11260 * config/arm/arm-cpu-data.h: Regenerated.
11262 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11264 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
11265 (armv5te, armv5tej): Likewise.
11266 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
11267 (armv7): Add options fp and vfpv3-d16.
11268 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
11269 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
11271 (armv7ve): Likewise.
11272 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
11273 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
11274 (armv8-a): Add nocrypto option.
11275 (armv8.1-a, armv8.2-a): Likewise.
11276 (armv8-m.main): add options fp, fp.dp and nofp.
11278 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11280 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
11282 (armv8-a+crc): Delete.
11283 (armv8.1-a): Add options simd, crypto and nofp.
11284 (armv8.2-a): Add options fp16, simd, crypto and nofp.
11285 (armv8.2-a+fp16): Delete.
11286 (armv8-m.main): Add option dsp.
11287 (armv8-m.main+dsp): Delete.
11288 (cortex-a8): Add fpu. Add nofp option.
11289 (cortex-a9): Add fpu. Add nofp and nosimd options.
11290 * config/arm/parsecpu.awk (gen_data): Generate option tables and
11291 link to main cpu and architecture data structures.
11292 (gen_comm_data): Only put isa attributes from the main architecture
11294 (option): New statement for architecture and CPU entries.
11295 * arm.c (struct cpu_option): New structure.
11296 (struct processors): Add entry for options.
11297 (arm_unrecognized_feature): New function.
11298 (arm_parse_arch_cpu_name): Ignore any characters after the first
11300 (arm_parse_arch_cpu_feature): New function.
11301 (arm_configure_build_target): Separate out any CPU and architecture
11302 features and parse separately. Don't error out if -mfpu=auto is
11303 used with only an architecture string.
11304 (arm_print_asm_arch_directives): New function.
11305 (arm_file_start): Call it.
11306 * config/arm/arm-cpu-cdata.h: Regenerated.
11307 * config/arm/arm-cpu-data.h: Likewise.
11308 * config/arm/arm-tables.opt: Likewise.
11310 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11312 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
11313 assembler when it is not -mfpu=auto.
11315 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11317 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
11318 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
11319 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
11320 (ASM_CPU_SPEC): Rewrite.
11321 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
11322 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
11323 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
11324 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
11325 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
11326 copied string is NUL-terminated. Also strip any characters prefixed
11328 (arm_rewrite_selected_arch): New function.
11329 (arm_rewrite_march): New function.
11331 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11333 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
11334 (x_arm_cpu_string, x_arm_tune_string): Likewise.
11335 (march, mcpu, mtune): Convert to string-based options.
11336 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
11337 (arm_parse_arch_cpu_name): New function.
11338 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
11339 identify selected architecture or CPU.
11340 (arm_option_save): New function.
11341 (TARGET_OPTION_SAVE): Redefine.
11342 (arm_option_restore): Restore string options.
11343 (arm_option_print): Print string options.
11345 2017-06-16 Martin Sebor <msebor@redhat.com>
11347 PR tree-optimization/80933
11348 PR tree-optimization/80934
11349 * builtins.c (fold_builtin_3): Do not handle bcmp here.
11350 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
11351 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
11352 (gimple_fold_builtin): Call them.
11354 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11356 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
11357 as unlikely; update profile.
11359 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11361 * predict.c (force_edge_cold): Handle declaring edges impossible
11364 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11366 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
11368 (try_unroll_loop_completely): Fix reporting.
11370 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11372 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
11374 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
11377 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
11378 if given a non-constant argument for an intrinsic which requires a
11381 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11383 * profile.c (compare_freqs): New function.
11384 (branch_prob): Sort edge list.
11385 (find_spanning_tree): Assume that the list is priority sorted.
11387 2017-06-16 Richard Biener <rguenther@suse.de>
11389 PR tree-optimization/81090
11390 * passes.def (pass_record_bounds): Remove.
11391 * tree-pass.h (make_pass_record_bounds): Likewise.
11392 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
11393 make_pass_record_bounds): Likewise.
11394 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
11395 not free niter estimates at the beginning but at the end.
11396 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
11398 2017-06-16 Richard Biener <rguenther@suse.de>
11400 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
11401 initializer to workaround ICE in host GCC 4.8.
11403 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11405 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
11407 (clone_inlined_nodes): Update.
11409 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11411 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
11412 prefetch settings, and enable prefetching by default at -O3.
11414 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11416 * config/aarch64/aarch64.c (aarch64_override_options_internal):
11417 Set flag_prefetch_loop_arrays according to tuning data.
11419 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11421 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
11422 New tune structure.
11423 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
11424 [Unrelated to main purpose of the patch] Place the pointer field last
11425 to enable type checking errors when tune structure are wrongly merged.
11426 * config/aarch64/aarch64.c (generic_prefetch_tune,)
11427 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
11428 (thunderx2t99_prefetch_tune): New tune constants.
11429 (tune_params *_tunings): Update all tunings (no functional change).
11430 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
11431 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
11432 from tunings structures.
11434 2017-06-16 Jakub Jelinek <jakub@redhat.com>
11437 * ubsan.c (instrument_null): Add T argument, use it instead
11438 of computing it based on IS_LHS.
11439 (instrument_object_size): Likewise.
11440 (pass_ubsan::execute): Adjust instrument_null and
11441 instrument_object_size callers to pass gimple_get_lhs or
11442 gimple_assign_rhs1 result to it. Use instrument_null instead of
11443 calling get_base_address and instrument_mem_ref. Handle
11444 aggregate call arguments for object-size sanitization.
11446 2017-06-16 Yury Gribov <tetra2005@gmail.com>
11448 PR tree-optimization/81089
11449 * tree-vrp.c (is_masked_range_test): Validate operands of
11452 2017-06-15 Martin Sebor <msebor@redhat.com>
11455 * dumpfile.c (dump_register): Avoid calling memset to initialize
11456 a class with a default ctor.
11457 * gcc.c (struct compiler): Remove const qualification.
11458 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
11459 * hash-table.h: Ditto.
11460 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
11462 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
11463 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
11465 * params.h (struct param_info): Make struct members non-const.
11466 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
11467 with default initialization.
11468 * vec.h (vec_copy_construct, vec_default_construct): New helper
11470 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
11471 with vec_copy_construct.
11472 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
11473 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
11474 * doc/invoke.texi (-Wclass-memaccess): Document.
11476 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11478 * emit-rtl.h (is_leaf): Update comment about local
11479 register allocator.
11481 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11484 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
11485 for a variable to have a section before checking if the section has a
11487 Set section to.persistent if persistent attribute is set.
11488 Warn if .persistent attribute is used on an automatic variable.
11490 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
11492 PR rtl-optimization/80474
11493 * reorg.c (update_block): Do not ignore instructions in a delay slot.
11495 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
11497 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
11500 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
11502 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
11503 (casesi): Emit bounds checking as RTL.
11504 (casesi_internal_mips16_<mode>): Remove bounds checking.
11506 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
11508 * config/xtensa/xtensa.c (xtensa_option_override): Append
11509 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
11510 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
11511 xtensa_doloop_hooks): Define unconditionally.
11512 (xtensa_reorg_loops): Only call reorg_loops in the presence of
11514 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
11515 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
11516 for it in xtensa_option_override.
11517 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
11518 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
11520 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
11522 * doc/cppopts.texi: Document '-' special value to -MF.
11524 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
11526 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
11527 (cortex_a53_fconst): Likewise.
11528 (cortex_a53_fpmul): Likewise.
11529 (cortex_a53_f_load_64): Likewise.
11530 (cortex_a53_f_load_many): Likewise.
11531 (cortex_a53_advsimd_alu): Likewise.
11532 (cortex_a53_advsimd_alu_q): Likewise.
11533 (cortex_a53_advsimd_mul): Likewise.
11534 (cortex_a53_advsimd_mul_q): Likewise.
11535 (fpmac bypass): Add new bypass for fpmac-fpmac case.
11536 Add missing fmul, r2f_cvt and fconst cases.
11538 2017-06-14 Richard Biener <rguenther@suse.de>
11540 PR middle-end/81088
11541 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
11543 (fold_binary_loc): When associating do not treat pre-existing
11544 TREE_OVERFLOW on literal constants as a reason to allow
11545 TREE_OVERFLOW on associated literal constants.
11547 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11549 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
11550 (MASK_FEATURES): New macro.
11551 * config/sparc/sparc.c (sparc_option_override): Remove the special
11552 handling of -mfpu and generalize it to all MASK_FEATURES switches.
11554 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11556 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
11557 a division of 0 if non-call exceptions are enabled.
11559 2017-06-14 Andrew Pinski <apinski@cavium.com>
11560 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11563 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
11564 Improve vector initialization code gen for only variable case.
11566 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11568 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
11570 2017-06-14 Richard Biener <rguenther@suse.de>
11572 PR tree-optimization/81083
11573 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
11576 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11578 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
11579 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
11580 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
11581 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
11582 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
11583 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
11585 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11587 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
11588 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
11590 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11592 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
11594 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11596 * config/rs6000/t-rtems: Don't handle SPE.
11598 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11600 * config/rs6000/t-linux: Don't handle SPE.
11602 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11604 * config/rs6000/eabispe.h: Delete file.
11606 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11608 * config/rs6000/t-spe: Delete file.
11610 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11612 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
11613 (rs6000_legitimate_offset_address_p): Return false for anything in
11614 V2SImode or V2SFmode.
11616 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11618 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
11619 except V2SF and V2SI. Rearrange the vector modes, and add comments.
11620 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
11622 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
11623 (rs6000_legitimate_offset_address_p): Ditto.
11624 (rs6000_emit_move): Ditto.
11625 (rs6000_init_builtins): Remove V4HI_type_node.
11627 2017-06-13 Martin Liska <mliska@suse.cz>
11630 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
11631 (gate_asan): Likewise.
11632 * asan.h (asan_no_sanitize_address_p): Remove the function.
11633 (sanitize_flags_p): New function.
11634 * builtins.def: Fix coding style.
11635 * common.opt: Use renamed enum value.
11636 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
11637 * doc/extend.texi: Document no_sanitize attribute.
11638 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
11639 to SANITIZE_UNDEFINED_NONDEFAULT.
11640 * gcc.c (sanitize_spec_function): Use the renamed enum value.
11641 * gimple-fold.c (optimize_atomic_compare_exchange_p):
11642 Use sanitize_flags_p.
11643 * gimplify.c (gimplify_function_tree): Likewise.
11644 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
11645 * opts.c (parse_no_sanitize_attribute): New function.
11646 (common_handle_option): Use renamed enum value.
11647 * opts.h (parse_no_sanitize_attribute): Declare.
11648 * tree.c (sanitize_flags_p): New function.
11649 * tree.h: Declared here.
11650 * tsan.c: Use sanitize_flags_p.
11651 * ubsan.c (ubsan_expand_null_ifn): Likewise.
11652 (instrument_mem_ref): Likewise.
11653 (instrument_bool_enum_load): Likewise.
11654 (do_ubsan_in_current_function): Remove the function.
11655 (pass_ubsan::execute): Use sanitize_flags_p.
11656 * ubsan.h: Remove do_ubsan_in_current_function
11657 * tree-cfg.c (print_no_sanitize_attr_value): New function.
11658 (dump_function_to_file): Use it here.
11660 2017-06-13 Martin Jambor <mjambor@suse.cz>
11662 PR tree-optimization/80803
11663 PR tree-optimization/81063
11664 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
11665 (propagate_subaccesses_across_link): Enqueue subtree whenever
11666 necessary instead of relying on the caller.
11668 2017-06-13 Martin Jambor <mjambor@suse.cz>
11670 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
11671 that have a first_link.
11672 (sort_and_splice_var_accesses): Do not check first_link before
11674 (subtree_mark_written_and_enqueue): Likewise.
11675 (propagate_all_subaccesses): Likewise and do not stop at first
11676 parent with a first_link.
11678 2017-06-13 Martin Jambor <mjambor@suse.cz>
11680 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
11683 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11685 * match.pd: New pattern.
11687 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11689 * tree-vrp.c (is_masked_range_test): New function.
11690 (register_edge_assert_for): Determine ranges for
11693 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11695 PR tree-optimization/67328
11696 * fold-const.c (maskable_range_p): New function.
11697 (build_range_check): Generate bittests if possible.
11699 2017-06-13 Martin Liska <mliska@suse.cz>
11701 * gimple-pretty-print.c (dump_probability): Add new argument.
11702 (dump_edge_probability): Dump both probability and count.
11703 (dump_gimple_label): Likewise.
11704 (dump_gimple_bb_header): Likewise.
11706 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
11709 * config/avr/avr-devices.c: Fix indentation.
11710 * config/avr/gen-avr-mmcu-specs.c: Dito.
11712 2017-06-13 Richard Biener <rguenther@suse.de>
11714 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
11715 instead get vector type from stmt_info.
11716 (vectorizable_reduction): Adjust. Remove dead code.
11718 2017-06-13 Richard Biener <rguenther@suse.de>
11720 PR middle-end/81065
11721 * fold-const.c (extract_muldiv_1): Remove bogus distribution
11722 case of C * (x * C2 + C3).
11723 (fold_addr_of_array_ref_difference): Properly fold index difference.
11725 2017-06-12 David S. Miller <davem@davemloft.net>
11728 * config/sparc/sparc.md (return expander): Emit frame blockage if
11729 function uses alloca.
11731 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
11733 * combine.c (make_field_assignment): Check len rather than the mode
11734 precision when calling force_to_mode.
11736 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
11738 Support multilibs and devices that see flash in RAM address range.
11741 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
11742 (avr_mcu_t) <flash_pm_offset>: New field.
11743 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
11744 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
11745 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
11746 (AVR_TINY_PM_OFFSET): Remove macro.
11747 * config/avr/avr.opt (-mshort-calls): New option.
11748 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11749 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
11750 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
11751 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
11752 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
11753 instead of avr_arch->have_jmp_call.
11754 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
11755 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
11756 avr_arch->flash_pm_offset to define.
11757 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
11758 new field flash_pm_offset. Add entry for avrxmega3.
11759 (avr_texinfo): Add entry for avrxmega3.
11760 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
11761 attiny212, attiny214,
11762 attiny412, attiny414, attiny416, attiny417,
11763 attiny814, attiny816, attiny817,
11764 attiny1614, attiny1616, attiny1617,
11765 attiny3214, attiny3216, attiny3217.
11766 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
11767 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
11768 (avr_print_operand_address) [AVR_TINY]: Same.
11769 (avr_asm_init_sections) <readonly_data_section>: Only patch
11770 callback if avr_arch->flash_pm_offset = 0.
11771 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
11772 for rodata if avr_arch->flash_pm_offset != 0.
11773 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
11774 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
11775 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
11776 (m_options): Append opt_rcall.
11777 (m_dirnames): Append dir_rcall.
11778 * config/avr/t-multilib: Regenerate.
11780 * configure.ac [target=avr]: Check whether avrxmega3 default
11781 linker description file works as needed.
11782 * configure: Regenerate.
11783 * doc/avr-mmcu.texi: Regenerate.
11784 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
11785 <__AVR_ARCH__>: Document avrxmega3 and 103.
11786 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
11787 <__AVR_SHORT_CALLS__>: Document it.
11788 <__AVR_PM_BASE_ADDRESS__>: Document it.
11789 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
11790 (AVR Variable Attributes) <progmem>: Document this is
11791 not needed for avrxmega3.
11792 (AVR Named Address Spaces) <__flash>: Dito.
11794 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
11796 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
11798 2017-06-12 Doug Rupp <rupp@adacore.com>
11800 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
11801 Append vxworks-stdint.h to the tm_file list.
11802 * config/vxworks-stdint.h: New file.
11804 2017-06-12 Martin Liska <mliska@suse.cz>
11806 PR tree-optimization/81041
11807 * tree-profile.c (gimple_gen_ic_func_profiler):
11808 Create an extra BB in profile-generate
11809 (gimple_gen_time_profiler): Likewise.
11811 2017-06-12 Jakub Jelinek <jakub@redhat.com>
11813 PR tree-optimization/81003
11814 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
11815 (update_range_test): Use it instead of force_gimple_operand_gsi.
11817 2017-06-12 Richard Biener <rguenther@suse.de>
11819 PR tree-optimization/81053
11820 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
11821 with backedge value not defined in loop. Simplify def stmt
11824 2017-06-11 Tom de Vries <tom@codesourcery.com>
11827 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
11829 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
11830 nvptx_cannot_force_const_mem.
11832 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11834 * opts.c (finish_options): Move test for flag_split_stack after
11835 it has been initialized.
11837 2017-06-11 Jason Merrill <jason@redhat.com>
11839 * tree.h (id_equal): New.
11840 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
11841 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
11842 instead of strcmp of IDENTIFIER_POINTER.
11844 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11846 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
11847 (mark_all_inlined_calls_cdtor): Fix formating.
11848 (inline_transform): Rescale profile before inlining.
11850 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11852 * cgraph.h (cgraph_edge::clone): Update prototype.
11853 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
11854 (cgraph_node::create_clone): Update.
11855 (cgraph_node::create_version_clone): Update.
11856 * tree-inline.c (copy_bb): Update.
11857 (expand_call_inline): Update.
11859 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11861 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
11862 factored out from ...
11863 (rs6000_emit_prologue): ... here.
11865 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11867 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
11868 factored out from ...
11869 (rs6000_emit_prologue): ... here.
11871 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11873 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
11875 (handle_missing_profiles): Fix computation of tp_first_run.
11876 (counts_to_freqs): Do not touch freqs when count is 0.
11878 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11880 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
11883 2017-06-10 Tom de Vries <tom@codesourcery.com>
11885 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
11886 attributes): Document signal effective target.
11888 2017-06-10 Tom de Vries <tom@codesourcery.com>
11890 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
11891 Document effective target stack_size.
11893 2017-06-09 David Malcolm <dmalcolm@redhat.com>
11895 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
11896 to the edit_context if they can be auto-applied.
11898 2017-06-9 Ian Lance Taylor <iant@golang.org>
11900 * opts.c (finish_options): If -fsplit-stack, disable implicit
11901 -forder-blocks-and-partition.
11902 * doc/invoke.texi (Optimize Options): Document that when using
11903 -fsplit-stack -forder-blocks-and-partition is not implicitly
11906 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11908 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
11909 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
11910 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
11911 * builtins.def (abort, trap, unreachable): Declare cold.
11912 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
11913 * tree-core.h (ECF_COLD): New.
11914 * tree.c (set_call_expr_flags): Handle ECF_COLD.
11915 (build_common_builtin_nodes): Mark unreachable and abort as cold.
11917 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11919 * predict.c (unlikely_executed_stmt_p): Cleanup.
11921 2017-06-09 Richard Biener <rguenther@suse.de>
11923 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
11924 model if the ref is always written to.
11926 2017-06-09 Tamar Christina <tamar.christina@arm.com>
11928 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
11930 2017-06-09 Tamar Christina <tamar.christina@arm.com>
11932 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
11935 2017-06-09 Tom de Vries <tom@codesourcery.com>
11938 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
11939 "target cannot support label values" when encountering LABEL_REF.
11941 2017-06-09 Martin Liska <mliska@suse.cz>
11943 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
11944 (gimple_gen_ic_func_profiler): Emit direct comparison
11945 of __gcov_indirect_call_callee with NULL.
11946 (gimple_gen_time_profiler): Change probability from
11947 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
11949 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11951 * profile.c (edge_gcov_counts): Turn to pointer.
11952 (compute_branch_probabilities, compute_branch_probabilities): Update.
11953 (branch_prob): Do not clear edge_gcov_count.
11954 * profile.h (edge_gcov_counts): Turn to pointer.
11955 (edge_gcov_count): Update.
11957 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11959 * gimple.h (gimple_check_failed): Mark cold.
11961 2017-06-09 Richard Biener <rguenther@suse.de>
11963 PR tree-optimization/66623
11964 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
11965 refactor check_reduction into two parts, properly computing
11966 whether we have to check reduction validity for outer loop
11969 2017-06-09 Richard Biener <rguenther@suse.de>
11971 PR tree-optimization/79483
11972 * graphite-scop-detection.c (order): New global.
11973 (get_order): Compute bb to order mapping that satisfies code
11974 generation constraints.
11975 (cmp_pbbs): New helper.
11976 (build_scops): Start domwalk at entry block, sort generated
11979 2017-06-09 Richard Biener <rguenther@suse.de>
11981 PR middle-end/81007
11982 * ipa-polymorphic-call.c
11983 (ipa_polymorphic_call_context::restrict_to_inner_class):
11984 Skip FIELD_DECLs with error_mark_node type.
11985 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
11988 2017-06-09 Martin Liska <mliska@suse.cz>
11990 * predict.c (struct branch_predictor): New struct.
11991 (test_prediction_value_range): New test.
11992 (predict_c_tests): New function.
11993 * selftest-run-tests.c (selftest::run_tests): Run the function.
11994 * selftest.h: Declare new tests.
11996 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
11999 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
12000 gen_add3_insn did not fail.
12001 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
12002 r0, construct that number in a temporary reg and add that reg to r0.
12003 If asked to put the result in r0 as well, fail.
12005 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
12007 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12008 for early expansion of vec_eqv.
12010 2017-06-08 Jakub Jelinek <jakub@redhat.com>
12012 PR middle-end/81005
12013 * ubsan.c (instrument_null): Avoid pointless code temporary.
12014 (pass_ubsan::execute): Instrument aggregate arguments of calls.
12016 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
12020 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
12023 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
12026 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12028 * predict.c (unlikely_executed_edge_p): Move ahead.
12029 (probably_never_executed_edge_p): Use it.
12031 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12033 PR middle-end/79988
12034 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
12035 gimple_call_builtin_p call.
12037 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12039 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
12040 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
12041 rtl_check_failed_type2, rtl_check_failed_code1,
12042 rtl_check_failed_code2, rtl_check_failed_code_mode,
12043 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
12044 rtvec_check_failed_bounds, rtl_check_failed_flag,
12045 _fatal_insn_not_found, _fatal_insn): Likewise.
12046 * tree.h (tree_contains_struct_check_failed,
12047 tree_check_failed, tree_not_check_failed,
12048 tree_class_check_failed, tree_range_check_failed,
12049 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
12050 tree_vec_elt_check_failed, phi_node_elt_check_failed,
12051 tree_operand_check_failed, omp_clause_check_failed,
12052 omp_clause_operand_check_failed, omp_clause_range_check_failed):
12055 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12057 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
12058 flag_branch_probabilities.
12059 * ipa-inline.c (edge_badness): Likewise.
12060 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12061 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
12062 * predict.c (maybe_hot_frequency_p): Likewise.
12063 (probably_never_executed): Likewise.
12064 * sched-ebb.c (schedule_ebbs): Likewise.
12065 * sched-rgn.c (find_single_block_region): Likewise.
12066 * tracer.c (tail_duplicate): Likewise.
12068 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12070 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
12071 longer requires x_flag_profile_use.
12073 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12075 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
12076 instead of flag_reorder_blocks_and_partition.
12077 * dbxout.c (dbxout_function_end): Likewise.
12078 * dwarf2out.c (gen_subprogram_die): Likewise.
12079 * haifa-sched.c (sched_create_recovery_edges): Likewise.
12080 * hw-doloop.c (reorg_loops): Likewise.
12081 * varasm.c (assemble_start_function,
12082 assemble_end_function): Likewise.
12083 (decide_function_section): Do not check for
12084 flag_reorder_blocks_and_partition.
12086 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12088 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12090 (chkp_get_hard_register_fake_addr_expr): Ditto.
12091 (chkp_build_addr_expr): Add check for hard reg case.
12092 (chkp_parse_array_and_component_ref): Ditto.
12093 (chkp_find_bounds_1): Ditto.
12094 (chkp_process_stmt): Don't generate bounds store for
12097 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12099 * predict.c (maybe_hot_bb_p): Do not check profile status.
12100 (maybe_hot_edge_p): Likewise.
12101 (probably_never_executed): Check for zero counts even if profile
12103 (unlikely_executed_edge_p): New function.
12104 (unlikely_executed_stmt_p): New function.
12105 (unlikely_executed_bb_p): New function.
12106 (set_even_probabilities): Use unlikely predicates.
12107 (combine_predictions_for_bb): Likewise.
12108 (predict_paths_for_bb): Likewise.
12109 (predict_paths_leading_to_edge): Likewise.
12110 (determine_unlikely_bbs): New function.
12111 (estimate_bb_frequencies): Use it.
12112 (compute_function_frequency): Use zero counts even if profile is
12114 * profile-count.h: Fix typo.
12116 2017-08-08 Julia Koval <julia.koval@intel.com>
12118 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
12119 _mm512_mask_cvtsepi16_storeu_epi8,
12120 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
12121 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
12122 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
12123 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
12124 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
12125 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
12126 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
12127 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
12128 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
12129 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
12130 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
12131 __builtin_ia32_pmovuswb256mem_mask,
12132 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
12133 __builtin_ia32_pmovwb512mem_mask): New builtins.
12135 2017-08-08 Julia Koval <julia.koval@intel.com>
12137 PR target/73350,80862
12138 * config/i386/subst.md (round): Fix round pattern.
12139 * config/i386/i386.c (ix86_erase_embedded_rounding):
12140 Fix erasing rounding for the fixed pattern.
12142 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12144 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
12146 2017-06-08 Martin Liska <mliska@suse.cz>
12148 PR gcov-profile/80911
12149 * gcov.c (block_info::block_info): New constructor.
12151 2017-06-07 Carl Love <cel@us.ibm.com>
12153 * config/rs6000/rs6000-c: The return type of the following
12154 built-in functions was implemented as int not long long. Fix sign
12155 of return value for the unsigned version of vec_mulo and vec_mule.
12156 vector unsigned long long vec_bperm (vector unsigned long long,
12157 vector unsigned char)
12158 vector signed long long vec_mule (vector signed int,
12160 vector unsigned long long vec_mule (vector unsigned int,
12161 vector unsigned int)
12162 vector signed long long vec_mulo (vector signed int,
12164 vector unsigned long long vec_mulo (vector unsigned int,
12165 vector unsigned int)
12166 * doc/extend.texi: Fix the documentation for the built-in
12169 2017-06-07 Carl Love <cel@us.ibm.com>
12172 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
12175 2017-06-07 Carl Love <cel@us.ibm.com>
12177 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
12178 support, Generate doublehv for signed int/float for BE case only.
12180 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
12182 * doc/invoke.texi (mcx16): Rewrite.
12184 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12186 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
12187 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
12188 *mov<mode>_softfloat, and an anonymous splitter): Use
12189 nonimmediate_operand instead of rs6000_nonimmediate_operand.
12191 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12193 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
12195 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12196 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
12197 (rs6000_debug_reg_global): Adjust.
12198 (rs6000_init_hard_regno_mode_ok): Adjust.
12199 (rs6000_dbx_register_number): Adjust.
12200 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
12201 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12202 Remove SPE_ACC and SPEFSCR.
12203 (REG_ALLOC_ORDER): Ditto.
12204 (FRAME_POINTER_REGNUM): Change to 111.
12205 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
12206 (REG_CLASS_NAMES): Ditto.
12207 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
12208 (REGISTER_NAMES): Ditto.
12209 (ADDITIONAL_REG_NAMES): Ditto.
12210 (rs6000_reg_names): Ditto.
12211 * config/rs6000/rs6000.md: Renumber some register number
12214 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12216 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
12218 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12219 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
12221 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
12222 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12223 Delete the SPE high registers.
12224 (REG_ALLOC_ORDER): Ditto.
12225 (enum reg_class): Remove SPE_HIGH_REGS.
12226 (REG_CLASS_NAMES): Ditto.
12227 (REG_CLASS_CONTENTS): Delete the SPE high registers.
12228 (REGISTER_NAMES): Ditto.
12229 (rs6000_reg_names): Ditto.
12230 * doc/tm.texi.in: Remove SPE as example.
12231 * doc/tm.texi: Regenerate.
12233 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12235 * config/rs6000/8540.md (ppc8540_brinc): Delete.
12236 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
12237 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
12238 * config/rs6000/rs6000.md (type): Remove "brinc".
12240 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12242 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
12243 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
12244 * config/rs6000/linuxspe.h: Delete file.
12245 * config/rs6000/rs6000.md: Don't include spe.md.
12246 * config/rs6000/spe.h: Delete file.
12247 * config/rs6000/spe.md: Delete file.
12248 * config/rs6000/t-rs6000: Remove spe.md.
12250 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12252 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
12253 (reg_or_none500mem_operand): Delete.
12254 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
12255 instead of reg_or_none500mem_operand.
12257 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12259 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
12260 handling of SPE flags.
12261 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
12263 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12265 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
12267 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
12268 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
12269 paired_divv2sf3): Similar.
12270 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
12271 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
12272 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
12274 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
12275 Rename the paired_* instruction patterns.
12276 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
12278 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
12279 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
12280 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
12281 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
12282 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
12283 PAIRED_VECTOR_MODE.
12284 (struct machine_function): Delete field spe_insn_chain_scanned_p.
12285 (spe_func_has_64bit_regs_p): Delete.
12286 (spe_expand_predicate_builtin): Delete.
12287 (spe_expand_evsel_builtin): Delete.
12288 (TARGET_DWARF_REGISTER_SPAN): Do not define.
12289 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
12290 (invalid_e500_subreg): Delete.
12291 (rs6000_legitimize_address): Always force_reg op2 as well, for
12292 paired single memory accesses.
12293 (rs6000_member_type_forces_blk): Delete.
12294 (rs6000_spe_function_arg): Delete.
12295 (rs6000_expand_unop_builtin): Delete SPE handling.
12296 (rs6000_expand_binop_builtin): Ditto.
12297 (spe_expand_stv_builtin): Delete.
12298 (bdesc_2arg_spe): Delete.
12299 (spe_expand_builtin): Delete.
12300 (spe_expand_predicate_builtin): Delete.
12301 (spe_expand_evsel_builtin): Delete.
12302 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
12303 (spe_init_builtins): Delete.
12304 (spe_func_has_64bit_regs_p): Delete.
12305 (savres_routine_name): Delete "info" parameter. Adjust callers.
12306 (rs6000_emit_stack_reset): Ditto.
12307 (rs6000_dwarf_register_span): Delete.
12308 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
12309 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
12310 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
12312 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
12314 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
12315 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
12316 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
12317 mulv2sf3, divv2sf3): Delete expanders.
12319 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12321 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
12323 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12325 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
12326 * config/rs6000/rs6000.c: Ditto.
12328 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12330 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
12331 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
12332 comparison_operator.
12334 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12336 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
12337 * config/rs6000/rs6000.opt: Ditto.
12338 * config/rs6000/t-rtems: Ditto.
12340 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12342 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
12343 TARGET_E500_SINGLE by 0, simplify.
12344 * config/rs6000/rs6000.c: Ditto.
12345 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
12346 (spe_build_register_parallel): Delete.
12347 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
12348 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
12349 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
12350 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
12351 (E500_CONVERT): Delete.
12352 * config/rs6000/spe.md: Remove many patterns and all define_constants.
12354 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12356 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
12357 * config/rs6000/dfp.md: Ditto.
12358 (negdd2, *negdd2_fpr): Merge.
12359 (absdd2, *absdd2_fpr): Merge.
12360 (negtd2, *negtd2_fpr): Merge.
12361 (abstd2, *abstd2_fpr): Merge.
12362 * config/rs6000/e500.h: Delete file.
12363 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
12364 TARGET_FPRS by 1 and simplify.
12365 * config/rs6000/rs6000-c.c: Ditto.
12366 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
12367 TARGET_DF_SPE by 0.
12368 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
12370 * config/rs6000/rs6000.md: Ditto.
12371 (floatdidf2, *floatdidf2_fpr): Merge.
12372 (move_from_CR_gt_bit): Delete.
12373 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
12374 (E500_CR_IOR_COMPARE): Delete.
12375 (All patterns that require !TARGET_FPRS): Delete.
12376 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
12378 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12380 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
12382 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12384 * graphds.c (add_edge): Intitialize edge's attached data.
12385 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
12386 pointer parameter. Call pointed function on each edge during
12387 graph traversing. Skip traversing the edge when the function
12389 (graphds_dfs, graphds_scc): Ditto.
12390 (for_each_edge): New parameter. Pass the new parameter to callback
12392 * graphds.h (skip_edge_callback): New function pointer type.
12393 (graphds_dfs, graphds_scc): New function pointer parameter.
12394 (graphds_edge_callback, for_each_edge): New parameter.
12396 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12398 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
12399 out code checking if runtime alias check is possible to below ...
12400 Call the new function.
12401 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
12402 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
12404 2017-06-07 Marek Polacek <polacek@redhat.com>
12407 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
12408 TYPE_OVERFLOW_WRAPS check.
12410 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12412 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
12413 if versioning is required.
12414 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
12415 peeling with the check for versioning.
12417 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12419 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
12420 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
12421 Set true to new parameter if new ssa variable is defined.
12422 (vect_gen_vector_loop_niters): Refactor. Set range information
12423 for the new vector loop bound variable.
12424 (vect_do_peeling): Ditto.
12426 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12428 * tree-affine.c (ssa.h): Include header file.
12429 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
12430 has wrapping overflow behavior.
12432 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12434 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
12436 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12438 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
12439 (tree_to_aff_combination): ... here.
12441 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12443 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
12444 reg_pressure model function.
12445 (ivopts_global_cost_for_size): Delete.
12446 (determine_set_costs, iv_ca_recount_cost): Call new model function
12447 ivopts_estimate_reg_pressure.
12449 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12451 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
12452 expensive than udiv. Remove floating point cases from mod.
12454 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12456 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
12457 Increase idiv cost.
12459 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12461 * config/aarch64/aarch64.md
12462 (copysignsf3): Fix mask generation.
12464 2017-06-07 Jakub Jelinek <jakub@redhat.com>
12466 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
12468 (class dump_manager): Add register_dumps method.
12469 * dumpfile.c: Include langhooks.h.
12470 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
12471 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
12472 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
12473 (dump_manager::dump_register): Start with 512 entries instead of 32.
12474 (dump_manager::register_dumps): New method.
12475 * toplev.c (general_init): Instead of invoking register_dumps
12476 langhook, invoke register_dumps method on the dump manager.
12477 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
12480 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
12482 * doc/md.texi: Clarify the restrictions on a define_insn condition.
12483 Say that # requires an associated define_split to exist, and that
12484 the define_split must be suitable for use after register allocation.
12486 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12488 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
12489 (compute_outgoing_frequencies): Also initialize zero counts.
12490 (find_many_sub_basic_blocks): Do not produce uninitialized profile
12491 around loops; preserve more of profile when nothing changes.
12493 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
12495 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
12497 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
12498 * config/arm/arm-cpu-cdata.h: Regenerate.
12499 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
12500 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
12501 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
12502 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
12503 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
12505 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
12507 * config/arm/t-rmprofile: Likewise.
12508 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
12510 2017-06-06 David S. Miller <davem@davemloft.net>
12513 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
12514 blockage if function uses alloca.
12516 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12518 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
12519 New "uid" fields to hold pretty-print IDs of group and ref.
12520 Memory references are now identified as <group_id>:<ref_id>
12521 instead of using [random] addresses.
12522 (dump_mem_details): Simplify, no functional change.
12523 (dump_mem_ref): Simplify and make output more concise.
12524 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
12525 (find_or_create_group): Initialize group uid.
12526 (record_ref): Initialize ref uid. Improve debug output.
12527 (prune_group_by_reuse, should_issue_prefetch_p,)
12528 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
12529 (mark_nontemporal_store, determine_loop_nest_reuse):
12530 Improve debug output.
12532 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12534 * dbgcnt.def (prefetch): New debug counter.
12535 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
12536 (schedule_prefetches): Stop issueing prefetches if debug counter
12539 2017-06-06 Tom de Vries <tom@codesourcery.com>
12541 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
12542 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
12544 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12546 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
12547 Use aarch64_reg_or_zero predicate for operand 4.
12548 (aarch64_compare_and_swap<mode> define_insn_and_split):
12549 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
12550 (aarch64_store_exclusive<mode>): Likewise for operand 2.
12552 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12554 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
12555 (arm_compute_save_core_reg_mask): This.
12556 (thumb1_compute_save_reg_mask): Rename into ...
12557 (thumb1_compute_save_core_reg_mask): This.
12558 (arm_compute_save_reg0_reg12_mask): Adapt comment.
12559 (arm_compute_frame_layout): Likewise.
12561 2017-06-06 Richard Biener <rguenther@suse.de>
12563 PR tree-optimization/80974
12564 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
12565 keep or clear leaders SSA info.
12567 2017-06-06 Tom de Vries <tom@codesourcery.com>
12569 * config/nvptx/nvptx.c (split_mode_p): New function.
12570 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
12572 2017-06-06 Tom de Vries <tom@codesourcery.com>
12574 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
12576 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12579 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
12582 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12584 * shrink-wrap.c (handle_simple_exit): Update profile.
12585 (try_shrink_wrapping): Upate profile.
12587 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12589 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
12590 (tree_guess_outgoing_edge_probabilities): New.
12591 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
12592 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
12594 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12596 * ipa-split.c (split_function): Initialize return bb profile.
12598 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12600 * profile.c (compute_branch_probabilities): Also initialize
12601 EXIT_BLOCK profile.
12603 2017-06-06 Richard Biener <rguenther@suse.de>
12605 PR tree-optimization/80928
12606 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
12607 (vect_analyze_loop_operations): Properly guard analysis for
12609 (vect_transform_loop): Likewise.
12610 (vect_analyze_loop_2): Also reset SLP type on PHIs.
12611 (vect_model_induction_cost): Do not cost for pure SLP.
12612 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
12613 of induction in inner loop vectorization.
12614 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
12615 (vect_get_and_check_slp_defs): Handle vect_induction_def.
12616 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
12618 (vect_analyze_slp_cost_1): Cost induction.
12619 (vect_detect_hybrid_slp_stmts): Handle PHIs.
12620 (vect_get_slp_vect_defs): Likewise.
12621 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
12622 (vect_transform_stmt): Handle SLP reductions.
12623 * tree-vectorizer.h (vectorizable_induction): Adjust.
12625 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12627 * config/rs6000/rs6000.c (make_resolver_func): Update
12628 init_lowered_empty_function call.
12630 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
12632 * doc/invoke.texi: Document the -fprofile-abs-path option.
12633 * common.opt (fprofile-abs-path): New option.
12634 * gcov-io.h (gcov_write_filename): Declare.
12635 * gcov-io.c (gcov_write_filename): New function.
12636 * coverage.c (coverage_begin_function): Use gcov_write_filename.
12637 * profile.c (output_location): Likewise.
12639 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12641 * shring-wrap.c: Revert accidental commit.
12643 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
12645 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
12647 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12649 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
12651 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
12652 profile in callgraph edge.
12653 * profile-count.h (apply_probability): If THIS is 0, then result is 0
12654 (apply_scale): Likewise.
12655 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
12656 Also scale profile when inlining function with zero profile.
12657 (initialize_cfun): Update exit block profile even when it is zero.
12658 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
12659 when profile is read.
12661 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12663 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
12664 (CLONE_*): New constants to define the processors we can generate
12665 code for with the target_clone attribute.
12666 (rs6000_clone_map): New array to identify which clone processors
12667 the current program is running on.
12668 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
12669 target_clone attribute.
12670 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
12671 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
12672 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
12673 (cpu_expand_builtin): Add support for target_clone attribute.
12674 (rs6000_valid_attribute_p): Allow "default" attribute.
12675 (get_decl_name): New debug function to simplify printing the
12676 current function name in debugging statements.
12677 (rs6000_clone_priority): New functions to support the target_clone
12678 attribute, and be able to generate code to switch between ISA 2.05
12679 through ISA 3.0 (power6 through power9).
12680 (rs6000_compare_version_priority): Likewise.
12681 (rs6000_get_function_versions_dispatcher): Likewise.
12682 (make_resolver_func): Likewise.
12683 (add_condition_to_bb): Likewise.
12684 (dispatch_function_versions): Likewise.
12685 (rs6000_generate_version_dispatcher_body): Likewise.
12686 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
12687 (fusion_gpr_load_p): Fix a spacing issue.
12688 * doc/extend.texi (Common Function Attributes): Document that the
12689 PowerPC supports the target_clone attribute.
12691 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
12693 * config/arm/arm.h: explain F symbol found in description of ARM
12694 register allocation in its legend.
12696 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12698 * config/mips/frame-header-opt.c: Include profile-count.h.
12699 * config/riscv/riscv.c: Include profile-count.h
12701 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12703 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
12705 (sm_set_flag_if_changed): Add bbs field.
12706 (execute_sm_if_changed_flag_set): Pass BBS.
12707 (execute_sm): Update.
12709 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12711 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
12714 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12716 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
12717 (peephole2): New peephole2 to emit the above.
12718 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
12720 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12722 * config/aarch64/aarch64.c (define_peephole2 above
12723 *sub_<shift>_<mode>): New peephole.
12725 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12727 * config/i386/i386.c (make_resolver_func): Update.
12728 * Makefile.in: Add profile-count.h and profile-count.o
12729 * auto-profile.c (afdo_indirect_call): Update to new API.
12730 (afdo_set_bb_count): Update.
12731 (afdo_propagate_edge): Update.
12732 (afdo_propagate_circuit): Update.
12733 (afdo_calculate_branch_prob): Update.
12734 (afdo_annotate_cfg): Update.
12735 * basic-block.h: Include profile-count.h
12736 (struct edge_def): Turn count to profile_count.
12737 (struct basic_block_def): Likewie.
12738 (REG_BR_PROB_BASE): Move to profile-count.h
12739 (RDIV): Move to profile-count.h
12740 * bb-reorder.c (max_entry_count): Turn to profile_count.
12741 (find_traces): Update.
12742 (rotate_loop):Update.
12743 (connect_traces):Update.
12744 (sanitize_hot_paths):Update.
12745 * bt-load.c (migrate_btr_defs): Update.
12746 * cfg.c (RDIV): Remove.
12747 (init_flow): Use alloc_block.
12748 (alloc_block): Uninitialize count.
12749 (unchecked_make_edge): Uninitialize count.
12750 (check_bb_profile): Update.
12751 (dump_edge_info): Update.
12752 (dump_bb_info): Update.
12753 (update_bb_profile_for_threading): Update.
12754 (scale_bbs_frequencies_int): Update.
12755 (scale_bbs_frequencies_gcov_type): Update.
12756 (scale_bbs_frequencies_profile_count): New.
12757 * cfg.h (update_bb_profile_for_threading): Update.
12758 (scale_bbs_frequencies_profile_count): Declare.
12759 * cfgbuild.c (compute_outgoing_frequencies): Update.
12760 (find_many_sub_basic_blocks): Update.
12761 * cfgcleanup.c (try_forward_edges): Update.
12762 (try_crossjump_to_edge): Update.
12763 * cfgexpand.c (expand_gimple_tailcall): Update.
12764 (construct_exit_block): Update.
12765 * cfghooks.c (verify_flow_info): Update.
12766 (dump_bb_for_graph): Update.
12767 (split_edge): Update.
12768 (make_forwarder_block): Update.
12769 (duplicate_block): Update.
12770 (account_profile_record): Update.
12771 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
12772 (get_estimated_loop_iterations): Update.
12773 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
12774 (single_likely_exit): Update.
12775 * cfgloopmanip.c (scale_loop_profile): Update.
12777 (set_zero_probability): Update.
12778 (lv_adjust_loop_entry_edge): Update.
12779 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12780 (purge_dead_edges): Update.
12781 (rtl_account_profile_record): Update.
12782 * cgraph.c (cgraph_node::create): Uninitialize count.
12783 (symbol_table::create_edge): Uninitialize count.
12784 (cgraph_update_edges_for_call_stmt_node): Update.
12785 (cgraph_edge::dump_edge_flags): Update.
12786 (cgraph_node::dump): Update.
12787 (cgraph_edge::maybe_hot_p): Update.
12788 * cgraph.h: Include profile-count.h
12789 (create_clone), create_edge, create_indirect_edge): Update.
12790 (cgraph_node): Turn count to profile_count.
12791 (cgraph_edge0: Likewise.
12792 (make_speculative, clone): Update.
12793 (create_edge): Update.
12794 (init_lowered_empty_function): Update.
12795 * cgraphclones.c (cgraph_edge::clone): Update.
12796 (duplicate_thunk_for_node): Update.
12797 (cgraph_node::create_clone): Update.
12798 * cgraphunit.c (cgraph_node::analyze): Update.
12799 (cgraph_node::expand_thunk): Update.
12800 * final.c (dump_basic_block_info): Update.
12801 * gimple-streamer-in.c (input_bb): Update.
12802 * gimple-streamer-out.c (output_bb): Update.
12803 * graphite.c (print_global_statistics): Update.
12804 (print_graphite_scop_statistics): Update.
12805 * hsa-brig.c: Include basic-block.h.
12806 * hsa-dump.c: Include basic-block.h.
12807 * hsa-gen.c (T sum_slice): Update.
12808 (convert_switch_statements):Update.
12809 * hsa-regalloc.c: Include basic-block.h.
12810 * ipa-chkp.c (chkp_produce_thunks): Update.
12811 * ipa-cp.c (struct caller_statistics): Update.
12812 (init_caller_stats): Update.
12813 (gather_caller_stats): Update.
12814 (ipcp_cloning_candidate_p): Update.
12815 (good_cloning_opportunity_p): Update.
12816 (get_info_about_necessary_edges): Update.
12817 (dump_profile_updates): Update.
12818 (update_profiling_info): Update.
12819 (update_specialized_profile): Update.
12820 (perhaps_add_new_callers): Update.
12821 (decide_about_value): Update.
12822 (ipa_cp_c_finalize): Update.
12823 * ipa-devirt.c (struct odr_type_warn_count): Update.
12824 (struct decl_warn_count): Update.
12825 (struct final_warning_record): Update.
12826 (possible_polymorphic_call_targets): Update.
12827 (ipa_devirt): Update.
12828 * ipa-fnsummary.c (redirect_to_unreachable): Update.
12829 * ipa-icf.c (sem_function::merge): Update.
12830 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
12831 * ipa-inline.c (compute_uninlined_call_time): Update.
12832 (compute_inlined_call_time): Update.
12833 (want_inline_small_function_p): Update.
12834 (want_inline_self_recursive_call_p): Update.
12835 (edge_badness): Update.
12836 (lookup_recursive_calls): Update.
12837 (recursive_inlining): Update.
12838 (inline_small_functions): Update.
12839 (dump_overall_stats): Update.
12840 (dump_inline_stats): Update.
12841 * ipa-profile.c (ipa_profile_generate_summary): Update.
12842 (ipa_propagate_frequency): Update.
12843 (ipa_profile): Update.
12844 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12845 * ipa-utils.c (ipa_merge_profiles): Update.
12846 * loop-doloop.c (doloop_modify): Update.
12847 * loop-unroll.c (report_unroll): Update.
12848 (unroll_loop_runtime_iterations): Update.
12849 * lto-cgraph.c (lto_output_edge): Update.
12850 (lto_output_node): Update.
12851 (input_node): Update.
12852 (input_edge): Update.
12853 (merge_profile_summaries): Update.
12854 * lto-streamer-in.c (input_cfg): Update.
12855 * lto-streamer-out.c (output_cfg): Update.
12856 * mcf.c (create_fixup_graph): Update.
12857 (adjust_cfg_counts): Update.
12858 (sum_edge_counts): Update.
12859 * modulo-sched.c (sms_schedule): Update.
12860 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
12861 * predict.c (maybe_hot_count_p): Update.
12862 (probably_never_executed): Update.
12863 (dump_prediction): Update.
12864 (combine_predictions_for_bb): Update.
12865 (propagate_freq): Update.
12866 (handle_missing_profiles): Update.
12867 (counts_to_freqs): Update.
12868 (rebuild_frequencies): Update.
12869 (force_edge_cold): Update.
12870 * predict.h: Include profile-count.h
12871 (maybe_hot_count_p, counts_to_freqs): UPdate.
12872 * print-rtl-function.c: Do not include cfg.h
12873 * print-rtl.c: Include basic-block.h
12874 * profile-count.c: New file.
12875 * profile-count.h: New file.
12876 * profile.c (is_edge_inconsistent): Update.
12877 (correct_negative_edge_counts): Update.
12878 (is_inconsistent): Update.
12879 (set_bb_counts): Update.
12880 (read_profile_edge_counts): Update.
12881 (compute_frequency_overlap): Update.
12882 (compute_branch_probabilities): Update; Initialize and deinitialize
12884 (branch_prob): Update.
12885 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
12886 (edge_gcov_count): New.
12887 (bb_gcov_count): New.
12888 * shrink-wrap.c (try_shrink_wrapping): Update.
12889 * tracer.c (better_p): Update.
12890 * trans-mem.c (expand_transaction): Update.
12891 (ipa_tm_insert_irr_call): Update.
12892 (ipa_tm_insert_gettmclone_call): Update.
12893 * tree-call-cdce.c: Update.
12894 * tree-cfg.c (gimple_duplicate_sese_region): Update.
12895 (gimple_duplicate_sese_tail): Update.
12896 (gimple_account_profile_record): Update.
12897 (execute_fixup_cfg): Update.
12898 * tree-inline.c (copy_bb): Update.
12899 (copy_edges_for_bb): Update.
12900 (initialize_cfun): Update.
12901 (freqs_to_counts): Update.
12902 (copy_cfg_body): Update.
12903 (expand_call_inline): Update.
12904 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
12905 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
12906 (try_unroll_loop_completely): Update.
12907 (try_peel_loop): Update.
12908 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12909 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
12910 * tree-ssa-loop-split.c (connect_loops): Update.
12911 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
12912 * tree-ssa-reassoc.c (branch_fixup): Update.
12913 * tree-ssa-tail-merge.c (replace_block_by): Update.
12914 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
12915 (compute_path_counts): Update.
12916 (update_profile): Update.
12917 (recompute_probabilities): Update.
12918 (update_joiner_offpath_counts): Update.
12919 (estimated_freqs_path): Update.
12920 (freqs_to_counts_path): Update.
12921 (clear_counts_path): Update.
12922 (ssa_fix_duplicate_block_edges): Update.
12923 (duplicate_thread_path): Update.
12924 * tree-switch-conversion.c (case_bit_test_cmp): Update.
12925 (struct switch_conv_info): Update.
12926 * tree-tailcall.c (decrease_profile): Update.
12927 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
12928 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12929 * value-prof.c (check_counter): Update.
12930 (gimple_divmod_fixed_value): Update.
12931 (gimple_mod_pow2): Update.
12932 (gimple_mod_subtract): Update.
12933 (gimple_ic_transform): Update.
12934 (gimple_stringop_fixed_value): Update.
12935 * value-prof.h (gimple_ic): Update.
12937 2017-06-02 Carl Love <cel@us.ibm.com>
12939 * config/rs6000/rs6000-c: Add support for built-in functions
12940 vector double vec_doublee (vector signed int);
12941 vector double vec_doublee (vector unsigned int);
12942 vector double vec_doublee (vector float);
12943 vector double vec_doubleh (vector signed int);
12944 vector double vec_doubleh (vector unsigned int);
12945 vector double vec_doubleh (vector float);
12946 vector double vec_doublel (vector signed int);
12947 vector double vec_doublel (vector unsigned int);
12948 vector double vec_doublel (vector float);
12949 vector double vec_doubleo (vector signed int);
12950 vector double vec_doubleo (vector unsigned int);
12951 vector double vec_doubleo (vector float);.
12952 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
12953 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
12955 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
12956 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
12957 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
12959 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
12960 vec_doublel, vec_doubleh.
12961 * doc/extend.texi: Update the built-in documentation file for the
12962 new built-in functions.
12964 2017-06-02 David Malcolm <dmalcolm@redhat.com>
12967 * ipa-inline-analysis.c (free_growth_caches): Set
12968 edge_removal_hook_holder to NULL after removing it.
12970 2017-06-02 Sudakshina Das <sudi.das@arm.com>
12972 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
12973 comparision with zero.
