1 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
4 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
5 (*mulsi3_sp64): New instruction.
6 (mulsi3): New expander.
8 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
10 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
12 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14 * sancov.c: Include memmodel.h.
16 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
19 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
22 2017-09-07 Carl Love <cel@us.ibm.com>
24 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
27 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
29 * sancov.c: Include tm_p.h.
31 2017-09-07 Jakub Jelinek <jakub@redhat.com>
34 * output.h (switch_to_other_text_partition): New declaration.
35 * varasm.c (switch_to_other_text_partition): New function.
36 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
37 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
38 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
39 to the other text partition before emitting LCL label and switch back
40 after emitting the word after it.
42 2017-09-07 Richard Biener <rguenther@suse.de>
44 * passes.def (pass_split_crit_edges): Remove instance before PRE.
45 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
46 critical edges here, after loop init.
47 (pass_data_pre): Remove PROP_no_crit_edges flags.
48 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
49 for valueization of call args to avoid leaking VN_TOP.
50 (visit_use): Assert we do not visit default defs.
51 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
52 Use error_mark_node to more easily detect leaking VN_TOP.
53 All default-defs are varying, not VN_TOP. Mark them visited.
54 (run_scc_vn): Make code match comment.
56 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
58 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
59 OPTION_MASK_FLOAT128_KEYWORD.
60 (POWERPC_MASKS): Likewise.
61 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
62 support for the -mfloat128-type option, and make -mfloat128
63 default on PowerPC Linux systems. Define or undefine
64 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
65 Define __float128 to be __ieee128 if IEEE 128-bit support is
66 enabled, or undefine it.
67 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
68 Delete defining __FLOAT128_TYPE__.
69 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
70 -mfloat128-type option and make -mfloat128 default on PowerPC
72 (TARGET_FLOAT128_TYPE): Likewise.
73 (-mfloat128-type): Likewise.
74 * config/rs6000/rs6000.c (rs6000_option_override_internal):
75 Delete the -mfloat128-type option and make -mfloat128 default on
76 PowerPC Linux systems. Always use __ieee128 to be the keyword for
77 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
78 128-bit floating point is enabled. Change tests from using
79 -mfloat128-type to -mfloat128.
80 (rs6000_mangle_type): Use the correct mangling for the __float128
81 type even if normal long double is restricted to 64-bits.
82 (floatn_mode): Enable the _Float128 type by default on VSX Linux
84 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
85 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
86 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
88 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
89 documentation for -mfloat128.
91 2017-09-06 Olivier Hainque <hainque@adacore.com>
93 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
95 2017-09-06 Wish Wu <wishwu007@gmail.com>
96 Jakub Jelinek <jakub@redhat.com>
98 * asan.c (initialize_sanitizer_builtins): Add
99 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
100 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
101 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
102 BT_FN_VOID_UINT64_PTR variables.
103 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
104 (BT_FN_VOID_UINT16_UINT16): Likewise.
105 (BT_FN_VOID_UINT32_UINT32): Likewise.
106 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
107 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
108 (BT_FN_VOID