12975 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
12976 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12977 for early expansion of vec_min and vec_max builtins.
12978 (builtin_function_type): Add min/max unsigned variants to those
12979 identified as having unsigned arguments.
12981 2017-06-02 Olivier Hainque <hainque@adacore.com>
12983 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
12985 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12987 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
12988 Use VALL_F16 iterator rather than VALL.
12990 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12992 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
12993 Emit CBNZ inside loop when doing a strong exchange and comparing
12994 against zero. Generate the CC flags after the loop.
12996 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
12998 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
12999 (dl_section_ref): New.
13000 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
13001 On AIX, append an expression to subtract the size of the
13002 section length to dl_section_ref.
13004 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13006 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13007 for early expansion of vector absolute builtins.
13009 2017-06-02 Richard Biener <rguenther@suse.de>
13011 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
13012 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
13014 2017-06-02 Richard Biener <rguenther@suse.de>
13016 PR tree-optimization/80948
13017 * tree-tailcall.c (find_tail_calls): Track stmts to move in
13018 stmt order as well.
13020 2017-06-02 Richard Biener <rguenther@suse.de>
13022 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
13024 * tree-vect-stmts.c (process_use): Do not mark backedge defs
13025 for inductions as relevant.
13027 2017-06-02 Richard Biener <rguenther@suse.de>
13029 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
13030 (vectorizable_induction): ... this. Remove dead code.
13032 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
13034 * builtins. (expand_builtin_alloca): Remove second parameter and
13035 infer its value from the first parameter instead.
13036 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
13038 2017-06-02 Jakub Jelinek <jakub@redhat.com>
13040 PR rtl-optimization/80903
13041 * loop-doloop.c (add_test): Unshare sequence.
13043 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
13045 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
13047 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
13049 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
13051 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
13052 xlogue_layout::get_instance, logue_layout::xlogue_layout,
13053 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
13054 (xlogue_layout::get_stub_rtx): Make static.
13055 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
13056 (xlogue_layout::compute_stub_managed_regs): Rename to...
13057 (xlogue_layout::count_stub_managed_regs): ...this.
13058 (xlogue_layout::is_stub_managed_reg): New function.
13059 (xlogue_layout::m_stub_names): Rename to...
13060 (xlogue_layout::s_stub_names): ...this, make static.
13061 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
13062 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
13063 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
13064 xlogue_layout::s_stub_names): Instantiate statics.
13065 (stub_managed_regs): Remove.
13066 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
13067 (disable_call_ms2sysv_xlogues): Rename to...
13068 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
13069 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
13071 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
13072 change after reload_completed.
13073 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
13075 (ix86_expand_prologue): Likewise.
13076 (ix86_expand_epilogue): Likewise.
13077 (ix86_expand_split_stack_prologue): Likewise.
13078 (ix86_compute_frame_layout): Remove frame parameter ...
13079 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
13080 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
13082 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
13083 (ix86_frame): Move from here ...
13084 * config/i386/i386.h (ix86_frame): ... to here.
13085 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
13086 complete ix86_frame data structure instead. Remove some_ld_name.
13088 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13090 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
13091 symbols that hold a DECL_VALUE_EXPR.
13093 2017-06-01 Martin Jambor <mjambor@suse.cz>
13095 PR tree-optimization/80898
13096 * tree-sra.c (process_subtree_disqualification): Removed.
13097 (disqualify_candidate): Do not acll
13098 process_subtree_disqualification.
13099 (subtree_mark_written_and_enqueue): New function.
13100 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
13101 RHS has been disqualified and re-queue LHS if necessary. Apart
13102 from that, ignore disqualified RHS.
13104 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13106 * config/s390/s390.c (s390_emit_epilogue): Disable early return
13107 address fetch for z10 or later.
13109 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13111 * config/arc/arc.md (tst_movb): Add guard when splitting.
13113 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13115 * config/arc/arc.c (arc_can_eliminate): Test against
13116 arc_frame_pointer_needed.
13118 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13120 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
13121 to prevent store reordering.
13122 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
13123 (type): Add block type.
13124 (stack_tie): Define special instruction to be used in
13127 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13129 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
13130 constraint. It is not valid for the pattern.
13131 (noncommutative_binary_comparison): Likewise.
13133 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13135 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
13138 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13140 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
13141 be used by the reg-alloc.
13143 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13145 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
13146 reg-alloc when having mul64 or mul32x16 instructions.
13147 (mulsidi3): Likewise.
13148 (umulsidi3): Likewise.
13149 (mulsi32x16): New pattern.
13150 (mulsi64): Likewise.
13151 (mulsidi64): Likewise.
13152 (umulsidi64): Likewise.
13153 (MUL32x16_REG): Define.
13154 (mul64_600): Use MUL32x16_REG.
13155 (mac64_600): Likewise.
13156 (umul64_600): Likewise.
13157 (umac64_600): Likewise.
13159 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13161 * config/arc/arc.md (mulsi3_700): Make it commutative.
13163 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
13165 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
13167 (*sign_extendsidi2_insn): Likewise for movstosw.
13169 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13171 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
13172 the type of the input discriminant value. Convert the
13173 discriminant value of signedness vary.
13175 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
13177 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
13178 Add to -Wall section.
13180 2017-06-01 Richard Biener <rguenther@suse.de>
13182 PR middle-end/66313
13183 * fold-const.c (fold_plusminus_mult_expr): If the factored
13184 factor may be zero use a wrapping type for the inner operation.
13185 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
13186 and handle moved defs.
13187 (process_assignment): Properly guard the unary op case. Return a
13188 tri-state indicating that moving the stmt before the call may allow
13189 to continue. Pass through to_move.
13190 (find_tail_calls): Handle moving unrelated defs before
13193 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
13196 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
13197 splitter result in the canonical way.
13199 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
13201 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
13202 also for 32bit target. Update insn attributes.
13203 (zero-extendsidi2 splitter): Allow all registers for operand 1.
13205 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
13207 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
13208 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
13209 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
13210 (_mm_maskz_min_ss): New intrinsics.
13212 2017-05-31 Martin Liska <mliska@suse.cz>
13214 * tree-vect-loop.c (vect_create_epilog_for_reduction):
13215 Change comment style to one we normally use.
13216 (vectorizable_reduction): Likewise.
13217 (vectorizable_induction): Likewise.
13218 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13219 (vectorizable_call): Likewise.
13220 (vectorizable_simd_clone_call): Likewise.
13221 (vectorizable_conversion): Likewise.
13222 (vectorizable_assignment): Likewise.
13223 (vectorizable_shift): Likewise.
13224 (vectorizable_operation): Likewise.
13225 (vectorizable_store): Likewise.
13226 (vectorizable_load): Likewise.
13227 * tree-vectorizer.h: Likewise.
13229 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
13231 * passes.c (emergency_dump_function): New.
13232 * tree-pass.h (emergency_dump_function): Declare.
13233 * plugin.c (plugins_internal_error_function): Remove.
13234 * plugin.h (plugins_internal_error_function): Remove declaration.
13235 * toplev.c (internal_error_function): New static function. Use it...
13236 (general_init): ...here.
13238 2017-05-31 Graham Markall <graham.markall@embecosm.com>
13240 * config/arc/arc.c (arc_print_operand): Handle constant operands.
13241 (arc_rtx_costs): Add costs for new patterns.
13242 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
13243 * config/arc/predicates.md: Add _1_2_3_operand predicate.
13245 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13247 * tree-ssa-strlen.c (get_next_strinfo): New function.
13248 (get_stridx_plus_constant): Use it.
13249 (zero_length_string): Likewise.
13250 (adjust_related_strinfos): Likewise.
13251 (adjust_last_stmt): Likewise.
13253 2017-05-31 Richard Biener <rguenther@suse.de>
13256 * config/i386/i386.c (ix86_expand_builtin): Remove assert
13257 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
13259 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13261 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
13262 loop_vinfo argument and use of dependence distance vectors.
13263 Check instead whether the two references differ only in their
13264 initial value and assume that they have the same alignment if the
13265 difference is a multiple of the vector alignment.
13266 (vect_analyze_data_refs_alignment): Update call accordingly.
13268 2017-05-31 Martin Liska <mliska@suse.cz>
13271 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
13273 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13275 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
13276 (create_intersect_range_checks): Move from ...
13277 * tree-data-ref.c (create_intersect_range_checks_index)
13278 (create_intersect_range_checks): ... to here.
13279 (create_runtime_alias_checks): New function factored from ...
13280 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
13281 here. Call above function.
13282 * tree-data-ref.h (create_runtime_alias_checks): New function.
13284 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13286 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
13287 segment length for dr_b and compute it in wide_int.
13289 2017-05-31 Richard Biener <rguenther@suse.de>
13291 PR tree-optimization/80906
13292 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
13293 and pass through iv_map.
13294 (copy_bb_and_scalar_dependences): Adjust.
13295 (translate_pending_phi_nodes): Likewise.
13296 (copy_loop_close_phi_args): Handle code-generating IVs instead
13299 2017-05-30 David Malcolm <dmalcolm@redhat.com>
13301 * diagnostic-color.c (color_dict): Add "type-diff".
13302 (parse_gcc_colors): Update comment.
13303 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
13304 -fdiagnostics-show-template-tree and -fno-elide-type.
13305 (GCC_COLORS): Add type-diff to example.
13307 (-fdiagnostics-show-template-tree): New.
13308 (-fno-elide-type): New.
13309 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
13310 the pp_format_decoder callback. Call any m_format_postprocessor's
13312 (pretty_printer::pretty_printer): Initialize
13313 m_format_postprocessor.
13314 (pretty_printer::~pretty_printer): Delete any
13315 m_format_postprocessor.
13316 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
13317 (class format_postprocessor): New class.
13318 (struct pretty_printer::format_decoder): Document the new parameters.
13319 (struct pretty_printer::m_format_postprocessor): New field.
13320 * tree-diagnostic.c (default_tree_printer): Update for new
13321 bool and const char ** params.
13322 * tree-diagnostic.h (default_tree_printer): Likewise.
13324 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
13326 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
13327 (lwa_operand): Delete rs6000_gen_cell_microcode test.
13328 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13329 rs6000_gen_cell_microcode code.
13330 (rs6000_final_prescan_insn): Delete.
13331 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
13332 "warn-cell-microcode" entries.
13333 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
13334 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
13335 throughout. Change cc_reg_not_micro_cr0_operand to
13336 cc_reg_not_cr0_operand throughout.
13337 (*extendhi<mode>2_noload): Delete.
13338 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
13339 (mwarn-cell-microcode): Delete.
13340 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
13341 -mgen-cell-microcode and -mwarn-cell-microcode.
13343 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
13346 * config/i386/constraints.md (Yd): New constraint.
13348 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
13349 and (?Yd, r) alternatives. Update insn attributes.
13350 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
13351 and (?*Yd, r) alternatives. Update insn attributes.
13352 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
13354 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
13356 * gimplify.c (gimplify_modify_expr): Don't create a
13357 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
13360 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
13362 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
13364 2017-05-30 Richard Biener <rguenther@suse.de>
13366 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
13367 and reduc_def fields.
13368 (STMT_VINFO_REDUC_TYPE): New define.
13369 (STMT_VINFO_REDUC_DEF): Likewise.
13370 (vect_force_simple_reduction): Adjust prototype.
13371 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
13372 (vect_is_simple_reduction): Remove check_reduction argument.
13373 (vect_force_simple_reduction): Adjust and set
13374 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13375 (vectorizable_reduction): Do not re-do reduction analysis
13376 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13377 * tree-parloops.c (gather_scalar_reductions): Adjust.
13379 2017-05-30 Richard Biener <rguenther@suse.de>
13381 PR middle-end/80901
13382 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
13385 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
13387 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
13388 Introduce unknown_misalignment parameter and remove vf.
13389 (vect_peeling_hash_get_lowest_cost):
13390 Pass unknown_misalignment parameter.
13391 (vect_enhance_data_refs_alignment):
13392 Fix unsupportable data ref treatment.
13394 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13396 * tree-vect-data-refs.c (vect_get_data_access_cost):
13397 Workaround for SLP handling.
13398 (vect_enhance_data_refs_alignment):
13399 Compute costs for doing no peeling at all, compare to the best
13400 peeling costs so far and avoid peeling if cheaper.
13402 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13404 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
13405 Return peeling info and set costs to zero for unlimited cost
13407 (vect_enhance_data_refs_alignment): Also inspect all datarefs
13408 with unknown misalignment. Compute and costs for unknown
13409 misalignment, compare them to the costs for known misalignment
13410 and choose the cheapest for peeling.
13412 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13414 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
13415 (vect_get_peeling_costs_all_drs): Create function.
13416 (vect_peeling_hash_get_lowest_cost):
13417 Use vect_get_peeling_costs_all_drs.
13418 (vect_peeling_supportable): Create function.
13419 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
13421 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13423 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
13424 DR_HAS_NEGATIVE_STEP.
13425 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
13426 (vect_enhance_data_refs_alignment): Use.
13427 (vect_duplicate_ssa_name_ptr_info): Use.
13428 * tree-vectorizer.h (dr_misalignment): Use.
13429 (known_alignment_for_access_p): Use.
13431 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13434 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
13436 (has_section_name): New function.
13438 2017-05-30 Martin Liska <mliska@suse.cz>
13441 * auto-profile.c (get_function_decl_from_block): Fix
13444 2017-05-30 Richard Biener <rguenther@suse.de>
13446 PR middle-end/80876
13447 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
13449 2017-05-30 Martin Liska <mliska@suse.cz>
13451 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
13452 * dumpfile.h (struct dump_file_info): Remove ctors.
13454 2017-05-30 Martin Liska <mliska@suse.cz>
13456 * predict.def: Fix GNU coding style.
13458 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13460 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13461 Mark 'to' argument with ATTRIBUTE_UNUSED.
13463 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13465 * config/xtensa/xtensa.c (xtensa_emit_call): Use
13466 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
13467 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
13470 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
13472 * doc/install.texi (Options specification): Restore entry of
13473 --enable-sjlj-exceptions.
13475 2017-05-27 Michael Eager <eager@eagercon.com>
13478 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
13480 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
13482 * config/microblaze/microblaze.h
13483 (FIXED_REGISTERS): Update in macro.
13484 (CALL_USED_REGISTERS): Update in macro.
13486 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
13488 * doc/install.texi: Add links to macOS binary distributions.
13490 2017-05-27 Jakub Jelinek <jakub@redhat.com>
13494 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13496 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13498 2017-05-26 Martin Liska <mliska@suse.cz>
13500 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
13502 2017-05-26 Martin Liska <mliska@suse.cz>
13504 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
13505 always leading ';; '.
13506 (dump_bb_info): Likewise.
13507 (brief_dump_cfg): Likewise.
13508 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
13509 * dumpfile.c: Remove usage of TDF_VERBOSE.
13510 * dumpfile.h (enum dump_kind): Likewise.
13511 (dump_gimple_bb_header): Do not use TDF_COMMENT.
13512 * print-tree.c (debug_verbose): Remove.
13513 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
13514 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
13515 * tree-diagnostic.c (default_tree_printer): Replace
13516 TDF_DIAGNOSTIC with TDF_SLIM.
13518 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13520 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
13521 in parameter loop, rather than loop_vinfo.
13522 (create_intersect_range_checks): Ditto.
13523 (vect_create_cond_for_alias_checks): Update call to above functions.
13525 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13527 PR tree-optimization/80815
13528 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
13529 for merging runtime alias checks. Handle negative DR_STEPs.
13531 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13533 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
13535 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
13536 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
13537 out code pruning runtime alias checks.
13538 * tree-data-ref.c (prune_runtime_alias_test_list): New function
13539 factored out from above.
13540 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13542 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13544 (prune_runtime_alias_test_list): New decalaration.
13546 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13548 * tree-vect-data-refs.c (compare_tree): Rename and move ...
13549 * tree-data-ref.c (data_ref_compare_tree): ... to here.
13550 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
13551 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
13552 (operator==, comp_dr_with_seg_len_pair): Ditto.
13553 (vect_prune_runtime_alias_test_list): Ditto.
13555 2017-05-26 Martin Liska <mliska@suse.cz>
13558 * params.def: Bound partial-inlining-entry-probability param.
13560 2017-05-26 Marek Polacek <polacek@redhat.com>
13563 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
13566 2017-05-26 Richard Biener <rguenther@suse.de>
13568 PR tree-optimization/80842
13569 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
13572 2017-05-26 Richard Biener <rguenther@suse.de>
13574 PR tree-optimization/80844
13575 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
13577 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
13579 * doc/md.texi (Machine Constraints): Update x86 family
13580 machine constraints section to match 'config/i386/constraints.md'.
13582 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
13584 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
13586 2017-05-25 Nathan Sidwell <nathan@acm.org>
13588 * doc/invoke.texi (--enable-languages): Update documentation.
13590 2017-05-25 Martin Liska <mliska@suse.cz>
13592 * dumpfile.c: Add TDF_FOLDING.
13593 * dumpfile.h (enum dump_kind): Likewise.
13594 * genmatch.c (dt_simplify::gen_1): Use it.
13596 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13598 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
13600 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13602 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13603 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
13605 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13607 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
13608 * match.pd (X == C): Rewrite it here.
13609 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
13610 with_certain_nonzero_bits2): New predicates.
13611 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
13613 2017-05-24 Nathan Sidwell <nathan@acm.org>
13615 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
13618 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
13621 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13623 * config/powerpcspe: New port. Files are copied from the rs6000
13624 port, with "rs6000" in filenames replaced by "powerpcspe".
13626 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
13628 PR rtl-optimization/80754
13629 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
13631 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
13633 * config/sparc/sparc.md (length): Return the correct value for -mflat
13634 sibcalls to match output_sibcall.
13636 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13640 * config/rs6000/rs6000.c (struct machine_function): Add new field
13642 (rs6000_get_separate_components): Init that field, use it.
13643 (rs6000_components_for_bb): Use the field.
13645 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
13647 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
13649 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
13651 PR middle-end/80823
13652 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
13654 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13657 * config/s390/s390.c (s390_check_qrst_address): Check incoming
13658 address against address_operand predicate.
13659 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
13661 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
13663 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
13664 parameters passed indirectly.
13666 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
13668 * config/i386/i386.md (*movdi_internal): Remove SSE4
13669 alternative 18 (?r, *v). Update insn attributes.
13670 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
13671 Update insn attributes.
13672 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
13673 Update insn attributes.
13674 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
13675 alternative 1 (r, v). Remove isa attribute.
13676 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
13677 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
13678 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
13680 2017-05-23 Tom de Vries <tom@codesourcery.com>
13682 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
13685 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13687 * cgraphunit.c (symbol_table::process_new_functions): Update.
13688 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
13689 (inline_generate_summary): Rename to ...
13690 (ipa_fn_summary_generate): ... this one.
13691 (inline_read_summary): Rename to ...
13692 (ipa_fn_summary_read): ... this one.
13693 (inline_write_summary): Rename to ...
13694 (ipa_fn_summary_write): ... this one.
13695 (inline_free_summary): Rename to ...
13696 (ipa_free_fn_summary): ... this one.
13697 (pass_data_local_fn_summary, pass_local_fn_summary,
13698 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
13699 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
13700 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
13701 make_pass_ipa_fn_summary): New.
13702 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
13703 inline_write_summary, inline_free_summary): Remove.
13704 (ipa_free_fn_summary) : New.
13705 * ipa-inline.c (ipa_inline): Update.
13706 (pass_ipa_inline): Do not generate summaries.
13707 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
13709 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
13710 and add pass_ipa_fn_summary.
13711 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
13713 (make_pass_inline_parameters): Remove.
13715 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
13717 * omp-low.c (struct omp_context): Remove "default_kind" member.
13720 * omp-offload.c (execute_oacc_device_lower): Remove the
13721 parallelism dimensions function attributes for unparallelized
13722 OpenACC kernels constructs.
13724 2017-05-23 Martin Liska <mliska@suse.cz>
13726 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
13728 (cgraph_edge::make_speculative): Likewise.
13729 (cgraph_edge::resolve_speculation): Likewise.
13730 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
13731 (cgraph_node::dump): Likewise.
13732 * cgraph.h: Likewise.
13733 * cgraphunit.c (analyze_functions): Likewise.
13734 (symbol_table::compile): Likewise.
13735 * ipa-cp.c (print_all_lattices): Likewise.
13736 (determine_versionability): Likewise.
13737 (initialize_node_lattices): Likewise.
13738 (ipcp_verify_propagated_values): Likewise.
13739 (estimate_local_effects): Likewise.
13740 (update_profiling_info): Likewise.
13741 (create_specialized_node): Likewise.
13742 (perhaps_add_new_callers): Likewise.
13743 (decide_about_value): Likewise.
13744 (decide_whether_version_node): Likewise.
13745 (identify_dead_nodes): Likewise.
13746 (ipcp_store_bits_results): Likewise.
13747 * ipa-devirt.c (dump_targets): Likewise.
13748 (ipa_devirt): Likewise.
13749 * ipa-icf.c (sem_item::dump): Likewise.
13750 (sem_function::equals): Likewise.
13751 (sem_variable::equals): Likewise.
13752 (sem_item_optimizer::read_section): Likewise.
13753 (sem_item_optimizer::execute): Likewise.
13754 (congruence_class::dump): Likewise.
13755 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
13756 (dump_inline_summary): Likewise.
13757 (estimate_node_size_and_time): Likewise.
13758 (inline_analyze_function): Likewise.
13759 * ipa-inline-transform.c (inline_call): Likewise.
13760 * ipa-inline.c (report_inline_failed_reason): Likewise.
13761 (want_early_inline_function_p): Likewise.
13762 (edge_badness): Likewise.
13763 (update_edge_key): Likewise.
13764 (inline_small_functions): Likewise.
13765 * ipa-profile.c (ipa_profile): Likewise.
13766 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
13767 (ipa_make_edge_direct_to_target): Likewise.
13768 (remove_described_reference): Likewise.
13769 (ipa_impossible_devirt_target): Likewise.
13770 (propagate_controlled_uses): Likewise.
13771 (ipa_print_node_params): Likewise.
13772 (ipcp_transform_function): Likewise.
13773 * ipa-pure-const.c (pure_const_read_summary): Likewise.
13774 (propagate_pure_const): Likewise.
13775 * ipa-reference.c (generate_summary): Likewise.
13776 (read_write_all_from_decl): Likewise.
13777 (propagate): Likewise.
13778 (ipa_reference_read_optimization_summary): Likewise.
13779 * ipa-utils.c (ipa_merge_profiles): Likewise.
13780 * ipa.c (walk_polymorphic_call_targets): Likewise.
13781 (symbol_table::remove_unreachable_nodes): Likewise.
13782 (ipa_single_use): Likewise.
13783 * passes.c (execute_todo): Likewise.
13784 * predict.c (drop_profile): Likewise.
13785 * symtab.c (symtab_node::get_dump_name): New function.
13786 (symtab_node::dump_name): Likewise.
13787 (symtab_node::dump_asm_name): Likewise.
13788 (symtab_node::dump_references): Likewise.
13789 (symtab_node::dump_referring): Likewise.
13790 (symtab_node::dump_base): Likewise.
13791 (symtab_node::debug_symtab): Likewise.
13792 * tree-sra.c (convert_callers_for_node): Likewise.
13793 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
13794 * value-prof.c (init_node_map): Likewise.
13796 2017-05-23 Martin Liska <mliska@suse.cz>
13798 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
13799 and symtab_node::debug_symtab to symbol_table::debug.
13800 * cgraphunit.c (analyze_functions): Use the renamed function.
13801 (symbol_table::compile): Likewise.
13802 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
13803 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
13804 * passes.c (execute_todo): Likewise.
13805 * symtab.c (symbol_table::dump): New function.
13806 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
13808 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13810 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
13811 that nonconst implies exec.
13813 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13815 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
13816 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
13817 (inline_edge_summary_vec): Turn into ...
13818 (ipa_call_summaries): ... this one.
13819 (redirect_to_unreachable, edge_set_predicate,
13820 evaluate_properties_for_edge, inline_summary_alloc,
13821 reset_ipa_call_summary, reset_inline_summary,
13822 inline_summary_t::duplicate): Update.
13823 (inline_edge_duplication_hook): Turn to ...
13824 (ipa_call_summary_t::duplicate): ... this one.
13825 (inline_edge_removal_hook): Turn to ...
13826 (ipa_call_summary_t::remove): ... this one.
13827 (dump_inline_edge_summary): Turn to ...
13828 (dump_ipa_call_summary): ... this one.
13829 (estimate_function_body_sizes): Update.
13830 (inline_update_callee_summaries): Update.
13831 (remap_edge_change_prob): Update.
13832 (remap_edge_summaries): Update.
13833 (inline_merge_summary): Update.
13834 (do_estimate_edge_time): Update.
13835 (inline_generate_summary): Update.
13836 (inline_read_section): Update.
13837 (inline_read_summary): Update.
13838 (inline_free_summary): Update.
13839 * ipa-inline.c (can_inline_edge_p): Update.
13840 (compute_inlined_call_time): Update.
13841 (want_inline_small_function_p): Update.
13842 (edge_badness): Update.
13843 (early_inliner): Update.
13844 * ipa-inline.h (inline_edge_summary): Turn to ...
13845 (ipa_call_summary): ... this one.
13846 (ipa_call_summary_t): New class.
13847 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
13848 (ipa_call_summaries): New.
13849 (inline_edge_summary): Remove.
13850 (estimate_edge_growth): Update.
13851 * ipa-profile.c (ipa_propagate_frequency_1): Update.
13852 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13853 * ipa-split.c (execute_split_functions): Update.
13854 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
13856 2017-05-23 Tom de Vries <tom@codesourcery.com>
13858 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13859 attributes): Document rdrand effective target.
13861 2017-05-23 Tom de Vries <tom@codesourcery.com>
13863 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13864 attributes): Sort alphabetically.
13866 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
13868 * config/avr/genmultilib.awk: Use gsub instead of gensub.
13870 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
13873 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
13874 V2DF/V2DI splat into two separate patterns, one that handles
13875 registers, and the other that only handles memory. Drop support
13876 for splatting from a GPR on ISA 2.07 and then splitting the
13877 splat into direct move and splat.
13878 (vsx_splat_<mode>_reg): Likewise.
13879 (vsx_splat_<mode>_mem): Likewise.
13881 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
13883 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
13885 2017-05-22 Jakub Jelinek <jakub@redhat.com>
13887 PR middle-end/80809
13888 * omp-low.c (finish_taskreg_remap): New function.
13889 (finish_taskreg_scan): If unit size of ctx->record_type
13890 is non-constant, unshare the size expression and replace
13891 decls in it with possible outer var refs.
13893 PR middle-end/80809
13894 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
13895 GOVD_SHARED rather than GOVD_PRIVATE with it.
13896 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
13897 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
13899 PR middle-end/80853
13900 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
13901 as last argument to build_outer_var_ref for pointer bases of array
13902 section reductions.
13904 2017-05-19 Martin Sebor <msebor@redhat.com>
13906 * print-tree.c (print_node): Print DECL_READ_P flag.
13908 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13910 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
13911 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
13912 * cgraph.c: Likewise.
13913 * cgraphunit.c: Likewise.
13914 * gengtype.c: Likewise.
13915 * ipa-cp.c: Likewise.
13916 * ipa-devirt.c: Likewise.
13917 * ipa-icf.c: Likewise.
13918 * ipa-predicate.c: Likewise.
13919 * ipa-profile.c: Likewise.
13920 * ipa-prop.c: Likewise.
13921 * ipa-split.c: Likewise.
13923 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
13924 edge_predicate_pool, dump_inline_hints,
13925 inline_summary::account_size_time, redirect_to_unreachable,
13926 edge_set_predicate, set_hint_predicate,
13927 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
13928 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
13929 inline_summary_t::remove, remap_hint_predicate_after_duplication,
13930 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
13931 ipa_call_summary_t::remove, initialize_growth_caches,
13932 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
13933 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
13934 mark_modified, unmodified_parm_1, unmodified_parm,
13935 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
13936 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
13937 compute_bb_predicates, will_be_nonconstant_expr_predicate,
13938 will_be_nonconstant_predicate, record_modified_bb_info,
13939 get_minimal_bb, record_modified, param_change_prob,
13940 phi_result_unknown_predicate, predicate_for_phi_result,
13941 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
13942 estimate_function_body_sizes, compute_inline_parameters,
13943 compute_inline_parameters_for_curren, pass_data_inline_parameters,
13944 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
13945 inline_update_callee_summaries, remap_edge_change_prob,
13946 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
13947 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
13948 inline_analyze_function, inline_summary_t::insert,
13949 inline_generate_summary, read_ipa_call_summary, inline_read_section,
13950 inline_read_summary, write_ipa_call_summary, inline_write_summary,
13951 inline_free_summary): Move to ipa-fnsummary.h
13952 (predicate_t): Remove.
13953 * ipa-fnsummary.c: New file.
13954 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
13955 (enum inline_hints_vals, inline_hints, agg_position_info,
13956 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
13957 inline_summaries, ipa_call_summary, ipa_call_summary_t,
13958 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
13959 dump_inline_summary, dump_inline_hints, inline_generate_summary,
13960 inline_read_summary, inline_write_summary, inline_free_summary,
13961 inline_analyze_function, initialize_inline_failed,
13962 inline_merge_summary, inline_update_overall_summary,
13963 compute_inline_parameters): Move to ipa-fnsummary.h
13964 * ipa-fnsummary.h: New file.
13965 * ipa-inline-transform.h: Include ipa-inline.h.
13966 * ipa-inline.c: LIkewise.
13968 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13970 * ipa-inline.c (edge_badness): Use inlined_time instead of
13971 inline_summaries->get.
13973 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13975 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
13977 2017-05-22 Nathan Sidwell <nathan@acm.org>
13979 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
13980 (fdump-lang): Document 'raw' option.
13981 * dumpfile.h (TDI_tu): Delete.
13982 * dumpfile.c (dump_files): Remove translation-unit.
13983 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
13985 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
13987 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
13988 command option from $(AWK) call.
13989 * config/avr/genmultilib.awk: Simplify and rewrite so that it
13990 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
13991 [FORMAT]: Remove handling of variable.
13992 * config/avr/t-multilib: Regenerate.
13994 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13996 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
13998 (dump_inline_summary): Do not print self_time.
13999 (estimate_function_body_sizes): Do not set self_time.
14000 (compute_inline_parameters): Likewise.
14001 (inline_read_section, inline_write_summary): Do not stream self_time.
14002 * ipa-inline.h (inline_summary): Drop self_time.
14004 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14006 * ipa-inline-analysis.c (account_size_time): Rename to ...
14007 (inline_summary::account_size_time): ... this one.
14008 (reset_ipa_call_summary): Turn to ...
14009 (ipa_call_summary::reset): ... this one.
14010 (reset_inline_summary): Turn to ...
14011 (inline_summary::reset): ... this one.
14012 (inline_summary_t::remove): Update.
14013 (inline_summary_t::duplicate): Update.
14014 (ipa_call_summary_t::remove): Update.
14015 (dump_inline_summary): Update.
14016 (estimate_function_body_sizes): Update.
14017 (compute_inline_parameters): Update.
14018 (estimate_node_size_and_time): Update.
14019 (inline_merge_summary): Update.
14020 (inline_update_overall_summary): Update.
14021 (inline_read_section): Update.
14022 (inline_write_summary): Update.
14023 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
14024 add account_size_time and reset member functions.
14025 (ipa_call_summary): Add reset function.
14026 * ipa-predicate.h (predicate::operator &): Constify.
14028 2017-05-22 Richard Biener <rguenther@suse.de>
14030 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
14032 2017-05-19 Jason Merrill <jason@redhat.com>
14034 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
14036 2017-05-19 Marek Polacek <polacek@redhat.com>
14039 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
14040 TYPE_OVERFLOW_WRAPS checks.
14042 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
14044 * tree-core.h (enum omp_clause_default_kind): Add
14045 "OMP_CLAUSE_DEFAULT_PRESENT".
14046 * tree-pretty-print.c (dump_omp_clause): Handle it.
14047 * gimplify.c (enum gimplify_omp_var_data): Add
14048 "GOVD_MAP_FORCE_PRESENT".
14049 (gimplify_adjust_omp_clauses_1): Map it to
14050 "GOMP_MAP_FORCE_PRESENT".
14051 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
14053 * gimplify.c (oacc_default_clause): Clarify.
14055 2017-05-19 Nathan Sidwell <nathan@acm.org>
14057 LANG_HOOK_REGISTER_DUMPS
14058 * toplev.c (general_init): Call register dump lang hook.
14059 * doc/invoke.texi: Document -fdump-lang option family.
14060 * dumpfile.c (dump_files): Remove class dump here.
14061 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
14062 * dumpfile.h (tree_dump_index): Remove TDI_class.
14063 * langhooks-def.h (lhd_register_dumps): Declare.
14064 (LANG_HOOKS_REGISTER_DUMPS): Define.
14065 (LANG_HOOKS_INITIALIZER): Add it.
14066 * langhooks.c (lhd_register_dumps): Define.
14067 * langhooks.h (struct lang_hooks): Add register_dumps.
14069 2017-05-19 Nathan Sidwell <nathan@acm.org>
14071 * context.h (context::set_passes): New.
14072 * context.c (context::context): Do not create pass manager.
14073 * toplev.c (general_init): Create pass manager here.
14075 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
14077 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
14078 use this splitter if two add or or instructions would also work for
14079 the constant we want to generate.
14081 2017-05-19 Richard Biener <rguenther@suse.de>
14084 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
14085 predicate evaluation.
14087 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14089 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
14091 * ipa-inline.c (want_inline_small_function_p): Do not cast to
14094 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14096 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14097 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14098 (inline_edge_summary_vec): Turn into ...
14099 (ipa_call_summaries): ... this one.
14100 (redirect_to_unreachable, edge_set_predicate,
14101 evaluate_properties_for_edge, inline_summary_alloc,
14102 reset_ipa_call_summary, reset_inline_summary,
14103 inline_summary_t::duplicate): Update.
14104 (inline_edge_duplication_hook): Turn to ...
14105 (ipa_call_summary_t::duplicate): ... this one.
14106 (inline_edge_removal_hook): Turn to ...
14107 (ipa_call_summary_t::remove): ... this one.
14108 (dump_inline_edge_summary): Turn to ...
14109 (dump_ipa_call_summary): ... this one.
14110 (estimate_function_body_sizes): Update.
14111 (inline_update_callee_summaries): Update.
14112 (remap_edge_change_prob): Update.
14113 (remap_edge_summaries): Update.
14114 (inline_merge_summary): Update.
14115 (do_estimate_edge_time): Update.
14116 (inline_generate_summary): Update.
14117 (inline_read_section): Update.
14118 (inline_read_summary): Update.
14119 (inline_free_summary): Update.
14120 * ipa-inline.c (can_inline_edge_p): Update.
14121 (compute_inlined_call_time): Update.
14122 (want_inline_small_function_p): Update.
14123 (edge_badness): Update.
14124 (early_inliner): Update.
14125 * ipa-inline.h (inline_edge_summary): Turn to ...
14126 (ipa_call_summary): ... this one.
14127 (ipa_call_summary_t): New class.
14128 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14129 (ipa_call_summaries): New.
14130 (inline_edge_summary): Remove.
14131 (estimate_edge_growth): Update.
14132 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14133 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14134 * ipa-split.c (execute_split_functions): Update.
14135 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14137 2017-05-19 Richard Biener <rguenther@suse.de>
14139 PR middle-end/80764
14140 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
14142 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
14144 * config/rs6000/rs6000.c (struct machine_function): Add field
14145 fpr_is_wrapped_separately.
14146 (rs6000_get_separate_components): Use 64 components. Handle the
14147 new FPR components.
14148 (rs6000_components_for_bb): Handle the FPR components.
14149 (rs6000_emit_prologue_components): Handle the FPR components.
14150 (rs6000_emit_epilogue_components): Handle the FPR components.
14151 (rs6000_set_handled_components): Handle the FPR components.
14152 (rs6000_emit_prologue): Don't output prologue code for those FPRs
14153 that are already separately shrink-wrapped.
14154 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
14155 that are already separately shrink-wrapped.
14157 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
14160 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
14163 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
14164 (define_peephole2 for Altivec d-form load): Add peepholes to catch
14165 cases where the register allocator uses a move and an offsettable
14166 memory operation to/from a FPR register on ISA 2.06/2.07.
14167 (define_peephole2 for Altivec d-form store): Likewise.
14169 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
14172 * config/i386/mmx.md (*mov<mode>_internal): Enable
14173 alternatives 11, 12, 13 and 14 also for 32bit targets.
14174 Remove alternatives 15, 16, 17 and 18.
14175 * config/i386/sse.md (vec_concatv2di): Change
14176 alternative (!x, *y) to (x, ?!*Yn).
14178 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
14180 * dumpfile.h (enum dump_kind): Remove stray comma.
14182 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14184 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
14185 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
14186 predicate::num_conditions
14187 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
14188 (CHANGED): turn into predicate::changed.
14189 (agg_position_info): Move to ipa-predicate.h
14190 (add_condition, predicate::add_clause, predicate::operator &=,
14191 predicate::or_with, predicate::evaluate, predicate::probability,
14192 dump_condition, dump_clause, predicate::dump,
14193 predicate::remap_after_duplication, predicate::remap_after_inlining,
14194 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
14195 (evaluate_conditions_for_known_args): Update.
14196 (set_cond_stmt_execution_predicate): Update.
14197 * ipa-inline.h: Include ipa-predicate.h
14198 (condition, inline_param_summary, conditions, agg_position_info,
14199 predicate): Move to ipa-predicate.h
14200 * ipa-predicate.c: New file.
14201 * ipa-predicate.h: New file.
14203 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
14205 * final.c (leaf_function_p): Check we are not in a sequence.
14207 2017-05-18 Martin Liska <mliska@suse.cz>
14209 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
14210 * dumpfile.c (dump_register): Use new enum dump_kind.
14211 (get_dump_file_name): Likewise.
14212 (dump_enable_all): Likewise.
14213 (dump_switch_p_1): Likewise.
14214 (enable_rtl_dump_file): Remove usage of TDF_RTL.
14215 * dumpfile.h (enum dump_kind): New enum type.
14216 (struct dump_file_info): Create constructor and
14217 format fields and comments.
14218 * passes.c (pass_manager::register_one_dump_file):
14220 * statistics.c (statistics_early_init): Likewise.
14221 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
14222 TDF_TREE with TDF_SLIM.
14223 (gather_memory_references_ref): Likewise.
14225 2017-05-18 Martin Liska <mliska@suse.cz>
14227 * vec.h (struct vnull): Use it.
14229 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14231 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
14232 (true_predicate, false_predicate, true_predicate_p,
14233 false_predicate_p): Remove.
14234 (single_cond_predicate, not_inlined_predicate): Turn to member function
14236 (add_condition): Update.
14237 (add_clause): Turn to...
14238 (predicate::add_clause): ... this one; update; allow passing NULL
14240 (and_predicates): Turn to ...
14241 (predicate::operator &=): ... this one.
14242 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
14243 (or_predicates): Turn to ...
14244 (predicate::or_with): ... this one.
14245 (evaluate_predicate): Turn to ...
14246 (predicate::evaluate): ... this one.
14247 (predicate_probability): Turn to ...
14248 (predicate::probability): ... this one.
14249 (dump_condition): Update.
14250 (dump_predicate): Turn to ...
14251 (predicate::dump): ... this one.
14252 (account_size_time): Update.
14253 (edge_set_predicate): Update.
14254 (set_hint_predicate): UPdate.
14255 (evaluate_conditions_for_known_args): Update.
14256 (evaluate_properties_for_edge): Update.
14257 (remap_predicate_after_duplication): Turn to...
14258 (predicate::remap_after_duplication): ... this one.
14259 (remap_hint_predicate_after_duplication): Update.
14260 (inline_summary_t::duplicate): UPdate.
14261 (dump_inline_edge_summary): Update.
14262 (dump_inline_summary): Update.
14263 (set_cond_stmt_execution_predicate): Update.
14264 (set_switch_stmt_execution_predicate): Update.
14265 (compute_bb_predicates): Update.
14266 (will_be_nonconstant_expr_predicate): Update.
14267 (will_be_nonconstant_predicate): Update.
14268 (phi_result_unknown_predicate): Update.
14269 (predicate_for_phi_result): Update.
14270 (array_index_predicate): Update.
14271 (estimate_function_body_sizes): Update.
14272 (estimate_node_size_and_time): Update.
14273 (estimate_ipcp_clone_size_and_time): Update.
14274 (remap_predicate): Rename to ...
14275 (predicate::remap_after_inlining): ... this one.
14276 (remap_hint_predicate): Update.
14277 (inline_merge_summary): Update.
14278 (inline_update_overall_summary): Update.
14279 (estimate_size_after_inlining): Update.
14280 (read_predicate): Rename to ...
14281 (predicate::stream_in): ... this one.
14282 (read_inline_edge_summary): Update.
14283 (write_predicate): Rename to ...
14284 (predicate::stream_out): ... this one.
14285 (write_inline_edge_summary): Update.
14286 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
14287 (clause_t): Turn to uint32_t
14288 (predicate): Turn to class; implement constructor and operators
14290 (size_time_entry): Update.
14291 (inline_summary): Update.
14292 (inline_edge_summary): Update.
14294 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
14296 * fold-const.c (fold_binary_loc): Move transformation...
14297 * match.pd (C - X CMP X): ... here.
14299 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
14301 * config/sparc/sparc.c (sparc_option_override): Set function
14302 alignment for -mcpu=niagara7 to 64 to match the I$ line.
14303 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
14305 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
14307 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
14309 2017-05-18 Marek Polacek <polacek@redhat.com>
14312 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
14313 (pass_ubsan::execute): Call gimple_assign_single_p instead of
14314 gimple_assign_load_p.
14316 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
14318 PR middle-end/80692
14319 * real.c (do_compare): Give decimal_do_compare preference over
14320 comparing just the signs.
14322 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
14324 * doc/md.texi (Canonicalization of Instructions): Describe the
14325 canonical form of instructions that inherently set a condition
14328 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
14330 PR middle-end/80775
14331 * tree-cfg.c: Move deletion of unreachable case statements to after
14332 the merging of consecutive case labels.
14334 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
14336 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
14337 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
14338 restoring of callee-saved registers.
14340 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
14342 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
14343 * config/visium/visium.c (single_set_and_flags): Likewise.
14344 * config/visium/visium.md (Substitutions): Likewise.
14346 2017-05-17 Martin Liska <mliska@suse.cz>
14348 * cfg.c: Introduce dump_flags_t type and
14349 use it instead of int type.
14351 * cfghooks.c: Likewise.
14352 * cfghooks.h (struct cfg_hooks): Likewise.
14353 * cfgrtl.c: Likewise.
14354 * cfgrtl.h: Likewise.
14355 * cgraph.c (cgraph_node::get_body): Likewise.
14356 * coretypes.h: Likewise.
14357 * domwalk.c: Likewise.
14358 * domwalk.h: Likewise.
14359 * dumpfile.c (struct dump_option_value_info): Likewise.
14360 (dump_enable_all): Likewise.
14361 (dump_switch_p_1): Likewise.
14362 (opt_info_switch_p): Likewise.
14363 * dumpfile.h (enum tree_dump_index): Likewise.
14364 (struct dump_file_info): Likewise.
14365 * genemit.c: Likewise.
14366 * generic-match-head.c: Likewise.
14367 * gengtype.c (open_base_files): Likewise.
14368 * gimple-pretty-print.c: Likewise.
14369 * gimple-pretty-print.h: Likewise.
14370 * graph.c (print_graph_cfg): Likewise.
14371 * graphite-scop-detection.c (dot_all_sese): Likewise.
14372 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
14373 * loop-unroll.c (report_unroll): Likewise.
14374 * passes.c (pass_manager::register_one_dump_file): Likewise.
14375 * print-tree.c: Likewise.
14376 * statistics.c: Likewise.
14377 * tree-cfg.c: Likewise.
14378 * tree-cfg.h: Likewise.
14379 * tree-dfa.c: Likewise.
14380 * tree-dfa.h: Likewise.
14381 * tree-dump.c (dump_function): Likewise.
14382 * tree-dump.h (struct dump_info): Likewise.
14383 * tree-pretty-print.c: Likewise.
14384 * tree-pretty-print.h: Likewise.
14385 * tree-ssa-live.c: Likewise.
14386 * tree-ssa-live.h: Likewise.
14387 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
14388 * tree-vect-loop.c: Likewise.
14389 * tree-vect-slp.c: Likewise.
14391 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
14392 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14394 PR tree-optimization/80457
14395 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
14396 of all arguments to a statement as scalar_to_vec operations.
14397 (vectorizable_call): Adjust call to vect_model_simple_cost for
14399 (vectorizable_conversion): Likewise.
14400 (vectorizable_assignment): Likewise.
14401 (vectorizable_shift): Likewise.
14402 (vectorizable_operation): Likewise.
14403 (vectorizable_comparison): Likewise.
14404 (vect_is_simple_cond): Record the def types for operands.
14405 (vectorizable_condition): Likewise, call vect_model_simple_cost.
14406 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
14407 for statement argument count.
14409 2017-05-16 Carl Love <cel@us.ibm.com>
14411 * config/rs6000/rs6000-c: Add support for built-in functions
14412 vector unsigned long long vec_bperm (vector unsigned long long,
14413 vector unsigned char)
14414 vector signed long long vec_mule (vector signed int,
14416 vector unsigned long long vec_mule (vector unsigned int,
14417 vector unsigned int)
14418 vector signed long long vec_mulo (vector signed int,
14420 vector unsigned long long vec_mulo (vector unsigned int,
14421 vector unsigned int)
14422 vector signed char vec_sldw (vector signed char,
14423 vector signed char,
14425 vector unsigned char vec_sldw (vector unsigned char,
14426 vector unsigned char,
14428 vector signed short vec_sldw (vector signed short,
14429 vector signed short,
14431 vector unsigned short vec_sldw (vector unsigned short,
14432 vector unsigned short,
14434 vector signed int vec_sldw (vector signed int,
14437 vector unsigned int vec_sldw (vector unsigned int,
14438 vector unsigned int,
14440 vector signed long long vec_sldw (vector signed long long,
14441 vector signed long long,
14443 vector unsigned long long vec_sldw (vector unsigned long long,
14444 vector unsigned long long,
14446 * config/rs6000/rs6000-c: Add support for built-in functions
14447 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
14448 * config/rs6000/altivec.h: Add defintion for vec_sldw.
14449 * doc/extend.texi: Update the built-in documentation for the
14450 new built-in functions.
14452 2017-05-16 Marek Polacek <polacek@redhat.com>
14456 * tree.c (save_expr): Don't fold the expression.
14458 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
14460 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
14461 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
14462 and (?*y,m). Update insn attributes.
14464 2017-05-16 Martin Liska <mliska@suse.cz>
14466 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
14467 flags argument of print_gimple_stmt, print_gimple_expr,
14468 print_generic_stmt and print_generic_expr.
14469 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
14470 * coretypes.h: Likewise.
14471 * except.c (dump_eh_tree): Likewise.
14472 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14473 * gimple-pretty-print.h: Likewise.
14474 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
14475 (backprop::push_to_worklist): Likewise.
14476 (backprop::pop_from_worklist): Likewise.
14477 (backprop::process_use): Likewise.
14478 (backprop::intersect_uses): Likewise.
14479 (note_replacement): Likewise.
14480 * gimple-ssa-store-merging.c
14481 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
14482 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
14483 (pass_store_merging::execute): Likewise.
14484 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
14485 (ssa_base_cand_dump_callback): Likewise.
14486 (dump_incr_vec): Likewise.
14487 (replace_refs): Likewise.
14488 (replace_mult_candidate): Likewise.
14489 (create_add_on_incoming_edge): Likewise.
14490 (create_phi_basis): Likewise.
14491 (insert_initializers): Likewise.
14492 (all_phi_incrs_profitable): Likewise.
14493 (introduce_cast_before_cand): Likewise.
14494 (replace_one_candidate): Likewise.
14495 * gimplify.c (gimplify_expr): Likewise.
14496 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
14497 (set_rename): Likewise.
14498 (rename_uses): Likewise.
14499 (copy_loop_phi_nodes): Likewise.
14500 (add_close_phis_to_merge_points): Likewise.
14501 (copy_loop_close_phi_args): Likewise.
14502 (copy_cond_phi_args): Likewise.
14503 (graphite_copy_stmts_from_block): Likewise.
14504 (translate_pending_phi_nodes): Likewise.
14505 * graphite-poly.c (print_pdr): Likewise.
14506 (dump_gbb_cases): Likewise.
14507 (dump_gbb_conditions): Likewise.
14508 (print_scop_params): Likewise.
14509 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
14510 (build_cross_bb_scalars_use): Likewise.
14511 (gather_bbs::before_dom_children): Likewise.
14512 * hsa-dump.c (dump_hsa_immed): Likewise.
14513 * ipa-cp.c (print_ipcp_constant_value): Likewise.
14514 (get_replacement_map): Likewise.
14515 * ipa-inline-analysis.c (dump_condition): Likewise.
14516 (estimate_function_body_sizes): Likewise.
14517 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
14518 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
14519 * ipa-prop.c (ipa_dump_param): Likewise.
14520 (ipa_print_node_jump_functions_for_edge): Likewise.
14521 (ipa_modify_call_arguments): Likewise.
14522 (ipa_modify_expr): Likewise.
14523 (ipa_dump_param_adjustments): Likewise.
14524 (ipa_dump_agg_replacement_values): Likewise.
14525 (ipcp_modif_dom_walker::before_dom_children): Likewise.
14526 * ipa-pure-const.c (check_stmt): Likewise.
14527 (pass_nothrow::execute): Likewise.
14528 * ipa-split.c (execute_split_functions): Likewise.
14529 * omp-offload.c (dump_oacc_loop_part): Likewise.
14530 (dump_oacc_loop): Likewise.
14531 * trans-mem.c (tm_log_emit): Likewise.
14532 (tm_memopt_accumulate_memops): Likewise.
14533 (dump_tm_memopt_set): Likewise.
14534 (dump_tm_memopt_transform): Likewise.
14535 * tree-cfg.c (gimple_verify_flow_info): Likewise.
14536 (print_loop): Likewise.
14537 * tree-chkp-opt.c (chkp_print_addr): Likewise.
14538 (chkp_gather_checks_info): Likewise.
14539 (chkp_get_check_result): Likewise.
14540 (chkp_remove_check_if_pass): Likewise.
14541 (chkp_use_outer_bounds_if_possible): Likewise.
14542 (chkp_reduce_bounds_lifetime): Likewise.
14543 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
14544 (chkp_mark_completed_bounds): Likewise.
14545 (chkp_register_incomplete_bounds): Likewise.
14546 (chkp_mark_invalid_bounds): Likewise.
14547 (chkp_maybe_copy_and_register_bounds): Likewise.
14548 (chkp_build_returned_bound): Likewise.
14549 (chkp_get_bound_for_parm): Likewise.
14550 (chkp_build_bndldx): Likewise.
14551 (chkp_get_bounds_by_definition): Likewise.
14552 (chkp_generate_extern_var_bounds): Likewise.
14553 (chkp_get_bounds_for_decl_addr): Likewise.
14554 * tree-chrec.c (chrec_apply): Likewise.
14555 * tree-data-ref.c (dump_data_reference): Likewise.
14556 (dump_subscript): Likewise.
14557 (dump_data_dependence_relation): Likewise.
14558 (analyze_overlapping_iterations): Likewise.
14559 * tree-inline.c (expand_call_inline): Likewise.
14560 (tree_function_versioning): Likewise.
14561 * tree-into-ssa.c (dump_defs_stack): Likewise.
14562 (dump_currdefs): Likewise.
14563 (dump_names_replaced_by): Likewise.
14564 (dump_update_ssa): Likewise.
14565 (update_ssa): Likewise.
14566 * tree-object-size.c (pass_object_sizes::execute): Likewise.
14567 * tree-parloops.c (build_new_reduction): Likewise.
14568 (try_create_reduction_list): Likewise.
14569 (ref_conflicts_with_region): Likewise.
14570 (oacc_entry_exit_ok_1): Likewise.
14571 (oacc_entry_exit_single_gang): Likewise.
14572 * tree-pretty-print.h: Likewise.
14573 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
14574 (get_scalar_evolution): Likewise.
14575 (add_to_evolution): Likewise.
14576 (get_loop_exit_condition): Likewise.
14577 (analyze_evolution_in_loop): Likewise.
14578 (analyze_initial_condition): Likewise.
14579 (analyze_scalar_evolution): Likewise.
14580 (instantiate_scev): Likewise.
14581 (number_of_latch_executions): Likewise.
14582 (gather_chrec_stats): Likewise.
14583 (final_value_replacement_loop): Likewise.
14584 (scev_const_prop): Likewise.
14585 * tree-sra.c (dump_access): Likewise.
14586 (disqualify_candidate): Likewise.
14587 (create_access): Likewise.
14588 (reject): Likewise.
14589 (maybe_add_sra_candidate): Likewise.
14590 (create_access_replacement): Likewise.
14591 (analyze_access_subtree): Likewise.
14592 (analyze_all_variable_accesses): Likewise.
14593 (sra_modify_assign): Likewise.
14594 (initialize_constant_pool_replacements): Likewise.
14595 (find_param_candidates): Likewise.
14596 (decide_one_param_reduction): Likewise.
14597 (replace_removed_params_ssa_names): Likewise.
14598 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
14599 * tree-ssa-copy.c (dump_copy_of): Likewise.
14600 (copy_prop_visit_cond_stmt): Likewise.
14601 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
14602 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
14603 (record_equivalences_from_stmt): Likewise.
14604 * tree-ssa-dse.c (compute_trims): Likewise.
14605 (delete_dead_call): Likewise.
14606 (delete_dead_assignment): Likewise.
14607 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
14608 (forward_propagate_into_cond): Likewise.
14609 (pass_forwprop::execute): Likewise.
14610 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14611 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
14613 (move_computations_worker): Likewise.
14614 (execute_sm): Likewise.
14615 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
14616 (remove_exits_and_undefined_stmts): Likewise.
14617 (remove_redundant_iv_tests): Likewise.
14618 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
14619 (adjust_iv_update_pos): Likewise.
14620 * tree-ssa-math-opts.c (bswap_replace): Likewise.
14621 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
14622 (value_replacement): Likewise.
14623 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
14624 * tree-ssa-pre.c (print_pre_expr): Likewise.
14625 (get_representative_for): Likewise.
14626 (create_expression_by_pieces): Likewise.
14627 (insert_into_preds_of_block): Likewise.
14628 (eliminate_insert): Likewise.
14629 (eliminate_dom_walker::before_dom_children): Likewise.
14630 (eliminate): Likewise.
14631 (remove_dead_inserted_code): Likewise.
14632 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
14633 * tree-ssa-reassoc.c (get_rank): Likewise.
14634 (eliminate_duplicate_pair): Likewise.
14635 (eliminate_plus_minus_pair): Likewise.
14636 (eliminate_not_pairs): Likewise.
14637 (undistribute_ops_list): Likewise.
14638 (eliminate_redundant_comparison): Likewise.
14639 (update_range_test): Likewise.
14640 (optimize_range_tests_var_bound): Likewise.
14641 (optimize_vec_cond_expr): Likewise.
14642 (rewrite_expr_tree): Likewise.
14643 (rewrite_expr_tree_parallel): Likewise.
14644 (linearize_expr): Likewise.
14645 (break_up_subtract): Likewise.
14646 (linearize_expr_tree): Likewise.
14647 (attempt_builtin_powi): Likewise.
14648 (attempt_builtin_copysign): Likewise.
14649 (transform_stmt_to_copy): Likewise.
14650 (transform_stmt_to_multiply): Likewise.
14651 (dump_ops_vector): Likewise.
14652 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
14653 (print_scc): Likewise.
14654 (set_ssa_val_to): Likewise.
14655 (visit_reference_op_store): Likewise.
14656 (visit_use): Likewise.
14657 (sccvn_dom_walker::before_dom_children): Likewise.
14658 (run_scc_vn): Likewise.
14659 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14661 (expr_hash_elt::print): Likewise.
14662 (const_and_copies::pop_to_marker): Likewise.
14663 (const_and_copies::record_const_or_copy_raw): Likewise.
14664 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
14665 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
14666 (dump_predicates): Likewise.
14667 (find_uninit_use): Likewise.
14668 (warn_uninitialized_phi): Likewise.
14669 (pass_late_warn_uninitialized::execute): Likewise.
14670 * tree-ssa.c (verify_vssa): Likewise.
14671 (verify_ssa): Likewise.
14672 (maybe_optimize_var): Likewise.
14673 * tree-vrp.c (dump_value_range): Likewise.
14674 (dump_all_value_ranges): Likewise.
14675 (dump_asserts_for): Likewise.
14676 (register_edge_assert_for_2): Likewise.
14677 (vrp_visit_cond_stmt): Likewise.
14678 (vrp_visit_switch_stmt): Likewise.
14679 (vrp_visit_stmt): Likewise.
14680 (vrp_visit_phi_node): Likewise.
14681 (simplify_cond_using_ranges_1): Likewise.
14682 (fold_predicate_in): Likewise.
14683 (evrp_dom_walker::before_dom_children): Likewise.
14684 (evrp_dom_walker::push_value_range): Likewise.
14685 (evrp_dom_walker::pop_value_range): Likewise.
14686 (execute_early_vrp): Likewise.
14688 2017-05-16 Richard Biener <rguenther@suse.de>
14690 * dwarf2out.c (loc_list_from_tree_1): Do not create
14691 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
14693 2017-05-16 Richard Biener <rguenther@suse.de>
14695 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
14697 (note_variable_value_in_expr): If we resolved the decl ref
14698 do not push to the stack.
14700 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
14702 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
14703 operations in fast-math mode.
14704 (vaddq_f16): Likewise.
14705 (vmul_f16): Likewise.
14706 (vmulq_f16): Likewise.
14707 (vsub_f16): Likewise.
14708 (vsubq_f16): Likewise.
14709 * config/arm/neon.md (add<mode>3): New.
14711 (fma:<VH:mode>3): New. Also remove outdated comment.
14714 2017-05-16 Martin Liska <mliska@suse.cz>
14718 * ipa-devirt.c (warn_types_mismatch): Fix typo.
14719 (odr_types_equivalent_p): Likewise.
14721 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
14723 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
14725 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
14728 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
14729 non-interunit SSE move alternatives with '?'.
14730 (zero-extendsidi peephole2): New peephole to skip intermediate
14731 general register in SSE zero-extend sequence.
14733 2017-05-15 Jeff Law <law@redhat.com>
14735 * reorg.c (relax_delay_slots): Create a new variable to hold
14736 the temporary target rather than clobbering TARGET_LABEL.
14738 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
14739 missing argument to extract_bit_field call.
14740 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
14742 2017-05-15 Martin Liska <mliska@suse.cz>
14745 * gcc.c (process_command): Do not allow empty argument of -o option.
14747 2017-05-15 Renlin Li <renlin.li@arm.com>
14749 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
14750 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
14751 * config/aarch64/constraints.md (Usf): Add long call check.
14752 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
14753 (call_value): Likewise.
14754 (sibcall): Likewise.
14755 (sibcall_value): Likewise.
14757 (call_value_insn): New.
14758 (sibcall_insn): Update rtx pattern.
14759 (sibcall_value_insn): Likewise.
14760 (call_internal): Remove.
14761 (call_value_internal): Likewise.
14762 (sibcall_internal): Likewise.
14763 (sibcall_value_internal): Likewise.
14764 (call_reg): Likewise.
14765 (call_symbol): Likewise.
14766 (call_value_reg): Likewise.
14767 (call_value_symbol): Likewise.
14769 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
14772 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
14774 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
14776 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
14777 compatible with CCGOCmode and with CCZmode.
14779 2017-05-14 Martin Sebor <msebor@redhat.com>
14781 PR middle-end/77671
14782 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
14783 (gimple_fold_builtin_snprintf): Same.
14784 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
14785 (gimple_fold_builtin_snprintf): Same.
14786 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
14787 of character types.
14788 (is_call_safe): New function.
14789 (try_substitute_return_value): Call it.
14790 (try_simplify_call): New function.
14791 (pass_sprintf_length::handle_gimple_call): Call it.
14793 2017-05-14 Martin Sebor <msebor@redhat.com>
14795 PR middle-end/80669
14796 * builtins.c (expand_builtin_stpncpy): Simplify.
14798 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
14800 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
14801 * config/i386/i386.h
14802 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
14803 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
14804 (struct machine_function): Add new members call_ms2sysv,
14805 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
14806 (struct machine_frame_state): New fields sp_realigned and
14807 sp_realigned_offset.
14808 * config/i386/i386.c
14809 (enum xlogue_stub): New enum.
14810 (enum xlogue_stub_sets): New enum.
14811 (class xlogue_layout): New class.
14812 (struct ix86_frame): New fields stack_realign_allocate_offset,
14813 stack_realign_offset and outlined_save_offset. Modify comments to
14814 detail stack layout when using out-of-line stubs.
14815 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
14816 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
14817 -mcall-ms2sysv-xlogues.
14818 (stub_managed_regs): New static variable.
14819 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
14820 registers managed by out-of-line stub.
14821 (disable_call_ms2sysv_xlogues): New function.
14822 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
14823 m->call_ms2sysv when appropriate and compute frame layout for
14825 (sp_valid_at, fp_valid_at): New inline functions.
14826 (choose_basereg): New function.
14827 (choose_baseaddr): Add align parameter, use choose_basereg and modify
14829 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
14830 Use align parameter of choose_baseaddr to generated aligned SSE movs
14832 (pro_epilogue_adjust_stack): Modify to track
14833 machine_frame_state::sp_realigned.
14834 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
14835 (ix86_nsaved_sseregs): Likewise.
14836 (ix86_emit_save_regs): Likewise.
14837 (ix86_emit_save_regs_using_mov): Likewise.
14838 (ix86_emit_save_sse_regs_using_mov): Likewise.
14839 (get_scratch_register_on_entry): Likewise.
14840 (gen_frame_set): New function.
14841 (gen_frame_load): Likewise.
14842 (gen_frame_store): Likewise.
14843 (emit_outlined_ms2sysv_save): Likewise.
14844 (emit_outlined_ms2sysv_restore): Likewise.
14845 (ix86_expand_prologue): Modify stack re-alignment code and call
14846 emit_outlined_ms2sysv_save when appropriate.
14847 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
14848 parameter rtx_insn *insn, which allows the function to be used to only
14849 generate the notes.
14850 (ix86_expand_epilogue): Modify validity checks of frame and stack
14851 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
14852 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
14853 * config/i386/predicates.md
14854 (save_multiple): New predicate.
14855 (restore_multiple): Likewise.
14856 * config/i386/sse.md
14857 (save_multiple<mode>): New pattern.
14858 (save_multiple_realign<mode>): Likewise.
14859 (restore_multiple<mode>): Likewise.
14860 (restore_multiple_and_return<mode>): Likewise.
14861 (restore_multiple_leave_return<mode>): Likewise.
14862 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
14864 2017-05-14 Julia Koval <julia.koval@intel.com>
14866 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
14867 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
14868 (__builtin_ia32_xsetbv): New builtins.
14869 * config/i386/i386.c (ix86_expand_special_args_builtin):
14871 (ix86_expand_builtin): Special expand for new intrinsics.
14872 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
14873 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
14874 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
14876 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14878 * cfganal.c (inverted_post_order_compute): Change argument type
14880 * cfganal.h (inverted_post_order_compute): Adjust prototype.
14881 * df-core.c (rest_of_handle_df_initialize): Adjust.
14882 (rest_of_handle_df_finish): Likewise.
14883 (df_analyze_1): Likewise.
14884 (df_analyze): Likewise.
14885 (loop_inverted_post_order_compute): Change argument to be a vec *.
14886 (df_analyze_loop): Adjust.
14887 (df_get_n_blocks): Likewise.
14888 (df_get_postorder): Likewise.
14889 * df.h (struct df_d): Change field to be a vec.
14890 * lcm.c (compute_laterin): Adjust.
14891 (compute_available): Likewise.
14892 * lra-lives.c (lra_create_live_ranges_1): Likewise.
14893 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
14894 * tree-ssa-pre.c (compute_antic): Likewise.
14896 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14898 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
14899 (depth_first_search::depth_first_search): Change structure init
14900 function to this constructor.
14901 (depth_first_search::add_bb): Rename function to this member.
14902 (depth_first_search::execute): Likewise.
14903 (flow_dfs_compute_reverse_finish): Adjust.
14905 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14907 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
14908 (longest_simple_path): Likewise.
14909 * shrink-wrap.c (spread_components): Likewise.
14910 (disqualify_problematic_components): Likewise.
14911 (emit_common_heads_for_components): Likewise.
14912 (emit_common_tails_for_components): Likewise.
14913 (insert_prologue_epilogue_for_components): Likewise.
14915 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14917 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
14920 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14922 * df-core.c (df_set_blocks): Start using auto_bitmap.
14923 (df_compact_blocks): Likewise.
14924 * df-problems.c (df_rd_confluence_n): Likewise.
14925 * df-scan.c (df_insn_rescan_all): Likewise.
14926 (df_process_deferred_rescans): Likewise.
14927 (df_update_entry_block_defs): Likewise.
14928 (df_update_exit_block_uses): Likewise.
14929 (df_entry_block_bitmap_verify): Likewise.
14930 (df_exit_block_bitmap_verify): Likewise.
14931 (df_scan_verify): Likewise.
14932 * lra-constraints.c (lra_constraints): Likewise.
14933 (undo_optional_reloads): Likewise.
14934 (lra_undo_inheritance): Likewise.
14935 * lra-remat.c (calculate_gen_cands): Likewise.
14936 (do_remat): Likewise.
14937 * lra-spills.c (assign_spill_hard_regs): Likewise.
14938 (spill_pseudos): Likewise.
14939 * tree-ssa-pre.c (bitmap_set_and): Likewise.
14940 (bitmap_set_subtract_values): Likewise.
14942 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14944 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
14945 management with auto_bitmap.
14946 (fix_inter_tick): Likewise.
14947 (fix_recovery_deps): Likewise.
14948 * ira.c (add_store_equivs): Likewise.
14949 (find_moveable_pseudos): Likewise.
14950 (split_live_ranges_for_shrink_wrap): Likewise.
14951 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
14952 (rtx_reuse_manager::seen_def_p): Likewise.
14953 (rtx_reuse_manager::set_seen_def): Likewise.
14954 * print-rtl.h (class rtx_reuse_manager): Likewise.
14956 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14958 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
14960 (migrate_btr_def): Likewise.
14961 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
14962 * df-core.c (loop_post_order_compute): Likewise.
14963 (loop_inverted_post_order_compute): Likewise.
14964 * hsa-common.h: Likewise.
14965 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
14966 * init-regs.c (initialize_uninitialized_regs): Likewise.
14967 * ipa-inline.c (resolve_noninline_speculation): Likewise.
14968 (inline_small_functions): Likewise.
14969 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
14970 * ira.c (combine_and_move_insns): Likewise.
14971 (build_insn_chain): Likewise.
14972 * loop-invariant.c (find_invariants): Likewise.
14973 * lower-subreg.c (propagate_pseudo_copies): Likewise.
14974 * predict.c (tree_predict_by_opcode): Likewise.
14975 (predict_paths_leading_to): Likewise.
14976 (predict_paths_leading_to_edge): Likewise.
14977 (estimate_loops_at_level): Likewise.
14978 (estimate_loops): Likewise.
14979 * shrink-wrap.c (try_shrink_wrapping): Likewise.
14980 (spread_components): Likewise.
14981 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
14982 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
14983 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
14984 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
14985 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
14986 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
14987 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
14988 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
14989 (mark_threaded_blocks): Likewise.
14990 (thread_through_all_blocks): Likewise.
14991 * tree-ssa.c (verify_ssa): Likewise.
14992 (execute_update_addresses_taken): Likewise.
14993 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
14995 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14997 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
14999 (post_order_compute): Likewise.
15000 (inverted_post_order_compute): Likewise.
15001 (pre_and_rev_post_order_compute_fn): Likewise.
15003 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15005 * genrecog.c (int_set::int_set): Explicitly construct our
15006 auto_vec base class.
15007 * vec.h (auto_vec::auto_vec): New constructor.
15009 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15011 * bitmap.h (class auto_bitmap): New constructor taking
15012 bitmap_obstack * argument.
15014 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15016 * bitmap.h (class auto_bitmap): Change type of m_bits to
15017 bitmap_head, and adjust ctor / dtor and member operators.
15019 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
15021 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
15022 when returned register mode doesn't match original mode.
15024 2017-05-12 Jeff Law <law@redhat.com>
15025 Jakub Jelinek <jakub@redhat.com>
15027 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
15028 we look for cc setter after the compare-elim changes.
15029 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
15030 within the vector to match what compare-elim now expects.
15031 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
15032 (xorsi3_flags, one_cmplsi2_flags): Likewise.
15034 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
15035 after the compare-elim changes.
15036 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
15037 the vector to match what compare-elim now expects.
15038 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
15039 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
15040 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
15041 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
15042 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
15044 * config/visium/visium.c (single_set_and_flags): Fix where
15045 we look for cc setter after the compare-elim changes.
15046 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
15047 with the vector to match what compare-elim now expects.
15048 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
15049 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
15050 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
15051 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
15052 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
15053 (neg<mode>2_insn_set_overflow): Likewise.
15055 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
15057 PR middle-end/79794
15058 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
15059 maybe_expand_insn call, set ops[0].target. If still set after call,
15060 set alt_rtl. Add extra arg to recursive calls.
15061 (extract_bit_field): Add alt_rtl argument. Pass to
15063 * expmed.h (extract_bit_field): Fix prototype.
15064 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
15065 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
15066 to extract_bit_field_calls.
15067 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
15068 Pass alt_rtl to extract_bit_field calls.
15069 * calls.c (store_unaligned_arguments_into_psuedos)
15070 load_register_parameters): Pass extra NULL to extract_bit_field calls.
15071 * optabs.c (maybe_legitimize_operand): Clear op->target when call
15073 * optabs.h (struct expand_operand): Add target bitfield.
15075 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15077 * compare-elim.c (try_eliminate_compare): Canonicalize
15078 operation with embedded compare to
15079 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
15080 (set (reg) (operation)].
15082 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
15084 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15087 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
15088 cost of adding a carry flag for ADC instruction.
15089 [case MINUS]: Ignore the cost of subtracting a carry flag
15090 for SBB instruction.
15092 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
15094 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
15096 * config/rs6000/bmiintrin.h: New file.
15097 * config/rs6000/bmi2intrin.h: New file.
15098 * config/rs6000/x86intrin.h: New file.
15100 2017-05-12 Jeff Law <law@redhat.com>
15102 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
15105 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
15107 PR middle-end/80707
15108 * tree-cfg.c: Remove cfg edges of unreachable case statements.
15110 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15112 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15113 early expansion of vector divide builtins.
15114 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
15115 builtins identified as having unsigned arguments.
15117 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15119 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
15120 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
15121 expansion of vector logical operations (and, andc, or, xor,
15124 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15126 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
15127 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
15129 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15131 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15132 early GIMPLE expansion of vector multiplies.
15134 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15136 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
15137 TARGET_HAVE_MOVT conditional.
15138 (movt splitter): Likewise.
15140 2017-05-12 Richard Biener <rguenther@suse.de>
15142 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
15143 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15144 Fold all stmts not inplace.
15146 2017-05-12 Richard Biener <rguenther@suse.de>
15148 PR tree-optimization/80713
15149 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
15150 inserted_exprs bit for not removed stmts.
15152 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
15154 PR middle-end/69921
15155 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
15156 parallelized" attribute for parallelized OpenACC kernels.
15157 * omp-offload.c (execute_oacc_device_lower): Use it.
15159 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
15160 Set "oacc kernels" attribute.
15161 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
15162 parameter. Adjust all users.
15163 (oacc_fn_attrib_kernels_p): Remove function.
15164 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
15165 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
15166 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
15167 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
15168 assert "oacc kernels" attribute is set.
15170 2017-05-11 Carl Love <cel@us.ibm.com>
15172 * config/rs6000/rs6000-c: Add support for built-in functions
15173 vector unsigned char vec_popcnt (vector signed char)
15174 vector unsigned char vec_popcnt (vector unsigned char)
15175 vector unsigned short vec_popcnt (vector signed short)
15176 vector unsigned short vec_popcnt (vector unsigned short)
15177 vector unsigned int vec_popcnt (vector signed int)
15178 vector unsigned int vec_popcnt (vector unsigned int)
15179 vector unsigned long long vec_popcnt (vector signed long long)
15180 vector unsigned long long vec_popcnt (vector unsigned long long)
15181 vector signed long long vec_slo (vector signed long long,
15182 vector signed char)
15183 vector signed long long vec_slo (vector signed long long,
15184 vector unsigned char)
15185 vector unsigned long long vec_slo (vector unsigned long long,
15186 vector signed char)
15187 vector unsigned long long vec_slo (vector unsigned long long,
15188 vector unsigned char)
15189 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
15190 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
15191 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
15192 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
15193 * doc/extend.texi: Update the built-in documentation file for the
15194 new built-in functions.
15196 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
15198 * attribs.h (sorted_attr_string): Move machine independent
15199 functions for target clone support from the i386 port to common
15200 code. Rename ix86_function_versions to common_function_versions.
15201 Rename make_name to make_unique_name.
15202 (common_function_versions): Likewise.
15203 (make_unique_name): Likewise.
15204 (make_dispatcher_decl): Likewise.
15205 (is_function_default_version): Likewise.
15206 * attribs.c (attr_strcmp): Likewise.
15207 (sorted_attr_string): Likewise.
15208 (common_function_versions): Likewise.
15209 (make_unique_name): Likewise.
15210 (make_dispatcher_decl): Likewise.
15211 (is_function_default_version): Likewise.
15212 * config/i386/i386.c (attr_strcmp): Likewise.
15213 (sorted_attr_string): Likewise.
15214 (ix86_function_versions): Likewise.
15215 (make_name): Likewise.
15216 (make_dispatcher_decl): Likewise.
15217 (is_function_default_version): Likewise.
15218 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
15220 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15223 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15224 Account for direct move costs for vec_construct of integer
15227 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
15230 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
15231 (UNSPEC_STX_ATOMIC): Ditto.
15232 (loaddi_via_sse): New insn.
15233 (storedi_via_sse): Ditto.
15234 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
15235 Update corresponding peephole2 patterns.
15236 (atomic_storedi_fpu): Ditto.
15238 2017-05-11 Julia Koval <julia.koval@intel.com>
15240 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
15241 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
15243 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
15244 (__builtin_ia32_rsqrt14ss_mask): New builtins.
15245 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
15247 2017-05-11 Nathan Sidwell <nathan@acm.org>
15249 * graphite-poly.c: Include dumpfile.h.
15251 * dumpfle.h (dump_function): Declare here ...
15252 * tree-dump.h (dump_function): ... not here.
15253 * dumpfile.c: #include tree-cfg.h.
15254 (dump_function): Move here from ...
15255 * tree-dump.c (dump_function): ... here.
15256 * gimplify.c: #include splay-tree.h, not tree-dump.h.
15257 * graphite-poly.c: Don't include tree-dump.h.
15258 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
15259 * print-tree.c: Likewise.
15260 * stor-layout.c: Likewise.
15261 * tree-nested.c: Likewise.
15263 * dumpfile.c (dump_start): Use TDF_FLAGS.
15264 (dump_enable_all): Fix TDF_KIND check thinko.
15266 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
15268 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15269 array entries to represent two legal parameterizations of the
15270 overloaded __builtin_cmpb function, as represented by the
15271 P6_OV_BUILTIN_CMPB constant.
15272 (altivec_resolve_overloaded_builtin): Add special case handling
15273 for the __builtin_cmpb function, as represented by the
15274 P6_OV_BUILTIN_CMPB constant.
15275 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
15276 (BU_P6_64BIT_2): New macro.
15277 (BU_P6_OVERLOAD_2): New macro
15278 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
15279 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
15280 (CMPB): Add overload support to represent both 32-bit and 64-bit
15281 compare-bytes function.
15282 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15283 support for TARGET_CMPB.
15284 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
15285 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15286 documentation of the __builtin_cmpb overloaded built-in function.
15288 2017-05-11 Richard Biener <rguenther@suse.de>
15290 PR tree-optimization/80705
15291 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
15292 bases are not vectorizable.
15294 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15296 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
15297 when counting register pressure.
15299 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15301 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
15302 (iv_ca_more_deps): Renamed to ...
15303 (iv_ca_compare_deps): ... this.
15304 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
15306 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15308 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
15310 (determine_group_iv_costs): ... here.
15311 (find_inv_vars_cb): Record inv var if it's not recorded before.
15313 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15315 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
15316 (get_shiftadd_cost): Ditto.
15318 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15320 * tree-ssa-address.c: Include header file.
15321 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
15323 (add_to_parts): Refactor.
15324 (addr_to_parts): New parameter. Update use of move_hint_to_base.
15325 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
15328 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15330 PR tree-optimization/53090
15331 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
15333 (extract_cond_operands): Detect condition with IV on both sides
15334 and return COMP_IV_EXPR_2.
15335 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
15336 (rewrite_use_compare): Simplify by removing call to function
15337 extract_cond_operands.
15339 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15341 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
15342 (extract_cond_operands): Detect condition comparing against non-
15343 invariant bound and return appropriate enum value.
15344 (find_interesting_uses_cond): Update use of extract_cond_operands.
15345 Handle its return value accordingly.
15346 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
15348 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15350 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
15351 nonlinear iv_use computation in loop invariant sensitive way.
15353 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15355 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
15356 (find_iv_candidates): Call relate_compare_use_with_all_cands.
15358 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15360 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
15361 (dump_cand): Support iv_cand.inv_exprs.
15362 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
15364 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
15367 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15369 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
15371 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
15372 as local function. Include necessary header files.
15373 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
15375 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15377 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
15379 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15381 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
15382 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
15383 RSHIFT_EXPR and BIT_NOT_EXPR.
15385 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15387 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
15388 (adjust_setup_cost): New parameter supporting round up adjustment.
15389 (struct address_cost_data): Delete.
15390 (force_expr_to_var_cost): Don't bound cost with spill_cost.
15391 (split_address_cost, ptr_difference_cost): Delete.
15392 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
15393 (struct ainc_cost_data): New struct.
15394 (get_address_cost_ainc): New function.
15395 (get_address_cost, get_computation_cost): Reimplement.
15396 (determine_group_iv_cost_address): Record inv_expr for all uses of
15398 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
15399 (iv_ca_has_deps): Reimplemented to ...
15400 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
15402 (iv_ca_extend): Call iv_ca_more_deps.
15404 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15406 * tree-ssa-address.c (struct mem_address): Move to header file.
15407 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
15408 * tree-ssa-address.h (struct mem_address): Move from C file.
15409 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
15411 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15413 * tree-affine.h (aff_combination_type): New interface.
15414 (aff_combination_zero_p): Remove static.
15415 (aff_combination_const_p): New interface.
15416 (aff_combination_singleton_var_p): New interfaces.
15418 2017-05-11 Richard Biener <rguenther@suse.de>
15420 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15421 Skip unreachable blocks and destinations.
15422 (eliminate): Move stmt removal and fixup ...
15423 (fini_eliminate): ... here. Skip inserted exprs.
15424 (pass_pre::execute): Move fini_pre after fini_eliminate.
15425 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
15426 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
15427 PRE to get rid of dead code that has invalid SSA form and
15428 split critical edges again.
15430 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15432 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15434 2017-05-11 Richard Biener <rguenther@suse.de>
15436 * passes.c (execute_function_todo): Verify loops if they are
15437 said to be up-to-date.
15438 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
15439 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
15441 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
15444 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
15445 handle calling assemble_external ourself.
15448 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
15449 modes with zero size. Enhance comment.
15451 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15453 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
15454 built-ins for vec_xl and vec_xst with short and char pointer
15457 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15459 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
15460 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
15461 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
15462 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
15463 (_mm_maskz_min_round_ss): New intrinsics.
15464 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
15465 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
15466 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
15467 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
15468 (__builtin_ia32_minss_mask_round): New builtins.
15469 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15470 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15471 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
15473 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
15474 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
15476 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
15479 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15481 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
15482 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
15483 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
15484 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
15485 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
15486 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
15487 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
15488 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15489 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
15490 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
15491 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
15492 (__builtin_ia32_mulss_mask_round): New builtins.
15493 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15494 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15495 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
15497 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
15498 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
15500 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
15503 2017-05-10 Julia Koval <julia.koval@intel.com>
15505 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
15506 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
15507 (_mm256_setr_m128i): New intrinsics.
15509 2017-05-10 Julia Koval <julia.koval@intel.com>
15511 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
15512 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
15513 (_mm_maskz_rcp14_ss): New intrinsics.
15514 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
15515 (__builtin_ia32_rcp14ss_mask): New builtins.
15516 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
15518 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
15520 PR tree-optimization/51513
15521 * tree-cfg.c (gimple_seq_unreachable_p): New function.
15522 (assert_unreachable_fallthru_edge_p): Use it.
15523 (group_case_labels_stmt): Likewise.
15524 * tree-cfg.h: Prototype it.
15525 * stmt.c: Include cfghooks.h and tree-cfg.h.
15526 (emit_case_dispatch_table) <gap_label>: New local variable.
15527 Use it to fill dispatch table gaps.
15528 Test for default_label before updating probabilities.
15529 (expand_case) <default_label>: Remove unneeded initialization.
15530 Test for unreachable default case statement and remove its edge.
15531 Set default_label accordingly.
15532 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
15534 2017-05-10 Carl Love <cel@us.ibm.com>
15536 * config/rs6000/rs6000-c: Add support for built-in functions
15537 vector signed char vec_neg (vector signed char)
15538 vector signed short int vec_neg (vector short int)
15539 vector signed int vec_neg (vector signed int)
15540 vector signed long long vec_neg (vector signed long long)
15541 vector float vec_neg (vector float)
15542 vector double vec_neg (vector double)
15543 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
15545 * config/rs6000/altivec.h: Add define for vec_neg
15546 * doc/extend.texi: Update the built-in documentation for the
15547 new built-in functions.
15549 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15551 PR tree-optimization/77644
15552 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
15554 2017-05-10 Nathan Sidwell <nathan@acm.org>
15556 * dumpfile.h (TDI_lang_all): New.
15557 (TDF_KIND): New. Renumber others
15558 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
15560 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
15562 (get_dump_file_name): Adjust suffix generation.
15563 (dump_enable_all): Use TDF_KIND.
15564 * doc/invoke.texi (-fdump-lang-all): Document.
15566 * dumpfile.h: Tabify.
15568 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
15571 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
15572 Move member access before delete.
15574 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
15576 * tree-inline.c (expand_call_inline): Split block at stmt
15579 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
15582 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
15583 are now unused after splitting mov{sf,sd}_hardfloat.
15584 (f32_lr2): Likewise.
15585 (f32_lm): Likewise.
15586 (f32_lm2): Likewise.
15587 (f32_li): Likewise.
15588 (f32_li2): Likewise.
15589 (f32_lv): Likewise.
15590 (f32_sr): Likewise.
15591 (f32_sr2): Likewise.
15592 (f32_sm): Likewise.
15593 (f32_sm2): Likewise.
15594 (f32_si): Likewise.
15595 (f32_si2): Likewise.
15596 (f32_sv): Likewise.
15597 (f32_dm): Likewise.
15598 (f32_vsx): Likewise.
15599 (f32_av): Likewise.
15600 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
15601 For movsf, order stores so the VSX stores occur before the GPR
15602 store which encourages the register allocator to use a traditional
15603 FPR instead of a GPR. For movsd, order the stores so that the GPR
15604 store comes before the VSX stores to allow the power6 to work.
15605 This is due to the power6 not having a 32-bit integer store
15606 instruction from a FPR.
15607 (movsf_hardfloat): Likewise.
15608 (movsd_hardfloat): Likewise.
15610 2017-05-09 Martin Sebor <msebor@redhat.com>
15612 PR translation/80280
15613 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
15616 PR translation/80280
15617 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
15618 data member added in r247778.
15619 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
15621 2017-05-09 Nathan Sidwell <nathan@acm.org>
15623 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
15625 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
15628 2017-05-09 Marek Polacek <polacek@redhat.com>
15630 * doc/invoke.texi: Fix typo.
15632 2017-05-09 Richard Biener <rguenther@suse.de>
15634 * tree-vrp.c (vrp_val_is_max): Adjust comment.
15635 (vrp_val_is_min): Likewise.
15636 (set_value_range_to_value): Likewise.
15637 (set_value_range_to_nonnegative): Likewise.
15638 (gimple_assign_nonzero_p): Likewise.
15639 (gimple_stmt_nonzero_p): Likewise.
15640 (vrp_int_const_binop): Likewise. Remove unreachable case.
15641 (adjust_range_with_scev): Adjust comments.
15642 (compare_range_with_value): Likewise.
15643 (extract_range_from_phi_node): Likewise.
15644 (test_for_singularity): Likewise.
15646 2017-05-09 Richard Biener <rguenther@suse.de>
15648 * tree-vrp.c (get_single_symbol): Add assert that we don't
15649 get overflowed constants as invariant part.
15650 (compare_values_warnv): Add comment before the TREE_NO_WARNING
15651 checks. Use wi::cmp instead of recursing for integer constants.
15652 (compare_values): Just ignore whether we assumed undefined
15653 overflow instead of failing the compare.
15654 (extract_range_for_var_from_comparison_expr): Add comment before the
15655 TREE_NO_WARNING sets.
15656 (test_for_singularity): Likewise.
15657 (extract_range_from_comparison): Do not disable optimization
15658 when we assumed undefined overflow.
15659 (extract_range_basic): Remove init of unused var.
15661 2017-05-09 Richard Biener <rguenther@suse.de>
15663 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
15664 (extract_range_from_multiplicative_op_1): Adjust.
15665 (extract_range_from_binary_expr_1): Use int_const_binop.
15667 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
15670 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
15671 rs6000_store_data_bypass_p in seven define_bypass directives and
15672 in several comments.
15673 * config/rs6000/rs6000-protos.h: Add prototype for
15674 rs6000_store_data_bypass_p function.
15675 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
15676 function implements slightly different (rs6000-specific) semantics
15677 than store_data_bypass_p, returning false rather than aborting
15678 with assertion error when arguments do not satisfy the
15679 requirements of store data bypass.
15680 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
15681 rs6000_store_data_bypass_p.
15683 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
15685 * config/xtensa/xtensa-protos.h
15686 (xtensa_initial_elimination_offset): New declaration.
15687 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15688 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
15689 macro definition, add case for FRAME_POINTER_REGNUM when
15690 FRAME_GROWS_DOWNWARD.
15691 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
15692 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
15693 xtensa_initial_elimination_offset.
15695 2017-05-08 Nathan Sidwell <nathan@acm.org>
15697 * doc/invoke.texi: Alphabetize -fdump options.
15699 2017-05-08 Martin Sebor <msebor@redhat.com>
15701 PR translation/80280
15702 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
15704 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
15706 * target.def (compute_frame_layout): New optional target hook.
15707 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
15708 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
15709 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
15711 * reload1.c (verify_initial_elim_offsets): Likewise.
15712 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
15713 (use_simple_return_p): Call arm_compute_frame_layout if needed.
15714 (arm_get_frame_offsets): Split up into this ...
15715 (arm_compute_frame_layout): ... and this function.
15717 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
15719 * config/aarch64/constraints.md (Usa): New constraint.
15720 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
15722 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
15724 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
15725 with_multilib_list after it has been checked.
15727 2017-05-08 Richard Biener <rguenther@suse.de>
15729 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
15730 (bitmap_set_subtract_values): Likewise.
15732 2017-05-08 Richard Biener <rguenther@suse.de>
15734 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
15735 (gimple_assign_nonzero): ... this and remove strict_overflow_p
15737 (gimple_stmt_nonzero_warnv_p): Rename to ...
15738 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
15740 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
15741 (extract_range_basic): Adjust, do not disable propagation on
15742 strict overflow sensitive simplification.
15743 (vrp_visit_cond_stmt): Likewise.
15745 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
15747 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
15748 body size unconditionally.
15750 2017-05-07 Jeff Law <law@redhat.com>
15753 2017-05-06 Jeff Law <law@redhat.com>
15754 PR tree-optimization/78496
15755 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15758 PR tree-optimization/78496
15759 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15760 (simplify_stmt_using_ranges): Call it.
15761 (vrp_dom_walker::before_dom_children): Extract equivalences
15762 from an ASSERT_EXPR with an equality comparison against a
15765 2017-05-06 Jeff Law <law@redhat.com>
15767 PR tree-optimization/78496
15768 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15771 PR tree-optimization/78496
15772 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15773 (simplify_stmt_using_ranges): Call it.
15774 (vrp_dom_walker::before_dom_children): Extract equivalences
15775 from an ASSERT_EXPR with an equality comparison against a
15778 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15780 * lra-constraints.c (lra_copy_reg_equiv): New function.
15781 (split_reg): Use it to copy equivalence information from the
15782 original register to the spill register.
15784 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15786 PR rtl-optimization/75964
15787 * simplify-rtx.c (simplify_const_relational_operation): Remove
15788 invalid handling of comparisons of integer ABS.
15790 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
15792 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
15793 initialize to zero.
15794 (init_regs): Remove declaration.
15795 (function_arg_advance_32): Initialize error_p as boolean variable.
15797 2017-05-05 Nathan Sidwell <nathan@acm.org>
15799 * store-motion.c (remove_reachable_equiv_notes): Reformat long
15800 lines. Use for (;;).
15802 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15804 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
15805 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
15806 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
15807 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
15808 VF=2 that require versioning.
15810 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15812 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
15815 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15817 * diagnostic.h (diagnostic_override_option_index): Convert from
15818 macro to inline function.
15820 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15822 * diagnostic.c (last_module_changed_p): New function.
15823 (set_last_module): New function.
15824 (diagnostic_report_current_module): Convert macro usage to
15825 the above functions.
15826 * diagnostic.h (diagnostic_context::last_module): Strengthen
15827 from const line_map * to const line_map_ordinary *.
15828 (diagnostic_last_module_changed): Delete macro.
15829 (diagnostic_set_last_module): Delete macro.
15831 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15833 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
15834 with diagnostic_report_diagnostic.
15835 (diagnostic_n_impl_richloc): Likewise.
15836 * diagnostic.h (report_diagnostic): Delete macro.
15837 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
15838 with diagnostic_report_diagnostic.
15839 * substring-locations.c (format_warning_va): Likewise.
15841 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15843 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
15844 save/restor of format_spec. Move option-printing code to...
15845 (print_option_information): ...this new function, and
15846 reimplement by simply printing to the pretty_printer,
15847 rather than appending to the format string.
15849 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15851 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
15852 handling logic into...
15853 (update_effective_level_from_pragmas): ...this new function.
15855 2017-05-04 Andrew Waterman <andrew@sifive.com>
15857 * config/riscv/riscv.opt (mstrict-align): New option.
15858 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
15859 (SLOW_UNALIGNED_ACCESS): Define.
15860 (riscv_slow_unaligned_access): Declare.
15861 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
15863 (riscv_slow_unaligned_access): New variable.
15864 (rocket_tune_info): Set slow_unaligned_access to true.
15865 (optimize_size_tune_info): Set slow_unaligned_access to false.
15866 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
15867 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
15868 (riscv_option_override): Set riscv_slow_unaligned_access.
15869 * doc/invoke.texi: Add -mstrict-align to RISC-V.
15871 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
15873 * config/riscv/riscv.md: Unify indentation.
15875 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15880 * config/rs6000/rs6000.md (u code attribute): Add FIX and
15882 (extendsi<mode>2): Add support for doing sign extension via
15883 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
15884 don't have ISA 3.0 instructions.
15885 (extendsi<mode>2 splitter): Likewise.
15886 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
15887 generate the normal insns since SImode can now go in vector
15888 registers. Disallow the special UNSPECs needed for previous
15889 machines to hide SImode being used. Add new insns
15890 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
15891 (fix_trunc<mode>si2_stfiwx): Likewise.
15892 (fix_trunc<mode>si2_internal): Likewise.
15893 (fixuns_trunc<mode>si2): Likewise.
15894 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15895 (fctiw<u>z_<mode>_smallint): Likewise.
15896 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
15897 of floating point to 32-bit integer from doing a direct move to
15898 the GPR registers to do a store.
15899 (fctiwz_<mode>): Break long line.
15901 2017-05-05 Bin Cheng <bin.cheng@arm.com>
15903 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
15904 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
15905 (addr_list, addr_offset_valid_p): New.
15906 (split_address_groups): Check offset validity with above function.
15907 (gt-tree-ssa-loop-ivopts.h): Include header file.
15909 2017-05-05 Nathan Sidwell <nathan@acm.org>
15911 * config.gcc (arm*-*-*): Add missing 'fi'.
15913 2017-05-05 Steve Ellcey <sellcey@cavium.com>
15915 * doc/invoke.texi (-fopt-info): Explicitly say order of options
15916 included in -fopt-info does not matter.
15917 * doc/optinfo.texi (-fopt-info): Fix description of default
15918 behavour. Explicitly say order of options included in -fopt-info
15921 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
15923 * config.gcc: Allow combinations of aprofile and rmprofile values for
15924 --with-multilib-list.
15925 * config/arm/t-multilib: New file.
15926 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
15927 variables. Remove setting of ISA and floating-point ABI in
15928 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
15929 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
15930 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
15931 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
15933 * config/arm/t-rmprofile: Likewise except for the matches changes.
15934 * doc/install.texi (--with-multilib-list): Document the combination of
15935 aprofile and rmprofile values and warn about pitfalls in doing that.
15937 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
15939 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
15940 (movdi_aarch64): Likewise.
15942 2017-05-05 Jakub Jelinek <jakub@redhat.com>
15944 PR tree-optimization/80632
15945 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
15947 (build_arrays): Initialize it for virtual phis.
15948 (fix_phi_nodes): Use it for virtual phis.
15950 PR tree-optimization/80558
15951 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
15952 [x, y] op z into [x op, y op z] for op & or | if conditions
15955 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
15956 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15959 * config/arm/arm.md (use_literal_pool): Remove.
15960 (64-bit immediate split): No longer takes cost into consideration
15961 if arm_disable_literal_pool is enabled.
15962 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
15963 used when arm_disable_literal_pool is enabled.
15964 (arm_max_const_double_inline_cost): Remove use of
15965 arm_disable_literal_pool.
15966 (push_minipool_fix): Add assert.
15967 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
15968 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
15969 (no_literal_pool_sf_immediate): New.
15971 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15973 PR tree-optimization/80613
15974 * tree-ssa-dce.c (propagate_necessity): Remove cases for
15975 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
15977 2017-05-05 Richard Biener <rguenther@suse.de>
15979 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
15981 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
15983 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
15984 of this flag from insn conditions due to removal from r247495.
15986 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
15988 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
15990 (arm_early_store_addr_dep_ptr): Likewise.
15991 * config/arm/aarch-common-protos.h
15992 (arm_early_load_addr_dep_ptr): Add prototype.
15993 (arm_early_store_addr_dep_ptr): Likewise.
15994 * config/arm/cortex-a53.md: Add new bypasses.
15996 2017-05-05 Jakub Jelinek <jakub@redhat.com>
15998 * tree.c (next_type_uid): Change type to unsigned.
15999 (type_hash_canon): Decrement back next_type_uid if
16000 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
16001 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
16004 2017-05-04 Martin Sebor <msebor@redhat.com>
16006 * builtins.c: Fix a trivial typo in a comment.
16008 PR middle-end/79234
16009 * builtins.c (check_sizes): Adjust to handle reading past the end.
16010 Avoid printing excessive upper bound of ranges. Use %E to print
16011 tree nodes instead of converting them to %wu.
16012 (expand_builtin_memchr): New function.
16013 (compute_dest_size): Rename...
16014 (compute_objsize): ...to this.
16015 (expand_builtin_memcpy): Adjust.
16016 (expand_builtin_mempcpy): Adjust.
16017 (expand_builtin_strcat): Adjust.
16018 (expand_builtin_strcpy): Adjust.
16019 (check_strncat_sizes): Adjust.
16020 (expand_builtin_strncat): Adjust.
16021 (expand_builtin_strncpy): Adjust and simplify.
16022 (expand_builtin_memset): Adjust.
16023 (expand_builtin_bzero): Adjust.
16024 (expand_builtin_memcmp): Adjust.
16025 (expand_builtin): Handle memcmp.
16026 (maybe_emit_chk_warning): Check strncat just once.
16028 2017-05-04 Martin Sebor <msebor@redhat.com>
16030 PR preprocessor/79214
16031 PR middle-end/79222
16032 PR middle-end/79223
16033 * builtins.c (check_sizes): Add inlining context and issue
16034 warnings even when -Wno-system-headers is set.
16035 (check_strncat_sizes): Same.
16036 (expand_builtin_strncat): Same.
16037 (expand_builtin_memmove): New function.
16038 (expand_builtin_stpncpy): Same.
16039 (expand_builtin): Handle memmove and stpncpy.
16041 2017-05-04 Bin Cheng <bin.cheng@arm.com>
16043 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
16044 which is not used any more.
16046 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16048 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
16050 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16052 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
16053 (cortexa53_tunings): Likewise.
16054 (cortexa57_tunings): Likewise.
16055 (cortexa72_tunings): Likewise.
16056 (cortexa73_tunings): Likewise.
16058 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16060 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
16061 Set loop alignment to 8.
16063 2017-05-04 Martin Sebor <msebor@redhat.com>
16065 PR translation/80280
16066 * builtins.c (expand_builtin_object_size): Add missing quoting to
16067 %D and like directives.
16068 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
16069 (hsa_type_for_tree_type): Same.
16070 (verify_function_arguments): Same.
16071 * symtab.c (symbol_table::change_decl_assembler_name): Same.
16072 * varasm.c (get_section): Same.
16075 2017-05-04 Martin Sebor <msebor@redhat.com>
16077 PR translation/80280
16078 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
16080 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16082 * config/aarch64/aarch64.c (generic_addrcost_table):
16083 Change HI/TI mode setting.
16085 2017-05-04 Martin Jambor <mjambor@suse.cz>
16087 PR tree-optimization/80622
16088 * tree-sra.c (comes_initialized_p): New function.
16089 (build_accesses_from_assign): Only set write lazily when
16090 comes_initialized_p is false.
16091 (analyze_access_subtree): Use comes_initialized_p.
16092 (propagate_subaccesses_across_link): Assert !comes_initialized_p
16093 instead of testing for PARM_DECL.
16095 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16097 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
16098 constraint on operand 0 to allow more general addressing modes.
16099 Adjust output template.
16100 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
16102 * config/aarch64/aarch64-protos.h
16103 (aarch64_address_valid_for_prefetch_p): Declare prototype.
16104 * config/aarch64/constraints.md (Dp): New address constraint.
16105 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
16108 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
16110 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
16111 update use of estimate_ipcp_clone_size_and_time.
16112 (estimate_local_effects): Update use of
16113 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
16114 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
16115 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
16116 Return nonspecialized time.
16118 2017-05-04 Richard Biener <rguenther@suse.de>
16120 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
16121 for the last VUSE which def dominates the PHI. Directly call
16123 (get_continuation_for_phi_1): Remove.
16125 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
16127 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
16128 to explain the use of truncating division. Cap the number of
16129 iterations to the maximum given by nb_iterations_upper_bound,
16132 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16134 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
16135 * configure: Regenerate.
16136 * config.in: Regenerate.
16137 * config/i386/driver-mingw32.c: new file.
16138 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
16139 * config.host: Link driver-mingw32.o on MinGW host.
16140 * doc/install.texi: Document new --enable-mingw-wildcard configure
16143 2017-05-04 Marek Polacek <polacek@redhat.com>
16145 PR tree-optimization/80612
16146 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
16148 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16149 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
16151 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
16152 (movt splitter): Likewise.
16153 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
16154 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
16155 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
16156 block for Thumb-1 with MOVT.
16157 (thumb2_legitimate_address_p): Move code block ...
16158 (can_avoid_literal_pool_for_label_p): ... into this new function.
16159 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
16161 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
16162 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
16163 "M-profile targets with the MOVT instruction".
16165 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16167 * config/arm/arm-builtins.c (arm_init_builtins): Rename
16168 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
16169 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
16171 2017-05-04 Martin Liska <mliska@suse.cz>
16173 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
16174 variable cond_code.
16176 2017-05-04 Richard Biener <rguenther@suse.de>
16178 * tree.c (array_at_struct_end_p): Handle arrays at struct
16179 end with flexarrays more conservatively. Refactor and treat
16180 arrays of arrays or aggregates more strict. Fix
16181 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
16182 * tree.c (array_at_struct_end_p): Adjust prototype.
16183 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
16184 * gimple-fold.c (get_range_strlen): Likewise.
16185 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
16187 2017-05-04 Richard Biener <rguenther@suse.de>
16189 PR tree-optimization/31130
16190 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
16192 (supports_overflow_infinity): Likewise.
16193 (is_negative_overflow_infinity): Likewise.
16194 (is_positive_overflow_infinity): Likewise.
16195 (is_overflow_infinity): Likewise.
16196 (stmt_overflow_infinity): Likewise.
16197 (overflow_infinity_range_p): Likewise.
16198 (usable_range_p): Remove as always returning true.
16199 (make_overflow_infinity): Remove.
16200 (negative_overflow_infinity): Likewise.
16201 (positive_overflow_infinity): Likewise.
16202 (avoid_overflow_infinity): Likewise.
16203 (set_value_range): Adjust accordingly.
16204 (set_value_range_to_nonnegative): Likewise, remove now unused
16205 overflow_infinity arg.
16206 (vrp_operand_equal_p): Adjust.
16207 (update_value_range): Likewise.
16208 (range_int_cst_singleton_p): Likewise.
16209 (operand_less_p): Likewise.
16210 (compare_values_warnv): Likewise.
16211 (extract_range_for_var_from_comparison_expr): Likewise.
16212 (vrp_int_const_binop): Likewise.
16213 (zero_nonzero_bits_from_vr): Likewise.
16214 (extract_range_from_multiplicative_op_1): Likewise.
16215 (extract_range_from_binary_expr_1): Likewise.
16216 (extract_range_from_unary_expr): Likewise.
16217 (extract_range_from_comparison): Likewise.
16218 (extract_range_basic): Likewise.
16219 (adjust_range_with_scev): Likewise.
16220 (compare_ranges): Likewise.
16221 (compare_range_with_value): Likewise.
16222 (dump_value_range): Likewise.
16223 (test_for_singularity): Likewise, remove strict_overflow_p parameter
16225 (simplify_cond_using_ranges): Adjust.
16227 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16229 * brig-builtins.def: Added a builtin for class_f64.
16230 * builtin-types.def: Added a builtin type needed by class_f64.
16232 2017-05-03 Jason Merrill <jason@redhat.com>
16234 * timevar.def: Add TV_CONSTEXPR.
16236 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16238 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
16240 2017-05-03 Martin Jambor <mjambor@suse.cz>
16242 * ipa-prop.c (ipa_update_after_lto_read): Removed.
16243 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
16244 * ipa-cp.c (ipcp_propagate_stage): Do not call
16245 ipa_update_after_lto_read.
16246 * ipa-inline.c (ipa_inline): Likewise.
16248 2017-05-03 Martin Jambor <mjambor@suse.cz>
16250 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
16251 tag. Added a default constructor and a destructor.
16252 (ipa_edge_args_sum_t): New class;
16253 (ipa_edge_args_sum): Declare.
16254 (ipa_edge_args_vector): Remove declaration.
16255 (IPA_EDGE_REF): Use ipa_edge_args_sum.
16256 (ipa_free_edge_args_substructures): Remove declaration.
16257 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
16258 (ipa_edge_args_info_available_for_edge_p): Likewise.
16259 * ipa-prop.c (ipa_edge_args_vector): Removed.
16260 (edge_removal_hook_holder): Likewise.
16261 (edge_duplication_hook_holder): Likewise.
16262 (ipa_edge_args_sum): New variable.
16263 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
16264 ipa_edge_args_vector.
16265 (ipa_free_edge_args_substructures): Likewise.
16266 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
16267 ipa_edge_args_vector.
16268 (ipa_edge_removal_hook): Turned into method
16269 ipa_edge_args_sum_t::remove.
16270 (ipa_edge_duplication_hook): Turned into method
16271 ipa_edge_args_sum_t::duplicate.
16272 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
16273 registering edge hooks.
16274 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
16275 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
16276 ipa_edge_args_sum instead of ipa_edge_args_vector.
16277 * ipa-profile.c (ipa_profile): Likewise.
16279 2017-05-03 Martin Jambor <mjambor@suse.cz>
16281 * symbol-summary.h (function_summary): New method exists.
16282 (function_summary::symtab_removal): Deallocate through release.
16283 (call_summary): New class.
16284 (gt_ggc_mx): New overload.
16285 (gt_pch_nx): Likewise.
16286 (gt_pch_nx): Likewise.
16288 2017-05-03 Jeff Law <law@redhat.com>
16290 PR tree-optimization/78496
16291 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
16292 from simplify_cond_using_ranges. Split off code to walk
16293 backwards through casts into ...
16294 (simplify_cond_using_ranges_2): New function.
16295 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
16296 (execute_vrp): After identifying jump threads, call
16297 simplify_cond_using_ranges_2.
16299 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
16302 * ipa-inline.h (inline_summary): Add ctor.
16303 (create_ggc): Do not use ggc_cleared_alloc.
16305 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
16306 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16308 * gcc.c (handle_braces): Support escaping in switch matching
16310 * doc/invoke.texi (Spec Files): Document it.
16311 Remove superfluous @code markup in items.
16313 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16315 * diagnostic-show-locus.c (struct column_range): New struct.
16316 (get_affected_columns): New function.
16317 (get_printed_columns): New function.
16318 (struct correction): New struct.
16319 (correction::ensure_capacity): New function.
16320 (correction::ensure_terminated): New function.
16321 (struct line_corrections): New struct.
16322 (line_corrections::~line_corrections): New dtor.
16323 (line_corrections::add_hint): New function.
16324 (layout::print_trailing_fixits): Reimplement in terms of the new
16326 (selftest::test_overlapped_fixit_printing): New function.
16327 (selftest::diagnostic_show_locus_c_tests): Call it.
16329 2017-05-03 Nathan Sidwell <nathan@acm.org>
16331 Canonicalize canonical type hashing
16332 * tree.h (type_hash_canon_hash): Declare.
16333 * tree.c (type_hash_list, attribute_hash_list): Move into
16334 type_hash_canon_hash.
16335 (build_type_attribute_qual_variant): Break out hash code calc into
16336 type_hash_canon_hash.
16337 (type_hash_canon_hash): New. Generic type hash computation.
16338 (build_range_type_1, build_array_type_1, build_function_type,
16339 build_method_type_directly, build_offset_type, build_complex_type,
16340 make_vector_type): Call it.
16342 2017-05-03 Richard Biener <rguenther@suse.de>
16344 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
16345 When all DRs have unknown misaligned do not always peel
16346 when there is a store but apply the same costing model as if
16347 there were only loads.
16349 2017-05-03 Richard Biener <rguenther@suse.de>
16352 PR tree-optimization/80492
16353 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16354 compare_base_decls returning dont-know properly.
16356 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
16358 * config/arm/iterators.md (CCSI): New mode iterator.
16359 (arch): New mode attribute.
16360 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
16361 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
16362 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
16363 code iterator for success result mode.
16364 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
16365 the corresponding new insn generators.
16367 2017-05-03 Bin Cheng <bin.cheng@arm.com>
16370 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16371 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16373 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
16375 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
16376 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
16377 (DDR_A): Wrap DDR argument in brackets.
16378 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
16379 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
16380 (DDR_REVERSED_P): Likewise.
16382 2017-05-03 Jakub Jelinek <jakub@redhat.com>
16384 PR tree-optimization/79472
16385 * tree-switch-conversion.c (struct switch_conv_info): Add
16386 contiguous_range and default_case_nonstandard fields.
16387 (collect_switch_conv_info): Compute contiguous_range and
16388 default_case_nonstandard fields, don't clear final_bb if
16389 contiguous_range and only the default case doesn't have the required
16391 (check_all_empty_except_final): Set default_case_nonstandard instead
16392 of failing if contiguous_range and the default case doesn't have empty
16394 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
16395 and only the default case doesn't have the required constants. Skip
16397 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
16398 if default_case_nonstandard.
16399 (build_constructors): Build constant 1 just once. Assert that default
16400 values aren't inserted in between cases if contiguous_range. Skip
16402 (build_arrays): Skip virtual phis.
16403 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
16404 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
16405 Handle virtual phis.
16406 (gen_inbound_check): Handle default_case_nonstandard case.
16407 (process_switch): Adjust check_final_bb caller. Call
16408 gather_default_values with the first non-default case instead of
16409 default case if default_case_nonstandard.
16411 2017-05-02 Nathan Sidwell <nathan@acm.org>
16413 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
16414 check. Fix formatting.
16416 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
16418 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
16419 errors when comparing specialized and unspecialized times.
16421 2017-05-02 David Malcolm <dmalcolm@redhat.com>
16423 * diagnostic-show-locus.c
16424 (layout::should_print_annotation_line_p): Make private.
16425 (layout::print_annotation_line): Make private.
16426 (layout::annotation_line_showed_range_p): Make private.
16427 (layout::show_ruler): Make private.
16428 (layout::print_source_line): Make private. Pass in line and
16429 line_width, rather than calling location_get_source_line. Drop
16431 (layout::print_leading_fixits): New method.
16432 (layout::print_any_fixits): Rename to...
16433 (layout::print_trailing_fixits): ...this, and make private.
16434 Don't print newline fixits.
16435 (diagnostic_show_locus): Move logic for printing one row into...
16436 (layout::print_line): ...this new function. Move the
16437 location_get_source_line call and error-handling from
16438 print_source_line to here. Call print_leading_fixits, and rename
16439 print_any_fixits to print_trailing_fixits.
16440 (selftest::test_fixit_insert_containing_newline): Update now that
16441 newlines are partially supported.
16442 (selftest::test_fixit_insert_containing_newline_2): New test.
16443 (selftest::test_fixit_replace_containing_newline): Update comments.
16444 (selftest::diagnostic_show_locus_c_tests): Call the new test.
16445 * edit-context.c (class added_line): New class.
16446 (class edited_line): Describe newline handling in comment.
16447 (edited_line::actually_edited_p): New method.
16448 (edited_line::print_content): Delete redundant decl.
16449 (edited_line::m_predecessors): New field.
16450 (edited_file::print_content): Call edited_line::print_content.
16451 (edited_file::print_diff): Update to support newlines.
16452 (edited_file::print_diff_hunk): Likewise.
16453 (edited_file::print_run_of_changed_lines): New function.
16454 (edited_file::print_diff_line): Convert to...
16455 (print_diff_line): ...this.
16456 (edited_file::get_effective_line_count): New function.
16457 (edited_line::edited_line): Initialize new field m_predecessors.
16458 (edited_line::~edited_line): Clean up m_predecessors.
16459 (edited_line::apply_fixit): Handle newlines.
16460 (edited_line::get_effective_line_count): New function.
16461 (edited_line::print_content): New function.
16462 (edited_line::print_diff_lines): New function.
16463 (selftest::test_applying_fixits_insert_containing_newline): New
16465 (selftest::test_applying_fixits_replace_containing_newline): New
16467 (selftest::insert_line): New function.
16468 (selftest::test_applying_fixits_multiple_lines): Add example of
16470 (selftest::edit_context_c_tests): Call the new tests.
16472 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16474 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
16475 parameter cand. Update dump information.
16476 (get_computation_cost): Update uses.
16478 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16480 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
16481 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
16482 (get_computation_at, rewrite_use_address): Update use of
16483 get_computation_aff.
16485 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16487 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
16488 (get_computation): Delete.
16489 (get_computation_cost): Implement like get_computation_cost_at.
16490 Use get_computation_at.
16491 (get_computation_cost_at): Delete.
16492 (rewrite_use_nonlinear_expr): Use get_computation_at.
16493 (rewrite_use_compare, remove_unused_ivs): Ditto.
16495 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16497 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
16499 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16501 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
16502 (ivopts_global_cost_for_size): Rename parameter and update uses.
16503 (iv_ca_recount_cost): Update uses.
16504 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
16505 candidates seperately in n_invs and n_cands.
16506 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
16508 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16510 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
16511 (find_inv_vars_cb): New.
16512 (find_depends): Renamed to ...
16513 (find_inv_vars): ... this.
16514 (add_candidate_1, force_var_cost): Call find_inv_vars.
16515 (split_address_cost, determine_group_iv_cost_cond): Ditto.
16517 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16519 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
16520 inv_vars. Add inv_exprs.
16521 (struct iv_cand): Rename depends_on to inv_vars.
16522 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
16523 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
16524 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
16525 (dump_cand): Dump inv_vars.
16526 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
16527 (record_invariant, find_depends, add_candidate_1): Ditto.
16528 (set_group_iv_cost, force_var_cost): Ditto.
16529 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
16530 (get_computation_cost_at, get_computation_cost): Ditto.
16531 (determine_group_iv_cost_generic): Ditto.
16532 (determine_group_iv_cost_address): Ditto.
16533 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
16534 (determine_group_iv_costs): Ditto.
16535 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
16536 (iv_ca_set_remove_invariants): Renamed to ...
16537 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
16538 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
16539 (iv_ca_set_add_invariants): Renamed to ...
16540 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
16541 (iv_ca_set_cp): Use iv_ca_set_add_invs.
16542 (iv_ca_has_deps): Support inv_vars and inv_exprs.
16543 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
16544 (create_new_ivs): Remove useless dump.
16546 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16548 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
16550 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
16551 (iv_ca_set_no_cp, create_new_iv): Ditto.
16553 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16555 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16557 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16559 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
16560 function tree_check2.
16562 2017-05-02 Martin Liska <mliska@suse.cz>
16564 * doc/gcov.texi: Add missing preposition.
16565 * gcov.c (function_info::function_info): Properly fill up
16566 all member variables.
16568 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16570 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
16572 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16574 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
16576 2017-05-02 Martin Liska <mliska@suse.cz>
16579 * lto-streamer-in.c (lto_read_tree_1): Remove
16580 LTO_STREAMER_DEBUG.
16581 * lto-streamer.c (struct tree_hash_entry): Likewise.
16582 (struct tree_entry_hasher): Likewise.
16583 (tree_entry_hasher::hash): Likewise.
16584 (tree_entry_hasher::equal): Likewise.
16585 (lto_streamer_init): Likewise.
16586 (lto_orig_address_map): Likewise.
16587 (lto_orig_address_get): Likewise.
16588 (lto_orig_address_remove): Likewise.
16589 * lto-streamer.h: Likewise.
16590 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
16591 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
16593 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
16595 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
16596 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
16597 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
16598 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
16599 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
16600 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
16601 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
16602 (mm_maskz_sub_ss): New intrinsics.
16603 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16604 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16605 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
16606 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
16607 (__builtin_ia32_subss_mask_round): New builtins.
16608 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16609 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16610 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
16612 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
16613 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16615 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16618 2017-05-02 Martin Jambor <mjambor@suse.cz>
16620 PR tree-optimization/78687
16621 * tree-sra.c (access): New field parent.
16622 (process_subtree_disqualification): New function.
16623 (disqualify_candidate): Call it.
16624 (build_accesses_from_assign): Reset write flag if creating an
16626 (build_access_subtree): Fill in parent field and also prpagate
16627 down grp_write flag.
16628 (create_artificial_child_access): New parameter set_grp_write, set
16629 grp_write to its value.
16630 (propagate_subaccesses_across_link): Also propagate grp_write flag
16632 (propagate_all_subaccesses): Push the closest parent back to work
16633 queue if add_access_to_work_queue returned true.
16635 2017-05-02 Richard Biener <rguenther@suse.de>
16637 * common.opt (fstrict-overflow): Alias negative to fwrapv.
16638 * doc/invoke.texi (fstrict-overflow): Remove all traces of
16639 -fstrict-overflow documentation.
16640 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
16641 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
16642 flag_strict_overflow.
16643 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
16644 * lto-opts.c (lto_write_options): Do not stream it.
16645 * lto-wrapper.c (merge_and_complain): Do not handle it.
16646 * opts.c (default_options_table): Do not set -fstrict-overflow.
16647 (finish_options): Likewise do not clear it when sanitizing.
16648 * simplify-rtx.c (simplify_const_relational_operation): Do not
16649 test flag_strict_overflow.
16651 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16653 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
16654 using enabled attribute.
16655 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
16656 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
16657 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
16658 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
16659 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
16660 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
16661 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
16662 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
16663 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
16664 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
16666 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16668 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
16670 2017-05-02 Richard Biener <rguenther@suse.de>
16672 PR tree-optimization/80591
16674 2017-04-10 Richard Biener <rguenther@suse.de>
16676 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16679 2017-05-02 Richard Biener <rguenther@suse.de>
16681 PR tree-optimization/80549
16682 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
16683 (cleanup_tree_cfg_noloop): Create forwarders to known loop
16684 headers if they do not have a preheader.
16686 2017-05-02 Martin Liska <mliska@suse.cz>
16689 * common.opt: Fix typo.
16690 * doc/invoke.texi: Likewise.
16692 2017-05-01 Jan Beulich <jbeulich@suse.com>
16694 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
16695 swapping, add (x,x,m,x,n) alternative.
16697 2017-05-01 Nathan Sidwell <nathan@acm.org>
16699 * calls.c (combine_pending_stack_adjustment_and_call): Remove
16700 unnecessary unadjusted_alignment check.
16702 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16705 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
16707 * gimplify.c (gimplify_cilk_detach): New function.
16708 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
16709 * tree-core.h: Document EXPR_CILK_SPAWN.
16710 * tree.h (EXPR_CILK_SPAWN): Define.
16712 2017-05-01 David Malcolm <dmalcolm@redhat.com>
16714 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
16715 to use new fixit_hint representation, using the "replace" logic.
16716 (get_line_span_for_fixit_hint): Likewise.
16717 (layout::print_any_fixits): Likewise.
16718 (selftest::test_one_liner_many_fixits): Rename to...
16719 (selftest::test_one_liner_many_fixits_1): ...this, and update
16720 comment and expected output to reflect that the multiple fix-it
16721 hints are now consolidated into one insertion.
16722 (selftest::test_one_liner_many_fixits_2): New test.
16723 (selftest::test_diagnostic_show_locus_one_liner): Update for
16725 (selftest::test_fixit_consolidation): Update for fix-it API
16727 * diagnostic.c (print_parseable_fixits): Likewise.
16728 * edit-context.c (edited_line::m_line_events): Convert from
16729 auto_vec <line_event *> to auto_vec <line_event>.
16730 (class line_event): Convert from abstract base class to a concrete
16731 class, taking over the role of replace_event.
16732 (class insert_event): Delete.
16733 (class replace_event): Rename to class line_event. Convert to
16735 (edit_context::add_fixits): Reimplement.
16736 (edit_context::apply_insert): Delete.
16737 (edit_context::apply_replace): Rename to...
16738 (edit_context::apply_fixit): ...this. Convert to half-open range.
16739 (edited_file::apply_insert): Delete.
16740 (edited_file::apply_replace): Rename to...
16741 (edited_file::apply_fixit): ...this.
16742 (edited_line::~edited_line): Drop deletion of events.
16743 (edited_line::apply_insert): Delete.
16744 (edited_line::apply_replace): Rename to...
16745 (edited_line::apply_fixit): ...this. Convert to half-open range.
16746 Update for change to type of m_line_events.
16747 * edit-context.h (edit_context::apply_insert): Delete.
16748 (edit_context::apply_replace): Rename to...
16749 (edit_context::apply_fixit): ...this.
16751 2017-05-01 Martin Sebor <msebor@redhat.com>
16753 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
16756 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
16759 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
16760 __get_cpuid_max returns 0.
16761 (__get_cpuid_count): Ditto.
16763 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
16765 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
16766 replacement expression is another instance of one of its arguments.
16768 2017-05-01 Jakub Jelinek <jakub@redhat.com>
16771 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
16772 check for stack push/pop autoinc.
16773 * config/i386/i386.c (ix86_agi_dependent): Return false
16774 if the only reason why modified_in_p returned true is that
16775 addr is SP based and set_insn is a push or pop.
16777 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16779 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
16782 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16785 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
16786 (account_size_time): Use two predicates - exec_pred and
16788 (evaluate_conditions_for_known_args): Compute both clause and
16790 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
16791 (inline_summary_t::duplicate): Update.
16792 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
16794 (compute_inline_parameters): Likewise.
16795 (estimate_edge_size_and_time): Update caluclation of time.
16796 (estimate_node_size_and_time): Compute both time and nonspecialized
16798 (estimate_ipcp_clone_size_and_time): Update.
16799 (inline_merge_summary): Update.
16800 (do_estimate_edge_time): Update.
16801 (do_estimate_edge_size): Update.
16802 (do_estimate_edge_hints): Update.
16803 (inline_read_section, inline_write_summary): Stream both new predicates.
16804 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
16806 (compute_inlined_call_time): Cleanup.
16807 (big_speedup_p): Update.
16808 (edge_badness): Update.
16809 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
16810 (size_time_entry): Replace predicate by exec_predicate and
16811 nonconst_predicate.
16812 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
16813 (estimate_edge_time): Return also nonspec_time.
16814 (reset_edge_growth_cache): Update.
16816 2017-04-29 Jakub Jelinek <jakub@redhat.com>
16818 PR rtl-optimization/80491
16819 * ifcvt.c (noce_process_if_block): When looking for x setter
16820 with missing else_bb, don't check only the insn right before
16821 cond_earliest, but look for the last insn that x is modified in
16822 within the same bb.
16824 PR rtl-optimization/80491
16825 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
16827 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
16829 PR tree-optimization/80487
16830 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
16832 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16834 PR tree-optimization/79697
16835 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
16836 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
16837 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
16839 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
16840 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
16842 2017-04-28 Martin Sebor <msebor@redhat.com>
16844 PR tree-optimization/80523
16845 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
16846 (init_target_to_host_charmap, target_to_host, target_strtol10): New
16848 (maybe_warn, format_directive, parse_directive): Use new functions.
16849 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
16851 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
16853 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
16855 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
16857 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
16858 target_header_dir): Set correctly.
16859 * configure: Regenerated.
16860 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
16861 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
16862 instead of SYSTEM_HEADER_DIR.
16864 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
16866 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
16867 (estimate_local_effects): Likewise.
16868 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
16869 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
16870 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
16871 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
16872 do_estimate_edge_time, estimate_edge_time): Likewise.
16873 * ipa-inline-analysis.c (estimate_node_size_and_time,
16874 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
16875 (estimate_time_after_inlining): Remove.
16877 2017-04-28 Martin Liska <mliska@suse.cz>
16879 * doc/gcov.texi: Enhance documentation of gcov.
16881 2017-04-28 Martin Liska <mliska@suse.cz>
16883 * doc/gcov.texi: Sort options in alphabetic order.
16884 * doc/gcov-dump.texi: Likewise.
16885 * doc/gcov-tool.texi: Likewise.
16886 * gcov.c (print_usage): Likewise.
16887 * gcov-dump.c (print_usage): Likewise.
16888 * gcov-tool.c (print_merge_usage_message): Likewise.
16889 (print_rewrite_usage_message): Likewise.
16890 (print_overlap_usage_message): Likewise.
16892 2017-04-28 Martin Liska <mliska@suse.cz>
16894 PR gcov-profile/53915
16895 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
16897 2017-04-28 Martin Liska <mliska@suse.cz>
16899 PR gcov-profile/79891
16900 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
16901 is marked by compiler as living on a line.
16902 (get_cycles_count): Remove usage of the union.
16903 (output_intermediate_file): Likewise.
16904 (find_source): Fix GNU coding style.
16905 (accumulate_line_counts): Remove old non-all block mode.
16906 (output_lines): Remove usage of the union.
16907 * profile.c (output_location): Include all BBs, even if
16908 belonging to a same line (and file) as a previous BB.
16910 2017-04-28 Martin Liska <mliska@suse.cz>
16912 * gcov.c (process_args): Handle new argument 'w'.
16913 (read_graph_file): Assign ID to BBs.
16914 (output_branch_count): Display BB # if verbose flag is set.
16915 (output_lines): Likewise for arcs.
16916 (print_usage): Add '--verbose' option help.
16917 * doc/gcov.texi: Document --verbose (-w) option.
16919 2017-04-28 Martin Liska <mliska@suse.cz>
16921 * gcov.c (struct block_location_info): New struct.
16922 (process_file): Fill up the new structure.
16923 (read_graph_file): Replace usage of encoding by the newly added
16925 (add_line_counts): Likewise.
16926 (accumulate_line_counts): Remove usage of the union.
16927 (function_info::function_info): New function.
16928 (function_info::~function_info): Likewise.
16929 (process_file): Call delete instead of release_function.
16930 (release_function): Release the function.
16931 (release_structures): Call delete instead of release_function.
16932 (solve_flow_graph): Replace usage of num_blocks.
16933 (find_exception_blocks): Likewise.
16934 (output_lines): Fix GNU coding style.
16936 2017-04-28 Martin Liska <mliska@suse.cz>
16939 * coverage.c (coverage_remove_note_file): New function.
16940 * coverage.h: Declare the function.
16941 * toplev.c (finalize): Clean if an error has been seen.
16943 2017-04-28 Martin Liska <mliska@suse.cz>
16945 PR gcov-profile/80031
16946 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
16947 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
16948 * gcov.c (read_graph_file): Read just number of blocks.
16949 * profile.c (branch_prob): Do not stream 0 flags per a basic
16952 2017-04-28 Martin Liska <mliska@suse.cz>
16954 * gcov-dump.c (tag_*): Add new argument to declarations.
16955 (dump_gcov_file): Likewise.
16956 (tag_blocks): Add and use new argument depth.
16957 (tag_arcs): Likewise.
16958 (tag_lines): Likewise.
16959 (tag_counters): Likewise.
16960 (tag_summary): Likewise.
16961 (dump_working_sets): Use depth to do a proper indentation.
16963 2017-04-28 Jakub Jelinek <jakub@redhat.com>
16966 * cgraph.h (symtab_node::debug_symtab): No longer inline.
16967 * symtab.c (symtab_node::debug_symtab): Move definition here.
16969 2017-04-28 Richard Biener <rguenther@suse.de>
16971 * lto-streamer.h (LTO_major_version): Bump to 7.
16973 2017-04-28 Richard Biener <rguenther@suse.de>
16975 * tree-vrp.c (assert_info): New struct.
16976 (add_assert_info): New helper.
16977 (register_edge_assert_for_2): Refactor to add asserts to a vector
16979 (register_edge_assert_for_1): Likewise.
16980 (register_edge_assert_for): Likewise.
16981 (finish_register_edge_assert_for): New helper actually registering
16982 asserts where live on edge.
16983 (find_conditional_asserts): Adjust.
16984 (find_switch_asserts): Likewise.
16985 (evrp_dom_walker::try_find_new_range): Generalize.
16986 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
16988 2017-04-27 Marek Polacek <polacek@redhat.com>
16991 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
16992 arg10 and arg11 to itype.
16994 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
16996 * doc/extend.texi (Object Size Checking): Improve grammar.
16998 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
17001 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
17002 that the logic for permitting reciprocal estimates matches that
17005 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17008 * tree.c (type_cache_hasher::equal): Only compare
17009 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
17010 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
17011 non-aggregate element types.
17012 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
17013 about the flag on ARRAY_TYPEs in the comment, formatting fix.
17015 2017-04-27 Richard Biener <rguenther@suse.de>
17017 PR middle-end/80533
17018 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
17019 stripping ARRAY_REFs from MEM_EXPR make sure we're not
17020 keeping a reference to a trailing array.
17022 2017-04-27 Richard Biener <rguenther@suse.de>
17024 PR middle-end/80539
17025 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
17026 being in loop-closed SSA form conservatively.
17027 (chrec_fold_multiply_poly_poly): Likewise.
17029 2017-04-27 Tamar Christina <tamar.christina@arm.com>
17031 PR middle-end/79665
17032 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
17033 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
17035 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17038 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
17039 (aarch64_function_arg_alignment): Return unsigned int again, but still
17040 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
17041 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17042 Don't emit -Wpsabi note.
17043 (aarch64_function_arg_boundary): Likewise.
17044 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17047 2017-04-26 Nathan Sidwell <nathan@acm.org>
17049 * tree.h (crc32_unsigned_n): Declare.
17050 (crc32_unsigned, crc32_unsigned): Make inline.
17051 * tree.c (crc32_unsigned_bits): Replace with ...
17052 (crc32_unsigned_n): ... this.
17053 (crc32_unsigned, crc32_byte): Remove.
17054 (crc32_string): Remove unnecessary braces.
17056 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17058 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
17059 * ipa-inline-analysis.c (MAX_TIME): Remove.
17060 (account_size_time): Use sreal for time.
17061 (dump_inline_summary): Update.
17062 (estimate_function_body_sizes): Update.
17063 (estimate_edge_size_and_time): Update.
17064 (estimate_calls_size_and_time): Update.
17065 (estimate_node_size_and_time): Update.
17066 (inline_merge_summary): Update.
17067 (inline_update_overall_summary): Update.
17068 (estimate_time_after_inlining): Update.
17069 (inline_read_section): Update.
17070 (inline_write_summary): Update.
17071 * ipa-inline.c (compute_uninlined_call_time): Update.
17072 (compute_inlined_call_time): Update.
17073 (recursive_inlining): Update.
17074 (inline_small_functions): Update.
17075 (dump_overall_stats): Update.
17076 * ipa-inline.h: Include sreal.h.
17077 (size_time_entry): Turn time to sreal.
17078 (inline_summary): Turn self_time nad time to sreal.
17080 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17082 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
17084 (sreal::stream_out, sreal::stream_in): New.
17085 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
17087 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17089 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
17092 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
17095 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
17097 (dimode_scalar_chain::compute_convert_gain): Ditto.
17098 (dimode_scalar_chain::make_vector_copies): Ditto.
17099 (dimode_scalar_chain::convert_reg): Ditto.
17100 (dimode_scalar_chain::convert_insn): Ditto.
17101 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
17102 (VI248_AVX512BW_1): New mode iterator.
17103 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
17104 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
17107 2017-04-25 Martin Sebor <msebor@redhat.com>
17109 PR tree-optimization/80497
17110 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
17111 constants are representable in HOST_WIDE_INT.
17112 (parse_directive): Ditto.
17114 2017-04-25 Martin Sebor <msebor@redhat.com>
17117 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
17118 (new_zero_array): Adjust signature.
17119 (dom_info::dom_init): Used unsigned rather that size_t.
17120 (dom_info::dom_info): Same.
17122 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17123 Jakub Jelinek <jakub@redhat.com>
17126 * config/arm/arm.c: Include gimple.h.
17127 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17128 returns negative, increment ncrn only if it returned positive.
17129 (arm_needs_doubleword_align): Return int instead of bool,
17130 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
17131 members, but if there is any such non-FIELD_DECL
17132 > PARM_BOUNDARY aligned decl, return -1 instead of false.
17133 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17134 returns negative, increment nregs only if it returned positive.
17135 (arm_setup_incoming_varargs): Likewise.
17136 (arm_function_arg_boundary): Emit -Wpsabi note if
17137 arm_needs_doubleword_align returns negative, return
17138 DOUBLEWORD_ALIGNMENT only if it returned positive.
17140 2017-04-25 Marek Polacek <polacek@redhat.com>
17143 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17144 first argument to type.
17146 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
17149 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
17150 type checks to test for compatibility instead of equality.
17152 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17153 Jakub Jelinek <jakub@redhat.com>
17156 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
17158 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
17159 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
17160 the alignment computation, but return their maximum in warn_alignment.
17161 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17162 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
17164 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
17165 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17168 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17170 * config/arc/simdext.md (dmpyh): Fix typo.
17172 2017-04-25 Richard Biener <rguenther@suse.de>
17174 PR tree-optimization/80492
17175 * alias.c (compare_base_decls): Handle registers with asm
17176 specification conservatively.
17177 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17178 compare_base_decls returning dont-know properly.
17180 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17182 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
17183 (legitimate_offset_address_p): New function.
17184 (arc_legitimate_address_p): Use above function.
17186 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17188 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
17190 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17192 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
17193 ACCH registers whenever they are available.
17195 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17197 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
17198 double regs fix when not used.
17200 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17202 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
17204 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
17205 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
17207 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17209 * config/arc/arc.c (arc_output_addsi): Check for h-register class
17210 when emitting short ADD instructions.
17212 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17214 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
17216 (cmpsi_cc_c_insn): Likewise.
17217 (cbranchsi4_scratch): Compute proper instruction length using
17218 compact_hreg_operand.
17219 * config/arc/predicates.md (compact_hreg_operand): New predicate.
17221 2017-04-25 Richard Biener <rguenther@suse.de>
17223 PR middle-end/80509
17224 * passes.c (pass_manager::pass_manager): Initialize
17225 m_name_to_pass_map.
17227 2017-04-25 Richard Biener <rguenther@suse.de>
17229 PR tree-optimization/79201
17230 * tree-ssa-sink.c (statement_sink_location): Handle calls.
17232 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17235 * config/s390/vector.md: Split MEM->GPR vector moves for
17236 non-s_operand addresses.
17238 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17241 * config/s390/predicates.md (reload_const_wide_int_operand): New
17243 * config/s390/s390.md ("movti"): Remove d/P alternative.
17244 ("movti_bigconst"): New pattern definition.
17246 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17249 * s390-protos.h (s390_expand_cs_hqi): Removed.
17250 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
17251 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
17252 modes as well as CCZ1mode and CCZmode.
17253 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
17254 signature of s390_emit_compare_and_swap.
17255 (s390_expand_cs_hqi): Likewise, make static.
17256 (s390_expand_cs_tdsi): Generate an explicit compare before trying
17257 compare-and-swap, in some cases.
17258 (s390_expand_cs): Wrapper function.
17259 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
17261 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
17262 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
17263 patterns for small and large integers. Forbid symref memory operands.
17264 Move expander to s390.c. Require cc register.
17265 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
17266 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
17267 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
17268 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
17269 symref memory operands. Remove CC mode and call s390_match_ccmode
17271 ("atomic_exchange<mode>"): Allow and implement all integer modes.
17273 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17275 * config/s390/s390.md (define_peephole2): New peephole to help
17276 combining the load-and-test pattern with volatile memory.
17278 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17280 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
17281 with CCZmode for TARGET_Z196.
17283 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17285 PR rtl-optimization/80501
17286 * combine.c (make_compound_operation_int): Set subreg_code to SET
17287 even for AND with mask of the sign bit of mode.
17289 PR rtl-optimization/80500
17290 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
17291 sum's initial value.
17293 2017-04-25 Julian Brown <julian@codesourcery.com>
17294 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17296 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
17298 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
17300 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
17302 2017-04-25 Julian Brown <julian@codesourcery.com>
17303 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17305 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
17306 (thunderx2t99_sha): New Reservation.
17308 2017-04-25 Julian Brown <julian@codesourcery.com>
17309 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17311 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
17312 type for 1-element load.
17314 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
17316 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
17318 2017-04-24 Martin Jambor <mjambor@suse.cz>
17320 PR tree-optimization/80293
17321 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
17322 char arrays not totally scalarizable if it is false.
17323 (analyze_all_variable_accesses): Pass correct value in the new
17324 parameter. Add a statistics counter.
17326 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
17328 PR middle-end/79931
17329 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
17331 2017-04-24 Richard Biener <rguenther@suse.de>
17333 PR tree-optimization/80494
17334 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
17335 out for complex types.
17337 2017-04-24 Richard Biener <rguenther@suse.de>
17339 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
17340 * tree-ssa-sccvn.c (print_scc): Print SCC size.
17341 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
17342 (DFS): Adjust and never fail.
17343 (sccvn_dom_walker::fail): Remove.
17344 (sccvn_dom_walker::before_dom_children): Adjust.
17345 (run_scc_vn): Likewise and never fail.
17346 * tree-ssa-pre.c (pass_pre::execute): Adjust.
17347 (pass_fre::execute): Likewise.
17349 2017-04-24 Richard Biener <rguenther@suse.de>
17351 PR tree-optimization/79725
17352 * tree-ssa-sink.c (statement_sink_location): Return whether
17353 failure reason was zero uses. Move that check later.
17354 (sink_code_in_bb): Deal with zero uses by removing the stmt
17357 2017-04-24 Richard Biener <rguenther@suse.de>
17360 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
17361 pointer-based references.
17363 2017-04-24 Richard Biener <rguenther@suse.de>
17366 * pass_manager.h (pass_manager::operator new): Remove.
17367 (pass_manager::operator delete): Likewise.
17368 * passes.c (pass_manager::operator new): Remove.
17369 (pass_manager::operator delete): Likewise.
17370 (pass_manager::pass_manager): Zero individual pass members.
17372 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
17375 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
17376 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
17377 Check "XEXP (src, 1)" operand here.
17378 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
17379 Check "XEXP (src, 1)" operand here.
17380 (dimode_scalar_chain::make_vector_copies): Detect count register
17381 of a shift instruction. Zero extend count register from QImode
17382 to DImode to satisfy vector shift pattern count operand predicate.
17383 Substitute vector shift count operand with a DImode copy.
17384 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
17387 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
17389 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
17390 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17391 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
17392 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17393 (UNSPEC_NOREX_MEM): Remove definition.
17395 2017-04-21 Richard Biener <rguenther@suse.de>
17397 PR tree-optimization/79547
17398 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17399 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
17400 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
17401 without any constraints.
17403 2017-04-21 Richard Biener <rguenther@suse.de>
17405 PR tree-optimization/78847
17406 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
17408 2017-04-21 Richard Biener <rguenther@suse.de>
17410 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
17411 (build_distinct_type_copy): Likewise.
17412 (build_variant_type_copy): Likewise.
17413 * tree.c (build_qualified_type): Pass down mem-stat info.
17414 (build_distinct_type_copy): Likewise.
17415 (build_variant_type_copy): Likewise.
17417 2017-04-21 Richard Biener <rguenther@suse.de>
17419 PR tree-optimization/80237
17420 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
17422 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
17423 for a simplified result.
17425 2016-04-21 Richard Biener <rguenther@suse.de>
17427 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
17428 sth as strict as a simple_iv but a chrec without symbols and an
17429 operand defined in the loop we are peeling (and not some subloop).
17430 (propagate_constants_for_unrolling): Propagate all constants.
17432 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17435 * config/i386/i386.c (print_reg): Remove assert for disalowed
17436 regno values, call output_operand_lossage instead.
17438 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17441 * config/i386/constraints.md (Yc): New register constraint.
17442 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
17443 Use Yc constraint for alternative 2 of operand 0. Remove
17444 preferred_for_speed attribute.
17446 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17448 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
17449 lastprivate clauses in SIMT case.
17451 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
17453 * doc/invoke.texi (-Wextra-semi): Document new warning option.
17455 2017-04-20 Richard Biener <rguenther@suse.de>
17457 PR tree-optimization/57796
17458 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
17459 as N scalar stores.
17460 (vect_model_load_cost): Cost gathers as N scalar loads.
17462 2017-04-20 Richard Biener <rguenther@suse.de>
17464 * ggc-page.c (ggc_allocated_p): Rename to ...
17465 (safe_lookup_page_table_entry): ... this and return the lookup
17467 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
17469 2017-04-20 Richard Biener <rguenther@suse.de>
17471 PR tree-optimization/80453
17472 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
17473 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
17474 from the conditions.
17475 (vn_phi_eq): Pass them down.
17476 (vn_phi_lookup): Record them.
17477 (vn_phi_insert): Likewise.
17479 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
17481 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
17482 uninitialized variable warning to avoid buffer overrun.
17484 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17487 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
17488 is suppressed for '{ 0 }' in C.
17490 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17492 * BASE-VER: Set to 8.0.0.
17494 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
17496 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
17497 priority .init_array and .fini_array section with SECTION_NOTYPE
17500 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17502 PR middle-end/80423
17503 * tree.h (build_array_type): Add typeless_storage default argument.
17504 * tree.c (type_cache_hasher::equal): Also compare
17505 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
17506 (build_array_type): Add typeless_storage argument, set
17507 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
17509 (build_nonshared_array_type): Adjust build_array_type_1 caller.
17510 (build_array_type): Likewise. Add typeless_storage argument.
17512 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17513 Jakub Jelinek <jakub@redhat.com>
17515 PR tree-optimization/80426
17516 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
17517 operation on symbolic operands, also compute the overflow for the
17518 invariant part when the operation degenerates into a negation.
17520 2017-04-19 Jakub Jelinek <jakub@redhat.com>
17523 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
17524 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
17527 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
17529 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
17532 * config/avr/avr.c (tree.h): Include it.
17533 (cgraph.h): Include it.
17534 (avr_encode_section_info): Don't warn for uninitialized progmem
17535 variable if it's just an alias.
17537 2017-04-19 Richard Biener <rguenther@suse.de>
17540 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
17541 when needed by AutoPGO.
17543 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
17546 * doc/lto.texi: Remove an extra 'that'.
17548 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
17550 PR rtl-optimization/80429
17551 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
17552 are only used in debug insns.
17554 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17555 Vladimir Makarov <vmakarov@redhat.com>
17557 * config/sparc/predicates.md (input_operand): Add comment. Return
17558 true for any memory operand when LRA is in progress.
17559 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
17561 2017-04-18 Jeff Law <law@redhat.com>
17564 * mips.md ({return,simple_return}_internal): Do not overwrite
17567 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17569 PR tree-optimization/80443
17570 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
17571 instead of adding 1, subtract -1 and similarly instead of subtracting
17574 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
17576 PR rtl-optimization/80357
17577 * haifa-sched.c (tmp_bitmap): New variable.
17578 (model_recompute): Handle duplicate use records.
17579 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
17580 (free_global_sched_pressure_data): Free it.
17582 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
17585 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
17586 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17587 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17588 instead of SYSTEM_HEADER_DIR.
17590 2017-04-18 Jeff Law <law@redhat.com>
17592 PR middle-end/80422
17593 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
17594 predecessors after walking up the insn chain.
17596 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17599 * dwarf2out.c (modified_type_die): Try harder not to emit internal
17600 sizetype type into debug info.
17602 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17605 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
17606 unneeded test for TARGET_UPPER_REGS_SF.
17607 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
17609 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17612 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
17613 instead of gsi_after_labels.
17615 2017-04-18 Jeff Law <law@redhat.com>
17617 * regcprop.c (maybe_mode_change): Avoid creating copies of the
17621 2017-04-13 Jeff Law <law@redhat.com>
17622 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17623 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17625 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
17628 * config/avr/avr.c (intl.h): Include it.
17629 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
17631 2017-04-18 Martin Liska <mliska@suse.cz>
17633 PR gcov-profile/78783
17634 * gcov-tool.c (gcov_output_files): Validate that destination
17635 file is either removed by the tool or by a user.
17637 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17638 Guy Benyei <guybe@mellanox.com>
17640 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
17641 block, and do not negate it, the stored id is already negative.
17643 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17645 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
17647 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17650 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
17651 masks of options that should be turned off if the VSX vector
17652 options are turned off.
17653 (OTHER_P8_VECTOR_MASKS): Likewise.
17654 (OTHER_VSX_VECTOR_MASKS): Likewise.
17655 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
17656 rs6000_disable_incompatible_switches to validate no type switches
17658 (rs6000_incompatible_switch): New function to disallow turning on
17659 other vector options if -mno-vsx, -mno-power8-vector, or
17660 -mno-power9-vector are specified.
17662 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17664 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
17666 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17668 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
17669 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
17670 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
17671 (ARG_POINTER_CFA_OFFSET): Likewise.
17673 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17675 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
17676 conditions to take advantage of various optimizations.
17678 2017-04-13 Jeff Law <law@redhat.com>
17680 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17681 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17682 (zero_extendsidi2_dext): Likewise.
17684 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17687 * fold-const.c (fold_ternary_loc): Revert
17688 use op0 instead of fold_convert_loc (loc, type, arg0) part of
17691 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
17693 PR rtl-optimization/80343
17694 * lra-remat.c (update_scratch_ops): Assign original hard reg to
17695 new scratch pseudo.
17697 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
17700 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
17701 to ubsan_encode_value.
17703 2017-04-13 Jeff Law <law@redhat.com>
17705 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
17706 appearing in DEBUG_INSNs.
17708 2017-04-13 Martin Liska <mliska@suse.cz>
17710 PR gcov-profile/80413
17711 * gcov-io.c (gcov_write_string): Copy to buffer just when
17712 allocated size is greater than zero.
17714 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17717 * dwarf2out.c (decls_for_scope): Ignore declarations of
17718 current_function_decl in BLOCK_NONLOCALIZED_VARS.
17720 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
17723 * ipa-visibility.c (non_local_p): Fix typos.
17724 (localize_node): When localizing symbol in same comdat group,
17725 dissolve the group only when we know external symbols are going
17727 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
17729 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17731 PR tree-optimization/79390
17732 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
17733 order does not result in usable sequence, retry with reversed operand
17739 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
17740 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
17741 op0 instead of fold_convert_loc (loc, type, arg0).
17743 2017-04-12 Jeff Law <law@redhat.com>
17745 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
17746 has a delay slot in the generated code.
17748 * config/cris/cris.md (cris_preferred_reload_class): Return
17749 GENNONACR_REGS rather than GENERAL_REGS.
17751 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17754 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
17755 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
17756 signedness of the result type.
17758 2017-04-12 Richard Biener <rguenther@suse.de>
17759 Jeff Law <law@redhat.com>
17761 PR tree-optimization/80359
17762 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
17763 trim stores to TARGET_MEM_REFs.
17765 2017-04-12 Richard Biener <rguenther@suse.de>
17767 PR tree-optimization/79390
17768 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
17769 threading case even more.
17771 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
17774 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
17775 for quad_address_p for TImode, instead of just not indexed_address.
17777 2017-04-12 Richard Biener <rguenther@suse.de>
17778 Bernd Edlinger <bernd.edlinger@hotmail.de>
17780 PR middle-end/79671
17781 * alias.c (component_uses_parent_alias_set_from): Handle
17782 TYPE_TYPELESS_STORAGE.
17783 (get_alias_set): Likewise.
17784 * tree-core.h (tree_type_common): Add typeless_storage flag.
17785 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
17786 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
17787 for types containing members with TYPE_TYPELESS_STORAGE.
17788 (place_field): Likewise.
17789 (layout_type): Likewise for ARRAY_TYPE.
17790 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
17791 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
17792 TYPE_TYPELESS_STORAGE.
17793 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
17795 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17798 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17799 first argument to type.
17801 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17805 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
17806 CONST0_RTX (mode) rather than const0_rtx where appropriate.
17807 (rs6000_expand_binop_builtin): Likewise.
17808 (rs6000_expand_ternop_builtin): Likewise; also add missing
17809 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
17810 vshasigma built-ins.
17811 * doc/extend.texi: Document that vec_xxpermdi's third argument
17812 must be a constant.
17814 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
17816 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
17817 Use shift_const cost parameter when calculating gain of STV shifts.
17819 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
17821 PR rtl-optimization/70478
17822 * lra-constraints.c (process_alt_operands): Check memory for
17823 disfavoring memory insn operand.
17825 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17827 PR middle-end/80100
17828 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
17829 left shift in unsigned HOST_WIDE_INT type.
17831 PR rtl-optimization/80385
17832 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
17833 (not (neg X)) into (plus X -1) for complex or non-integral modes.
17836 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
17837 if they have any depend clauses.
17839 2017-04-11 Martin Liska <mliska@suse.cz>
17842 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
17843 * ipa-split.c (split_function): Create a local comdat symbol
17844 if caller is in a comdat group.
17846 2017-04-11 Martin Liska <mliska@suse.cz>
17849 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
17852 2017-04-11 Martin Sebor <msebor@redhat.com>
17854 PR middle-end/80364
17855 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
17856 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
17858 (directive::set_width, directive::set_precision, format_character):
17860 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
17863 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
17866 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
17867 conflict, set target->arch_name instead of target->cpu_name.
17869 2017-04-11 Richard Biener <rguenther@suse.de>
17871 PR tree-optimization/80374
17872 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
17873 build_zero_cst, remove fold_convertible_p check again.
17875 2017-04-11 Martin Liska <mliska@suse.cz>
17878 * ubsan.c (instrument_object_size): Do not instrument register
17881 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17884 * config/i386/i386-builtin-types.def
17885 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
17886 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
17887 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
17888 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
17889 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
17890 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
17891 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
17892 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
17893 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
17894 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
17895 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
17896 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
17897 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
17898 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
17899 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
17900 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
17901 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
17902 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
17903 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
17904 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
17905 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
17906 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
17907 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
17908 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
17909 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
17910 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
17911 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
17912 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
17913 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
17914 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
17915 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
17916 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
17917 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
17918 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
17919 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
17920 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
17921 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
17922 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
17923 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
17924 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
17925 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
17926 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
17927 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
17928 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
17929 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
17930 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
17932 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
17933 flag to second_arg_count, handle 4 argument function type _COUNT
17934 aliases, handle second_arg_count on second argument rather than last.
17936 2017-04-10 Jeff Law <law@redhat.com>
17938 PR tree-optimization/80374
17939 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
17940 record anything if we can not convert integer_zero_node to the
17943 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
17946 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17947 Enhance special handling given to the TARGET_P9_MINMAX option in
17948 relation to certain other options.
17950 2017-04-10 Bin Cheng <bin.cheng@arm.com>
17952 PR tree-optimization/80153
17953 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
17954 remove POINTER_PLUS_EXPR's base part directly, rather than through
17957 2017-04-10 Richard Biener <rguenther@suse.de>
17958 Bin Cheng <bin.cheng@arm.com>
17960 PR tree-optimization/80153
17961 * tree-affine.c (aff_combination_to_tree): Get base pointer from
17962 the first element of pointer type aff_tree. Build result expr in
17964 (add_elt_to_tree): Convert to type unconditionally. Remove other
17965 fold_convert calls.
17966 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
17967 (rewrite_use_nonlinear_expr): Check invariant using iv information.
17969 2017-04-10 Richard Biener <rguenther@suse.de>
17971 * tree-ssa-structalias.c (find_func_aliases): Properly handle
17974 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
17976 PR rtl-optimization/70478
17977 * lra-constraints.c (curr_small_class_check): New.
17978 (update_and_check_small_class_inputs): New.
17979 (process_alt_operands): Update curr_small_class_check. Disfavor
17980 alternative insn memory operands. Check available regs for small
17983 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
17986 * config/mips/mips.opt (-mvirt): Update description.
17987 * doc/invoke.texi (-mvirt): Likewise.
17989 2017-04-10 Richard Biener <rguenther@suse.de>
17991 PR middle-end/80362
17992 * fold-const.c (fold_binary_loc): Look at unstripped ops when
17993 looking for NEGATE_EXPR in -A / -B to A / B folding.
17995 2017-04-10 Martin Liska <mliska@suse.cz>
17997 PR gcov-profile/80224
17998 * gcov.c (print_usage): Fix usage string.
17999 (get_gcov_intermediate_filename): Remove.
18000 (output_gcov_file): Use both for normal and intermediate format.
18001 (generate_results): Do not initialize special file for
18002 intermediate format.
18004 2017-04-10 Richard Biener <rguenther@suse.de>
18006 PR tree-optimization/80304
18007 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
18010 2017-04-10 Nathan Sidwell <nathan@acm.org>
18013 * config/rs6000/rs6000.c (rs6000_vector_type): New.
18014 (rs6000_init_builtins): Use it.
18016 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18018 * config/arm/arm.md (<mrc>): Add mode to SET source.
18019 (<mrrc>): Likewise.
18021 2017-04-10 Richard Biener <rguenther@suse.de>
18023 PR middle-end/80344
18024 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
18026 2017-04-10 Jakub Jelinek <jakub@redhat.com>
18029 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
18030 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
18031 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
18032 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
18033 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
18034 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
18035 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
18036 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
18037 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
18038 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
18039 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
18040 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
18041 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
18042 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
18043 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
18044 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
18045 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
18046 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
18047 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
18048 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
18049 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
18050 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
18051 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
18053 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
18055 PR rtl-optimization/70478
18056 * lra-constraints.c: Reverse the last patch.
18058 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
18060 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
18061 Add comment for WCHAR_T.
18063 2017-04-08 Martin Liska <mliska@suse.cz>
18066 2017-04-07 Martin Liska <mliska@suse.cz>
18069 * ipa-split.c (split_function): Add function part to a same comdat
18072 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18075 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
18077 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
18079 * rs6000/rs6000.c (vec_load_pendulum): Rename...
18080 (vec_pairing): ...to this.
18081 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
18082 (rs6000_sched_init): Adjust for name change.
18083 (struct rs6000_sched_context): Likewise.
18084 (rs6000_init_sched_context): Likewise.
18085 (rs6000_set_sched_context): Likewise.
18087 2017-04-07 Jakub Jelinek <jakub@redhat.com>
18093 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
18095 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
18096 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
18097 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
18099 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
18101 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
18103 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18105 PR rtl-optimization/70703
18106 * ira-color.c (update_conflict_hard_regno_costs): Use
18107 int64_t instead of HOST_WIDE_INT.
18109 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18111 PR rtl-optimization/70478
18112 * lra-constraints.c (process_alt_operands): Disfavor alternative
18113 insn memory operands.
18115 2017-04-07 Jeff Law <law@redhat.com>
18117 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
18118 CALL and NOTE_INSN_CALL_ARG_LOCATION.
18120 2017-04-07 Martin Liska <mliska@suse.cz>
18123 * config/aarch64/aarch64.c (aarch64_process_target_attr):
18124 Show error message instead of an ICE.
18126 2017-04-07 Martin Liska <mliska@suse.cz>
18129 * ipa-split.c (split_function): Add function part to a same comdat
18132 2017-04-07 Richard Biener <rguenther@suse.de>
18134 PR middle-end/80341
18135 * tree.c (get_unwidened): Also handle ! for_type case for
18137 * convert.c (do_narrow): Split out from ...
18138 (convert_to_integer_1): ... here. Do not pass final truncation
18139 type to get_unwidened for TRUNC_DIV_EXPR.
18141 2017-04-07 Richard Biener <rguenther@suse.de>
18143 * tree-affine.c (wide_int_ext_for_comb): Take type rather
18145 (aff_combination_const): Adjust.
18146 (aff_combination_scale): Likewise.
18147 (aff_combination_add_elt): Likewise.
18148 (aff_combination_add_cst): Likewise.
18149 (aff_combination_convert): Likewise.
18150 (add_elt_to_tree): Likewise. Remove unused argument.
18151 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
18153 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
18155 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
18157 * config/arm/arm.c (arm_default_short_enums): Use
18158 ARM_DEFAULT_SHORT_ENUMS.
18159 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
18161 2017-04-06 Jakub Jelinek <jakub@redhat.com>
18164 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
18165 members with redundant out-of-class redeclaration.
18167 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18170 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
18171 * config/i386/i386.md (*zero_extendsidi2):
18172 Add (?*x,*x) and (?*v,*v) alternatives.
18174 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18177 * config/i386/i386.c (ix86_expand_builtin)
18178 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
18179 mode from insn data. Convert operands to insn operand mode.
18180 Copy operands that don't satisfy insn predicate to a register.
18182 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
18184 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
18187 2017-04-06 Richard Biener <rguenther@suse.de>
18189 PR tree-optimization/80334
18190 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
18191 preserve alignment of accesses.
18193 2017-04-06 Richard Biener <rguenther@suse.de>
18195 PR tree-optimization/80262
18196 * tree-sra.c (build_ref_for_offset): Preserve address-space
18198 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
18199 Drop useless address-space information on MEM_REF offsets.
18201 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
18203 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
18205 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18207 PR rtl-optimization/70703
18208 * ira-color.c (update_conflict_hard_regno_costs): Use
18209 HOST_WIDE_INT instead of long.
18211 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
18214 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
18215 not defined for x86_64 target. Add -mmmx target option when __SSE2__
18217 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
18218 for x86_64 target. Handle -m3dnowa option.
18220 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18222 PR rtl-optimization/70703
18223 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
18224 (update_conflict_hard_regno_costs): Use long instead of unsigned
18225 arithmetic for cost calculation.
18227 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18228 Bernd Edlinger <bernd.edlinger@hotmail.de>
18231 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
18234 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
18237 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
18238 ptr_mode with Pmode throughout.
18239 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
18240 into probe_stack_range and use DImode.
18242 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
18245 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
18246 call_eh_return is true.
18248 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18250 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
18251 Initialize last_match_fntype_index.
18253 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18256 * tree-nvr.c: Include internal-fn.h.
18257 (pass_return_slot::execute): Ignore internal calls without
18260 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18261 Richard Biener <rguenther@suse.de>
18264 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
18265 captures used multiple times, except for the last use.
18266 * generic-match-head.c: Include gimplify.h.
18268 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18270 PR tree-optimization/79390
18271 * target.h (struct noce_if_info): Declare.
18272 * targhooks.h (default_noce_conversion_profitable_p): Declare.
18273 * target.def (noce_conversion_profitable_p): New target hook.
18274 * ifcvt.h (struct noce_if_info): New type, moved from ...
18275 * ifcvt.c (struct noce_if_info): ... here.
18276 (noce_conversion_profitable_p): Renamed to ...
18277 (default_noce_conversion_profitable_p): ... this. No longer
18279 (noce_try_store_flag_constants, noce_try_addcc,
18280 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
18281 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
18282 instead of noce_conversion_profitable_p.
18283 * config/i386/i386.c: Include ifcvt.h.
18284 (ix86_option_override_internal): Don't override
18285 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
18286 (ix86_noce_conversion_profitable_p): New function.
18287 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
18288 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
18289 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
18290 * doc/tm.texi: Regenerated.
18292 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18294 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
18297 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18300 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
18301 instructions for small multiply cores.
18303 2017-04-04 Jeff Law <law@redhat.com>
18305 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
18307 (mips_expand_vec_perm_const): Initialize elements in orig_perm
18308 that are not set by the loop over the elements.
18310 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18313 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
18314 int mode, convert_modes it to mode as unsigned, otherwise use
18315 lowpart_subreg to mode rather than SImode.
18316 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
18317 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
18318 Use DImode instead of SImode for the shift count operand.
18319 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
18322 2017-04-04 Richard Biener <rguenther@suse.de>
18324 PR middle-end/80281
18325 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
18326 arithmetic done for the negate or the plus. Simplify.
18327 (A - (-B) -> A + B): Likewise.
18328 * fold-const.c (split_tree): Make sure to not negate pointers.
18330 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
18332 PR rtl-optimization/60818
18333 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
18334 a compare of comparisons with the thing compared if this results
18335 in a different machine mode.
18337 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
18339 * alias.c (base_alias_check): Fix typo in comment.
18340 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
18341 * cgraphunit.c (symbol_table::compile): Likewise.
18342 * collect2.c (maybe_run_lto_and_relink): Likewise.
18343 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
18344 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
18345 * config/avr/avr.c (avr_map_op_t): Likewise.
18346 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
18347 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
18348 * config/epiphany/epiphany.md (movcc): Likewise.
18349 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
18350 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
18352 * config/mips/mips.c (mips_save_restore_reg): Likewise.
18353 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
18354 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
18355 * config/sh/sh.c (sh_rtx_costs): Likewise.
18356 * fold-const.c (fold_truth_andor): Likewise.
18357 * genautomata.c (collapse_flag): Likewise.
18358 * gengtype.h (struct type::u::s): Likewise.
18359 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
18360 * input.c (FORMAT_AMOUNT): Likewise.
18361 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
18362 (known_aggs_to_agg_replacement_list): Likewise.
18363 * ipa-inline-analysis.c: Likewise.
18364 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
18365 * ipa-polymorphic-call.c
18366 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
18367 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
18368 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
18370 * modulo-sched.c (apply_reg_moves): Likewise.
18371 * omp-expand.c (build_omp_regions_1): Likewise.
18372 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
18373 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
18374 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
18375 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
18376 * value-prof.c: Likewise.
18377 * var-tracking.c (val_reset): Likewise.
18379 2017-04-03 Richard Biener <rguenther@suse.de>
18381 PR tree-optimization/80275
18382 * fold-const.c (split_address_to_core_and_offset): Handle
18385 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
18387 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
18388 descriptors is at least equal to that of functions.
18390 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18392 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
18394 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18397 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
18398 (mov<IMOD4:mode>): New expander.
18399 (*mov<IMOD4:mode>_internal): New insn and split pattern.
18401 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
18403 PR rtl-optimization/79405
18404 * fwprop.c (propagations_left): New variable.
18405 (forward_propagate_into): Decrement it.
18406 (fwprop_init): Initialize it.
18407 (fw_prop): If the variable has reached zero, stop propagating.
18408 (fwprop_addr): Ditto.
18410 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18413 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
18414 a FUNCTION_DECL, pass it as decl instead of origin to
18417 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
18419 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
18422 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
18425 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
18426 TARGET_VSX_SMALL_INTEGER.
18428 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18430 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18431 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
18433 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18435 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
18436 extraction from odd-numbered MSA register.
18438 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18440 PR middle-end/80173
18441 * expmed.c (store_bit_field_1): Don't attempt to create
18442 a word subreg out of hard registers wider than word if they
18443 have HARD_REGNO_NREGS of 1 for their mode.
18445 PR middle-end/80163
18446 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
18447 conversions to integer types wider than word and pointer.
18450 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
18451 (rtx_equal_for_cselib_p): Pass 0 to it.
18452 * cselib.c (cselib_hasher::equal): Likewise.
18453 (rtx_equal_for_cselib_1): Add depth argument. If depth
18454 is 128, don't look up VALUE locs and punt. Increment
18455 depth in recursive calls when walking VALUE locs.
18457 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
18459 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
18460 (make_gcov_file_name): Use the canonical path name for generating
18462 (read_line): Fix handling of files with ascii null bytes.
18464 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
18466 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
18467 to initialise a vector register instead
18468 of using a const_int.
18470 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18472 PR translation/80189
18473 * gimplify.c (omp_default_clause): Use %qs instead of %s in
18474 diagnostic messages.
18476 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
18479 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
18480 (dfp_diex_<mode>): Update mode of operand 1.
18481 * doc/extend.texi (dxex, dxexq): Document change to return type.
18482 (diex, diexq): Document change to argument type.
18484 2017-03-30 Martin Jambor <mjambor@suse.cz>
18487 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
18488 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
18489 it reflects the signature changes performed at the callee side.
18490 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
18491 to cgraph_build_function_type_skip_args.
18492 (build_function_decl_skip_args): Adjust call to the above function.
18494 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18497 * config/i386/sse.md
18498 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
18499 register as dest whenever it is a MEM not rtx_equal_p to the
18500 corresponding dup operand, and when forcing into reg move the
18501 reg into the memory afterwards.
18502 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
18503 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
18504 for the force_reg mode.
18505 (avx512vl_vextractf128<mode>): Use register as dest either
18506 always when a MEM, or when it is a MEM not rtx_equal_p to the
18507 corresponding dup operand, or even not when it is a CONST_VECTOR
18508 depending on the mode and lo vs. hi.
18509 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
18511 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
18512 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
18513 Likewise. Require that operands[2] is even.
18514 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
18515 Remove extraneous parens. Require that operands[2] is a multiple
18517 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
18518 operands[0] is a MEM if <mask_applied>, the predicates/constraints
18519 disallow memory then.
18521 2017-03-30 Richard Biener <rguenther@suse.de>
18523 PR tree-optimization/77498
18524 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
18525 to non-constants over backedges.
18527 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
18529 PR rtl-optimization/80233
18530 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
18531 as last_combined_insn. Do not test for BARRIER_P separately.
18533 2017-03-29 Andreas Schwab <schwab@suse.de>
18536 * calls.c (prepare_call_address): Convert funexp to Pmode before
18537 copying to temp reg.
18539 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18541 PR tree-optimization/80158
18542 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
18543 Handle possible future case of more than one alternate
18545 (replace_rhs_if_not_dup): Likewise.
18546 (replace_one_candidate): Likewise.
18548 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
18550 PR rtl-optimization/80193
18551 * ira.c (ira): Do not check allocation for LRA.
18553 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
18555 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
18556 (nvptx_output_simt_exit): Declare.
18557 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
18558 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
18559 (init_softstack_frame): Move initialization of crtl->is_leaf to...
18560 (nvptx_declare_function_name): ...here. Emit declaration of local
18561 memory space buffer for omp_simt_enter insn.
18562 (nvptx_output_unisimt_switch): New.
18563 (nvptx_output_softstack_switch): New.
18564 (nvptx_output_simt_enter): New.
18565 (nvptx_output_simt_exit): New.
18566 * config/nvptx/nvptx.h (struct machine_function): New fields
18567 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
18568 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
18569 (UNSPECV_SIMT_EXIT): Ditto.
18570 (omp_simt_enter_insn): New insn.
18571 (omp_simt_enter): New expansion.
18572 (omp_simt_exit): New insn.
18573 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
18575 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
18576 (expand_GOMP_SIMT_ENTER_ALLOC): New.
18577 (expand_GOMP_SIMT_EXIT): New.
18578 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
18579 (GOMP_SIMT_ENTER_ALLOC): Ditto.
18580 (GOMP_SIMT_EXIT): Ditto.
18581 * target-insns.def (omp_simt_enter): New insn.
18582 (omp_simt_exit): Ditto.
18583 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
18585 (lower_rec_simd_input_clauses): Implement SIMT privatization.
18586 (lower_rec_input_clauses): Likewise.
18587 (lower_lastprivate_clauses): Handle SIMT privatization.
18589 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
18590 (ompdevlow_adjust_simt_enter): New.
18591 (find_simtpriv_var_op): New.
18592 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
18593 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
18595 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
18596 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
18597 (copy_decl_for_dup_finish): Ditto.
18599 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
18601 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
18604 * config/i386/i386.c (ix86_option_override_internal): Always
18605 allow -mpreferred-stack-boundary=3 for 64-bit targets.
18607 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18609 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
18611 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18613 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
18614 mark new edge's irreducible flag accordign to it.
18615 (vect_do_peeling): Check loop preheader edge's irreducible flag
18616 and pass it to function slpeel_add_loop_guard.
18618 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
18620 PR tree-optimization/80218
18621 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18622 Update block frequencies and counts.
18624 2017-03-28 Richard Biener <rguenther@suse.de>
18626 PR tree-optimization/78644
18627 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
18628 of a simplification result we may not use it at all.
18630 2017-03-28 Richard Biener <rguenther@suse.de>
18633 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
18634 without arguments, generate default definition of a SSA name.
18636 2017-03-28 Richard Biener <rguenther@suse.de>
18638 PR middle-end/80222
18639 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
18640 TYPE_REF_CAN_ALIAS_ALL references.
18641 * fold-const.c (fold_indirect_ref_1): Likewise.
18643 2017-03-28 Martin Liska <mliska@suse.cz>
18646 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
18647 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
18649 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18650 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
18652 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
18653 (EXTRA_SPECS): Define.
18654 (SUBTARGET_EXTRA_SPECS): Likewise.
18655 (SUBTARGET_CPP_SPEC): Likewise.
18656 * config/arc/elf.h (EXTRA_SPECS): Renamed to
18657 SUBTARGET_EXTRA_SPECS.
18658 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
18660 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18662 * config/arc/simdext.md (vst64_insn): Update pattern.
18663 (vld32wh_insn): Likewise.
18664 (vld32wl_insn): Likewise.
18665 (vld64_insn): Likewise.
18666 (vld32_insn): Likewise.
18668 2017-03-28 Marek Polacek <polacek@redhat.com>
18671 * fold-const.c (fold_comparison): Use protected_set_expr_location
18672 instead of SET_EXPR_LOCATION.
18674 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
18676 * tree.c (add_expr): Avoid name lookup warning.
18678 2017-03-27 Jeff Law <law@redhat.com>
18680 PR tree-optimization/80216
18681 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
18682 function name. Limit recursion depth.
18683 (record_temporary_equivalences): Corresponding changes.
18685 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
18687 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
18690 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18693 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
18695 * cfgcleanup.c (reg_note_cfa_p): New array.
18696 (insns_have_identical_cfa_notes): New function.
18697 (old_insns_match_p): Don't cross-jump in between /f
18698 and non-/f instructions. If both i1 and i2 are frame related,
18699 verify all CFA notes, their order and content.
18701 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
18704 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
18705 HImode and SImode with zero extend to DImode to one insn.
18706 (bswap<mode>2_extenddi): Likewise.
18707 (bswapsi2_extenddi): Likewise.
18708 (bswaphi2_extendsi): Likewise.
18709 (bswaphi2): Combine bswap HImode and SImode into one insn.
18710 Separate memory insns from swapping register.
18711 (bswapsi2): Likewise.
18712 (bswap<mode>2): Likewise.
18713 (bswaphi2_internal): Delete, no longer used.
18714 (bswapsi2_internal): Likewise.
18715 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
18716 store, and gpr<-gpr swap insns.
18717 (bswap<mode>2_store): Likewise.
18718 (bswaphi2_reg): Register only splitter, combine with the splitter.
18719 (bswaphi2 splitter): Likewise.
18720 (bswapsi2_reg): Likewise.
18721 (bswapsi2 splitter): Likewise.
18722 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
18723 the insns into load, store, and register/register insns.
18724 (bswapdi2_ldbrx): Likewise.
18725 (bswapdi2_load): Likewise.
18726 (bswapdi2_store): Likewise.
18727 (bswapdi2_reg): Likewise.
18729 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
18731 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
18732 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
18734 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
18737 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
18739 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18740 special handling for target option conflicts between dform
18741 options (-mpower9-dform, -mpower9-dform-vector,
18742 -mpower9-dform-scalar) and -mno-direct-move.
18744 2017-03-27 Richard Biener <rguenther@suse.de>
18746 PR tree-optimization/80181
18747 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
18749 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18751 * config/arc/predicates.md (move_double_src_operand): Replace the
18752 call to move_double_src_operand with a call to address_operand.
18754 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18756 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
18757 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
18758 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
18760 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18762 * config/arc/predicates.md (long_immediate_loadstore_operand):
18763 Consider scaled addresses cases.
18765 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18767 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
18768 restored when in interrupt.
18769 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
18770 doesn't have delay slot.
18772 2017-03-27 Richard Biener <rguenther@suse.de>
18775 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
18776 inlined thunk clones.
18778 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18781 * asan.c (instrument_derefs): Copy over last operand from
18782 original COMPONENT_REF to the new COMPONENT_REF with
18783 DECL_BIT_FIELD_REPRESENTATIVE.
18784 * ubsan.c (instrument_object_size): Likewise.
18786 2017-03-27 Richard Biener <rguenther@suse.de>
18788 PR tree-optimization/80170
18789 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
18790 sure DR/SCEV didnt fold in constants we do not see when looking
18791 at the reference base alignment.
18793 2017-03-27 Richard Biener <rguenther@suse.de>
18795 PR middle-end/80171
18796 * gimple-fold.c (fold_ctor_reference): Properly guard against
18797 NULL return value from canonicalize_constructor_val.
18799 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
18802 * config/i386/i386.c (ix86_expand_builtin)
18803 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
18804 flags reg setting and flags reg using instructions.
18805 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
18806 clobbering instructions to zero extend op2.
18808 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
18810 * doc/install.texi (Configuration) <--with-aix-soname>:
18811 Update link to AIX ld.
18813 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
18815 PR rtl-optimization/80160
18816 PR rtl-optimization/80159
18817 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
18818 reg_alternate_class into account.
18820 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
18823 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
18824 to consider in curr_insn_transform.
18826 2017-03-24 Jakub Jelinek <jakub@redhat.com>
18828 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
18829 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
18830 and emit_mode_inner.
18832 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18834 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
18835 argument to the overloaded builtin variants. Use the new flag to
18836 deprecate certain builtin variants.
18837 * config/s390/s390-builtin-types.def: Add new builtin types.
18838 * config/s390/s390-builtins.h: Support new flags field for
18839 overloaded builtins.
18840 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
18841 (s390_macro_to_expand): Enable vector float data type.
18842 (s390_cpu_cpp_builtins_internal): Indicate support of the new
18843 builtins by incrementing the __VEC__ version number.
18844 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
18846 (s390_resolve_overloaded_builtin): Emit error messages depending
18847 on the builtin flags.
18848 * config/s390/s390.c (s390_expand_builtin): Support additional
18849 flags argument. Change error message to match the messages
18850 emitted in s390-c.c.
18851 * config/s390/s390.md: New UNSPEC_* constants.
18852 (op_type): Add new instruction types.
18853 * config/s390/vecintrin.h: Add new builtins and test data class
18855 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
18856 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
18857 (VEC_INEXACT, VEC_NOINEXACT): New constants.
18858 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
18859 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
18860 ("vec_mergel<mode>"): V_HW -> VEC_HW.
18862 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
18863 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
18864 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
18865 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
18867 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
18868 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
18869 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
18870 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
18872 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
18873 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
18874 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
18875 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
18876 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
18877 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
18878 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
18880 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
18881 ("vec_scatter_element<V_HW_4:mode>_DI")
18882 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
18883 ("vec_fpint<mode>", "vflls")
18884 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
18885 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
18886 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
18887 ("*vec_cmphe<mode>_cc"): ... these.
18889 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
18890 mode constant instead of magic value.
18892 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18894 * config/s390/s390.c (s390_expand_vec_compare): Support other
18895 vector floating point modes than just V2DF.
18896 (s390_expand_vcond): Likewise.
18897 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
18898 (s390_cannot_change_mode_class): Prevent mode changes between TF
18899 and V1TF in vector registers.
18900 * config/s390/s390.md (DF, SF): New mode attributes.
18901 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
18902 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
18903 SFmode support for VRs.
18904 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
18906 (VFT, VF_HW): New mode iterators.
18907 (vw, sdx): New mode attributes.
18908 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
18909 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
18910 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
18911 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
18912 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
18913 also the new vector floating point modes. Renaming to ...
18915 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
18916 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
18917 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
18918 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
18919 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
18920 ("vec_unordered<mode>"): ... these.
18922 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
18923 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
18924 ("*vec_extendv2df"): New insn definitions.
18926 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18928 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
18929 ("mulditi3_2", "*muldi3_sign"): New patterns.
18930 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
18931 rename the pattern definition.
18933 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18935 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
18937 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
18939 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18941 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
18942 instruction if possible.
18943 * config/s390/vector.md (vec_halfnumelts): New mode
18945 ("*vec_vllezlf<mode>"): New pattern.
18947 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18949 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
18950 ("popcountv4si2", "popcountv2di2"): Rename to ...
18951 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
18952 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
18954 ("popcount<mode>2_vxe"): New pattern.
18956 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18958 * common/config/s390/s390-common.c (processor_flags_table): Add
18960 * config.gcc: Add arch12.
18961 * config/s390/driver-native.c (s390_host_detect_local_cpu):
18962 Default to arch12 for unknown CPU model numbers.
18963 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
18964 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
18965 PROCESSOR_max sanity check.
18966 * config/s390/s390-opts.h (enum processor_type): Add
18968 * config/s390/s390.c (processor_table): Add arch12.
18969 (s390_expand_builtin): Add check for B_VXE flag.
18970 (s390_issue_rate): Add PROCESSOR_ARCH12.
18971 (s390_get_sched_attrmask): Likewise.
18972 (s390_get_unit_mask): Likewise.
18973 (s390_sched_score): Enable z13 scheduling for arch12.
18974 (s390_sched_reorder): Likewise.
18975 (s390_sched_variable_issue): Likewise.
18976 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
18978 (s390_tune_attr): Use z13 scheduling also for arch12.
18979 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
18980 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
18981 (TARGET_VXE_P): New macros.
18982 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
18983 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
18984 * config/s390/s390.opt: Add arch12 as processor_type.
18986 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18988 * config/s390/s390.md
18989 ("fixuns_truncdddi2", "fixuns_trunctddi2")
18990 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
18991 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
18993 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
18994 Rename expanders to ...
18996 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
18997 ("fixuns_truncdddi2_emu"): ... these.
18999 ("fixuns_trunc<mode>si2_emu"): New expander.
19001 ("*fixuns_truncdfdi2_z13"): Rename to ...
19002 ("*fixuns_truncdfdi2_vx"): ... this.
19004 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19006 * config/s390/2964.md: Remove the single element vector compare
19007 instructions which are no longer used.
19008 * config/s390/s390.c (s390_select_ccmode): Remove handling of
19010 (s390_canonicalize_comparison): Remove handling of DFmode
19012 (s390_expand_vec_compare_scalar): Remove function.
19013 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
19014 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
19016 ("*cmp<mode>_ccs"): Add wfcdb instruction.
19018 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19020 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
19022 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
19023 will anyway by matched by mov<mode>_64dfp.
19025 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19027 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
19028 vlef/vstef. Add missing operand to vleif.
19030 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19032 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
19033 pair for all vector types with 64 bit elements.
19034 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
19035 * config/s390/vector.md (V_HW_64): ... here.
19036 (V_128_NOSINGLE): New mode iterator.
19037 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
19038 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
19039 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
19040 ("*vec_load_pairv2di"): Change to ...
19041 ("*vec_load_pair<mode>"): ... this one.
19043 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19045 * config/s390/constraints.md: Add comments.
19046 (jKK): Reject element sizes > 8 bytes.
19047 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
19049 * config/s390/s390.md: Add the s_operand checks formerly in
19050 s390_split_ok_p to various splitters where they are still
19052 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
19053 for 128 bit vectors. Plus two splitters.
19055 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19057 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
19060 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19063 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
19064 error if the boundary argument is not constant.
19066 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19068 PR rtl-optimization/80112
19069 * loop-doloop.c (doloop_condition_get): Don't check condition
19070 if cmp isn't SET with IF_THEN_ELSE src.
19072 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19074 PR tree-optimization/80158
19075 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
19076 replacing a candidate statement, also replace it for the
19077 candidate's alternate interpretation.
19078 (replace_rhs_if_not_dup): Likewise.
19079 (replace_one_candidate): Likewise.
19081 2017-03-24 Richard Biener <rguenther@suse.de>
19083 PR tree-optimization/80167
19084 * graphite-isl-ast-to-gimple.c
19085 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
19087 (translate_isl_ast_to_gimple::get_rename): Likewise.
19089 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
19091 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
19092 handling of certain combinations of target options, including the
19093 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
19094 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
19096 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19099 * config/arm/arm.md (*load_multiple): Add reload_completed to
19100 matching condition.
19102 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19103 Richard Biener <rguenth@suse.de>
19105 PR tree-optimization/79908
19106 PR tree-optimization/80136
19107 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19108 been cast away, gimplify_and_add suffices.
19110 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
19112 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
19114 2017-03-23 Richard Biener <rguenther@suse.de>
19116 PR tree-optimization/80032
19117 * gimplify.c (gimple_push_cleanup): Forced unconditional
19118 cleanups still have to go to the conditional_cleanups
19121 2017-03-22 Jakub Jelinek <jakub@redhat.com>
19123 PR tree-optimization/80072
19124 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
19126 (next_operand_entry_id): Change type to unsigned int.
19127 (sort_by_operand_rank): Make sure to return the right return value
19128 even if unsigned fields are bigger than INT_MAX.
19129 (struct oecount): Change cnt and id type to unsigned int.
19130 (oecount_hasher::equal): Formatting fix.
19131 (oecount_cmp): Make sure to return the right return value
19132 even if unsigned fields are bigger than INT_MAX.
19133 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
19136 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
19137 TREE_READONLY on result if writing it more than once.
19140 * doc/invoke.texi (-fsanitize=thread): Document that with
19141 -fnon-call-exceptions atomics are not able to throw
19145 * tsan.c: Include tree-eh.h.
19146 (instrument_builtin_call): Call maybe_clean_eh_stmt or
19147 maybe_clean_or_replace_eh_stmt where needed.
19148 (instrument_memory_accesses): Add cfg_changed argument.
19149 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
19150 if it returned true.
19151 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
19153 PR rtl-optimization/63191
19154 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
19155 wrapper function, moved the whole old content into ...
19156 (ix86_delegitimize_address_1): ... this. New inline function.
19157 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
19158 true as last argument instead of ix86_delegitimize_address.
19160 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19162 * config/aarch64/aarch64.c (generic_branch_cost): Copy
19163 cortexa57_branch_cost.
19165 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19167 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
19169 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19172 * doc/md.texi (Constraints): Document wA constraint.
19173 * config/rs6000/constraints.md (wA): New.
19174 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
19175 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
19176 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
19177 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
19179 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
19182 * gimplify.c (is_oacc_declared): New function.
19183 (oacc_default_clause): Use it to set default flags for acc declared
19184 variables inside parallel regions.
19185 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
19186 declared variables.
19187 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
19188 declare attribute to any decl as necessary.
19190 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
19193 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
19194 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
19195 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
19196 (arm_arch_lpae): This.
19197 * config/arm/arm.c (arm_arch7ve): Rename into ...
19198 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
19199 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
19202 2017-03-22 Martin Liska <mliska@suse.cz>
19205 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
19206 error message instead of an ICE.
19208 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19210 * doc/extend.texi (6.11 Additional Floating Types): Revise.
19212 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
19214 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19216 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19219 2017-03-21 Martin Sebor <msebor@redhat.com>
19221 * doc/extend.texi: Use "cannot" instead of "can't."
19222 * doc/hostconfig.texi: Same.
19223 * doc/install.texi: Same.
19224 * doc/invoke.texi: Same.
19225 * doc/loop.texi: Same.
19226 * doc/md.texi: Same.
19227 * doc/objc.texi: Same.
19228 * doc/rtl.texi: Same.
19229 * doc/tm.texi: Same.
19230 * doc/tm.texi.in: Same.
19231 * doc/trouble.texi: Same.
19233 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
19236 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
19237 (collect_checksum_attributes): Set it.
19238 (die_checksum_ordered): Use it.
19240 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19242 PR tree-optimization/79908
19243 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
19244 change: For a VA_ARG whose LHS has been cast away, use
19245 force_gimple_operand to construct the side effects.
19247 2017-03-21 David Malcolm <dmalcolm@redhat.com>
19249 PR translation/80001
19250 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
19251 more amenable to translation.
19252 (oacc_loop_auto_partitions): Likewise.
19254 2017-03-21 Marek Polacek <polacek@redhat.com>
19255 Martin Sebor <msebor@redhat.com>
19257 PR tree-optimization/80109
19258 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
19259 on INTEGRAL_TYPE_P.
19261 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19262 Segher Boessenkool <segher@kernel.crashing.org>
19265 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
19266 check reg_used_between_p between insn and one of succ or succ2
19267 depending on if succ is artificial insn not inserted into insn
19270 2017-03-21 Martin Liska <mliska@suse.cz>
19272 PR gcov-profile/80081
19273 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
19274 * doc/gcc.texi: Include gcov-dump stuff.
19275 * doc/gcov-dump.texi: New file.
19277 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
19279 PR rtl-optimization/79150
19280 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
19281 conditional jump, if the jump is the last insn of the loop.
19283 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19284 Richard Biener <rguenth@suse.de>
19286 PR tree-optimization/79908
19287 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19288 been cast away, use force_gimple_operand to construct the side
19291 2017-03-21 Martin Liska <mliska@suse.cz>
19293 PR libfortran/79956
19294 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
19297 2017-03-21 Brad Spengler <spender@grsecurity.net>
19300 * plugin.c (htab_hash_plugin): New function.
19301 (add_new_plugin): Use it and adjust.
19302 (parse_plugin_arg_opt): Adjust.
19303 (init_one_plugin): Likewise.
19305 2017-03-21 Richard Biener <rguenther@suse.de>
19307 PR tree-optimization/80032
19308 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
19309 if set force the cleanup to happen unconditionally.
19310 (gimplify_target_expr): Push inserted clobbers with force_uncond
19311 to avoid them being removed by control-dependent DCE.
19313 2017-03-21 Richard Biener <rguenther@suse.de>
19315 PR tree-optimization/80122
19316 * tree-inline.c (copy_bb): Do not expans va-arg packs or
19317 va_arg_pack_len when the inlined call stmt requires pack
19319 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
19321 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19324 * tsan.c (instrument_builtin_call): If the memory model argument
19325 is not a constant, assume it is valid.
19328 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
19331 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19333 PR rtl-optimization/79910
19334 * combine.c (can_combine_p): Do not allow combining an I0 or I1
19335 if its dest is used by an insn before I2 (other than the combined
19336 insns themselves, which are properly handled already).
19338 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19341 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19343 * combine.c (record_used_regs): New static function.
19344 (try_combine): Handle situations where there is an additional
19345 instruction between I2 and I3 which needs to have a LOG_LINK
19349 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19351 * combine.c (try_combine): Delete redundant i1 test. Call
19352 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19354 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19357 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
19358 alternatives 13/14.
19360 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19362 PR tree-optimization/80054
19363 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
19364 the optimization if a PHI or any of its arguments is not dominated
19365 by the candidate's basis. Use gphi* rather than gimple* as
19367 (replace_profitable_candidates): Clean up a gimple* variable that
19368 should be a gphi* variable.
19370 2017-03-20 Martin Sebor <msebor@redhat.com>
19373 * doc/extend.texi (attribute constructor): Document present limitation.
19375 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
19378 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
19379 __POWER9_VECTOR__ #ifdef control, change template definition to
19380 use Power9-specific built-in function.
19381 (vec_any_eq): Likewise.
19382 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
19383 to control outcomes from this test.
19384 (vector_ae_<mode>p): For VEC_F modes, likewise.
19386 2017-03-20 Ian Lance Taylor <iant@google.com>
19388 * config/i386/i386.c (ix86_function_regparm): Save an extra
19389 register for -fsplit-stack with DECL_STATIC_CHAIN.
19391 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19394 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
19395 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
19397 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19399 * config/riscv/riscv.c (riscv_print_operand): Use "fence
19401 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
19404 2017-03-20 Marek Polacek <polacek@redhat.com>
19407 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
19409 2017-03-20 Richard Biener <rguenther@suse.de>
19411 PR tree-optimization/80113
19412 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
19413 allocate extra SSA name for PHI def.
19414 (add_close_phis_to_outer_loops): Likewise.
19415 (add_close_phis_to_merge_points): Likewise.
19416 (copy_loop_close_phi_args): Likewise.
19417 (copy_cond_phi_nodes): Likewise.
19419 2017-03-20 Martin Liska <mliska@suse.cz>
19421 PR middle-end/79753
19422 * tree-chkp.c (chkp_build_returned_bound): Do not build
19423 returned bounds for a LHS that's not a BOUNDED_P type.
19425 2017-03-20 Martin Liska <mliska@suse.cz>
19429 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
19430 COMPLEX_CST and VECTOR_CST.
19432 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19435 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
19436 target operand. A new splitter adds the clobber statement in case
19437 the target operand is dead anyway.
19439 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
19441 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
19442 to age-old versions of binutils and glibc.
19444 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
19446 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
19448 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19450 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
19452 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19454 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
19455 requirement for binutils 2.13.
19457 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19459 * combine.c (try_combine): Delete redundant i1 test. Call
19460 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19462 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
19464 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
19465 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
19467 <riscv64-*-elf>: Re-arrange section
19468 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
19469 <riscv32-*-linux>: Likewise.
19470 <riscv64-*-elf>: Likewise
19471 <riscv64-*-linux>: Likewise.
19473 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
19476 * aarch64.opt(verbose-cost-dump): Fix typo.
19478 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
19481 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
19482 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
19484 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19486 * reload.c (find_reloads): When reloading a nonoffsettable address,
19487 use RELOAD_OTHER for it and its address reloads.
19489 PR rtl-optimization/79910
19490 * combine.c (record_used_regs): New static function.
19491 (try_combine): Handle situations where there is an additional
19492 instruction between I2 and I3 which needs to have a LOG_LINK
19495 2017-03-17 Jeff Law <law@redhat.com>
19497 PR tree-optimization/71437
19498 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
19499 conditional in the hash table first.
19500 (vrp_dom_walker::before_dom_children): Extract condition from
19501 ASSERT_EXPR. Record condition, its inverion and any implied
19502 conditions as well.
19504 2017-03-17 Marek Polacek <polacek@redhat.com>
19505 Markus Trippelsdorf <markus@trippelsdorf.de>
19507 PR tree-optimization/80079
19508 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
19511 2017-03-17 Richard Biener <rguenther@suse.de>
19513 PR middle-end/80075
19514 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
19515 Properly verify the LHS before the RHS possibly claims to be
19517 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
19520 2017-03-17 Martin Jambor <mjambor@suse.cz>
19522 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
19523 (List of -O2 options): Likewise.
19524 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
19527 2017-03-17 Tom de Vries <tom@codesourcery.com>
19529 * gcov-dump.c (print_usage): Print bug_report_url.
19531 2017-03-17 Richard Biener <rguenther@suse.de>
19533 PR middle-end/80050
19534 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
19535 (parser::peek): Likewise.
19537 2017-03-17 Richard Biener <rguenther@suse.de>
19539 PR tree-optimization/80048
19540 * sese.c (free_sese_info): Properly release rename_map and
19541 copied_bb_map elements.
19543 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
19545 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
19546 Add linked-list forward and backlinks. Insert on
19547 construction, remove on destruction.
19548 (class pass_store_merging): Add m_stores_head field.
19549 (pass_store_merging::terminate_and_process_all_chains):
19550 Iterate over m_stores_head list.
19551 (pass_store_merging::terminate_all_aliasing_chains):
19553 (pass_store_merging::execute): Check for debug stmts first.
19554 Push new chains onto the m_stores_head stack.
19556 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
19559 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
19560 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
19561 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
19563 2017-03-16 Jeff Law <law@redhat.com>
19565 PR tree-optimization/71437
19566 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
19567 member function. Implementation moved into after_dom_children
19568 member function and into the threader's thread_outgoing_edges
19570 (dom_opt_dom_walker::after_dom_children): Simplify by moving
19571 some code into new thread_outgoing_edges.
19572 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
19573 definition. Simplify marker handling (do it here). Assume we always
19574 have the available expression and the const/copies tables.
19575 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
19577 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
19578 * tree-vrp.c (equiv_stack): No longer file scoped.
19579 (vrp_dom_walker): New class.
19580 (vrp_dom_walker::before_dom_children): New member function.
19581 (vrp_dom_walker::after_dom_children): Likewise.
19582 (identify_jump_threads): Setup domwalker. Use it rather than
19583 walking edges in a random order by hand. Simplify setup/finalization.
19584 (finalize_jump_threads): Remove.
19585 (vrp_finalize): Do not call identify_jump_threads here.
19586 (execute_vrp): Do it here instead and call thread_through_all_blocks
19589 PR tree-optimization/71437
19590 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
19592 (simplify_stmt_for_jump_threading): Add basic_block argument. All
19594 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
19595 (dom_opt_dom_walker::thread_across_edge): Remove
19596 handle_dominating_asserts argument. All callers changed.
19597 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
19598 changes. Remove calls to lhs_of_dominating_assert. Other
19599 uses of handle_dominating_asserts turn into unconditional code
19600 (simplify_control_stmt_condition_1): Likewise.
19601 (simplify_control_stmt_condition): Likewise.
19602 (thread_through_normal_block, thread_across_edge): Likewise.
19603 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
19604 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
19605 object if it is not an SSA_NAME.
19606 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
19607 before calling into the VRP specific simplifiers.
19608 (identify_jump_threads): Remove handle_dominating_asserts
19611 2017-03-16 Jakub Jelinek <jakub@redhat.com>
19614 * tree-diagnostic.c (default_tree_printer): No longer static.
19615 * tree-diagnostic.h (default_tree_printer): New prototype.
19617 2017-03-16 Tamar Christina <tamar.christina@arm.com>
19619 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
19620 Change ins into fmov.
19622 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19624 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
19625 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
19626 Use h_con constraint for operand 1.
19627 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
19628 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
19630 2017-03-15 Jeff Law <law@redhat.com>
19632 PR tree-optimization/71437
19633 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
19634 (record_temporary_equivalences): Use it.
19636 PR tree-optimization/71437
19637 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
19638 tree-ssa-scopedtables.
19639 (lookup_avail_expr, build_and_record_new_cond): Likewise.
19640 (record_conditions, record_cond, vuse_eq): Likewise.
19641 (record_edge_info): Adjust to API tweak of record_conditions.
19642 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
19643 (record_temporary_equivalences, optimize_stmt): Likewise.
19644 (eliminate_redundant_computations): Likewise.
19645 (record_equivalences_from_stmt): Likewise.
19646 * tree-ssa-scopedtables.c: Include options.h and params.h.
19647 (vuse_eq): New function, moved from tree-ssa-dom.c
19648 (build_and_record_new_cond): Likewise.
19649 (record_conditions): Likewise. Accept vector of conditions rather
19650 than edge_equivalence structure for first argument.
19651 for the first argument.
19652 (avail_exprs_stack::lookup_avail_expr): New member function, moved
19653 from tree-ssa-dom.c.
19654 (avail_exprs_stack::record_cond): Likewise.
19655 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
19656 from tree-ssa-dom.c.
19657 (avail_exprs_stack): Add new member functions lookup_avail_expr
19659 (record_conditions): Declare.
19661 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
19664 * lra-constraints.c (process_alt_operands): Increase reject for
19665 reloading an input/output operand.
19667 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
19670 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
19671 insns to convert from signed/unsigned char/short to IEEE 128-bit
19673 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
19675 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
19678 * config/i386/i386.c (ix86_vector_duplicate_value): Create
19679 subreg of inner mode for values already in registers.
19681 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
19683 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
19684 iteration reg is used after the loop.
19686 2017-03-14 Martin Sebor <msebor@redhat.com>
19688 PR tree-optimization/79800
19689 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
19690 precision in negative-positive range.
19691 (format_floating): Call non-const overload with adjusted precision.
19693 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19696 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
19699 2017-03-14 Martin Sebor <msebor@redhat.com>
19701 PR middle-end/80020
19702 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
19703 * builtins.def (aligned_alloc): Use it.
19706 * Makefile.in (GTFILES): Add calls.c.
19707 * calls.c: Include "gt-calls.h".
19709 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
19711 PR rtl-optimization/79728
19712 * regs.h (struct target_regs): New field
19713 x_contains_allocatable_regs_of_mode.
19714 (contains_allocatable_regs_of_mode): New macro.
19715 * reginfo.c (init_reg_sets_1): Initialize it, and change
19716 contains_reg_of_mode so it includes global regs as well.
19717 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
19718 rather than contains_regs_of_mode.
19720 2017-03-14 Martin Liska <mliska@suse.cz>
19722 * doc/invoke.texi: Document options that can't be combined with
19723 -fcheck-pointer-bounds.
19725 2017-03-14 Martin Liska <mliska@suse.cz>
19727 PR middle-end/79831
19728 * doc/invoke.texi (-Wchkp): Document the option.
19730 2017-03-14 Martin Liska <mliska@suse.cz>
19732 * Makefile.in: Install gcov-dump.
19734 2017-03-14 Martin Liska <mliska@suse.cz>
19736 * multiple_target.c (expand_target_clones): Bail out for
19737 an invalid attribute.
19739 2017-03-14 Richard Biener <rguenther@suse.de>
19741 * alias.c (struct alias_set_entry): Pack properly.
19742 * cfgloop.h (struct loop): Likewise.
19743 * cse.c (struct set): Likewise.
19744 * ipa-utils.c (struct searchc_env): Likewise.
19745 * loop-invariant.c (struct invariant): Likewise.
19746 * lra-remat.c (struct cand): Likewise.
19747 * recog.c (struct change_t): Likewise.
19748 * rtl.h (struct address_info): Likewise.
19749 * symbol-summary.h (function_summary): Likewise.
19750 * tree-loop-distribution.c (struct partition): Likewise.
19751 * tree-object-size.c (struct object_size_info): Likewise.
19752 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
19753 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
19754 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
19755 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
19756 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
19757 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
19758 (struct _stmt_vec_info): Likewise.
19760 2017-03-14 Martin Liska <mliska@suse.cz>
19763 * multiple_target.c (create_dispatcher_calls): Check that
19764 a target can create a function dispatcher.
19766 2017-03-14 Martin Liska <mliska@suse.cz>
19769 * multiple_target.c (expand_target_clones): Drop local.local
19770 flag for default implementation.
19772 2017-03-14 Richard Biener <rguenther@suse.de>
19774 PR tree-optimization/80030
19775 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
19777 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
19779 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
19780 gcc_fallthrough() instead of __attribute__((fallthrough));
19782 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19784 * doc/gcc.texi: Remove "up" link to (DIR).
19785 * doc/gccint.texi: Ditto.
19787 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19789 * doc/install.texi (Specific) <avr>: Remove reference to
19792 2017-03-13 Jeff Law <law@redhat.com>
19794 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
19795 attribute rather than comments.
19797 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
19798 match_scratch operand is highest.
19800 2017-03-13 Martin Liska <mliska@suse.cz>
19802 PR middle-end/78339
19803 * ipa-pure-const.c (warn_function_noreturn): If the declarations
19804 is a CHKP clone, use original declaration.
19806 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19808 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
19809 (arc_conditional_register_usage): Use a different allocation order
19810 when optimizing for size.
19811 * common/config/arc/arc-common.c (arc_option_optimization_table):
19812 Section anchors default on when optimizing for size.
19814 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19816 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
19818 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19820 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
19821 * config/arc/arc.md (cpu_facility): Add cd variant.
19822 (*movqi_insn): Add code density variant.
19823 (*movhi_insn): Likewise.
19824 (*movqi_insn): Likewise.
19825 (*addsi3_mixed): Likewise.
19826 (subsi3_insn): Likewise.
19828 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19830 * config/arc/arc.md (movsi_cond_exec): Update constraint.
19832 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19834 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
19835 expressions with MINUS and UNARY ops.
19837 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19840 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
19842 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
19843 between vec_select and vector argument.
19844 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
19845 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
19846 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
19847 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
19848 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
19849 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
19851 2017-03-13 Richard Biener <rguenther@suse.de>
19854 * params.def (vect-max-peeling-for-alignment): Fix typo.
19856 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19858 * doc/install.texi (Specific) <mips-*-*>: Remove description of
19859 issue that only occurred with binutils below 2.18.
19861 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19863 * doc/install.texi (Specific) <cris-axis-elf>: No longer
19864 refer to binutils 2.11/2.12 minimum.
19866 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19868 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
19869 ftp.kernel.org and simplify binutils requirement.
19871 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
19873 * doc/invoke.texi (Warning Options): Fix spelling of link-time
19875 (Optimize Options): Ditto. Also remove redundancy.
19877 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19879 PR translation/79848
19880 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
19882 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
19883 to G_ to avoid double translation.
19885 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19887 PR translation/79923
19888 * auto-profile.c (get_combined_location): Convert leading
19889 character of diagnostics to lower case and remove trailing period.
19890 (read_profile): Likewise for various diagnostics.
19891 * config/arm/arm.c (arm_option_override): Remove trailing period
19892 from various diagnostics.
19893 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
19894 (msp430_expand_delay_cycles): Likewise.
19896 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19899 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
19900 full command-line argument, rather than just "str".
19901 (aarch64_validate_march): Likewise.
19902 (aarch64_validate_mtune): Likewise.
19904 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
19906 PR rtl-optimization/78911
19907 * lra-assigns.c (must_not_spill_p): New function.
19908 (spill_for): Use it.
19910 2017-03-10 Jakub Jelinek <jakub@redhat.com>
19912 PR tree-optimization/79981
19913 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
19914 ATOMIC_COMPARE_EXCHANGE ifn result.
19915 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
19916 IFN_ATOMIC_COMPARE_EXCHANGE.
19918 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19921 * opts.c (parse_sanitizer_options): Add missing question mark to
19922 "did you mean" message.
19924 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19926 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
19928 (VMULEUH_UNS): Likewise.
19929 (VMULOUB_UNS): Likewise.
19930 (VMULOUH_UNS): Likewise.
19931 * config/rs6000/rs6000.c (builtin_function_type): Remove
19932 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
19934 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19937 * read-rtl-function.c (function_reader::read_rtx_operand): Update
19938 x with result of extra_parsing_for_operand_code_0.
19939 (function_reader::extra_parsing_for_operand_code_0): Convert
19940 return type from void to rtx, returning x. When reading
19941 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
19942 larger size containing struct block_symbol.
19944 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
19946 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
19947 -mfloat128-hardware without -m64.
19949 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
19952 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
19953 entries to the case statement that marks unsigned arguments to
19954 overloaded functions.
19956 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19958 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
19959 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
19961 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
19964 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
19965 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
19967 2017-03-10 Martin Liska <mliska@suse.cz>
19971 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
19972 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
19975 2017-03-10 Olivier Hainque <hainque@adacore.com>
19977 * tree-switch-conversion (array_value_type): Start by resetting
19978 candidate type to it's main variant.
19980 2017-03-10 Jakub Jelinek <jakub@redhat.com>
19982 PR rtl-optimization/79909
19983 * combine.c (try_combine): Use simplify_replace_rtx on individual
19984 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
19985 of the whole CALL_INSN_FUNCTION_USAGE.
19987 PR tree-optimization/79972
19988 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
19989 get_range_info on SSA_NAMEs. Formatting fixes.
19991 2017-03-10 Richard Biener <rguenther@suse.de>
19992 Jakub Jelinek <jakub@redhat.com>
19994 PR tree-optimization/77975
19995 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
19996 edge to be constant.
19997 (get_val_for): For constant x return it. Formatting fix.
19998 (loop_niter_by_eval): Avoid pointless looping if the next iteration
19999 would use the same bases as the current one.
20001 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20003 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
20004 instead of vec_select for V1TImode.
20005 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
20007 (VSX_LE_128): Add V1TI to this mode iterator.
20008 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
20009 (*vsx_le_perm_store_<mode>): Likewise.
20010 (pre-reload splitter for VSX stores): Likewise.
20011 (post-reload splitter for VSX stores): Likewise.
20012 (*vsx_xxpermdi2_le_<mode>): Likewise.
20013 (*vsx_lxvd2x2_le_<mode>): Likewise.
20014 (*vsx_stxvd2x2_le_<mode>): Likewise.
20016 2017-03-09 Michael Eager <eager@eagercon.com>
20018 Correct failures with --enable-checking=yes,rtl.
20020 * config/microblaze/microblaze.c (microblaze_expand_shift):
20021 Replace GET_CODE test with CONST_INT_P and INTVAL test with
20022 test for const0_rtx.
20023 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
20024 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
20026 2017-03-09 Richard Biener <rguenther@suse.de>
20028 PR tree-optimization/79977
20029 * graphite-scop-detection.c (scop_detection::merge_sese):
20030 Handle the case of extra exits to blocks dominating the entry.
20032 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
20034 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20037 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
20039 PR rtl-optimization/79949
20040 * lra-constraints.c (process_alt_operands): Check memory when
20041 trying to predict a cycle. Print about the overall increase.
20043 2017-03-09 Richard Biener <rguenther@suse.de>
20045 PR middle-end/79971
20046 * gimple-expr.c (useless_type_conversion_p): Preserve
20047 TYPE_SATURATING for fixed-point types.
20049 2017-03-09 Richard Biener <rguenther@suse.de>
20052 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
20053 alignment of BLKmode params.
20055 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20058 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
20059 (VALL_NO_V2Q): Likewise.
20061 * config/aarch64/aarch64-simd.md
20062 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
20064 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
20065 VALL_NO_V2Q mode iterator.
20066 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
20068 2017-03-09 Martin Liska <mliska@suse.cz>
20070 PR tree-optimization/79631
20071 * tree-chkp-opt.c (chkp_is_constant_addr): Call
20072 tree_int_cst_sign_bit just for INTEGER constants.
20074 2017-03-09 Martin Liska <mliska@suse.cz>
20078 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
20081 2017-03-09 Marek Polacek <polacek@redhat.com>
20084 * tree.c (inchash::add_expr): Handle TREE_VEC.
20086 2017-03-09 Martin Liska <mliska@suse.cz>
20089 (chkp_narrow_size_and_offset): New function.
20090 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
20091 (void chkp_parse_bit_field_ref): New function.
20092 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
20093 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
20095 2017-03-09 Martin Liska <mliska@suse.cz>
20098 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
20099 (chkp_find_bounds_1): Remove gcc_unreachable.
20101 2017-03-09 Jakub Jelinek <jakub@redhat.com>
20104 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
20105 BUILT_IN_SYNC*, determine the access type from the size suffix and
20106 always build a MEM_REF with that type. Handle forgotten
20107 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
20110 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
20111 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
20112 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
20113 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
20114 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
20115 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
20116 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
20117 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
20118 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
20119 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
20120 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
20121 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
20122 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
20123 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
20124 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
20125 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
20126 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
20127 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
20128 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
20129 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
20130 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
20131 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
20132 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
20133 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
20134 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
20135 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
20136 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
20137 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
20138 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
20139 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
20140 definitions outside of __OPTIMIZE__ guarded section.
20143 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
20144 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
20145 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
20146 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
20149 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20151 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
20152 ("vfenez<mode>"): Add missing constraints.
20154 2017-03-08 Martin Sebor <msebor@redhat.com>
20157 * config/nds32/nds32.c (nds32_option_override):
20158 Fix misspelled diagnostic.
20160 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20163 * gimplify.c (gimplify_omp_for): Replace index var in outer
20164 taskloop statement with an artificial variable and add
20165 OMP_CLAUSE_PRIVATE clause for it.
20167 2017-03-08 Richard Biener <rguenther@suse.de>
20169 PR tree-optimization/79955
20170 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
20171 for accesses that are completely outside of the variable.
20173 2017-03-08 Andrew Haley <aph@redhat.com>
20175 PR tree-optimization/79943
20176 * tree-ssa-loop-split.c (compute_new_first_bound): When
20177 calculating the new upper bound, (END-BEG) should be added, not
20180 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20182 * config/avr/avr.md (setmemhi): Make sure match_dup
20183 operand number comes before match_scratch.
20185 2017-03-08 Richard Biener <rguenther@suse.de>
20187 PR tree-optimization/79920
20188 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
20189 with ncopies == 1 to ...
20190 (vect_transform_slp_perm_load): ... here. Properly compute
20191 all element loads by iterating VF times over the group. Do
20192 not handle ncopies (computed in a broken way) in
20193 vect_create_mask_and_perm.
20195 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20198 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
20199 is a uniform vector, use uniform_vector_p return value instead of
20200 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
20202 2017-03-07 Marek Polacek <polacek@redhat.com>
20204 PR middle-end/79809
20205 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
20206 (alloca_call_type): Likewise.
20208 2017-03-07 Martin Liska <mliska@suse.cz>
20210 * gcov.c (process_args): Put comment to correct location.
20212 2017-03-07 Martin Liska <mliska@suse.cz>
20214 PR middle-end/68270
20215 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
20216 Use array_at_struct_end_p instead of DECL_CHAIN (field).
20217 (chkp_narrow_bounds_for_field): Likewise.
20218 (chkp_parse_array_and_component_ref): Pass one more argument to
20221 2017-03-07 Richard Biener <rguenther@suse.de>
20223 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
20226 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
20228 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
20229 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
20231 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20234 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
20235 to end of description.
20236 (PARAM_MAX_STORES_TO_MERGE): Likewise.
20238 2017-03-07 Jakub Jelinek <jakub@redhat.com>
20240 PR rtl-optimization/79901
20241 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
20243 (*avx512f_<code><mode>3<mask_name>): ... this.
20244 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
20245 iterator instead of VI8_AVX2_AVX512BW.
20247 PR rtl-optimization/79901
20248 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
20249 min/max expander, expand it using expand_vec_cond_expr.
20252 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
20255 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20258 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
20259 to void * for PCH reasons.
20260 * dwarf2out.c (output_loc_operands, output_die): Cast
20261 v.val_vec.array to unsigned char *.
20263 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
20266 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
20269 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
20271 PR rtl-optimization/79571
20272 * lra-constraints.c (process_alt_operands): Calculate static
20273 reject and subtract it from overall when only addresses will be
20276 2017-03-06 Julia Koval <julia.koval@intel.com>
20279 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
20280 incoming stack boundary to 128 for 64-bit targets.
20282 2017-03-06 Richard Biener <rguenther@suse.de>
20284 PR tree-optimization/79894
20285 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
20286 to NULL after folding it.
20288 2017-03-06 Richard Biener <rguenther@suse.de>
20290 PR tree-optimization/79824
20291 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
20292 check disabling peeling for gaps.
20294 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
20296 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20297 attributes): Document gettimeofday.
20299 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20301 * config/s390/s390.c (s390_option_override_internal): Set
20302 PARAM_MIN_VECT_LOOP_BOUND
20304 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20306 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
20307 * config/s390/s390.md: Likewise.
20309 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20312 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
20313 (<avx2_avx512>_perm<mode>): Rename to ...
20314 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
20316 (<avx512>_perm<mode>_mask): Rename to ...
20317 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
20319 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
20320 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
20321 instead of VI8F_256_512.
20322 (avx512f_perm<mode>): New define_expand.
20323 (avx512f_perm<mode>_mask): Likewise.
20324 (avx512f_perm<mode>_1<mask_name>): New define_insn.
20325 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
20327 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20329 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
20330 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
20332 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
20334 2017-03-06 Martin Liska <mliska@suse.cz>
20337 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
20338 when having a SSA NAME w/o VAR_DECL assigned to it.
20340 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20342 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
20343 msa_dpsub_<su>_d): Fix MODE for vec_select.
20345 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20347 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
20349 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
20351 2017-03-06 Richard Biener <rguenther@suse.de>
20353 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
20354 * plugin.c (register_plugin_info): Likewise.
20355 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
20357 2017-03-05 Jakub Jelinek <jakub@redhat.com>
20359 * config/i386/sse.md (sse_storehps, sse_storelps,
20360 avx_<castmode><avxsizesuffix>_<castmode>,
20361 avx512f_<castmode><avxsizesuffix>_<castmode>,
20362 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
20363 in condition that at least one operand is not a MEM.
20365 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20367 PR middle-end/79805
20368 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
20369 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
20371 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
20372 gimple_call_nothrow_p flag based on whether original builtin can throw.
20373 If it can, emit following stmts on the fallthrough edge.
20374 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
20375 don't create new bb if inserting just debug stmts on the edge, try to
20376 insert them on the fallthru bb or just reset debug stmts.
20378 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
20381 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
20382 restore recog_data (including the operand rtxes inside it) around
20383 the call to get_insn_template.
20385 2017-03-03 Martin Sebor <msebor@redhat.com>
20387 PR tree-optimization/79699
20388 * context.c (context::~context): Free MPFR caches to avoid
20389 a memory leak on program exit.
20391 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20393 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
20394 Use wide_int::ulow () instead of .elt (0).
20396 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20398 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
20399 (*pushxf): Limit oF constraint to 32bit targets and add oC
20400 constraint for 64bit targets.
20401 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
20402 (*pushdf): Change rmF constraint to rmC.
20404 2017-03-03 Martin Liska <mliska@suse.cz>
20406 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
20407 Remove unused variable.
20409 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20412 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
20413 is a memory operand, increase num_memory.
20414 (ix86_expand_args_builtin): Likewise.
20416 2017-03-03 Jan Hubicka <jh@suse.cz>
20419 * ipa-devirt.c (maybe_record_node): Properly handle
20420 __cxa_pure_virtual visibility.
20422 2017-03-03 Martin Liska <mliska@suse.cz>
20424 PR tree-optimization/79803
20425 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
20427 (pass_loop_prefetch::execute): Disabled optimization if an
20428 assumption about L1 cache size is not met.
20430 2017-03-03 Martin Liska <mliska@suse.cz>
20432 PR rtl-optimization/79574
20433 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
20434 (hash_scan_set): Likewise.
20435 (dump_hash_table): Likewise.
20436 (hoist_code): Likewise.
20438 2017-03-03 Richard Biener <rguenther@suse.de>
20440 * fixed-value.c (fixed_from_string): Restore use of elt (1)
20441 in place of uhigh ().
20442 (fixed_convert_from_real): Likewise.
20444 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20447 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
20449 2017-03-03 Richard Biener <rguenther@suse.de>
20451 PR middle-end/79818
20452 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
20453 TYPE_OVERFLOW_UNDEFINED check.
20455 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20457 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
20459 (vector_ae_<mode>_p): Likewise.
20460 (vector_nez_<mode>_p): Likewise.
20461 (vector_ne_v2di_p): Likewise.
20462 (vector_ae_v2di_p): Likewise.
20463 (vector_ne_<mode>_p): Likewise.
20464 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
20466 (vsx_tsqrt<mode>2_fe): Likewise.
20468 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
20471 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
20473 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20475 PR rtl-optimization/79780
20476 * cprop.c (one_cprop_pass): When second and further conditional trap
20477 in a single basic block is turned into an unconditional trap, turn it
20478 into a deleted note to avoid RTL verification failures.
20480 2017-03-02 Richard Biener <rguenther@suse.de>
20482 * fold-const.c (const_binop): Use ulow () instead of elt (0).
20484 2017-03-02 Richard Biener <rguenther@suse.de>
20486 PR tree-optimization/79345
20488 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
20489 param and abort the walk, returning -1 if it is hit.
20490 (walk_aliased_vdefs): Take a limit param and pass it on.
20491 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
20492 defaulting to 0 and return a signed int.
20493 * tree-ssa-uninit.c (struct check_defs_data): New struct.
20494 (check_defs): New helper.
20495 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
20496 about uninitialized memory.
20497 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
20498 bogus uninitialized warning.
20499 (fixed_convert_from_real): Likewise.
20501 2017-03-02 Bin Cheng <bin.cheng@arm.com>
20503 PR tree-optimization/66768
20504 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
20505 iv_use if base object can't be determined.
20507 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20509 PR tree-optimization/79345
20510 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
20511 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
20512 (get_pattern_stats): Initialize it.
20513 * genemit.c (gen_expand): Verify match_scratch numbers come after
20514 match_operand/match_dup numbers.
20515 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
20516 match_scratch numbers.
20517 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
20519 * config/s390/s390.md (trunctdsd2): Likewise.
20521 2017-03-02 Richard Biener <rguenther@suse.de>
20523 * wide-int.h (wide_int_storage::operator=): Implement in terms
20526 2017-03-02 Richard Biener <rguenther@suse.de>
20528 PR tree-optimization/79777
20529 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
20530 the to insert expression to sth existing.
20532 2017-03-01 Martin Sebor <msebor@redhat.com>
20534 PR middle-end/79692
20535 * gimple-ssa-sprintf.c
20536 (directive::known_width_and_precision): New function.
20537 (format_integer): Use it.
20538 (get_mpfr_format_length): Consider the full range of precision
20539 when computing %g output with the # flag. Set the likely byte
20540 count to 3 rather than 1 when precision is indeterminate.
20541 (format_floating): Correct the lower bound of precision.
20543 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20545 * doc/invoke.texi: Document default code model for 64-bit Linux.
20547 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20550 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
20551 udiv rather than div since input pattern is unsigned.
20553 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
20555 * config/i386/i386.c (print_reg): Warn for values of
20556 unsupported size in integer register.
20558 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
20561 * config/rs6000/predicates.md (current_file_function_operand): Do
20562 not allow self calls to be local if the function is replaceable.
20564 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
20567 * config/rs6000/altivec.h (vec_ctz and others): Change the
20568 preprocessor macro that controls conditional compilation from
20569 _ARCH_PWR9 to __POWER9_VECTOR__.
20570 (vec_all_ne): Change parameterization of __altivec_scalar_pred
20571 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
20572 control (instead of _ARCH_PWR9 control) so that template
20573 definition uses power9-specific function.
20574 (vec_any_eq): Likewise.
20575 (vec_all_ne): Change macro definition to use a power9-specific
20576 expansion under #ifdef __POWER9_VECTOR__ control (instead of
20577 _ARCH_PWR9 control).
20578 (vec_any_eq) Likewise.
20579 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
20580 expansion for CMPNEF to remove support for xvcmpnesp instruction.
20581 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
20582 support for xvcmpnedp instruction.
20583 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
20584 macro expansion so that Power9 implementation of vec_all_ne does
20585 not use the AltiVec predicate framework.
20586 (VCMPNEH_P): Likewise.
20587 (VCMPNEW_P): Likewise.
20588 (VCMPNED_P): Likewise.
20589 (VCMPNEFP_P): Likewise.
20590 (VCMPNEDP_P): Likewise.
20591 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
20592 implementation of vec_any_eq to not use AltiVec predicate
20594 (VCMPAEH_P): Likewise.
20595 (VCMPAEW_P): Likewise.
20596 (VCMPAED_P): Likewise.
20597 (VCMPAEFP_P): Likewise.
20598 (VCMPAEDP_P): Likewise.
20599 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
20600 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
20601 not use the AltiVec predicate framework.
20602 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
20603 of vec_any_eq to not use AltiVec predicate framework.
20604 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20605 support for predefined __POWER9_VECTOR__ macro to indicate that
20606 Power9 instruction selection is enabled.
20607 (altivec_overloaded_builtins): Remove extraneous
20608 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
20609 function argument types RS6000_BTI_bool_V16QI and
20610 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
20611 entry for overloaded function argument types RS6000_BTI_bool_V4SI
20612 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
20613 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
20614 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
20615 Power9 for implementations of vec_cmpne. Change the signature for
20616 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
20617 (representing vec_all_ne) to remove the previously described first
20618 argument of type RS6000_BTI_INTSI, as this was an artifact of
20619 reliance on the AltiVec predicate framework, which is no longer
20620 used in the implementation of these functions. Add
20621 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
20622 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
20623 since, unlike the AltiVec predicate framework implementation, we
20624 do not share function descriptors between vec_alle and vec_anyeq.
20625 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
20626 set of modes that receive special treatment even when
20627 TARGET_P9_VECTOR is true. The special treatment emits code that
20628 does not depend on Power9 instructions.
20629 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
20630 define_expand to not rely on AltiVec predicate framework.
20631 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20633 (vector_ne_v2di_p): Change this define_expand to not rely on
20634 AltiVec predicate framework.
20635 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
20637 (vector_ne_<mode>_p): Change this define_expand to not rely on
20638 AltiVec predicate framework.
20639 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20641 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
20642 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
20643 define_insn pattern.
20644 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
20645 define_insn pattern because the xvcmpne<VSs>. instruction is not
20647 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
20648 instruction is not supported.
20650 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20652 * config/nvptx/nvptx.c: Include intl.h.
20654 2017-03-01 Martin Jambor <mjambor@suse.cz>
20657 * ipa-prop.h (ipa_bits): Removed field known.
20658 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
20659 to pointers. Adjusted their comments to warn about their sharing.
20660 (ipcp_transformation_summary): Change bits to a vector of pointers.
20661 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
20662 (ipa_get_ipa_bits_for_value): Declare.
20663 * tree-vrp.h (value_range): Mark as GTY((for_user)).
20664 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
20665 (ipa_bits_hash_table): Likewise.
20666 (ipa_vr_ggc_hash_traits): Likewise.
20667 (ipa_vr_hash_table): Likewise.
20668 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
20669 being pointers and vr_known being removed.
20670 (ipa_set_jf_unknown): Likewise.
20671 (ipa_get_ipa_bits_for_value): New function.
20672 (ipa_set_jfunc_bits): Likewise.
20673 (ipa_get_value_range): New overloaded functions.
20674 (ipa_set_jfunc_vr): Likewise.
20675 (ipa_compute_jump_functions_for_edge): Use the above functions to
20676 construct bits and vr parts of jump functions.
20677 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
20678 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20680 (ipcp_grow_transformations_if_necessary): Also allocate
20681 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20683 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
20684 them. Fix too long lines.
20685 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
20686 vr_known being removed.
20687 (ipa_read_jump_function): Use new setter functions to construct bits
20688 and vr parts of jump functions or set them to NULL.
20689 (write_ipcp_transformation_info): Adjust for bits being pointers.
20690 (read_ipcp_transformation_info): Likewise.
20691 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
20693 Include gt-ipa-prop.h.
20694 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
20696 (ipcp_store_bits_results): Likewise.
20697 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
20698 Do not write to existing jump functions but use a temporary instead.
20700 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20703 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
20704 attempt to use its first operand as BIT_FIELD_REF base.
20706 2017-03-01 Richard Biener <rguenther@suse.de>
20708 PR middle-end/79721
20709 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
20710 interpolating formula in wrapping arithmetic.
20711 (chrec_apply): Convert chrec_evaluate return value to wanted type.
20713 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20715 PR tree-optimization/79734
20716 * tree-vect-generic.c (expand_vector_condition): Optimize
20717 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
20718 Handle VEC_COND_EXPR where comparison has different inner width from
20719 type's inner width.
20721 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
20723 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
20724 markup, and similar issues. Remove @opindex entries for things
20725 that aren't options. Add missing -mmpy-option entries.
20727 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20729 PR tree-optimization/79737
20730 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
20731 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
20732 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
20733 instead of byte_size. Formatting fix.
20734 (shift_bytes_in_array_right): Formatting fix.
20736 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
20739 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
20740 condition on optimize for the leaf function test.
20742 2017-02-28 Martin Liska <mliska@suse.cz>
20745 * read-rtl-function.c (function_reader::handle_unknown_directive):
20746 Bail out when one uses -flto.
20748 2017-02-28 Martin Liska <mliska@suse.cz>
20750 * common.opt: Replace space with tabular for options of <number>
20752 * config/i386/i386.opt: Show <number> value for
20753 -mlarge-data-threshold.
20754 * opts.c (print_filtered_help): Do not display number in hexadecimal
20757 2017-02-28 Martin Liska <mliska@suse.cz>
20759 * common.opt: Fix --help=option -Q for options which are of
20762 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
20764 * config/i386/i386.c (print_reg): Error out for values
20765 of 8-bit size in invalid integer register.
20767 2017-02-28 Martin Sebor <msebor@redhat.com>
20769 PR tree-optimization/79691
20770 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
20772 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20775 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
20776 gcc_unreachable with output_operand_lossage.
20778 2017-02-28 Richard Biener <rguenther@suse.de>
20780 PR tree-optimization/79740
20781 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
20783 (visit_nary_op): Insert the nary into the hashtable if we
20784 pattern-matched sth.
20785 * tree-ssa-pre.c (eliminate_insert): Robustify.
20787 2017-02-28 Richard Biener <rguenther@suse.de>
20789 PR middle-end/79731
20790 * fold-const.c (decode_field_reference): Reject out-of-bound
20793 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20795 * config/i386/i386.c: Include intl.h.
20796 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
20797 instead of just cond ? "..." : "...".
20798 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
20799 * coverage.c (read_counts_file): Likewise.
20800 * omp-offload.c: Include intl.h.
20801 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
20802 of just cond ? "..." : "...".
20803 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
20804 of just cond ? "..." : "...".
20806 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
20809 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
20811 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
20812 'tune for' CPU name.
20813 * config/arm/arm-cpu-data.h: Regenerated.
20815 2017-02-28 Richard Biener <rguenther@suse.de>
20817 PR tree-optimization/79732
20818 * tree-inline.c (expand_call_inline): Do not shadow var.
20820 2017-02-28 Richard Biener <rguenther@suse.de>
20822 PR tree-optimization/79723
20823 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
20824 address-space properly.
20826 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
20828 * doc/optinfo.texi (Optimization groups): Fix option used for
20830 * doc/invoke.texi (-fopt-info): Document "omp".
20831 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
20832 (OPTGROUP_ALL): Add OPTGROUP_OMP.
20833 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
20834 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
20835 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
20837 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
20839 * dumpfile.c (optgroup_options): Instead of "openmp", associate
20840 OPTGROUP_OMP with "omp".
20842 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
20845 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
20846 for arithmetic shift of unsigned V2DI.
20848 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
20850 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
20851 arc/linux.h headers.
20852 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
20853 (LINK_SPEC): Likewise.
20854 (ARC_TLS_EXTRA_START_SPEC): Likewise.
20855 (EXTRA_SPECS): Likewise.
20856 (STARTFILE_SPEC): Likewise.
20857 (ENDFILE_SPEC): Likewise.
20858 (LIB_SPEC): Likewise.
20859 (TARGET_SDATA_DEFAULT): Likewise.
20860 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
20861 (MULTILIB_DEFAULTS): Likewise.
20862 (DWARF2_UNWIND_INFO): Likewise.
20863 * config/arc/big.h: New file.
20864 * config/arc/elf.h: Likewise.
20865 * config/arc/linux.h: Likewise.
20866 * config/arc/t-uClibc: Remove.
20868 2017-02-27 Bin Cheng <bin.cheng@arm.com>
20870 PR tree-optimization/77536
20871 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
20872 (tree_transform_and_unroll_loop): Use above function to compute the
20873 estimated niter of unrolled loop and use it when scaling profile.
20874 Also use count info rather than frequency if it's non-zero.
20875 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
20876 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
20877 (vect_transform_loop): Call above function.
20879 2017-02-27 Richard Biener <rguenther@suse.de>
20881 PR tree-optimization/45397
20882 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
20883 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
20884 (visit_nary_op): Add pattern matching for CSEing sign-changed
20885 or truncated operations with wider ones.
20887 2017-02-27 Richard Biener <rguenther@suse.de>
20889 PR tree-optimization/79690
20890 * tree-vect-stmts.c (vectorizable_store): Use vector type
20891 built from the DR with address-space.
20893 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
20895 * doc/invoke.texi (Optimize Options): Refine the description
20896 of asan-use-after-return.
20898 2017-02-25 Alan Modra <amodra@gmail.com>
20900 PR rtl-optimization/79584
20901 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
20902 base, not ad->base_term, the reg within base. Remove assertion
20903 that ad->base == ad->base_term. Replace gen_int_mode using
20904 bogus mode with const0_rtx.
20906 2017-02-25 Jakub Jelinek <jakub@redhat.com>
20908 PR middle-end/79396
20909 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
20910 FMA_EXPR like tcc_binary or tcc_unary.
20912 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
20915 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
20917 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
20918 (output_loc_operands): Handle DW_OP_call_ref and
20919 DW_OP_GNU_variable_value.
20920 (struct variable_value_struct): New type.
20921 (struct variable_value_hasher): Likewise.
20922 (variable_value_hash): New variable.
20923 (string_types): Remove.
20924 (copy_loc_descr): New function.
20925 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
20926 (prepend_loc_descr_to_each): New function.
20927 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
20928 instead of add_loc_descr_to_each if the first argument is single
20929 location list and the second has multiple.
20930 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
20931 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
20932 when looking for variable value which doesn't have other location info.
20933 (loc_list_from_tree): Formatting fix.
20934 (gen_array_type_die): Simplify DW_AT_string_length handling.
20935 (adjust_string_types): Remove.
20936 (gen_subprogram_die): Don't call adjust_string_types nor test/set
20937 string_types. Call resolve_variable_values.
20938 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
20939 (resolve_addr_in_expr): Likewise. Add A argument.
20940 (copy_deref_exprloc): Remove deref argument. Adjust for the
20941 original expression being DW_OP_GNU_variable_value with optionally
20942 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
20943 optionally after it.
20944 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
20945 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
20946 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
20947 (variable_value_hasher::hash, variable_value_hasher::equal): New
20949 (resolve_variable_value_in_expr, resolve_variable_value,
20950 resolve_variable_values, note_variable_value_in_expr,
20951 note_variable_value): New functions.
20952 (dwarf2out_early_finish): Call note_variable_value on all toplevel
20955 2017-02-24 Jakub Jelinek <jakub@redhat.com>
20958 * opts.h (handle_generated_option): Add GENERATED_P argument.
20959 * opts-common.c (handle_option): Adjust function comment.
20960 (handle_generated_option): Add GENERATED_P argument, pass it to
20962 (control_warning_option): Pass false to handle_generated_option
20964 * opts.c (maybe_default_option): Pass true to handle_generated_option
20966 * optc-gen.awk: Likewise.
20968 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20970 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
20971 a REG, look at the REG it is a SUBREG of.
20972 (splitter for cmpeqsi_t): Ditto.
20974 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20976 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
20977 the special USEs with the pattern of the insn, not the insn itself.
20979 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
20982 * doc/invoke.texi: Document -mload-store-pairs.
20984 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20985 Sandra Loosemore <sandra@codesourcery.com>
20987 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
20988 argument isn't a CONST_INT.
20989 (nios2_alternate_compare_const): Assert op is a CONST_INT.
20990 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
20991 (nios2_validate_compare): Bypass alternate compare logic if *op2
20992 is not a CONST_INT.
20993 (ldstwm_operation_p): Return false if first_base is not a REG or
20994 if first_offset is not a CONST_INT.
20996 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20998 * config/cris/cris.md: Use correct operand in a define_peephole2.
21000 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21002 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
21004 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21006 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
21007 this_insn if it is an INSN or JUMP_INSN.
21008 (force_offsettable): Look at base, not at addr.
21009 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
21010 on things that aren't necessarily CONST_INTs.
21012 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
21014 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
21015 -mfpmath=sse is the default also for x86-32 targets with SSE2
21016 instruction set when @option{-ffast-math} is enabled
21018 2017-02-24 Jeff Law <law@redhat.com>
21020 PR rtl-optimizatoin/79286
21021 * ira.c (update_equiv_regs): Drop may_trap_p exception to
21024 2017-02-24 Richard Biener <rguenther@suse.de>
21026 PR tree-optimization/79389
21027 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
21030 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
21032 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
21033 function comment to reflect reality.
21034 (loop_exits_before_overflow): Fix typo in function description.
21036 2017-02-24 Richard Biener <rguenther@suse.de>
21038 PR tree-optimization/79389
21039 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
21040 properly that a threading opportunity exists. Detect conditional
21041 copy/constant propagation opportunities.
21043 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
21045 * config/visium/visium.md (type): Add trap.
21046 (b): New mode attribute.
21047 (*btst): Rename into...
21048 (*btst<mode>): ...this and adjust.
21049 (*cbranchsi4_btst_insn): Rename into...
21050 (*cbranch<mode>4_btst_insn): ...this and adjust.
21051 (trap): New define_insn.
21053 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21055 PR tree-optimization/79389
21056 * ifcvt.c (struct noce_if_info): Add rev_cond field.
21057 (noce_reversed_cond_code): New function.
21058 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
21059 reversed_comparison_code. Formatting fix.
21060 (noce_try_store_flag): Test rev_cond != NULL in addition to
21061 reversed_comparison_code.
21062 (noce_try_store_flag_constants): Likewise.
21063 (noce_try_store_flag_mask): Likewise.
21064 (noce_try_addcc): Use rev_cond if non-NULL instead of
21065 reversed_comparison_code.
21066 (noce_try_cmove_arith): Likewise. Formatting fixes.
21067 (noce_try_minmax, noce_try_abs): Clear rev_cond.
21068 (noce_find_if_block): Initialize rev_cond.
21069 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
21070 instead of false as last argument never attempt to reverse it
21073 2017-02-23 Bin Cheng <bin.cheng@arm.com>
21075 PR tree-optimization/79663
21076 * tree-predcom.c (combine_chains): Process refs in reverse order
21077 only for ZERO length chains, and add explaining comment.
21079 2017-02-23 Jeff Law <law@redhat.com>
21081 PR tree-optimization/79578
21082 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
21083 in call to operand_equal_p.
21085 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
21088 * config/i386/cpuid.h: Fix another undefined behavior.
21090 2017-02-23 Richard Biener <rguenther@suse.de>
21092 PR tree-optimization/79683
21093 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
21094 vector types for data-refs.
21096 2017-02-23 Martin Liska <mliska@suse.cz>
21098 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
21100 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21102 PR middle-end/79665
21103 * internal-fn.c (get_range_pos_neg): Moved to ...
21104 * tree.c (get_range_pos_neg): ... here. No longer static.
21105 * tree.h (get_range_pos_neg): New prototype.
21106 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
21107 are known to be in between 0 and signed maximum inclusive, try to
21108 expand both unsigned and signed divmod and use the cheaper one from
21111 2017-02-22 Jeff Law <law@redhat.com>
21113 PR tree-optimization/79578
21114 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
21115 to compare base operands.
21117 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
21120 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
21121 gpc_reg_operand instead of fpr_reg_operand.
21123 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
21125 * config/mips/mips.c (mips_return_in_memory): Force FP
21126 vector types to be returned in memory for o32 ABI.
21128 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21130 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
21131 instead of DW_TAG_member for static data member declarations and don't
21132 set no_linkage_name for static inline data members.
21133 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
21136 2017-02-22 Martin Liska <mliska@suse.cz>
21138 * doc/invoke.texi: Replace inequality signs with square brackets
21141 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21143 PR tree-optimization/68644
21144 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
21146 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
21149 * lra-constraints.c (simplify_operand_subreg): Handle
21150 WORD_REGISTER_OPERATIONS targets.
21152 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21155 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
21156 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
21157 elimination by swapping fld*.
21159 2017-02-22 Richard Biener <rguenther@suse.de>
21161 PR tree-optimization/79673
21162 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
21163 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
21164 irrelevant address-space qualifiers and avoiding a
21165 ADDR_SPACE_CONVERT_EXPR from fold_convert.
21167 2017-02-22 Richard Biener <rguenther@suse.de>
21169 PR tree-optimization/79666
21170 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
21171 to not symbolically negate if that may introduce undefined
21174 2017-02-22 Martin Liska <mliska@suse.cz>
21177 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
21178 * data-streamer-out.c (streamer_write_gcov_count_stream):
21180 * value-prof.c (stream_out_histogram_value): Make assert more
21181 precise based on type of counter.
21183 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
21186 * config/i386/i386.md (standard_x87sse_constant_load splitter):
21187 Use nonimmediate_operand instead of memory_operand for operand 1.
21188 (float-extend standard_x87sse_constant_load splitter): Ditto.
21190 2017-02-21 Jeff Law <law@redhat.com>
21192 PR tree-optimization/79621
21193 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
21194 blocks with edges to themselves.
21196 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21199 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
21200 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
21201 Use gimple_call_builtin_p.
21204 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
21205 on temporarily removed DEBUG_INSNs.
21207 PR tree-optimization/79649
21208 * tree-loop-distribution.c (classify_partition): Give up on
21209 non-generic address space loads/stores.
21211 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
21213 * doc/loop.texi (Loop manipulation): Remove nonexistent
21214 tree_ssa_loop_version from the documentation.
21215 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
21217 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21220 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
21221 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
21222 * config/rs6000/rs6000.c: Include except.h.
21223 (rs6000_expand_split_stack_prologue): Call
21224 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
21226 2017-02-21 Martin Jambor <mjambor@suse.cz>
21229 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
21230 have been analyzed.
21232 2017-02-21 Martin Jambor <mjambor@suse.cz>
21234 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
21235 for backward compatibility only.
21236 * doc/invoke.texi (Option Summary): Remove all references to
21237 -fipa-cp-alignment.
21239 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
21243 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21245 * lra-constraints.c (curr_insn_transform): Handle
21246 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21248 2017-02-21 Martin Liska <mliska@suse.cz>
21250 * config/i386/i386.opt: Replace -masm-dialect with -masm.
21252 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
21254 PR translation/79638
21255 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
21257 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
21260 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
21261 (arm_function_ok_for_sibcall): Return false for an indirect call by
21262 descriptor if all the argument registers are used.
21263 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
21264 alignment of the function.
21266 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21268 PR tree-optimization/61441
21269 * simplify-rtx.c (simplify_const_unary_operation): For
21270 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
21271 the sNaN unmodified.
21273 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21275 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21276 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21277 instead of SYSTEM_HEADER_DIR.
21279 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
21280 Martin Liška <mliska@suse.cz>
21282 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
21283 Fix typos and grammar, use active voice, and clarify.
21285 2017-02-20 Marek Polacek <polacek@redhat.com>
21287 PR middle-end/79537
21288 * gimplify.c (gimplify_expr): Handle unused *&&L;.
21291 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
21293 2017-02-20 Jakub Jelinek <jakub@redhat.com>
21296 * config/i386/i386.c (ix86_expand_builtin): Handle
21297 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
21298 ix86_builtins_isa[fcode].isa as a requirement of those
21299 flags and any other flag in the bitmask.
21300 (ix86_init_mmx_sse_builtins): Use 0 instead of
21301 ~OPTION_MASK_ISA_64BIT as mask.
21302 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
21303 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
21304 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
21305 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
21307 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21310 * lra-constraints.c (split_reg): Check requested split mode
21311 is supported by the register.
21313 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21315 * lra-constraints.c (simplify_operand_subreg): Remove early
21318 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21321 * lra-constraints.c (curr_insn_transform): Tighten condition
21322 for converting SUBREG reloads from OP_OUT to OP_INOUT.
21324 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21327 * lra-constraints.c (curr_insn_transform): Handle
21328 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21330 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
21333 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
21335 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
21337 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
21340 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
21343 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21345 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
21348 2017-02-18 Jakub Jelinek <jakub@redhat.com>
21350 * final.c (last_columnnum, override_columnnum): New variables.
21351 (final_start_function): Set last_columnnum, pass it to begin_prologue
21352 hook and pass 0 to dwarf2out_begin_prologue.
21353 (final_scan_insn): Update override_columnnum. Pass last_columnnum
21354 to source_line debug hook.
21355 (notice_source_line): Compute last_columnnum and for debug_column_info
21356 return true on column changes.
21357 * debug.h (struct gcc_debug_hooks): Add column argument to
21358 source_line and begin_prologue hooks.
21359 (debug_nothing_int_charstar_int_bool): Remove prototype.
21360 (debug_nothing_int_int_charstar,
21361 debug_nothing_int_int_charstar_int_bool): New prototypes.
21362 (dwarf2out_begin_prologue): Add column argument.
21363 * debug.c (do_nothing_debug_hooks): Adjust source_line and
21364 begin_prologue hooks.
21365 (debug_nothing_int_charstar_int_bool): Remove.
21366 (debug_nothing_int_int_charstar,
21367 debug_nothing_int_int_charstar_int_bool): New functions.
21368 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
21369 through to dwarf2out_source_line.
21370 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
21371 (dwarf2out_source_line): Add column argument, emit it if requested.
21372 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
21374 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21375 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21376 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
21377 through to dwarf2out_begin_prologue.
21378 (vmsdbgout_source_line): Add column argument, pass it through to
21379 dwarf2out_source_line.
21380 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
21381 dbxout_source_line caller.
21382 (dbxout_source_line): Add column argument.
21384 * common.opt (gno-column-info, gcolumn-info): New options.
21385 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
21386 (check_die): Also test for multiple DW_AT_decl_column attributes.
21387 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
21388 DW_AT_decl_column if requested.
21389 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
21391 (gen_variable_die): Likewise.
21392 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
21393 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
21396 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
21397 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
21398 (ix86_handle_option): Handle OPT_m3dnowa.
21399 * doc/invoke.texi (-m3dnowa): Document.
21400 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
21401 -m3dnowa instead of -m3dnow -march=athlon.
21404 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
21405 instead of gcc_assert for K, r and R code checks. Formatting fixes.
21407 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21410 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
21411 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
21412 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
21413 generator for vsx_xxpermdi_<mode>_be.
21414 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
21415 force big-endian semantics.
21416 (vsx_xxpermdi_<mode>_be): New define_expand with same
21417 implementation as previous version of vsx_xxpermdi_<mode>.
21419 2017-02-17 Jakub Jelinek <jakub@redhat.com>
21421 PR tree-optimization/79327
21422 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
21423 variable, its initialization and use.
21425 2017-02-17 Julia Koval <julia.koval@intel.com>
21427 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
21428 (OPTION_MASK_ISA_PKU_UNSET): New.
21429 (ix86_handle_option): Handle -mrdpid.
21430 * config/i386/cpuid.h (bit_RDPID): New.
21431 * config/i386/driver-i386.c (host_detect_local_cpu):
21432 Detect RDPID feature.
21433 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
21434 * config/i386/i386-c.c (ix86_target_macros_internal):
21436 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
21437 (ix86_valid_target_attribute_inner_p): Add "rdpid".
21438 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
21439 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
21440 * config/i386/i386.md (define_insn "rdpid"): New.
21441 * config/i386/i386.opt Add -mrdpid.
21442 * config/i386/immintrin.h (_rdpid_u32): New.
21444 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
21446 PR rtl-optimization/79541
21447 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
21448 instead of transforming it into USE.
21450 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
21452 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
21453 If HONOR_SNANS (SFmode) force the input to a register.
21454 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
21455 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
21456 an frsp or similar insn.
21458 2017-02-17 Martin Liska <mliska@suse.cz>
21460 PR rtl-optimization/79577
21461 * params.def (selsched-max-sched-times): Increase minimum to 1.
21463 2017-02-17 Martin Liska <mliska@suse.cz>
21465 PR rtl-optimization/79574
21466 * gcse.c (want_to_gcse_p): Prevent integer overflow.
21468 2017-02-17 Martin Liska <mliska@suse.cz>
21470 PR tree-optimization/79529
21471 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
21472 ssa_defined_default_def_p to handle cases which are implicitly
21474 * tree-ssa.c (ssa_defined_default_def_p): New function.
21475 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
21476 which are implicitly defined.
21477 * tree-ssa.h (ssa_defined_default_def_p): Declare.
21479 2017-02-17 Richard Biener <rguenther@suse.de>
21481 PR middle-end/79576
21482 * params.def (max-ssa-name-query-depth): Limit to 10.
21484 2017-02-17 Richard Biener <rguenther@suse.de>
21486 PR tree-optimization/79552
21487 * tree-ssa-structalias.c (visit_loadstore): Properly verify
21490 2017-02-17 Richard Biener <rguenther@suse.de>
21493 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
21495 2017-02-17 Marek Polacek <polacek@redhat.com>
21497 PR middle-end/79536
21498 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
21499 (fold_negate_expr): New wrapper.
21501 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
21503 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
21504 Correct terminology and de-emphasize pre-standard behavior.
21506 2017-02-16 Alan Modra <amodra@gmail.com>
21508 PR rtl-optimization/79286
21509 * ira.c (def_dominates_uses): New function.
21510 (update_equiv_regs): Don't create an equivalence for insns that
21511 may trap where the register def does not dominate the use.
21513 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
21515 PR rtl-optimization/78127
21516 * lra.c (lra): Call lra_eliminate before finish the loop after
21519 2017-02-16 Richard Biener <rguenther@suse.de>
21521 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
21523 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
21524 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
21525 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
21526 (isl_val_int_from_wi): New function.
21527 (extract_affine_gmp): Rename to ...
21528 (extract_affine_wi): ... this, take a widest_int.
21529 (extract_affine_int): Just wrap extract_affine_wi.
21530 (add_param_constraints): Use isl_val_int_from_wi.
21531 (add_loop_constraints): Likewise, and extract_affine_wi.
21533 2017-02-15 Jeff Law <law@redhat.com>
21535 PR middle-end/79521
21536 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
21537 ira_init_register_move_cost_if_necessary.
21539 2017-02-15 Martin Sebor <msebor@redhat.com>
21541 PR middle-end/32003
21542 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
21543 removed in a prior commit.
21545 2017-02-15 Bin Cheng <bin.cheng@arm.com>
21547 PR tree-optimization/79347
21548 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
21549 counters during peeling.
21551 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
21553 * Makefile.in (site.exp): Remove "set ISLVER".
21555 2017-02-15 Jakub Jelinek <jakub@redhat.com>
21558 * real.c (real_from_integer): Call real_convert even for decimal.
21560 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
21563 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
21565 2017-02-14 Andrew Pinski <apinski@cavium.com>
21567 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
21568 cores and change the partno/implementer to be correct.
21569 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
21570 the 'B" as the implementer.
21571 * config/aarch64/aarch64-tune.md: Regenerate.
21573 2017-02-14 Carl Love <cel@us.ibm.com>
21575 * config/rs6000/rs6000.c: Add case statement entry to make the
21576 xvcvuxdsp built-in argument unsigned.
21577 * config/rs6000/vsx.md: Fix the source and return operand types so they
21578 match the instruction definitions from the ISA document. Fix typo
21579 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
21582 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
21585 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
21586 member early_clobber_alts.
21587 * lra-lives.c (reg_early_clobber_p): New.
21588 (process_bb_lives): Use it.
21589 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
21590 (debug_operand_data): Initialize early_clobber_alts.
21591 (setup_operand_alternative): Set up early_clobber_alts.
21592 (collect_non_operand_hard_regs): Ditto. Pass early clobber
21593 alternatives to new_insn_reg.
21594 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
21596 (lra_update_insn_regno_info): Pass the new arg.
21598 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21600 PR middle-end/79505
21601 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
21602 (new_oacc_loop_raw): Don't clear already cleared fields.
21605 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
21606 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
21607 _mm512_prefetch_i64gather_ps): New inline functions and macros.
21609 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
21612 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
21614 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
21617 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
21618 the extra instruction to the right place to store 128-bit constant
21621 2017-02-14 Martin Sebor <msebor@redhat.com>
21623 PR middle-end/79448
21624 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
21625 warning for strings of unknown length.
21627 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
21629 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
21631 2017-02-14 Jeff Law <law@redhat.com>
21634 * ira-costs.c (scan_one_insn): Initialize register move costs
21635 for pseudos seen in USE/CLOBBER insns.
21637 PR tree-optimization/79095
21638 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
21639 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
21640 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
21641 if the operands are known to be not equal, then the resulting range
21643 (intersect_ranges): If the new range is ~[0,0] and the old range is
21644 wide, then prefer ~[0,0].
21645 * tree-vrp.c (overflow_comparison_p_1): New function.
21646 (overflow_comparison_p): New function.
21647 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
21648 if NAME is used in an overflow test.
21649 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
21650 overflow check that can be expressed as an equality test, then adjust
21651 ops to be that equality test.
21653 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21655 * config/s390/s390-builtin-types.def: Remove flags argument.
21656 * config/s390/s390.c (s390_init_builtins): Likewise.
21658 2017-02-14 Martin Liska <mliska@suse.cz>
21660 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
21661 vector. Fix trailing white spaces.
21663 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
21665 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
21668 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21670 PR rtl-optimization/68664
21671 * config/arm/arm.c (arm_sched_can_speculate_insn):
21672 New function. Declare prototype.
21673 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21675 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21677 PR rtl-optimization/68664
21678 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
21680 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21682 2017-02-14 Amit Pawar <amit.pawar@amd.com>
21684 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
21685 max skip bytes for function, loop and jump.
21687 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21689 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
21690 ABS_EXPR for gimple dump.
21692 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21695 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
21697 PR tree-optimization/79408
21698 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
21699 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
21700 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
21701 also if rhs1 is INTEGER_CST.
21703 2017-02-14 Richard Biener <rguenther@suse.de>
21705 PR middle-end/79432
21706 * tree-into-ssa.c (insert_phi_nodes): When the function can
21707 have abnormal edges rewrite SSA names with broken use-def
21708 dominance out of SSA and register them for PHI insertion.
21710 2017-02-13 Martin Sebor <msebor@redhat.com>
21712 PR middle-end/79496
21713 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
21714 clearing info.nowrite flag when snprintf size argument is a range.
21716 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21718 * cprop.c (cprop_jump): Add missing space in string literal.
21719 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
21720 (get_constraint_for_component_ref): Likewise.
21721 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
21722 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
21723 * lra-constraints.c (process_alt_operands): Likewise.
21724 * ipa-inline.c (inline_small_functions): Likewise.
21725 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
21726 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21727 * trans-mem.c (diagnose_tm_1_op): Likewise.
21728 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
21729 (grid_parallel_clauses_gridifiable): Likewise.
21731 * config/nvptx/mkoffload.c (process): Add space in between
21734 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
21735 "MOD4_SSE_REGS" and "ALL_REGS".
21737 * spellcheck.c (test_data): Add , in between "foo" and "food".
21739 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21742 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
21743 boundary crossing check and subsequent code generation agree.
21745 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21747 * config/aarch64/aarch64.c (has_memory_op): Delete.
21748 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
21751 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21753 PR rtl-optimization/79388
21754 PR rtl-optimization/79450
21755 * combine.c (distribute_notes): When removing TEM_INSN for which
21756 corresponding dest has last value recorded, invalidate that last
21759 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21761 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
21762 of explicit '@'. Add missing assembly comment marker on branch costs
21765 2017-02-13 Nathan Sidwell <nathan@acm.org>
21767 * gengtype-lex.l (<in_struct>): Add '/'.
21769 2017-02-13 Martin Liska <mliska@suse.cz>
21772 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
21774 2017-02-13 Richard Biener <rguenther@suse.de>
21776 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
21778 * configure: Re-generate.
21779 * config.in: Likewise.
21780 * graphite-dependences.c: Simplify as if
21781 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
21782 * graphite-isl-ast-to-gimple.c: Likewise.
21783 * graphite-optimize-isl.c: Likewise.
21784 * graphite-poly.c: Likewise.
21785 * graphite-sese-to-poly.c: Likewise.
21786 * graphite.h: Likewise.
21787 * toplev.c: Include isl/version.h and use isl_version () for
21788 printing the ISL version.
21789 * doc/install.texi: Update ISL requirement.
21791 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21793 * doc/standards.texi (Standards): Update reference to
21796 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21798 * doc/extend.texi (Named Address Spaces): sourceware.org now
21800 * doc/install.texi (Binaries): Ditto.
21803 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21805 * doc/cpp.texi: Replace "stringify"/"stringification" with C
21806 standard terminology "stringize"/"stringizing" throughout.
21807 * doc/cppinternals.texi: Likewise.
21809 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21811 * doc/extend.texi: Fix some spelling mistakes and typos.
21812 * doc/invoke.texi: Likewise.
21814 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21817 * params.def (inline-min-speedup) Change from 10 to 8.
21819 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21821 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
21824 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21827 * ipa-inline-analysis.c (get_minimal_bb): New function.
21828 (record_modified): Use it.
21829 (remap_edge_change_prob): Handle also ancestor functions.
21831 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
21833 * doc/contrib.texi (Contributors): Remove broken link into
21834 the Mauve CVS repository.
21836 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21838 PR middle-end/79454
21839 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
21840 result computation whenever lhs doesn't have vector mode, not
21841 just when it has BLKmode.
21843 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21845 * doc/makefile.texi (profiledbootstrap): Refer to the
21846 installation instructions only in textual form.
21848 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21851 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
21853 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21855 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
21856 (Specific): Update mingw-w64 reference.
21858 (Specific): Remove broken link to Renesas RX processor.
21860 2017-02-10 Richard Biener <rguenther@suse.de>
21862 * toplev.c (process_options): Do not mention obsolete graphite
21863 options when printing sorry message about missing graphite support.
21864 Mention -floop-nest-optimize.
21866 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
21868 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
21869 (vtst_p16): Likewise.
21870 (vtstq_p8): Likewise.
21871 (vtstq_p16): Likewise.
21873 (vtstq_p64): Likewise.
21874 * config/arm/arm_neon.h (vgetq_lane_p64): New.
21875 (vset_lane_p64): New.
21876 (vsetq_lane_p64): New.
21878 2017-02-10 Jakub Jelinek <jakub@redhat.com>
21880 PR tree-optimization/79411
21881 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
21882 stmt operands are SSA_NAMEs used in abnormal phis.
21883 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
21886 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21889 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
21892 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21894 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
21896 2017-02-09 Jakub Jelinek <jakub@redhat.com>
21898 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
21902 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
21903 not arbitrary TREE_CONSTANT.
21906 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
21907 "omp declare target link" attribute unless is_global_var.
21908 * omp-offload.c (find_link_var_op): Likewise.
21910 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
21911 Chung-Lin Tang <cltang@codesourcery.com>
21913 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
21915 (gimplify_adjust_omp_clauses): Don't delete TILE.
21916 (gimplify_omp_for): Deal with TILE.
21917 * internal-fn.c (expand_GOACC_TILE): New function.
21918 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
21920 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
21921 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
21923 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
21924 avoid DIV for outermost collapse var.
21925 (expand_oacc_for): Insert tile element loop as needed. Adjust.
21926 Remove out of date comments, fix whitespace.
21927 * omp-general.c (omp_extract_for_data): Deal with tiling.
21928 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
21929 adjust OLF_DIM_BASE value.
21930 (struct omp_for_data): Add tiling field.
21931 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
21932 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
21933 for auto loops. Remove default auto determining, moved to
21934 oacc_loop_fixed_partitions.
21935 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
21936 stmts, add e_mask field.
21937 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
21938 (oacc_thread_numbers): Use oacc_dim_call.
21939 (oacc_xform_tile): New.
21940 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
21941 (finish_oacc_loop): Adjust for ifns vector.
21942 (oacc_loop_discover_walk): Append loop abstraction sites to list,
21943 add case for GOACC_TILE fns.
21944 (oacc_loop_xform_loop): Delete.
21945 (oacc_loop_process): Iterate over call list directly, and add
21946 handling for GOACC_TILE fns.
21947 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
21949 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
21950 vector partitioning to outer loops. Assign 2 partitions to loops
21951 when available. Add TILE handling.
21952 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
21953 (execite_oacc_device_lower): Process GOACC_TILE fns,
21954 ignore unknown specs.
21955 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
21956 * tree.c (omp_clause_num_ops): Adjust TILE ops.
21957 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
21959 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
21961 * configure.ac (ACX_BUGURL): Update.
21962 * configure: Regenerate.
21964 2017-02-09 Richard Biener <rguenther@suse.de>
21966 PR tree-optimization/69823
21967 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
21968 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
21970 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
21972 * config/arc/arc-c.def: Add __NPS400__ definition.
21973 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
21974 (TARGET_NPS400): Define.
21976 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
21978 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
21980 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
21981 pointer, arch_info.
21982 (arc_cpu_types): Fill the arch_info field with a pointer into the
21983 arc_arch_types table.
21984 (arc_selected_cpu): Declare.
21985 * config/arc/arc.c (arc_selected_cpu): Make global.
21986 (arc_selected_arch): Delete.
21987 (arc_base_cpu): Delete.
21988 (arc_override_options): Remove references to deleted variables,
21989 update access to arch information.
21990 (ARC_OPT): Update access to arch information.
21991 (ARC_OPTX): Likewise.
21992 * config/arc/arc.h (arc_base_cpu): Remove declaration.
21993 (TARGET_ARC600): Update access to arch information.
21994 (TARGET_ARC601): Likewise.
21995 (TARGET_ARC700): Likewise.
21996 (TARGET_EM): Likewise.
21997 (TARGET_HS): Likewise.
21998 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
22001 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
22004 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
22005 condition/operands for integer GE/LE/GEU/LEU operations.
22007 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
22009 PR translation/79397
22010 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
22013 2017-02-08 Martin Jambor <mjambor@suse.cz>
22016 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
22017 whether allocation happened.
22018 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
22019 nothing was allocated.
22021 2017-02-08 Jakub Jelinek <jakub@redhat.com>
22023 PR tree-optimization/79408
22024 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
22025 constant, but SSA_NAME with a known integer range, use the minimum
22026 of that range instead of op1 to determine if modulo can be replaced
22027 with its first operand.
22029 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22031 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
22033 2017-02-08 Richard Biener <rguenther@suse.de>
22035 PR tree-optimization/71824
22036 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22037 Check all loops contained in the merged region.
22039 2017-02-07 Andrew Pinski <apinski@cavium.com>
22041 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
22043 2017-02-07 Andrew Pinski <apinski@cavium.com>
22045 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
22046 (thunderxt88): Likewise.
22047 (thunderxt81): Disable LSE and change v8.1 to v8.
22048 (thunderxt83): Likewise.
22050 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22051 Richard Biener <rguenther@suse.de>
22053 PR middle-end/79399
22054 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
22055 type from int to size_t.
22056 * ira-costs.c (struct_costs_size): Change type from int to size_t.
22058 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22060 PR rtl-optimization/79386
22061 * cprop.c (bypass_conditional_jumps): Initialize
22062 bypass_last_basic_block already before splitting bbs after
22063 unconditional traps...
22064 (bypass_conditional_jumps): ... rather than here.
22067 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
22068 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
22069 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
22070 fix -masm=intel patterns.
22072 2017-02-07 Richard Biener <rguenther@suse.de>
22074 PR tree-optimization/79256
22075 PR middle-end/79278
22076 * builtins.c (get_object_alignment_2): Use min_align_of_type
22077 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
22078 and ADJUST_FIELD_ALIGN.
22080 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
22082 * doc/tm.texi: Regenerate.
22083 * stor-layout.c (layout_decl): Adjust.
22084 (update_alignment_for_field): Likewise.
22085 (place_field): Likewise.
22086 (min_align_of_type): Likewise.
22087 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
22088 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
22089 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
22090 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
22091 * config/frv/frv.c (frv_adjust_field_align): Likewise.
22092 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
22093 * config/i386/i386.c (x86_field_alignment): Likewise.
22094 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
22095 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
22096 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
22097 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
22098 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
22099 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
22103 2017-01-30 Richard Biener <rguenther@suse.de>
22105 PR tree-optimization/79256
22106 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22107 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22110 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
22112 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
22113 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
22114 builtins to SImode and emit a zero-extend, if necessary.
22116 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22118 * docs/invoke.texi (RISC-V Options): Alphabetize.
22120 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22122 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
22125 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22127 * config/riscv/riscv.c: New file.
22128 * common/config/riscv/riscv-common.c: Likewise.
22129 * config.gcc: Likewise.
22130 * config/riscv/constraints.md: Likewise.
22131 * config/riscv/elf.h: Likewise.
22132 * config/riscv/generic.md: Likewise.
22133 * config/riscv/linux.h: Likewise.
22134 * config/riscv/multilib-generator: Likewise.
22135 * config/riscv/peephole.md: Likewise.
22136 * config/riscv/pic.md: Likewise.
22137 * config/riscv/predicates.md: Likewise.
22138 * config/riscv/riscv-builtins.c: Likewise.
22139 * config/riscv/riscv-c.c: Likewise.
22140 * config/riscv/riscv-ftypes.def: Likewise.
22141 * config/riscv/riscv-modes.def: Likewise.
22142 * config/riscv/riscv-opts.h: Likewise.
22143 * config/riscv/riscv-protos.h: Likewise.
22144 * config/riscv/riscv.h: Likewise.
22145 * config/riscv/riscv.md: Likewise.
22146 * config/riscv/riscv.opt: Likewise.
22147 * config/riscv/sync.md: Likewise.
22148 * config/riscv/t-elf-multilib: Likewise.
22149 * config/riscv/t-linux: Likewise.
22150 * config/riscv/t-linux-multilib: Likewise.
22151 * config/riscv/t-riscv: Likewise.
22152 * configure.ac: Likewise.
22153 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
22154 Waterman as RISC-V maintainers.
22155 * doc/install.texi: Add RISC-V entries.
22156 * doc/invoke.texi: Add RISC-V options section.
22157 * doc/md.texi: Add RISC-V constraints section.
22158 * configure: Regenerated.
22160 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
22163 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
22164 false values to be constant vectors with all 0 or all 1 bits set.
22165 (vcondu<mode><mode>): Likewise.
22166 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
22168 (fpmask_comparison_operator): Update comment.
22169 (vecint_comparison_operator): New predicate.
22170 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
22171 vector conditionals when the true and false values are constant
22172 vectors with all 0 bits or all 1 bits set.
22174 2017-02-06 Martin Sebor <msebor@redhat.com>
22176 PR tree-optimization/79376
22177 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
22179 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
22181 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
22182 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
22183 to simplify split condition.
22185 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22187 * omp-expand.c (oxpand_omp_atomic_fetch_op,
22188 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
22191 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
22193 PR rtl-optimization/68664
22194 * target.def (can_speculate_insn): New hook.
22195 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
22196 * doc/tm.texi: Regenerate.
22197 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
22198 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
22199 (rs6000_sched_can_speculate_insn): New function.
22201 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22203 PR tree-optimization/79284
22204 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
22205 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
22206 vectorizable_mask_load_store, vectorizable_operation,
22207 vect_is_simple_cond, get_same_sized_vectype): Use it instead
22208 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
22209 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
22210 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
22211 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
22212 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
22213 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
22214 is_gimple_assign (stmt). Replace another such test with
22215 is_gimple_assign (stmt).
22217 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22220 * config/avr/avr.c (rtl-iter.h): Include it.
22221 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
22222 (avr_legitimate_combined_insn): ...and implementation.
22224 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
22226 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
22227 * config/s390/s390.c (s390_const_operand_ok)
22228 (s390_canonicalize_comparison, s390_extract_part)
22229 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
22230 (s390_contiguous_bitmask_p, s390_rtx_costs)
22231 (legitimize_pic_address): Likewise.
22232 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
22233 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
22234 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
22235 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
22236 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
22238 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22240 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
22241 REGNO($0) == REGNO($1).
22243 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22245 * config/s390/linux.h(SIZE_TYPE): Add comment.
22247 2017-02-06 Julian Brown <julian@codesourcery.com>
22248 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22249 Virendra Pathak <virendra.pathak@broadcom.com>
22251 * config/aarch64/aarch64-cores.def: Change the scheduler
22253 * config/aarch64/aarch64.md: Include thunderx2t99.md.
22254 * config/aarch64/thunderx2t99.md: New file.
22256 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22258 * doc/standards.texi (Go Language): Update link to language
22261 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22263 * tree-eh.c (lower_resx): Sanitize profile.
22264 (cleanup_empty_eh_move_lp): Likewise.
22266 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22269 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
22271 * cfgloopmanip.h (loop_version): Update prototype.
22272 * modulo-sched.c (sms_schedule): Update call of loop_version.
22273 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
22274 * tree-parloops.c (gen_parallel_loop): Likewise.
22275 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
22276 * tree-ssa-loop-split.c (split_loop): Likewise.
22277 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
22278 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22280 2017-02-05 Martin Liska <mliska@suse.cz>
22283 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
22285 (print_operand_address): Initialize a struct to zero.
22287 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22289 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
22290 garbage collector only in textual form.
22292 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22294 * doc/extend.texi (x86 specific memory model extensions for
22295 transactional memory): Simplify a phrase.
22297 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
22300 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
22301 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
22302 (atomic_storedi_1): Likewise.
22304 2017-02-04 Jakub Jelinek <jakub@redhat.com>
22306 PR tree-optimization/79338
22307 * tree-parloops.c (gather_scalar_reductions): Don't call
22308 vect_analyze_loop_form for loop->inner before destroying loop's
22311 2017-02-03 Martin Sebor <msebor@redhat.com>
22313 PR tree-optimization/79327
22314 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
22315 when precision has resulted in leading zeros.
22316 (format_integer): Adjust the likely counter to assume an unknown
22317 argument that may be zero is non-zero.
22319 2017-02-03 Jason Merrill <jason@redhat.com>
22322 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
22323 avoid copying non-taken branch.
22325 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22327 PR tree-optimization/79340
22328 * tree-vect-loop.c (vectorizable_reduction): Release
22329 vec_defs elements after safe_splicing them into other vectors.
22332 PR tree-optimization/79327
22333 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
22334 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
22336 (format_integer): Use wide_int_to_tree instead of build_int_cst
22337 + to_?hwi. If argmin is NULL, just set argmin and argmax to
22338 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
22339 of shortest and longest sequence.
22341 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
22343 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
22344 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
22346 2017-02-03 Walter Lee <walt@tilera.com>
22349 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
22350 after initial stackframe link reg save.
22351 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
22353 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22356 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
22357 wu for stxssp alternative.
22359 2017-02-03 Martin Sebor <msebor@redhat.com>
22361 PR tree-optimization/79352
22362 * gimple-fold.c (get_range_strlen): Add argument.
22363 (get_range_strlen): Change return type to bool.
22364 (get_maxval_strlen): Pass in a dummy argument.
22365 * gimple-fold.h (get_range_strlen): Change return type to bool.
22366 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
22367 * tree.h (array_at_struct_end_p): Add argument.
22368 * tree.c (array_at_struct_end_p): Handle it.
22370 2017-02-03 Martin Liska <mliska@suse.cz>
22373 * multiple_target.c (create_dispatcher_calls): Redirect edge
22374 from a caller of a dispatcher.
22375 (expand_target_clones): Make the clones local.
22376 (ipa_target_clone): Do both target clones and resolvers.
22377 (ipa_dispatcher_calls): Remove the pass.
22378 (pass_dispatcher_calls::gate): Likewise.
22379 (make_pass_dispatcher_calls): Likewise.
22380 * passes.def (pass_target_clone): Put as very first IPA early
22383 2017-02-03 Martin Liska <mliska@suse.cz>
22385 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
22386 in case of a function with ifunc attribute.
22388 2017-02-03 Martin Liska <mliska@suse.cz>
22390 * cgraph.c (cgraph_node::dump): Dump function version info.
22391 * symtab.c (symtab_node::dump_base): Add missing new line.
22393 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22395 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
22396 (ifcombine_ifandif): Use it.
22398 2017-02-03 Martin Liska <mliska@suse.cz>
22400 * doc/invoke.texi: Document default value for
22401 use-after-scope-direct-emission-threshold.
22403 2017-02-03 Martin Liska <mliska@suse.cz>
22405 PR tree-optimization/79339
22406 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
22407 (format_floating): Likewise.
22409 2017-02-03 Martin Liska <mliska@suse.cz>
22412 * ipa-prop.c (ipa_node_params_t::insert): Remove current
22414 (ipa_node_params_t::remove): Likewise.
22415 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
22416 initialization from removed ipa_node_params_t::insert.
22417 (ipa_node_params::~ipa_node_params): Move from removed
22418 ipa_node_params_t::release.
22419 * symbol-summary.h (symbol_summary::m_released): New member.
22420 Do not release a summary twice. Do not allow to call finalizer
22421 for types of a summary that live in GGC memory.
22423 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22425 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
22428 2017-02-02 Martin Sebor <msebor@redhat.com>
22430 PR middle-end/79275
22431 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
22432 (format_string): Tighten up the range of output for non-constant
22433 strings and correct the expected range for wide non-constant strings.
22435 2017-02-02 Martin Sebor <msebor@redhat.com>
22437 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
22439 PR middle-end/32003
22440 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
22442 (-fdump-tree-@var): Add to index and document how to come up
22443 with pass-specific option and dump file names.
22444 (-fdump-passes): Clarify where to look for output.
22446 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22448 PR middle-end/77445
22449 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
22450 statistics of the analyzed path; allow threading for speed when
22451 any of BBs along the path are optimized for speed.
22453 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
22455 PR middle-end/78468
22456 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
22457 settings of the virtual registers.
22460 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
22462 * explow.c (get_dynamic_stack_size): Take known alignment of stack
22463 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
22466 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22468 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
22469 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
22471 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22473 * config/s390/s390.md: Add missing comments with the expanded
22475 * config/s390/vector.md: Likewise.
22476 * config/s390/vx-builtins.md: Likewise.
22478 2017-02-02 Jakub Jelinek <jakub@redhat.com>
22481 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
22482 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
22483 conditions on a single line.
22485 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22487 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22488 __S390_VX__ to __VX__.
22490 2017-02-01 Andrew Pinski <apinski@cavium.com>
22492 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
22493 stmt_info to record_stmt_cost.
22494 (vect_get_known_peeling_cost): Pass stmt_info if known to
22496 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
22497 cpu_vector_cost field into
22498 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
22499 field into vec_int_stmt_cost and vec_fp_stmt_cost.
22500 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
22501 splitting of scalar_stmt_cost and vec_stmt_cost.
22502 (thunderx_vector_cost): Likewise.
22503 (cortexa57_vector_cost): LIkewise.
22504 (exynosm1_vector_cost): Likewise.
22505 (xgene1_vector_cost): Likewise.
22506 (thunderx2t99_vector_cost): Improve after the splitting of the two
22508 (aarch64_builtin_vectorization_cost): Update for the splitting of
22509 scalar_stmt_cost and vec_stmt_cost.
22511 2017-02-01 Torvald Riegel <triegel@redhat.com>
22512 Richard Henderson <rth@redhat.com>
22514 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
22515 conditional on existance of a fast atomic load.
22516 * optabs-query.c (can_atomic_load_p): New function.
22517 * optabs-query.h (can_atomic_load_p): Declare it.
22518 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
22519 no fast atomic load is available for the particular size of access.
22520 (expand_atomic_compare_and_swap): Likewise.
22521 (expand_atomic_load): Likewise.
22522 (expand_atomic_store): Likewise.
22523 (expand_atomic_fetch_op): Likewise.
22524 * testsuite/lib/target-supports.exp
22525 (check_effective_target_sync_int_128): Remove x86 because it provides
22526 no fast atomic load.
22527 (check_effective_target_sync_int_128_runtime): Likewise.
22529 2017-02-01 Richard Biener <rguenther@suse.de>
22531 * graphite.c: Include tree-vectorizer.h for find_loop_location.
22532 (graphite_transform_loops): Provide opt-info for optimized nests.
22533 * tree-parloop.c (parallelize_loops): Provide opt-info for
22534 parallelized loops.
22536 2017-02-01 Richard Biener <rguenther@suse.de>
22538 PR middle-end/79315
22539 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
22540 was not set before.
22542 2017-02-01 Richard Biener <rguenther@suse.de>
22544 PR tree-optimization/71824
22545 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22546 Verify the loops are valid in the merged SESE region.
22547 (scop_detection::can_represent_loop_1): Check analyzing the
22548 evolution of the number of iterations in the region succeeds.
22550 2017-01-31 Ian Lance Taylor <iant@golang.org>
22552 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
22553 REG_ARGS_SIZE note to 32-bit push insns and call insn.
22555 2017-01-31 David Malcolm <dmalcolm@redhat.com>
22557 PR preprocessor/79210
22558 * input.c (get_substring_ranges_for_loc): Replace line_width
22559 assertion with error-handling.
22561 2017-01-31 Richard Biener <rguenther@suse.de>
22563 PR tree-optimization/77318
22564 * graphite-sese-to-poly.c (extract_affine): Fix assert.
22565 (create_pw_aff_from_tree): Take loop parameter.
22566 (add_condition_to_pbb): Pass loop of the condition to
22567 create_pw_aff_from_tree.
22569 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22571 * config/s390/s390.c (s390_asan_shadow_offset): New function.
22572 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
22574 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
22578 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
22580 (convert_int_to_float128): Likewise.
22581 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
22582 (convert_int_to_float128): Likewise.
22583 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
22584 (UNSPEC_IEEE128_CONVERT): Likewise.
22585 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
22586 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
22587 Use local variables for IBM extended format.
22588 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
22589 (fix_trunc<mode>si2_fprs): Likewise.
22590 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
22591 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
22592 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
22593 to know that we can now have integers of all sizes in vector
22595 (fix<uns>_<mode>di2_hw): Likewise.
22596 (float<uns>_<mode>si2_hw): Likewise.
22597 (fix_<mode>si2_hw): Likewise.
22598 (fixuns_<mode>si2_hw): Likewise.
22599 (float<uns>_<mode>di2_hw): Likewise.
22600 (float_<mode>di2_hw): Likewise.
22601 (float_<mode>si2_hw): Likewise.
22602 (floatuns_<mode>di2_hw): Likewise.
22603 (floatuns_<mode>si2_hw): Likewise.
22604 (xscvqp<su>wz_<mode>): Delete, no longer used.
22605 (xscvqp<su>dz_<mode>): Likewise.
22606 (xscv<su>dqp_<mode>): Likewise.
22607 (ieee128_mfvsrd_64bit): Likewise.
22608 (ieee128_mfvsrd_32bit): Likewise.
22609 (ieee128_mfvsrwz): Likewise.
22610 (ieee128_mtvsrw): Likewise.
22611 (ieee128_mtvsrd_64bit): Likewise.
22612 (ieee128_mtvsrd_32bit): Likewise.
22614 2017-01-31 Martin Liska <mliska@suse.cz>
22617 * ipa-prop.c (ipa_free_all_node_params): Call release method
22618 instead of ~sumbol_summary to not to trigger double times
22621 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
22623 PR tree-optimization/71691
22624 * bitmap.h (class auto_bitmap): New.
22625 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
22626 is_maybe_undefined instead of ssa_undefined_value_p.
22628 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22630 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22631 __S390_ARCH_LEVEL__ to __ARCH__.
22633 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22635 PR tree-optimization/79267
22636 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
22637 if should_remove_lhs_p is true.
22639 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
22642 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
22643 (add_alignment_attribute): New.
22644 (base_type_die): Add alignment attribute.
22645 (subrange_type_die): Likewise.
22646 (modified_type_die): Likewise.
22647 (gen_array_type_die): Likewise.
22648 (gen_descr_array_type_die: Likewise.
22649 (gen_enumeration_type_die): Likewise.
22650 (gen_subprogram_die): Likewise.
22651 (gen_variable_die): Likewise.
22652 (gen_field_die): Likewise.
22653 (gen_ptr_to_mbr_type_die): Likewise.
22654 (gen_struct_or_union_type_die): Likewise.
22655 (gen_subroutine_type_die): Likewise.
22656 (gen_typedef_die): Likewise.
22657 (base_type_cmp): Compare alignment attribute.
22659 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22662 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
22663 (setb_unsigned) New pattern for setb with CCUNS.
22664 * config/rs6000/rs6000.c (expand_block_compare): Use a different
22665 subfc./subfe sequence to avoid overflow problems. Generate a
22666 shorter sequence with cmpld/setb for power9.
22667 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
22668 for generating subfc. instruction.
22669 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
22670 now uses this instruction.
22672 2017-01-30 Ian Lance Taylor <iant@google.com>
22675 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
22676 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
22678 2017-01-30 Martin Sebor <msebor@redhat.com>
22680 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
22681 Move constant to the right of a relational operator.
22682 (get_mpfr_format_length, format_character, format_string): Ditto.
22683 (should_warn_p, maybe_warn): Same.
22685 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
22687 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
22690 * asan.c (get_translation_unit_decl): Remove function.
22691 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
22693 2017-01-30 Martin Liska <mliska@suse.cz>
22695 PR gcov-profile/79259
22696 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
22697 -fprofile-generate.
22699 2017-01-30 Martin Liska <mliska@suse.cz>
22702 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
22703 Initialize variables with NULL value.
22705 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
22708 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
22710 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
22712 2017-01-30 Richard Biener <rguenther@suse.de>
22714 PR tree-optimization/79276
22715 * tree-vrp.c (process_assert_insertions): Properly adjust common
22716 when removing a duplicate.
22718 * gcc.dg/torture/pr79276.c: New testcase.
22720 2017-01-30 Richard Biener <rguenther@suse.de>
22722 PR tree-optimization/79256
22723 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22724 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22726 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
22728 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
22731 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
22732 ("*r<noxa>sbg_<mode>_sll_bitmask")
22733 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
22734 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
22735 Use contiguous_bitmask_nowrap_operand.
22737 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22740 * config/rs6000/altivec.h (vec_xl): Revise #define.
22741 (vec_xst): Likewise.
22743 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
22745 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
22747 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
22749 PR rtl-optimization/79194
22750 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
22751 traps before call to bypass_conditional_jumps.
22753 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22755 PR tree-optimization/71374
22756 * lra-constraints.c (check_conflict_input_operands): New.
22757 (match_reload): Use it.
22759 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22762 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
22763 account to calculate conflict_set.
22765 2017-01-27 Bin Cheng <bin.cheng@arm.com>
22767 PR rtl-optimization/78559
22768 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
22769 other_insn in combine.
22771 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22773 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
22774 uint16_type_node for BT_UINT16.
22776 2017-01-27 David Malcolm <dmalcolm@redhat.com>
22778 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
22779 "RTL Tests" to menu.
22780 (GIMPLE Tests): New node.
22781 (RTL Tests): New node.
22783 2017-01-27 Richard Biener <rguenther@suse.de>
22785 PR tree-optimization/79245
22786 * tree-loop-distribution.c (distribute_loop): Apply cost
22787 modeling also to detected patterns.
22789 2017-01-27 Richard Biener <rguenther@suse.de>
22791 PR tree-optimization/71433
22792 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
22793 (compare_assert_loc): New function.
22794 (process_assert_insertions): Sort and optimize assert locations
22795 to remove duplicates and push down identical assertions on
22796 edges to their destination block.
22798 2017-01-27 Richard Biener <rguenther@suse.de>
22800 PR tree-optimization/79244
22801 * tree-vrp.c (remove_range_assertions): Forcefully propagate
22802 out SSA names even if abnormal.
22804 2017-01-27 Jakub Jelinek <jakub@redhat.com>
22806 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
22807 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
22808 instead of MPFR_RNDN.
22810 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
22813 * arm.c (arm_option_override): Don't call build_target_option_node
22814 until after doing all option overrides.
22815 (arm_valid_target_attribute_tree): Likewise.
22817 2017-01-27 Martin Liska <mliska@suse.cz>
22819 * doc/invoke.texi (-fprofile-arcs): Document profiling support
22820 for {cd}tors and C++ {cd}tors.
22822 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
22824 * config/s390/s390.md ("*setmem_long_and")
22825 ("*setmem_long_and_31z"): Use zero_extend instead of and.
22827 2017-01-26 Martin Sebor <msebor@redhat.com>
22829 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
22832 2017-01-26 Martin Sebor <msebor@redhat.com>
22834 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
22835 HAVE_DFmode before using XFmode or DFmode.
22836 (parse_directive): Avoid using the z length modifier to avoid
22837 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
22839 PR middle-end/78703
22840 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
22841 to accept adjustment as an array.
22842 (get_int_range): New function.
22843 (struct directive): Make width and prec arrays.
22844 (directive::set_width, directive::set_precision): Call get_int_range.
22845 (format_integer, format_floating): Handle width and precision ranges.
22846 (format_string, parse_directive): Same.
22848 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22851 * dwarf2out.c (generate_skeleton_bottom_up): For children with
22852 comdat_type_p set, just clone them, but keep the children in the
22856 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
22857 which have direct callers with -fvar-tracking-assignments enabled
22859 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
22860 inside of type units.
22862 2017-01-26 Martin Sebor <msebor@redhat.com>
22864 PR middle-end/78703
22865 * gimple-ssa-sprintf.c (struct result_range): Add likely and
22867 (struct format_result): Replace number_chars, number_chars_min,
22868 and number_chars_max with a single member of struct result_range.
22870 (format_result::operator+=): Adjust.
22871 (struct fmtresult): Remove bounded. Handle likely and unlikely
22873 (fmtresult::adjust_for_width_or_precision): New function.
22874 (fmtresult:type_max_digits): New function.
22875 (bytes_remaining): Handle likely and unlikely counters.
22876 (min_bytes_remaining): Remove.
22877 (format_percent): Simplify.
22878 (format_integer, format_floating): Set likely and unlikely counters.
22879 (get_string_length, format_character, format_string): Same.
22880 (format_plain, should_warn_p): New function.
22881 (maybe_warn): Call should_warn_p. Update diagnostic messages
22882 and handle those for all directives, including plain strings.
22883 (format_directive): Handle likely and unlikely counters.
22884 Remove unnecessary quoting from diagnostics. Add an informational
22886 (add_bytes): Remove.
22887 (pass_sprintf_length::compute_format_length): Simplify.
22888 (try_substitute_return_value): Handle likely and unlikely counters.
22890 2017-01-26 Carl Love <cel@us.ibm.com>
22892 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
22893 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
22895 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
22898 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
22899 endianess for subregs into account.
22900 * lra-constraints.c (lra_constraints): Do risky transformations
22901 always on the first iteration.
22902 * lra-lives.c (check_pseudos_live_through_calls): Add arg
22903 last_call_used_reg_set.
22904 (process_bb_lives): Define and use last_call_used_reg_set.
22905 * lra.c (lra): Always continue after lra_constraints on the first
22908 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
22910 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
22912 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
22914 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22916 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
22917 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
22918 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
22919 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
22920 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
22921 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
22922 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
22923 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
22924 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
22926 2017-01-26 Marek Polacek <polacek@redhat.com>
22929 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
22930 for the third operand.
22932 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22934 PR middle-end/79236
22935 * omp-low.c (struct omp_context): Add simt_stmt field.
22936 (scan_omp_for): Return omp_context *.
22937 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
22938 context to the _simt_ SIMD stmt.
22939 (lower_omp_for): For combined SIMD with sibling _simt_
22940 SIMD, make sure to use the same decls in _looptemp_
22941 clauses as in the sibling.
22943 2017-01-26 David Sherwood <david.sherwood@arm.com>
22945 PR middle-end/79212
22946 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
22949 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22952 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
22953 emit fld b; fld a; if possible.
22955 * brig-builtins.def: Update copyright years.
22956 * config/arm/arm_acle_builtins.def: Update copyright years.
22958 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
22961 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
22962 constraint instead of o for the stxsd instruction.
22964 2017-01-25 Carl Love <cel@us.ibm.com>
22966 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
22967 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
22969 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
22971 * doc/invoke.texi (C++ Dialect Options): Fix typo.
22973 2017-01-25 Richard Biener <rguenther@suse.de>
22975 PR tree-optimization/69264
22976 * target.def (vector_alignment_reachable): Improve documentation.
22977 * doc/tm.texi: Regenerate.
22978 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
22980 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
22981 earlier changes with respect to TYPE_USER_ALIGN.
22982 (vector_alignment_reachable_p): Likewise. Improve dumping.
22984 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22987 * config/arm/arm.md (xordi3): Force constant operand into a register
22990 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22992 * doc/invoke.texi (-fstore-merging): Correct default optimization
22993 levels at which it is enabled.
22994 (-O): Move -fstore-merging from list to...
22997 2017-01-25 Richard Biener <rguenther@suse.de>
23000 * omp-expand.c: Include debug.h.
23001 (expand_omp_taskreg): Make sure to generate early debug before
23002 outlining anything from a function.
23003 (expand_omp_target): Likewise.
23004 (grid_expand_target_grid_body): Likewise.
23006 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
23009 * asan.c (get_translation_unit_decl): New function.
23010 (asan_add_global): Extract modules file name from globals
23011 TRANSLATION_UNIT_DECL name.
23013 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
23016 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
23017 for long calls with APCS frame and VFP.
23019 2017-01-24 David Malcolm <dmalcolm@redhat.com>
23021 * cfg.c (original_copy_tables_initialized_p): New function.
23022 * cfg.h (original_copy_tables_initialized_p): New decl.
23023 * cfgrtl.c (relink_block_chain): Guard the call to
23024 free_original_copy_tables with a call to
23025 original_copy_tables_initialized_p.
23026 * cgraph.h (symtab_node::native_rtl_p): New decl.
23027 * cgraphunit.c (symtab_node::native_rtl_p): New function.
23028 (symtab_node::needed_p): Don't assert for early assembly output
23029 for __RTL functions.
23030 (cgraph_node::finalize_function): Set "force_output" for __RTL
23032 (cgraph_node::analyze): Bail out early for __RTL functions.
23033 (analyze_functions): Update assertion to support __RTL functions.
23034 (cgraph_node::expand): Bail out early for __RTL functions.
23035 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
23037 * function.h (struct function): Update comment for field
23039 * gimple-expr.c: Include "tree-pass.h".
23040 (gimple_has_body_p): Return false for __RTL functions.
23041 * Makefile.in (OBJS): Add run-rtl-passes.o.
23042 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
23044 (gcc::pass_manager::get_clean_slate): New accessor.
23045 * passes.c: Include "insn-addr.h".
23046 (should_skip_pass_p): Add logging. Update logic for running
23047 "expand" to be compatible with both __GIMPLE and __RTL. Guard
23048 property-provider override so it is only done for gimple passes.
23050 (skip_pass): New function.
23051 (execute_one_pass): Call skip_pass when skipping passes.
23052 * read-md.c (md_reader::read_char): Support filtering
23053 the input to a subset of line numbers.
23054 (md_reader::md_reader): Initialize fields
23055 m_first_line and m_last_line.
23056 (md_reader::read_file_fragment): New function.
23057 * read-md.h (md_reader::read_file_fragment): New decl.
23058 (md_reader::m_first_line): New field.
23059 (md_reader::m_last_line): New field.
23060 * read-rtl-function.c (function_reader::create_function): Only
23061 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
23062 curr_properties. Set DECL_INITIAL to a dummy block.
23063 (read_rtl_function_body_from_file_range): New function.
23064 * read-rtl-function.h (read_rtl_function_body_from_file_range):
23066 * run-rtl-passes.c: New file.
23067 * run-rtl-passes.h: New file.
23069 2017-01-24 Jeff Law <law@redhat.com>
23071 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
23074 2017-01-24 Bin Cheng <bin.cheng@arm.com>
23076 PR tree-optimization/79159
23077 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
23078 (record_nonwrapping_iv): Improve boundary using above function if no
23079 value range information.
23081 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
23082 Martin Jambor <mjambor@suse.cz>
23084 * brig-builtins.def: New file.
23085 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
23086 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
23087 (DEF_HSAIL_SAT_BUILTIN): Likewise.
23088 (DEF_HSAIL_INTR_BUILTIN): Likewise.
23089 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
23090 * builtin-types.def (BT_INT8): New.
23091 (BT_INT16): Likewise.
23092 (BT_UINT8): Likewise.
23093 (BT_UINT16): Likewise.
23094 (BT_FN_ULONG): Likewise.
23095 (BT_FN_UINT_INT): Likewise.
23096 (BT_FN_UINT_ULONG): Likewise.
23097 (BT_FN_UINT_LONG): Likewise.
23098 (BT_FN_UINT_PTR): Likewise.
23099 (BT_FN_ULONG_PTR): Likewise.
23100 (BT_FN_INT8_FLOAT): Likewise.
23101 (BT_FN_INT16_FLOAT): Likewise.
23102 (BT_FN_UINT32_FLOAT): Likewise.
23103 (BT_FN_UINT16_FLOAT): Likewise.
23104 (BT_FN_UINT8_FLOAT): Likewise.
23105 (BT_FN_UINT64_FLOAT): Likewise.
23106 (BT_FN_UINT16_UINT32): Likewise.
23107 (BT_FN_UINT32_UINT16): Likewise.
23108 (BT_FN_UINT16_UINT16_UINT16): Likewise.
23109 (BT_FN_INT_PTR_INT): Likewise.
23110 (BT_FN_UINT_PTR_UINT): Likewise.
23111 (BT_FN_LONG_PTR_LONG): Likewise.
23112 (BT_FN_ULONG_PTR_ULONG): Likewise.
23113 (BT_FN_VOID_UINT64_UINT64): Likewise.
23114 (BT_FN_UINT8_UINT8_UINT8): Likewise.
23115 (BT_FN_INT8_INT8_INT8): Likewise.
23116 (BT_FN_INT16_INT16_INT16): Likewise.
23117 (BT_FN_INT_INT_INT): Likewise.
23118 (BT_FN_UINT_FLOAT_UINT): Likewise.
23119 (BT_FN_FLOAT_UINT_UINT): Likewise.
23120 (BT_FN_ULONG_UINT_UINT): Likewise.
23121 (BT_FN_ULONG_UINT_PTR): Likewise.
23122 (BT_FN_ULONG_ULONG_ULONG): Likewise.
23123 (BT_FN_UINT_UINT_UINT): Likewise.
23124 (BT_FN_VOID_UINT_PTR): Likewise.
23125 (BT_FN_UINT_UINT_PTR: Likewise.
23126 (BT_FN_UINT32_UINT64_PTR): Likewise.
23127 (BT_FN_INT_INT_UINT_UINT): Likewise.
23128 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
23129 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
23130 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
23131 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
23132 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
23133 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
23134 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
23135 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
23136 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
23137 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
23138 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
23139 * doc/frontends.texi: List BRIG FE.
23140 * doc/install.texi (Testing): Add BRIG tesring requirements.
23141 * doc/invoke.texi (Overall Options): Mention BRIG.
23142 * doc/standards.texi (Standards): Doucment BRIG HSA version.
23144 2017-01-24 Richard Biener <rguenther@suse.de>
23146 PR translation/79208
23147 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
23149 2017-01-24 Martin Jambor <mjambor@suse.cz>
23152 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
23153 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
23154 and known_contexts.
23156 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
23158 PR middle-end/79123
23159 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
23160 casts from signed to unsigned really don't have a range.
23162 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
23164 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
23165 GMP_RNDx for compatiblity.
23167 2017-01-24 Martin Liska <mliska@suse.cz>
23170 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
23171 that would prevent us to call alloca with -1 as argument.
23173 2017-01-24 Jakub Jelinek <jakub@redhat.com>
23175 * dwarf2out.c (output_compilation_unit_header, output_file_names):
23176 Avoid -Wformat-security warning.
23178 2017-01-23 Andrew Pinski <apinski@cavium.com>
23180 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
23183 2017-01-23 Martin Sebor <msebor@redhat.com>
23185 PR middle-end/78703
23186 * gimple-ssa-sprintf.c (warn_level): New global.
23187 (format_integer): Use it here and throughout the rest of the file.
23188 Use the same switch to compute sign as base.
23189 (maybe_warn): New function.
23190 (format_directive): Factor out warnings into maybe_warn.
23191 Add debugging output. Use warn_level.
23192 (add_bytes): Use warn_level.
23193 (pass_sprintf_length::compute_format_length): Add debugging output.
23194 (try_substitute_return_value): Same.
23195 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
23197 PR middle-end/78703
23198 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
23199 (struct fmtresult, format_integer, format_floating): Adjust.
23200 (fmtresult::fmtresult): Set max correctly in two argument ctor.
23201 (get_string_length, format_string,format_directive): Same.
23202 (pass_sprintf_length::compute_format_length): Same.
23203 (try_substitute_return_value): Simplify slightly.
23205 PR middle-end/78703
23206 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
23207 (fmtresult::operator+=): Outlined.
23208 (struct fmtresult): Add ctors.
23209 (struct conversion_spec): Rename...
23210 (struct directive): ...to this. Add and remove data members.
23211 (directive::set_width, directive::set_precision): New functions.
23212 (format_percent): Use fmtresult ctor.
23213 (get_width_and_precision): Remove.
23214 (format_integer): Make naming changes. Avoid computing width and
23216 (format_floating): Same. Adjust indentation.
23217 (format_character, format_none): New functions.
23218 (format_string): Moved character handling to format_character.
23219 (format_directive): Remove arguments, change return type.
23220 (parse_directive): New function.
23221 (pass_sprintf_length::compute_format_length): Move directive
23222 parsing to parse_directive.
23224 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23226 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
23227 (assign_assembler_name_if_needed): ... this.
23228 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
23229 (assign_assembler_name_if_needed): ... this.
23230 (free_lang_data_in_cgraph): Adjust callers.
23231 * cgraphunit.c (cgraph_node::analyze): Likewise.
23232 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
23235 2017-01-23 Richard Biener <rguenther@suse.de>
23237 PR tree-optimization/79088
23238 PR tree-optimization/79188
23239 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
23240 resetting loop bounds after last path deletion. Reset loop
23241 bounds of the target loop, make code match the comments.
23242 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
23243 Make sure loops need no fixups.
23245 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23247 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
23248 exponent support with double type for first argument.
23249 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
23250 type returned by __builtin_vec_extract_sig,
23251 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
23252 functions from "vector int" to "vector unsigned int" or from
23253 "vector long long int" to "vector unsigned long long int".
23254 Changed type returned by __builtin_vec_extract_exp,
23255 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
23256 functions from "vector int" to "vector unsigned int" or from
23257 "vector long long int" to "vector unsigned long long int".
23258 Changed return type of __builtin_vec_test_data_class,
23259 __builtin_vec_test_data_class_sp, and
23260 __builtin_vec_test_data_class_dp from "vector int" to
23261 "vector bool int" or from "vector long long int" to "vector bool
23262 long long int" and changed second argument type from "unsigned
23263 int" to "int". Added new overloaded function forms "vector float
23264 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
23265 "vector float __builtin_vec_insert_exp_sp (vector float, vector
23266 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
23267 double, vector unsigned long long int)" and "vector double
23268 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
23269 long int)". Changed return type of
23270 __builtin_scalar_test_data_class and
23271 __builtin_scalar_test_data_class_sp and
23272 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
23273 int" and changed second argument from "unsigned int" to "int".
23274 Changed type returned by __builtin_scalar_test_neg,
23275 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
23276 from "int" to "bool int". Added new overloaded function form
23277 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
23278 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
23279 exponent double-precision with floating point first argument.
23280 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
23281 documentation of scalar_test_data_class, scalar_test_neg,
23282 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
23283 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
23284 vec_test_data_class built-in functions to reflect refinements in
23285 their type signatures.
23287 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
23289 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
23291 (aarch64_elf_asm_destructor): Likewise.
23293 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
23295 PR rtl-optimization/78634
23296 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
23297 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
23298 * ifcvt.c (noce_try_cmove): Add missing cost check.
23300 PR rtl-optimization/71724
23301 * combine.c (if_then_else_cond): Look for situations where it is
23302 beneficial to undo the work of one of the recursive calls.
23304 2017-01-23 Bin Cheng <bin.cheng@arm.com>
23306 PR tree-optimization/70754
23307 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
23308 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
23309 combined stmt before it if not NULL.
23310 (combine_chains): Process refs reversely and compute dominance point
23313 2017-01-23 Martin Liska <mliska@suse.cz>
23315 PR tree-optimization/79196
23316 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
23317 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
23319 (strlen_optimize_stmt): Call the renamed function.
23321 2017-01-23 Michael Matz <matz@suse.de>
23323 PR tree-optimization/78384
23324 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
23326 2017-01-23 Richard Biener <rguenther@suse.de>
23328 PR tree-optimization/79186
23329 * tree-vrp.c (register_new_assert_for): Make sure we've seen
23330 both incoming edges before moving an assert.
23332 2017-01-23 Martin Jambor <mjambor@suse.cz>
23334 * ipa-prop.c (load_from_param_1): Removed.
23335 (load_from_unmodified_param): Bits from load_from_param_1 put back
23337 (load_from_param): Removed.
23338 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
23339 with stmt. Reverted back to use of load_from_unmodified_param.
23341 2017-01-23 Martin Jambor <mjambor@suse.cz>
23344 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
23345 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
23346 field a pointer to garbage collected vector, mark lattices and
23347 ipcp_orig_node with GTY((skip)).
23348 (ipa_get_param_count): Adjust to descriptors being a pointer.
23349 (ipa_get_param): Likewise.
23350 (ipa_get_type): Likewise.
23351 (ipa_get_param_move_cost): Likewise.
23352 (ipa_set_param_used): Likewise.
23353 (ipa_get_controlled_uses): Likewise.
23354 (ipa_set_controlled_uses): Likewise.
23355 (ipa_is_param_used): Likewise.
23356 (ipa_node_params_t): Move into garbage collector. New methods insert
23358 (ipa_node_params_sum): Annotate wth GTY(()).
23359 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
23361 (ipa_load_from_parm_agg): Adjust declaration.
23362 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
23363 * ipa-profile.c (ipa_profile): Likewise.
23364 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
23365 (ipa_populate_param_decls): Make descriptors parameter garbage
23367 (ipa_dump_param): Adjust to descriptors being a pointer.
23368 (ipa_alloc_node_params): Likewise.
23369 (ipa_initialize_node_params): Likewise.
23370 (load_from_param_1): Make descriptors parameter garbage collected.
23371 (load_from_unmodified_param): Likewise.
23372 (load_from_param): Likewise.
23373 (ipa_load_from_parm_agg): Likewise.
23374 (ipa_node_params::~ipa_node_params): Removed.
23375 (ipa_free_all_node_params): Remove call to delete operator.
23376 (ipa_node_params_t::insert): New.
23377 (ipa_node_params_t::remove): Likewise.
23378 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
23379 copy known_csts and known_contexts vectors.
23380 (ipa_read_node_info): Adjust to descriptors being a pointer.
23381 (ipcp_modif_dom_walker): Make m_descriptors field garbage
23383 (ipcp_transform_function): Make descriptors variable garbage
23386 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
23388 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
23389 * config/i386/avx512dqintrin.h: Ditto.
23390 * config/i386/avx512fintrin.h: Ditto.
23391 * config/i386/i386.c: Handle new builtins.
23392 * config/i386/i386-builtin.def: Add new builtins.
23393 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
23394 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
23396 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23397 Martin Liska <mliska@suse.cz>
23399 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
23400 * asan.c (asan_expand_poison_ifn): Support stores and use
23401 appropriate ASAN report function.
23402 * internal-fn.c (expand_ASAN_POISON_USE): New function.
23403 * internal-fn.def (ASAN_POISON_USE): Declare.
23404 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
23405 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
23406 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
23407 ASAN_POISON calls w/o LHS.
23408 * tree-ssa.c (execute_update_addresses_taken): Create clobber
23409 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
23410 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
23411 * gimplify.c (asan_poison_variables): Add attribute
23412 use_after_scope_memory to variables that really needs to live
23414 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
23415 having the attribute.
23417 2017-01-23 Martin Liska <mliska@suse.cz>
23419 * asan.c (create_asan_shadow_var): New function.
23420 (asan_expand_poison_ifn): Likewise.
23421 * asan.h (asan_expand_poison_ifn): New declaration.
23422 * internal-fn.c (expand_ASAN_POISON): Likewise.
23423 * internal-fn.def (ASAN_POISON): New builtin.
23424 * sanopt.c (pass_sanopt::execute): Expand
23425 asan_expand_poison_ifn.
23426 * tree-inline.c (copy_decl_for_dup_finish): Make function
23428 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
23429 * tree-ssa.c (is_asan_mark_p): New function.
23430 (execute_update_addresses_taken): Rewrite local variables
23431 (identified just by use-after-scope as addressable) into SSA.
23433 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23435 * doc/install.texi (Specific): opensource.apple.com uses https
23436 now. Remove trailing slash.
23438 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23440 * README.Portability: Remove note on an Irix compatibility issue.
23442 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
23444 * gcov.c (INCLUDE_ALGORITHM): Define.
23445 (INCLUDE_VECTOR): Define.
23446 No longer include <vector> and <algorithm> directly.
23448 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
23450 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
23452 * doc/invoke.texi (Code Gen Options): Ditto.
23454 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
23457 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
23459 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
23461 rtl-optimization/79125
23462 * cprop.c (local_cprop_pass): Handle cases where we make an
23463 unconditional trap.
23465 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
23469 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
23470 read from, for big endian.
23472 2017-01-20 Jiong Wang <jiong.wang@arm.com>
23474 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
23475 register pauth builtins for LP64 only.
23477 2017-01-20 Marek Polacek <polacek@redhat.com>
23480 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
23483 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23485 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
23487 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
23488 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
23489 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
23491 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23494 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
23495 in big-endian mode when they are not a single duplicated value.
23497 2017-01-20 Richard Biener <rguenther@suse.de>
23499 * BASE-VER: Bump to 7.0.1.
23501 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23503 * omp-low.c (omplow_simd_context): New struct. Use it...
23504 (lower_rec_simd_input_clauses): ...here and...
23505 (lower_rec_input_clauses): ...here to hold common data. Adjust all
23506 references to idx, lane, max_vf, is_simt.
23508 2017-01-20 Graham Markall <graham.markall@embecosm.com>
23510 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
23513 2017-01-20 Martin Jambor <mjambor@suse.cz>
23515 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
23516 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
23518 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
23519 (GTFILES): Rename hsa.c to hsa-common.c.
23520 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
23521 * hsa-dump.c: Likewise.
23522 * hsa-gen.c: Likewise.
23523 * hsa-regalloc.c: Likewise.
23524 * ipa-hsa.c: Likewise.
23525 * omp-expand.c: Likewise.
23526 * omp-low.c: Likewise.
23527 * toplev.c: Likewise.
23529 2017-01-20 Marek Polacek <polacek@redhat.com>
23532 * doc/invoke.texi: Document -Wduplicated-branches.
23533 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
23534 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
23535 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
23536 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
23537 return 0 only when not OEP_LEXICOGRAPHIC.
23538 (fold_build_cleanup_point_expr): Use the expression
23539 location when building CLEANUP_POINT_EXPR.
23540 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
23541 * tree.c (add_expr): Handle error_mark_node.
23543 2017-01-20 Martin Liska <mliska@suse.cz>
23546 * tree-profile.c (init_ic_make_global_vars): Do not call
23548 (gimple_init_gcov_profiler): Likewise.
23550 2017-01-20 Martin Liska <mliska@suse.cz>
23553 * cgraph.h (maybe_create_reference): Remove argument and
23555 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
23557 * ipa-cp.c (create_specialized_node): Likewise.
23558 * symtab.c (symtab_node::maybe_create_reference): Handle
23559 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
23561 2017-01-20 Martin Liska <mliska@suse.cz>
23563 * read-rtl-function.c (function_reader::create_function): Use
23564 build_decl instread of build_decl_stat.
23566 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
23568 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
23569 * config/i386/avx512dqintrin.h: Ditto.
23570 * config/i386/avx512fintrin.h: Ditto.
23571 * config/i386/i386-builtin-types.def: Add new types.
23572 * config/i386/i386.c: Handle new types.
23573 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
23574 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
23575 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
23576 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
23577 (__builtin_ia32_kshiftridi): New.
23578 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
23580 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
23584 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
23585 define to rs6000_init_stack_protect_guard.
23586 (rs6000_init_stack_protect_guard): New function.
23588 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23589 Yunqiang Su <yunqiang.su@imgtec.com>
23591 * config.gcc (supported_defaults): Add madd4.
23592 (with_madd4): Add validation.
23593 (all_defaults): Add madd4.
23594 * config/mips/mips.opt (mmadd4): New option.
23595 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23597 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23599 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
23600 (ISA_HAS_FUSED_MADD4): Likewise.
23601 * doc/invoke.texi (-mmadd4): Document the new option.
23602 * doc/install.texi (--with-madd4): Document the new option.
23604 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23606 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
23607 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
23608 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
23609 (aarch64_init_pauth_hint_builtins): New.
23610 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
23611 (aarch64_expand_builtin): Expand new builtins.
23613 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23615 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
23616 * combine-stack-adj.c (no_unhandled_cfa): Handle
23617 REG_CFA_TOGGLE_RA_MANGLE.
23618 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
23619 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
23620 info for return address signing.
23621 (aarch64_expand_epilogue): Likewise.
23623 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23625 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
23626 * config/aarch64/aarch64-protos.h
23627 (aarch64_return_address_signing_enabled): New declaration.
23628 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
23630 (aarch64_expand_prologue): Sign return address before it's pushed onto
23632 (aarch64_expand_epilogue): Authenticate return address fetched from
23634 (aarch64_override_options): Sanity check for ILP32 and ISA level.
23635 (aarch64_attributes): New function attributes for "sign-return-address".
23636 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
23637 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
23638 ("*do_return"): Generate combined instructions according to key index.
23639 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
23640 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
23642 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
23643 * config/aarch64/aarch64.opt (msign-return-address=): New.
23644 * doc/extend.texi (AArch64 Function Attributes): Documents
23645 "sign-return-address=".
23646 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
23648 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23650 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
23651 overall option summary.
23653 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23655 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
23656 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
23657 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
23658 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
23660 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
23662 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
23663 -mpower9-minmax by default for -mcpu=power9.
23664 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
23665 128-bit floating point.
23667 2017-01-20 Alan Modra <amodra@gmail.com>
23669 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
23670 optimizing for size.
23672 2017-01-20 Alan Modra <amodra@gmail.com>
23675 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
23676 for strcmp and strncmp from corresponding builtin decl.
23678 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23680 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
23681 instead of i386/rtems-64.h.
23682 * config/i386/rtems-64.h: Remove.
23684 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23688 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
23690 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
23692 2017-01-19 Tamar Christina <tamar.christina@arm.com>
23694 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
23695 Change int to HOST_WIDE_INT.
23696 * config/aarch64/aarch64-protos.h
23697 (aarch64_simd_gen_const_vector_dup): Likewise.
23698 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
23700 2017-01-19 David Malcolm <dmalcolm@redhat.com>
23702 * langhooks-def.h (lhd_type_for_size): New decl.
23703 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
23704 * langhooks.c (lhd_type_for_size): New function, taken from
23707 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
23709 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
23710 define_bypass for CR latency.
23711 (power9-cracked-alu): Update bypass latency and remove power9-branch.
23712 (power9-alu2): Add define_bypass for CR latency.
23714 (power9-mul): Update insn latency.
23715 (power9-mul-compare): Update insn latency, bypass latency and remove
23718 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23720 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
23722 * config/aarch64/aarch64.md
23723 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
23724 aarch64_nopcrelative_literal_loads.
23725 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
23727 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
23729 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
23730 TARGET_LOONGSON_3A.
23731 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
23733 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
23736 * config.gcc (supported_defaults): Add lxc1-sxc1.
23737 (with_lxc1_sxc1): Add validation.
23738 (all_defaults): Add lxc1-sxc1.
23739 * config/mips/mips.opt (mlxc1-sxc1): New option.
23740 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23742 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23743 __mips_no_lxc1_sxc1.
23744 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
23745 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
23746 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
23748 2017-01-19 Richard Biener <rguenther@suse.de>
23750 PR tree-optimization/72488
23751 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
23752 sure to restore SSA info.
23753 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
23755 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
23757 PR rtl-optimization/79121
23758 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
23759 of the inner type when shifting an extended value.
23761 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
23764 * symtab.c (symtab_node::equal_address_to): Fix comparing of
23765 interposable aliases.
23767 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
23770 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
23771 Use the evmergelohi instruction.
23772 (mov_si<mode>_e500_subreg4_2_le): Likewise.
23773 (mov_sitf_e500_subreg8_2_be): Likewise.
23774 (mov_sitf_e500_subreg12_2_le): Likewise.
23775 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
23776 (mov_si<mode>_e500_subreg4_2_be): Likewise.
23777 (mov_sitf_e500_subreg8_2_le): Likewise.
23778 (mov_sitf_e500_subreg12_2_be): Likewise.
23780 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23782 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
23783 attribute from vecsimple to vecperm.
23784 (altivec_vbpermq2): Likewise.
23786 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23789 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
23791 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23792 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
23793 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
23794 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
23795 case where N arg is SIZE_MAX.
23796 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
23797 (cmpstrsi): Add pattern.
23799 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23801 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23802 __builtin_vec_revb builtins.
23803 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
23804 built-in functions to support generation of the ISA 3.0 XXBR<x>
23805 vector byte reverse instructions.
23806 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
23807 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
23808 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
23809 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
23810 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
23811 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
23812 (P9V_BUILTIN_VEC_REVB): Likewise.
23813 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
23814 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
23815 (p9_xxbrq_v16qi): Likewise.
23816 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
23817 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
23818 (p9_xxbrh_v8hi): Likewise.
23819 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
23820 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
23821 vec_revb built-in functions.
23823 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
23825 PR rtl-optimization/78952
23826 * config/i386/i386.md (any_extract): New code iterator.
23827 (*insvqi_2): Use any_extract for source operand.
23828 (*insvqi_3): Use any_shiftrt for source operand.
23830 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
23832 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
23834 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
23836 2017-01-18 Matthias Klose <doko@ubuntu.com>
23838 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
23840 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23842 * config/rs6000/altivec.h (vec_bperm): Change #define.
23843 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
23844 (altivec_vbpermq2): New define_insn.
23845 (altivec_vbpermd): Likewise.
23846 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
23847 function interface.
23848 (VBPERMD): Likewise.
23849 (VBPERM): New polymorphic function interface.
23850 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
23851 Add entries for P9V_BUILTIN_VEC_VBPERM.
23852 * doc/extend.texi: Add interfaces for vec_bperm.
23854 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23856 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
23857 first letter of error messages.
23858 (s390_resolve_overloaded_builtin): Likewise.
23859 * config/s390/s390.c (s390_expand_builtin): Likewise.
23860 (s390_invalid_arg_for_unprototyped_fn): Likewise.
23861 (s390_valid_target_attribute_inner_p): Likewise.
23862 * config/s390/s390.md ("tabort"): Likewise.
23864 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
23866 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
23867 (ISA_AVOID_DIV_HILO): New macro.
23868 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
23869 (ISA_HAS_DDIV): Likewise.
23871 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23873 * doc/invoke.texi (fabi-version): Correct number of occurrences.
23875 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23877 * doc/invoke.texi (fabi-version): Spelling fix.
23879 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23882 * doc/invoke.texi (fabi-version): Mention mangling fix for
23885 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23888 * doc/invoke.texi (fabi-version): Document discriminator mangling.
23890 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
23893 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
23894 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
23896 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
23898 (stack_protect_test): Ditto.
23899 * config/rs6000/rs6000.opt (mstack-protector-guard=,
23900 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
23902 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
23903 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
23904 -mstack-protector-guard-offset=.
23905 (RS/6000 and PowerPC Options): Ditto.
23907 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
23909 * config/i386/i386.h (MASK_CLASS_P): New define.
23910 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
23911 there are no registers from different register sets also when
23912 mask registers are used. Update function comment.
23913 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
23914 to (*k/*r) and (*k/*km) alternatives.
23916 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
23918 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
23919 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
23920 (EH_RETURN_HANDLER_RTX): New define.
23921 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
23922 Force frame pointer in EH return functions.
23923 (aarch64_expand_epilogue): Add barrier for eh_return.
23924 (aarch64_final_eh_return_addr): Remove.
23925 (aarch64_eh_return_handler_rtx): New function.
23926 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
23928 (aarch64_eh_return_handler_rtx): New prototype.
23930 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23932 * config/rs6000/altivec.h (vec_rlmi): New #define.
23933 (vec_vrlnm): Likewise.
23934 (vec_rlnm): Likewise.
23935 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
23936 (UNSPEC_VRLNM): Likewise.
23937 (VIlong): New mode iterator.
23938 (altivec_vrl<VI_char>mi): New define_insn.
23939 (altivec_vrl<VI_char>nm): Likewise.
23940 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
23942 (VRLDNM): Likewise.
23943 (RLNM): New polymorphic function entry.
23944 (VRLWMI): New monomorphic function entry.
23945 (VRLDMI): Likewise.
23946 (RLMI): New polymorphic function entry.
23947 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
23948 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
23949 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
23952 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23955 * dwarf2out.c (field_byte_offset): Restore the
23956 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
23957 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
23958 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
23961 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
23964 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
23966 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23969 * dwarf2out.c (add_data_member_location_attribute): For constant
23970 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
23971 instead of DW_AT_data_member_location, DW_AT_bit_offset and
23972 DW_AT_byte_size attributes.
23974 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
23976 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
23977 after forcing to constant memory when the code model is medium.
23979 2017-01-17 Julia Koval <julia.koval@intel.com>
23982 * config/i386/avx512fintrin.h
23983 (_mm512_i32gather_ps): Change __addr type to void const*.
23984 (_mm512_mask_i32gather_ps): Ditto.
23985 (_mm512_i32gather_pd): Ditto.
23986 (_mm512_mask_i32gather_pd): Ditto.
23987 (_mm512_i64gather_ps): Ditto.
23988 (_mm512_mask_i64gather_ps): Ditto.
23989 (_mm512_i64gather_pd): Ditto.
23990 (_mm512_mask_i64gather_pd): Ditto.
23991 (_mm512_i32gather_epi32): Ditto.
23992 (_mm512_mask_i32gather_epi32): Ditto.
23993 (_mm512_i32gather_epi64): Ditto.
23994 (_mm512_mask_i32gather_epi64): Ditto.
23995 (_mm512_i64gather_epi32): Ditto.
23996 (_mm512_mask_i64gather_epi32): Ditto.
23997 (_mm512_i64gather_epi64): Ditto.
23998 (_mm512_mask_i64gather_epi64): Ditto.
23999 (_mm512_i32scatter_ps): Change __addr type to void*.
24000 (_mm512_mask_i32scatter_ps): Ditto.
24001 (_mm512_i32scatter_pd): Ditto.
24002 (_mm512_mask_i32scatter_pd): Ditto.
24003 (_mm512_i64scatter_ps): Ditto.
24004 (_mm512_mask_i64scatter_ps): Ditto.
24005 (_mm512_i64scatter_pd): Ditto.
24006 (_mm512_mask_i64scatter_pd): Ditto.
24007 (_mm512_i32scatter_epi32): Ditto.
24008 (_mm512_mask_i32scatter_epi32): Ditto.
24009 (_mm512_i32scatter_epi64): Ditto.
24010 (_mm512_mask_i32scatter_epi64): Ditto.
24011 (_mm512_i64scatter_epi32): Ditto.
24012 (_mm512_mask_i64scatter_epi32): Ditto.
24013 (_mm512_i64scatter_epi64): Ditto.
24014 (_mm512_mask_i64scatter_epi64): Ditto.
24015 * config/i386/avx512pfintrin.h
24016 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
24017 (_mm512_mask_prefetch_i32gather_ps): Ditto.
24018 (_mm512_mask_prefetch_i64gather_pd): Ditto.
24019 (_mm512_mask_prefetch_i64gather_ps): Ditto.
24020 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
24021 (_mm512_prefetch_i32scatter_ps): Ditto.
24022 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
24023 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
24024 (_mm512_prefetch_i64scatter_pd): Ditto.
24025 (_mm512_prefetch_i64scatter_ps): Ditto.
24026 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
24027 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
24028 * config/i386/avx512vlintrin.h
24029 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
24030 (_mm_mmask_i32gather_ps): Ditto.
24031 (_mm256_mmask_i32gather_pd): Ditto.
24032 (_mm_mmask_i32gather_pd): Ditto.
24033 (_mm256_mmask_i64gather_ps): Ditto.
24034 (_mm_mmask_i64gather_ps): Ditto.
24035 (_mm256_mmask_i64gather_pd): Ditto.
24036 (_mm_mmask_i64gather_pd): Ditto.
24037 (_mm256_mmask_i32gather_epi32): Ditto.
24038 (_mm_mmask_i32gather_epi32): Ditto.
24039 (_mm256_mmask_i32gather_epi64): Ditto.
24040 (_mm_mmask_i32gather_epi64): Ditto.
24041 (_mm256_mmask_i64gather_epi32): Ditto.
24042 (_mm_mmask_i64gather_epi32): Ditto.
24043 (_mm256_mmask_i64gather_epi64): Ditto.
24044 (_mm_mmask_i64gather_epi64): Ditto.
24045 (_mm256_i32scatter_ps): Change __addr type to void*.
24046 (_mm256_mask_i32scatter_ps): Ditto.
24047 (_mm_i32scatter_ps): Ditto.
24048 (_mm_mask_i32scatter_ps): Ditto.
24049 (_mm256_i32scatter_pd): Ditto.
24050 (_mm256_mask_i32scatter_pd): Ditto.
24051 (_mm_i32scatter_pd): Ditto.
24052 (_mm_mask_i32scatter_pd): Ditto.
24053 (_mm256_i64scatter_ps): Ditto.
24054 (_mm256_mask_i64scatter_ps): Ditto.
24055 (_mm_i64scatter_ps): Ditto.
24056 (_mm_mask_i64scatter_ps): Ditto.
24057 (_mm256_i64scatter_pd): Ditto.
24058 (_mm256_mask_i64scatter_pd): Ditto.
24059 (_mm_i64scatter_pd): Ditto.
24060 (_mm_mask_i64scatter_pd): Ditto.
24061 (_mm256_i32scatter_epi32): Ditto.
24062 (_mm256_mask_i32scatter_epi32): Ditto.
24063 (_mm_i32scatter_epi32): Ditto.
24064 (_mm_mask_i32scatter_epi32): Ditto.
24065 (_mm256_i32scatter_epi64): Ditto.
24066 (_mm256_mask_i32scatter_epi64): Ditto.
24067 (_mm_i32scatter_epi64): Ditto.
24068 (_mm_mask_i32scatter_epi64): Ditto.
24069 (_mm256_i64scatter_epi32): Ditto.
24070 (_mm256_mask_i64scatter_epi32): Ditto.
24071 (_mm_i64scatter_epi32): Ditto.
24072 (_mm_mask_i64scatter_epi32): Ditto.
24073 (_mm256_i64scatter_epi64): Ditto.
24074 (_mm256_mask_i64scatter_epi64): Ditto.
24075 (_mm_i64scatter_epi64): Ditto.
24076 (_mm_mask_i64scatter_epi64): Ditto.
24077 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
24078 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
24079 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
24080 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
24081 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
24082 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
24083 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
24084 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
24085 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
24086 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
24087 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
24088 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
24089 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
24090 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
24091 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
24092 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
24093 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
24094 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
24095 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
24096 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
24097 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
24098 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
24099 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
24100 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
24101 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
24102 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
24103 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
24104 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
24105 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
24106 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
24107 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
24108 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
24109 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
24110 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
24111 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
24112 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
24113 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
24114 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
24115 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
24116 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
24117 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
24118 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
24119 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
24120 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
24121 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
24122 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
24123 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
24124 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
24125 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
24126 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
24127 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
24128 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
24129 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
24130 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
24131 definitions accordingly.
24133 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
24134 Kuan-Lin Chen <kuanlinchentw@gmail.com>
24137 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
24140 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
24143 * ira-conflicts.c (ira_build_conflicts): Update total conflict
24144 hard regs for inner regno.
24146 2017-01-17 Martin Liska <mliska@suse.cz>
24149 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
24150 assumption and add comment.
24152 2017-01-17 Nathan Sidwell <nathan@acm.org>
24154 * ipa-visibility.c (localize_node): New function, broken out of ...
24155 (function_and_variable_visibility): ... here. Call it.
24157 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24159 PR middle-end/77445
24160 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
24161 correctly set frequency of oudgoing edge.
24162 (duplicate_thread_path): Fix profile updating.
24164 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24167 * configure.ac: Add GCC_BASE_VER.
24168 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
24169 version from BASE-VER file.
24170 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
24171 (gcc.o): Depend on $(BASEVER).
24172 * common.opt (dumpfullversion): New option.
24173 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
24174 * doc/invoke.texi: Document -dumpfullversion.
24175 * doc/install.texi: Document --with-gcc-major-version-only.
24176 * configure: Regenerated.
24178 2017-01-17 Richard Biener <rguenther@suse.de>
24180 PR tree-optimization/71433
24181 * tree-vrp.c (register_new_assert_for): Merge same asserts
24182 on all incoming edges.
24183 (process_assert_insertions_for): Handle insertions at the
24186 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
24188 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
24189 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
24191 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
24194 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
24197 2017-01-17 Alan Modra <amodra@gmail.com>
24200 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
24201 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
24202 symbolic stack limit when pic.
24204 2017-01-16 Martin Sebor <msebor@redhat.com>
24206 PR tree-optimization/78608
24207 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
24209 2017-01-16 Jeff Law <law@redhat.com>
24212 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
24213 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
24214 for several include directories that may be relative to sysroot.
24215 * config/i386/x-mingw32 (gplus_includedir): Define.
24216 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
24217 (native_system_includedir): Likewise.
24218 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
24219 override if TARGET_SYSTEM_ROOT is defined.
24220 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
24222 PR tree-optimization/79090
24223 PR tree-optimization/33562
24224 PR tree-optimization/61912
24225 PR tree-optimization/77485
24226 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
24227 and computed trims into the dump file.
24229 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24231 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
24233 2017-01-16 Jakub Jelinek <jakub@redhat.com>
24236 * gimplify.c (gimplify_init_constructor): If want_value and
24237 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
24241 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
24242 sequence. Formatting fixes.
24243 (doloop_optimize): Formatting fixes.
24246 * gcc.c (debug_level_greater_than_spec_func): New function.
24247 (static_spec_functions): Add debug-level-gt spec function.
24248 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
24250 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24251 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24252 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
24253 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
24254 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
24255 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
24257 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
24259 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
24260 QImode fixups to general and mask registers only.
24262 2017-01-16 Carl Love <cel@us.ibm.com>
24264 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
24265 for built-in functions
24266 vector signed char vec_nabs (vector signed char)
24267 vector signed short vec_nabs (vector signed short)
24268 vector signed int vec_nabs (vector signed int)
24269 vector signed long long vec_nabs (vector signed long long)
24270 vector float vec_nabs (vector float)
24271 vector double vec_nabs (vector double)
24272 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
24274 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
24275 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
24276 * doc/extend.texi: Update the documentation file for the new built-in
24279 2017-01-16 Martin Sebor <msebor@redhat.com>
24281 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
24284 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24286 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
24287 UNSPEC_VSX__XXSPLTD to require special splat handling.
24289 2017-01-16 David Malcolm <dmalcolm@redhat.com>
24292 * system.h: Poison strndup.
24294 2017-01-16 Alan Modra <amodra@gmail.com>
24297 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
24300 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
24302 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
24304 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
24306 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
24307 call recog here. Assert that INSN_CODE (insn) is non-negative.
24309 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24312 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
24314 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
24315 in the currently scheduled RTL fragment.
24317 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24319 PR rtl-optimization/78751
24320 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
24323 2017-01-14 Jeff Law <law@redhat.com>
24325 PR tree-optimization/79090
24326 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
24327 variable length stores.
24328 (compute_trims): Delete dead assignment to *trim_tail.
24329 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
24332 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
24334 PR rtl-optimization/78626
24335 PR rtl-optimization/78727
24336 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
24337 of a block, and split such blocks after everything else is finished.
24339 2017-01-14 Alan Modra <amodra@gmail.com>
24342 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
24343 target legitimate_combined_insn.
24344 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
24345 (rs6000_legitimate_combined_insn): New function.
24346 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
24348 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
24349 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
24350 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
24352 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
24354 * doc/frontends.texi (G++ and GCC): Remove references to Java.
24356 2017-01-13 Jeff Law <law@redhat.com>
24358 PR tree-optimization/33562
24359 PR tree-optimization/61912
24360 PR tree-optimization/77485
24361 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
24363 (delete_dead_assignment): Likewise.
24364 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
24365 statement to delete_dead_call and delete_dead_assignment.
24367 2017-01-13 David Malcolm <dmalcolm@redhat.com>
24370 * substring-locations.c (format_warning_va): Strengthen case 1 so
24371 that both endpoints of the substring must be within the format
24372 range for just the substring to be printed.
24374 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
24376 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
24377 * config/i386/i386.c (ix86_target_string): Add missing options
24378 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
24379 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
24380 flags_other and ix86_target_other to flags2_other. Display unknown
24382 (ix86_valid_target_attribute_inner_p): Add missing options and
24383 reorder options by implied ISAs, as in ix86_target_string.
24385 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24387 * hash-table.h (hash_table::too_empty_p): New function.
24388 (hash_table::expand): Use it.
24389 (hash_table::traverse): Likewise.
24390 (hash_table::empty_slot): Use sizeof (value_type) instead of
24391 sizeof (PTR) to convert bytes to elements. Shrink the table
24392 if the current size is excessive for the current number of
24395 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24397 * ira-costs.c (record_reg_classes): Break from the inner loop
24398 early once alt_fail is known to be true. Update outer loop
24399 handling accordingly.
24401 2017-01-13 Jeff Law <law@redhat.com>
24403 * tree-ssa-dse.c (decrement_count): New function.
24404 (increment_start_addr, maybe_trim_memstar_call): Likewise.
24405 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
24406 when we know the partially dead statement is a mem* function.
24408 PR tree-optimization/61912
24409 PR tree-optimization/77485
24410 * tree-ssa-dse.c: Include expr.h.
24411 (maybe_trim_constructor_store): New function.
24412 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
24414 PR tree-optimization/33562
24415 PR tree-optimization/61912
24416 PR tree-optimization/77485
24417 * doc/invoke.texi: Document new dse-max-object-size param.
24418 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
24419 * tree-ssa-dse.c: Include params.h.
24420 (dse_store_status): New enum.
24421 (initialize_ao_ref_for_dse): New, partially extracted from
24423 (valid_ao_ref_for_dse, normalize_ref): New.
24424 (setup_live_bytes_from_ref, compute_trims): Likewise.
24425 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
24426 (maybe_trim_partially_dead_store): Likewise.
24427 (maybe_trim_complex_store): Likewise.
24428 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
24429 Track what bytes live from the original store. Return tri-state
24430 for dead, partially dead or live.
24431 (dse_dom_walker): Add constructor, destructor and new private members.
24432 (delete_dead_call, delete_dead_assignment): New extracted from
24434 (dse_optimize_stmt): Make a member of dse_dom_walker.
24435 Use initialize_ao_ref_for_dse.
24437 PR tree-optimization/33562
24438 PR tree-optimization/61912
24439 PR tree-optimization/77485
24440 * sbitmap.h (bitmap_count_bits): Prototype.
24441 (bitmap_clear_range, bitmap_set_range): Likewise.
24442 * sbitmap.c (bitmap_clear_range): New function.
24443 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
24445 2017-01-13 Martin Liska <mliska@suse.cz>
24448 * function.c (set_cfun): Add new argument force.
24449 * function.h (set_cfun): Likewise.
24450 * ipa-inline-transform.c (inline_call): Use the function when
24451 strict alising from is dropped for function we inline to.
24453 2017-01-13 Richard Biener <rguenther@suse.de>
24455 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
24456 for dumping GIMPLE INTEGER_CSTs.
24458 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24460 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
24461 to 201112L since C++17.
24463 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24466 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
24467 if -fsanitize=kernel-address is present.
24469 2017-01-13 Richard Biener <rguenther@suse.de>
24471 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
24472 as _Literal ( type ) number in case usual suffixes do not
24473 preserve all information.
24475 2017-01-13 Richard Biener <rguenther@suse.de>
24477 PR tree-optimization/77283
24478 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
24479 and ssa-iterators.h.
24480 (is_feasible_trace): Implement a cost model based on joiner
24483 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24486 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
24487 char or short to __float128/_Float128 directly.
24489 2017-01-12 Martin Sebor <msebor@redhat.com>
24491 to -Wformat-overflow.
24492 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
24493 (min_bytes_remaining): Same.
24494 (get_string_length): Same.
24495 (format_string): Same.
24496 (format_directive): Same.
24498 (pass_sprintf_length::handle_gimple_call): Same.
24500 2017-01-12 Jakub Jelinek <jakub@redhat.com>
24502 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
24503 info.nowrite calls with no lhs that can't throw. Return bool
24504 whether gsi_remove has been called or not.
24505 (pass_sprintf_length::handle_gimple_call): Return bool whether
24506 try_substitute_return_value called gsi_remove. Formatting fix.
24507 (pass_sprintf_length::execute): Don't use gsi_remove if
24508 handle_gimple_call returned true.
24511 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
24512 be removed due to side-effects, don't remove following barrier nor
24513 turn the successor edge into fallthru edge.
24515 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24518 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
24519 element-reversing loads and stores as not swappable.
24521 2017-01-12 Nathan Sidwell <nathan@acm.org>
24522 Nicolai Stange <nicstange@gmail.com>
24524 * combine.c (try_combine): Don't ignore result of overlap checking
24525 loop. Combine overlap & asm check into single loop.
24527 2017-01-12 Richard Biener <rguenther@suse.de>
24529 * tree-pretty-print.c (dump_generic_node): Provide -gimple
24530 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
24532 2017-01-12 Richard Biener <rguenther@suse.de>
24534 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
24535 and TS_TARGET_OPTION directly derive from TS_BASE.
24536 * tree-core.h (tree_optimization_option): Derive from tree_base.
24537 (tree_target_option): Likewise.
24539 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
24541 * config/i386/i386.c (memory_address_length): Increase len
24542 only when rip_relative_addr_p returns false.
24544 2017-01-11 Julia Koval <julia.koval@intel.com>
24546 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
24547 (OPTION_MASK_ISA_SGX_SET): New.
24548 (ix86_handle_option): Handle OPT_msgx.
24549 * config.gcc: Added sgxintrin.h.
24550 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
24551 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
24552 * config/i386/i386.c (ix86_target_string): Add -msgx.
24554 (ix86_option_override_internal): Handle new options.
24555 (ix86_valid_target_attribute_inner_p): Add sgx.
24556 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
24557 * config/i386/i386.opt: Add msgx.
24558 * config/i386/sgxintrin.h: New file.
24559 * config/i386/x86intrin.h: Add sgxintrin.h.
24561 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24564 * fold-const.c (maybe_nonzero_address): Return 1 for function
24566 (tree_single_nonzero_warnv_p): Don't handle function local objects
24570 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
24573 2017-01-11 David Malcolm <dmalcolm@redhat.com>
24576 * opts.c: Include "spellcheck.h"
24577 (struct string_fragment): New struct.
24578 (struct edit_distance_traits<const string_fragment &>): New
24580 (get_closest_sanitizer_option): New function.
24581 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
24583 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24585 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
24587 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
24588 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
24589 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
24590 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
24591 for initial die_offset if dwarf_split_debug_info.
24592 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
24593 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
24595 (output_skeleton_debug_sections): Formatting fix. Use
24596 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
24597 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
24599 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
24601 * config/arm/cortex-a53.md: Add bypasses for
24602 cortex_a53_r2f_cvt.
24603 (cortex_a53_r2f): Only use for transfers.
24604 (cortex_a53_f2r): Likewise.
24605 (cortex_a53_r2f_cvt): Add reservation for conversions.
24606 (cortex_a53_f2r_cvt): Likewise.
24608 2017-01-11 Tamar Christina <tamar.christina@arm.com>
24610 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
24611 to all inlined functions, change static to extern.
24613 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
24616 * config/arm/arm.c (legitimize_pic_address): Handle reference to
24618 (arm_assemble_integer): Likewise.
24620 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
24622 * config.gcc: Use new awk script to check CPU, FPU and architecture
24623 parameters for --with-... options.
24624 * config/arm/parsecpu.awk: New file
24625 * config/arm/arm-cpus.in: New file.
24626 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
24628 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
24630 * config/arm/t-arm: Update dependency rules.
24631 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
24632 of processing .def files.
24633 * config/arm/genopt.sh: Deleted.
24634 * config/arm/gentune.sh: Deleted.
24635 * config/arm/arm-cores.def: Deleted.
24636 * config/arm/arm-arches.def: Deleted.
24637 * config/arm/arm-fpus.def: Deleted.
24638 * config/arm/arm-tune.md: Regenerated.
24639 * config/arm/arm-tables.opt: Regenerated.
24640 * config/arm/arm-cpu.h: New generated file.
24641 * config/arm/arm-cpu-data.h: New generated file.
24642 * config/arm/arm-cpu-cdata.h: New generated file.
24644 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
24647 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
24649 (input_varpool_node): Unpack dynamically_initialized bit.
24651 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
24653 PR rtl-optimization/79032
24654 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
24655 the alignment of the adjusted memory reference against that of MODE,
24656 instead of the alignment of the original memory reference.
24658 2017-01-11 Martin Jambor <mjambor@suse.cz>
24660 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
24662 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
24663 decorated functions.
24665 2017-01-11 Richard Biener <rguenther@suse.de>
24667 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
24668 set range/nonnull info for PHI results. Do not set it on
24669 stmts marked for removal.
24671 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
24673 * expr.c (store_field): In the bitfield case, fetch the return value
24674 from the registers before applying a single big-endian adjustment.
24675 Always do a final load for a BLKmode value not larger than a word.
24677 2017-01-10 David Malcolm <dmalcolm@redhat.com>
24680 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
24681 that we correctly handle column numbers greater than
24682 LINE_MAP_MAX_COLUMN_NUMBER.
24684 2017-01-10 Martin Sebor <msebor@redhat.com>
24686 PR middle-end/78245
24687 * gimple-ssa-sprintf.c (get_destination_size): Call
24688 {init,fini}object_sizes.
24689 * tree-object-size.c (addr_object_size): Adjust.
24690 (pass_through_call): Adjust.
24691 (pass_object_sizes::execute): Adjust.
24692 * tree-object-size.h (fini_object_sizes): Declare.
24694 2017-01-10 Martin Sebor <msebor@redhat.com>
24696 PR tree-optimization/78775
24697 * builtins.c (get_size_range): Move...
24698 * calls.c: ...to here.
24699 (alloc_max_size): Accept zero argument.
24700 (operand_signed_p): Remove.
24701 (maybe_warn_alloc_args_overflow): Call get_size_range.
24702 * calls.h (get_size_range): Declare.
24704 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
24706 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
24707 from TI's devices.csv file as of September 2016.
24708 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
24710 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
24712 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
24713 * doc/invoke.texi: Likewise.
24714 * doc/md.texi: Likewise.
24715 * doc/objc.texi: Likewise.
24717 2017-01-10 Joshua Conner <joshconner@google.com>
24719 * config/arm/fuchsia-elf.h: New file.
24720 * config/fuchsia.h: New file.
24721 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
24722 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
24724 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
24726 2016-01-10 Richard Biener <rguenther@suse.de>
24728 PR tree-optimization/79034
24729 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24730 Propagate out degenerate PHIs in the joiner.
24732 2017-01-10 Martin Liska <mliska@suse.cz>
24734 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
24735 (sort_congruence_classes_by_decl_uid): Likewise.
24736 (sort_congruence_class_groups_by_decl_uid): Likewise.
24737 (sem_item_optimizer::merge_classes): Sort class, groups in these
24738 classes and members in the groups by DECL_UID of declarations.
24739 This would make merge operations stable.
24741 2017-01-10 Martin Liska <mliska@suse.cz>
24743 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
24744 usage of m_classes_vec.
24745 (sem_item_optimizer::~sem_item_optimizer): Likewise.
24746 (sem_item_optimizer::get_group_by_hash): Likewise.
24747 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
24748 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
24749 (sem_item_optimizer::verify_classes): Likewise.
24750 (sem_item_optimizer::process_cong_reduction): Likewise.
24751 (sem_item_optimizer::dump_cong_classes): Likewise.
24752 (sem_item_optimizer::merge_classes): Likewise.
24753 * ipa-icf.h (congruence_class_hash): Rename from
24754 congruence_class_group_hash. Remove declaration of m_classes_vec.
24756 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
24758 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
24759 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
24760 * config.gcc: Add avx512vpopcntdqintrin.h.
24761 * config/i386/avx512vpopcntdqintrin.h: New.
24762 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
24763 * config/i386/i386-builtin-types.def: Add new types.
24764 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
24765 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
24766 __builtin_ia32_vpopcountq_v8di_mask): New.
24767 * config/i386/i386-c.c (ix86_target_macros_internal): Define
24768 __AVX512VPOPCNTDQ__.
24769 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
24770 (PTA_AVX512VPOPCNTDQ): Define.
24771 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
24772 TARGET_AVX512VPOPCNTDQ_P): Define.
24773 * config/i386/i386.opt: Add mavx512vpopcntdq.
24774 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
24775 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
24777 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24779 PR middle-end/77484
24780 * predict.def (PRED_CALL): Set to 67.
24782 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
24784 * expr.c (store_field): In the bitfield case, if the value comes from
24785 a function call and is of an aggregate type returned in registers, do
24786 not modify the field mode; extract the value in all cases if the mode
24787 is BLKmode and the size is not larger than a word.
24789 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
24792 * config/i386/cpuid.h: Fix undefined behavior.
24794 2017-01-04 Jeff Law <law@redhat.com>
24796 PR tree-optimization/79007
24797 PR tree-optimization/67955
24798 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
24799 conservative for pt.null when flag_non_call_exceptions is on.
24801 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24803 PR translation/79019
24804 PR translation/79020
24805 * params.def (PARAM_INLINE_MIN_SPEEDUP,
24806 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
24807 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
24809 * config/avr/avr.opt (maccumulate-args): Likewise.
24810 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
24811 * common.opt (freport-bug): Likewise.
24812 * cif-code.def (CIF_FINAL_ERROR): Likewise.
24813 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
24814 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
24815 translatable string.
24816 * config/i386/i386.c (function_value_32): Likewise.
24817 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
24818 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
24820 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
24821 * common/config/msp430/msp430-common.c (msp430_handle_option):
24823 * symtab.c (symtab_node::verify_base): Likewise.
24824 * opts.c (set_debug_level): Likewise.
24825 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
24826 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
24827 missing whitespace to translatable strings.
24828 * config/avr/avr.md (bswapsi2): Fix typo in comment.
24829 * config/sh/superh.h: Likewise.
24830 * config/i386/xopintrin.h: Likewise.
24831 * config/i386/znver1.md: Likewise.
24832 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
24833 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
24834 * double-int.h (struct double_int): Likewise.
24835 * double-int.c (div_and_round_double): Likewise.
24836 * wide-int.cc: Likewise.
24837 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
24838 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
24839 * cfgcleanup.c (crossjumps_occured): Renamed to ...
24840 (crossjumps_occurred): ... this.
24841 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
24844 PR tree-optimization/78899
24845 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
24846 returning bool return struct loop *, NULL for failure and the new
24848 (versionable_outer_loop_p): Don't version outer loop if it has
24849 dont_vectorized bit set.
24850 (tree_if_conversion): When versioning outer loop, ensure
24851 tree_if_conversion is performed also on the inner loop of the
24852 non-vectorizable outer loop copy.
24853 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
24854 LOOP_VECTORIZED in inner loop of the scalar outer loop and
24855 prevent vectorization of it.
24856 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
24857 the outer loop vectorization of the non-scalar version is attempted
24858 before vectorization of the inner loop in scalar version. If
24859 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
24860 vectorization of its inner loop.
24861 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
24862 has 2 inner loops, rename also on edges from bb whose single pred
24863 is outer_loop->header. Fix typo in function comment.
24865 2017-01-09 Martin Sebor <msebor@redhat.com>
24868 * asan.c (asan_emit_stack_protection): Increase local buffer size
24869 to avoid snprintf truncation warning.
24871 2017-01-09 Andrew Pinski <apinski@cavium.com>
24873 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
24874 to reference thunderx2t99 for the tuning structure
24875 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
24877 (thunderx2t99_extra_costs): This.
24878 * config/aarch64/aarch64-tune.md: Regenerate.
24879 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
24880 (vulcan_addrcost_table): This.
24881 (vulcan_regmove_cost): Rename to ...
24882 (thunderx2t99_regmove_cost): This.
24883 (vulcan_vector_cost): Rename to ...
24884 (thunderx2t99_vector_cost): this.
24885 (vulcan_branch_cost): Rename to ...
24886 (thunderx2t99_branch_cost): This.
24887 (vulcan_tunings): Rename to ...
24888 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
24889 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
24891 2017-01-09 Martin Jambor <mjambor@suse.cz>
24895 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
24896 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
24897 (propagate_vr_accross_jump_function): Use the above function for all
24898 value range computations for pass-through jump functions and type
24899 converasion from explicit value range values.
24900 (ipcp_propagate_stage): Do not attempt to deduce types of formal
24901 parameters from TYPE_ARG_TYPES.
24902 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
24903 (ipa_write_node_info): Stream type of the actual argument.
24904 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
24906 2017-01-09 Martin Liska <mliska@suse.cz>
24909 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
24910 (lookup_compiler): Do not show error message with have_E.
24912 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24914 PR tree-optimization/78938
24915 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
24916 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
24917 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
24918 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
24921 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24923 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
24926 2017-01-09 Richard Biener <rguenther@suse.de>
24928 PR tree-optimization/78997
24929 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
24930 name condition properly.
24932 2017-01-09 Richard Biener <rguenther@suse.de>
24935 * dwarf2out.c (is_cxx): New overload with context.
24936 (is_naming_typedef_decl): Use it.
24938 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24940 * invoke.texi (Option Summary): Correct spacing in option lists
24941 and add line breaks to fix over-long lines.
24943 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24945 PR middle-end/17660
24947 * extend.texi (Common Variable Attributes): Add xref to GCC
24948 Internals manual to explain mode attribute keywords.
24950 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24953 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
24954 and Preprocessor Options.
24955 (Options for Linking): Document -pthread here....
24956 (RS/6000 and PowerPC Options): ...not here.
24957 (Solaris 2 Options): ...or here.
24958 * doc/cppopts.texi: Document -pthread.
24960 2017-01-08 Martin Sebor <msebor@redhat.com>
24962 PR middle-end/77708
24963 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
24964 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
24965 New member functions.
24966 (format_directive): Used them.
24968 (pass_sprintf_length::handle_gimple_call): Same.
24969 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
24970 to avoid truncation for any argument.
24971 (extract_affine_mul): Same.
24972 * tree.c (get_file_function_name): Same.
24974 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24976 PR middle-end/77484
24977 * predict.def (PRED_INDIR_CALL): Set to 86.
24979 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24981 PR preprocessor/54124
24982 * doc/cppopts.texi: Reformat -d subtable to list the full name
24983 of the options. Add cross-reference to the docs for the general
24984 compiler -d options.
24985 * doc/invoke.texi (Developer Options): Add cross-reference to the
24986 preprocessor-specific -d option documentation.
24988 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24990 PR preprocessor/13498
24991 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
24992 redudant material, and reflect new command-line options.
24993 (System Headers): Likewise.
24995 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24997 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
24998 -isystem, and -idirafter. Copy-edit.
24999 * doc/cppopts.texi: Copy-edit. Remove contradiction about
25000 default for -ftrack-macro-expansion. Delete obsolete and
25001 badly-formatted implementation details about -fdebug-cpp output.
25002 * doc/cppwarnopts.texi: Copy-edit.
25004 2017-01-07 David Malcolm <dmalcolm@redhat.com>
25007 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25008 that the transition from a max line width >= 1<<10 to narrower
25009 lines works correctly.
25011 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
25013 * doc/options.texi (PerFunction): New.
25014 * opt-functions.awk (switch_flags): Map both Optimization and
25015 PerFunction to CL_OPTIMIZATION.
25016 * opth-gen.awk: Test for PerFunction flag along with
25018 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
25019 it only when the latter is present. Skip those that don't in
25020 the hash function generator.
25021 * common.opt (fvar-tracking): Mark as PerFunction instead of
25023 (fvar-tracking-assignments): Likewise.
25024 (fvar-tracking-assignments-toggle): Likewise.
25025 (fvar-tracking-uninit): Likewise.
25027 2017-01-07 Jakub Jelinek <jakub@redhat.com>
25029 PR translation/79018
25030 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
25033 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
25036 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
25037 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
25038 TARGET_LONG_JUMP_TABLE_OFFSETS.
25039 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
25040 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
25041 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
25042 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
25043 * config/m68k/m68k.md (tablejump expander): Likewise.
25044 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
25045 TARGET_LONG_JUMP_TABLE_OFFSETS.
25046 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
25047 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
25049 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
25050 David Holsgrove <david.holsgrove@xilinx.com>
25052 * common/config/microblaze/microblaze-common.c
25053 (TARGET_EXCEPT_UNWIND_INFO): Remove.
25054 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
25056 * config/microblaze/microblaze.c (microblaze_must_save_register)
25057 (microblaze_expand_epilogue, microblaze_return_addr): Handle
25059 (microblaze_eh_return): New function.
25060 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
25061 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
25062 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
25063 * config/microblaze/microblaze.md (eh_return): New pattern.
25065 2017-01-06 Jakub Jelinek <jakub@redhat.com>
25067 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
25068 GCC_DIAGNOSTIC_STRINGIFY): Define.
25070 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
25072 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25074 * config/arm/arm.md (<mcrr>): New.
25076 * config/arm/arm.c (arm_arch5te): New.
25077 (arm_option_override): Set arm_arch5te.
25078 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
25080 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
25081 (arm_mcrr_qualifiers): ... this. New.
25082 (MRRC_QUALIFIERS): Define to...
25083 (arm_mrrc_qualifiers): ... this. New.
25084 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
25086 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
25087 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
25088 (MRRCI, mrrc, MRRC): New.
25089 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
25090 VUNSPEC_MRRC2): New.
25092 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25094 * config/arm/arm.md (<mcr>): New.
25096 * config/arm/arm.c (arm_coproc_builtin_available): Add
25097 support for mcr, mrc, mcr2 and mrc2.
25098 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
25099 (arm_mcr_qualifiers): ... this. New.
25100 (MRC_QUALIFIERS): Define to ...
25101 (arm_mrc_qualifiers): ... this. New.
25102 (MCR_QUALIFIERS): Define to ...
25103 (arm_mcr_qualifiers): ... this. New.
25104 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
25106 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
25107 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
25108 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
25109 VUNSPEC_MRC2): New.
25111 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25113 * config/arm/arm.md (*ldc): New.
25117 * config/arm/arm.c (arm_coproc_builtin_available): Add
25118 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
25119 (arm_coproc_ldc_stc_legitimate_address): New.
25120 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
25121 'qualifier_const_pointer'.
25122 (LDC_QUALIFIERS): Define to...
25123 (arm_ldc_qualifiers): ... this. New.
25124 (STC_QUALIFIERS): Define to...
25125 (arm_stc_qualifiers): ... this. New.
25126 * config/arm/arm-protos.h
25127 (arm_coproc_ldc_stc_legitimate_address): New.
25128 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
25129 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
25130 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
25131 stc2, stcl, stc2l): New.
25132 * config/arm/constraints.md (Uz): New.
25133 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
25134 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
25135 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
25136 VUNSPEC_STC2L): New.
25138 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25140 * config/arm/arm.md (<cdp>): New.
25141 * config/arm/arm.c (neon_const_bounds): Rename this ...
25142 (arm_const_bounds): ... this.
25143 (arm_coproc_builtin_available): New.
25144 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
25145 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
25146 (CDP_QUALIFIERS): Define to...
25147 (arm_cdp_qualifiers): ... this. New.
25149 (arm_expand_builtin_args): Add case for 6 arguments.
25150 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
25151 (arm_const_bounds): ... this.
25152 (arm_coproc_builtin_available): New.
25153 * config/arm/arm_acle.h (__arm_cdp): New.
25155 * config/arm/arm_acle_builtins.def (cdp): New.
25157 * config/arm/iterators.md (CDPI,CDP,cdp): New.
25158 * config/arm/neon.md: Rename all 'neon_const_bounds' to
25159 'arm_const_bounds'.
25160 * config/arm/types.md (coproc): New.
25161 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
25162 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
25163 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
25164 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
25166 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25168 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
25169 (UBINOP_QUALIFIERS): New.
25171 (acle_builtin_data): New. Change comment.
25172 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
25173 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
25174 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
25175 arm_acle_builtins.def.
25176 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
25177 (arm_init_acle_builtins): New.
25178 (CRC32_BUILTIN): Remove.
25179 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
25180 crc32cb, crc32ch and crc32cw.
25181 (arm_init_crc32_builtins): Remove.
25182 (arm_init_builtins): Use arm_init_acle_builtins rather
25183 than arm_init_crc32_builtins.
25184 (arm_expand_acle_builtin): New.
25185 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
25186 * config/arm/arm_acle_builtins.def: New.
25188 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25190 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
25191 (arm_builtin_datum): ... this.
25192 (arm_init_neon_builtin): Rename to ...
25193 (arm_init_builtin): ... this. Add a new parameters PREFIX
25194 and USE_SIG_IN_NAME.
25195 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
25196 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
25197 'arm_builtin_datum'.
25198 (arm_init_vfp_builtins): Likewise.
25199 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
25200 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
25201 (arm_expand_neon_args): Rename to ...
25202 (arm_expand_builtin_args): ... this. Rename builtin_arg
25203 enum values and differentiate between ARG_BUILTIN_MEMORY
25204 and ARG_BUILTIN_NEON_MEMORY.
25205 (arm_expand_neon_builtin_1): Rename to ...
25206 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
25207 values, arm_expand_builtin_args and add bool parameter NEON.
25208 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
25209 (arm_expand_vfp_builtin): Likewise.
25210 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
25212 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25214 PR middle-end/77484
25215 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
25216 * predict.c (tree_estimate_probability_bb): Reverse direction of
25217 polymorphic call predictor.
25219 2017-01-06 David Malcolm <dmalcolm@redhat.com>
25221 * passes.c (execute_one_pass): Split out pass-skipping logic into...
25222 (determine_pass_name_match): ...this new function and...
25223 (should_skip_pass_p): ...this new function.
25225 2017-01-06 Nathan Sidwell <nathan@acm.org>
25227 * ipa-visibility.c (function_and_variable_visibility): Reformat
25228 comments and long lines. Remove extrneous if.
25229 * symtab.c (symtab_node::make_decl_local): Fix code format.
25230 (symtab_node::set_section_for_node): Fix comment typo.
25232 2017-01-06 Martin Liska <mliska@suse.cz>
25235 * lra-constraints.c: Rename invariant to lra_invariant.
25236 * predict.c (set_even_probabilities): Initialize e to NULL.
25238 2017-01-05 Martin Sebor <msebor@redhat.com>
25240 PR tree-optimization/78910
25241 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
25242 (format_integer): Correct off-by-one error in the handling
25243 of precision with negative numbers in signed conversions..
25245 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
25247 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
25249 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25251 PR tree-optimization/71016
25252 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
25253 factor_out_conditional_conversion. Formatting fix.
25254 (factor_out_conditional_conversion): Add cond_stmt argument.
25255 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
25256 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
25259 2017-01-05 David Malcolm <dmalcolm@redhat.com>
25261 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
25262 read-rtl-function.o, and selftest-rtl.o.
25263 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
25264 (selftest::aarch64_test_loading_full_dump): New function.
25265 (selftest::aarch64_run_selftests): New function.
25266 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
25267 selftest::aarch64_run_selftests.
25268 * config/i386/i386.c
25269 (selftest::ix86_test_loading_dump_fragment_1): New function.
25270 (selftest::ix86_test_loading_call_insn): New function.
25271 (selftest::ix86_test_loading_full_dump): New function.
25272 (selftest::ix86_test_loading_unspec): New function.
25273 (selftest::ix86_run_selftests): Call the new functions.
25274 * emit-rtl.c (maybe_set_max_label_num): New function.
25275 * emit-rtl.h (maybe_set_max_label_num): New decl.
25276 * function.c (instantiate_decls): Guard call to
25277 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
25278 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
25280 * gensupport.c (gen_reader::gen_reader): Pass "false"
25281 for new "compact" param of rtx_reader.
25282 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
25283 rather than an empty string for NULL strings.
25284 * read-md.c: Potentially include config.h rather than bconfig.h.
25285 Wrap include of errors.h with #ifdef GENERATOR_FILE.
25286 (have_error): New global, copied from errors.c.
25287 (md_reader::read_name): Rename to...
25288 (md_reader::read_name_1): ...this, adding "out_loc" param,
25289 and converting "missing name or number" to returning false, rather
25291 (md_reader::read_name): Reimplement in terms of read_name_1.
25292 (md_reader::read_name_or_nil): New function.
25293 (md_reader::read_string): Handle "(nil)" by returning NULL.
25294 (md_reader::md_reader): Add new param "compact".
25295 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
25296 (md_reader::read_file): New method.
25297 * read-md.h (md_reader::md_reader): Add new param "compact".
25298 (md_reader::read_file): New method.
25299 (md_reader::is_compact): New accessor.
25300 (md_reader::read_name): Convert return type from void to file_location.
25301 (md_reader::read_name_or_nil): New decl.
25302 (md_reader::read_name_1): New decl.
25303 (md_reader::m_compact): New field.
25304 (noop_reader::noop_reader): Pass "false" for new "compact" param
25306 (rtx_reader::rtx_reader): Add new "compact" param.
25307 (rtx_reader::read_rtx_operand): Make virtual and convert return
25308 type from void to rtx.
25309 (rtx_reader::read_until): New decl.
25310 (rtx_reader::handle_any_trailing_information): New virtual function.
25311 (rtx_reader::postprocess): New virtual function.
25312 (rtx_reader::finalize_string): New virtual function.
25313 (rtx_reader::m_in_call_function_usage): New field.
25314 (rtx_reader::m_reuse_rtx_by_id): New field.
25315 * read-rtl-function.c: New file.
25316 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
25317 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
25318 (selftest::verify_three_block_rtl_cfg): New decl.
25319 * read-rtl-function.h: New file.
25320 * read-rtl.c: Potentially include config.h rather than bconfig.h.
25321 For host, include function.h, memmodel.h, and emit-rtl.h.
25322 (one_time_initialization): New function.
25323 (struct compact_insn_name): New struct.
25324 (compact_insn_names): New array.
25325 (find_code): Handle insn codes in compact dumps.
25326 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
25327 (bind_subst_iter_and_attr): Likewise.
25328 (add_condition_to_string): Likewise.
25329 (add_condition_to_rtx): Likewise.
25330 (apply_attribute_uses): Likewise.
25331 (add_current_iterators): Likewise.
25332 (apply_iterators): Likewise.
25333 (initialize_iterators): Guard usage of apply_subst_iterator with
25334 #ifdef GENERATOR_FILE.
25335 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
25336 (md_reader::read_mapping): Likewise.
25337 (add_define_attr_for_define_subst): Likewise.
25338 (add_define_subst_attr): Likewise.
25339 (read_subst_mapping): Likewise.
25340 (check_code_iterator): Likewise.
25341 (rtx_reader::read_rtx): Likewise. Move one-time initialization
25343 (one_time_initialization): New function.
25344 (rtx_reader::read_until): New method.
25345 (read_flags): New function.
25346 (parse_reg_note_name): New function.
25347 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
25348 Handle reuse_rtx ids.
25349 Wrap iterator lookup within #ifdef GENERATOR_FILE.
25350 Add parsing support for RTL dumps, mirroring the special-cases in
25351 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
25352 values, and calling handle_any_trailing_information.
25353 (rtx_reader::read_rtx_operand): Convert return type from void
25354 to rtx, returning return_rtx. Handle case 'e'. Call
25355 finalize_string on XSTR and XTMPL fields.
25356 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
25357 "(nil)" values were omitted. Call the postprocess vfunc on the
25359 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
25360 class ctor. Initialize m_in_call_function_usage. Call
25361 one_time_initialization.
25362 * rtl-tests.c (selftest::test_uncond_jump): Call
25363 set_new_first_and_last_insn.
25364 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
25365 * selftest-rtl.c: New file.
25366 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
25367 (selftest::get_insn_by_uid): New decl.
25368 * selftest-run-tests.c (selftest::run_tests): Call
25369 read_rtl_function_c_tests.
25370 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
25371 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
25374 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
25376 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
25377 operands in a special way. Assert that pos+len <= mode precision.
25379 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25381 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
25382 3 argument Alias with unlimited for the negative form.
25383 (fno-vect-cost-model): Removed.
25385 2017-01-05 Martin Liska <mliska@suse.cz>
25387 * hsa-gen.c (gen_hsa_divmod): New function.
25388 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
25390 2017-01-05 Martin Liska <mliska@suse.cz>
25393 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
25396 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25398 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
25399 small constant length operands.
25401 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25403 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
25404 between loop iterations.
25406 2017-01-05 Martin Liska <mliska@suse.cz>
25409 * gimplify.c (gimplify_decl_expr): Compare to
25410 asan_poisoned_variables instread of checking flags.
25411 (gimplify_target_expr): Likewise.
25412 (gimplify_expr): Likewise.
25413 (gimplify_function_tree): Conditionally initialize
25414 asan_poisoned_variables.
25416 2017-01-04 Jeff Law <law@redhat.com>
25418 PR tree-optimizatin/78812
25419 * rtl.h (contains_mem_rtx_p): Prototype.
25420 * ifcvt.c (containts_mem_rtx_p): Move from here to...
25421 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
25422 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
25423 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
25424 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
25426 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25428 * input.c (assert_char_at_range): Default-initialize actual_range.
25430 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25432 * df-scan.c (df_ref_create_structure): Make regno unsigned,
25433 to match the caller.
25435 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25437 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
25438 insns after final jump in test to emit dummy move.
25440 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25442 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
25443 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
25445 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25447 * multiple_target.c (create_dispatcher_calls): Init e_next.
25448 * tree-ssa-loop-split.c (split_loop): Init border.
25449 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
25452 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
25457 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
25458 (altivec_register_operand): Do not return true if the operand
25459 contains a SUBREG mixing SImode and SFmode.
25460 (vsx_register_operand): Likewise.
25461 (vsx_reg_sfsubreg_ok): New predicate.
25462 (vfloat_operand): Do not return true if the operand contains a
25463 SUBREG mixing SImode and SFmode.
25464 (vint_operand): Likewise.
25465 (vlogical_operand): Likewise.
25466 (gpc_reg_operand): Likewise.
25467 (int_reg_operand): Likewise.
25468 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
25469 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
25470 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
25472 (rs6000_emit_move_si_sf_subreg): New helper function.
25473 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
25474 fixup SUBREGs involving SImode and SFmode.
25475 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
25476 numbers for the new peephole2 optimization.
25477 (peephole2 for SFmode unions): New peephole2 to optimize cases in
25478 the GLIBC math library that do AND/IOR/XOR operations on single
25479 precision floating point.
25480 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
25481 target macros to say whether we need to avoid SUBREGs mixing
25483 (TARGET_ALLOW_SF_SUBREG): Likewise.
25484 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
25485 (UNSPEC_SI_FROM_SF): Likewise.
25486 (iorxor): Change spacing.
25487 (and_ior_xor): New iterator for AND, IOR, and XOR.
25488 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
25489 (movdi_from_sf_zero_ext): Likewise.
25490 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
25491 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
25492 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
25493 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
25494 (fms<mode>4): Likewise.
25495 (fnma<mode>4): Likewise.
25496 (fnms<mode>4): Likewise.
25497 (nfma<mode>4): Likewise.
25498 (nfms<mode>4): Likewise.
25500 2017-01-04 Marek Polacek <polacek@redhat.com>
25503 * doc/invoke.texi: Document -Wpointer-compare.
25505 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25507 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
25510 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
25511 descriptions for -gdwarf-5 and emit them as uleb128 instead of
25514 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
25517 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
25518 documentation of the powerpc_popcntb_ok attribute.
25519 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25520 code to issue warning messages if a requested CPU configuration is
25521 not supported by the binary (assembler and loader) toolchain.
25522 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
25523 made to define a built-in function that has been disabled.
25524 (paired_init_builtins): Add assertion to prevent ICE if attempt is
25525 made to define a built-in function that has been disabled.
25526 (altivec_init_builtins): Add comment explaining why definition
25527 of the DST built-in functions is not preceded by an assertion
25528 check. Add assertions to prevent ICE if attempts are made to
25529 define an altivec predicate or an abs* built-in function that has
25531 (htm_init_builtins): Add comment explaining why definition of the
25532 htm built-in functions is not preceded by an assertion check.
25534 2017-01-04 Jeff Law <law@redhat.com>
25536 PR tree-optimizatin/67955
25537 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
25538 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
25539 the points-to solution does not include pt_null. Use DECL_PT_UID
25542 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
25544 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
25545 Use gen_int_mode instead of gen_lopwart for const_int operands.
25547 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25549 PR tree-optimization/71563
25550 * match.pd: Simplify X << Y into X if Y is known to be 0 or
25551 out of range value - has low bits known to be zero.
25553 2017-01-04 Alan Modra <amodra@gmail.com>
25555 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
25556 * configure: Regenerate.
25557 * config.in: Regenerate.
25559 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25562 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
25563 a substring of the message, but strcmp with the whole message. Ifdef
25564 ENABLE_NLS, translate the message first using dgettext.
25566 2017-01-03 Jeff Law <law@redhat.com>
25568 PR tree-optimizatin/78856
25569 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
25570 (mark_threaded_blocks): Remove code to truncate thread paths that
25571 cross multiple loop headers. Instead invalidate the cached loop
25572 iteration information and handle case of a thread path walking
25573 into an irreducible region.
25575 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
25578 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
25579 assertions. Add support for doing the signbit if the IEEE 128-bit
25580 floating point value is in a GPR.
25581 * config/rs6000/rs6000.md (Fsignbit): Delete.
25582 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
25583 Update the length attribute if the value is in a GPR.
25584 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
25585 the sign or zero extension instruction, since the value is always 0/1.
25586 (signbit<mode>2_dm2): Delete using <Fsignbit>.
25589 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
25590 extracting SImode to a GPR register so that we can generate a
25591 store, limit the vector to be in a traditional Altivec register
25592 for the vextuwrx instruction.
25594 2017-01-03 Ian Lance Taylor <iant@google.com>
25596 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
25598 2017-01-03 Martin Sebor <msebor@redhat.com>
25600 PR tree-optimization/78696
25601 * gimple-ssa-sprintf.c (format_floating): Correct handling of
25602 precision. Use MPFR for %f for greater fidelity. Correct handling
25604 (pass_sprintf_length::compute_format_length): Set width and precision
25605 specified by asrerisk to void_node for vararg functions.
25606 (try_substitute_return_value): Adjust dump output.
25608 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
25610 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
25612 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
25614 * doc/invoke.texi (SPARC options): Document -mlra as the default.
25615 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
25616 -mlra/-mno-lra was passed to the compiler.
25618 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
25620 PR rtl-optimization/65618
25621 * emit-rtl.c (try_split): Move initialization of "before" and
25622 "after" to just before the call to emit_insn_after_setloc.
25624 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
25626 * doc/md.texi (Standard Names): Remove reference to Java frontend.
25628 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
25630 * dwarf2out.c (gen_enumeration_type_die): When
25631 -gno-strict-dwarf, add a DW_AT_encoding attribute.
25633 2017-01-03 Jakub Jelinek <jakub@redhat.com>
25635 PR tree-optimization/78965
25636 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
25637 Change first argument from const call_info & to call_info &. For %n
25638 set info.nowrite to false.
25640 PR middle-end/78901
25641 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
25642 possibly throwing calls.
25644 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
25645 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
25646 and fns handling, rather than in a separate case SSA_NAME.
25648 2017-01-02 Jeff Law <law@redhat.com>
25650 * config/darwin-driver.c (darwin_driver_init): Const-correctness
25651 fixes for first_period and second_period variables.
25653 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
25656 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
25657 (*insvqi_1): New insn pattern.
25658 (*insvqi_1_mem_rex64): Ditto.
25659 (*insvqi_2): Ditto.
25660 (*insvqi_3): Rename from *insvqi.
25662 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
25664 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
25666 * doc/cfg.texi (Edges): Remove reference to Java.
25667 (Maintaining the CFG): Ditto.
25669 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25671 PR middle-end/77674
25672 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
25673 transparent aliases.
25675 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25677 PR middle-end/77484
25678 * predict.def (PRED_CALL): Update hitrate.
25679 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
25680 * predict.c (tree_estimate_probability_bb): Split CALL predictor
25681 into direct/indirect/polymorphic variants.
25683 2017-01-01 Jakub Jelinek <jakub@redhat.com>
25685 Update copyright years.
25687 * gcc.c (process_command): Update copyright notice dates.
25688 * gcov-dump.c (print_version): Ditto.
25689 * gcov.c (print_version): Ditto.
25690 * gcov-tool.c (print_version): Ditto.
25691 * gengtype.c (create_file): Ditto.
25692 * doc/cpp.texi: Bump @copying's copyright year.
25693 * doc/cppinternals.texi: Ditto.
25694 * doc/gcc.texi: Ditto.
25695 * doc/gccint.texi: Ditto.
25696 * doc/gcov.texi: Ditto.
25697 * doc/install.texi: Ditto.
25698 * doc/invoke.texi: Ditto.
25700 Copyright (C) 2017 Free Software Foundation, Inc.
25702 Copying and distribution of this file, with or without modification,
25703 are permitted in any medium without royalty provided the copyright
25704 notice and this notice are preserved.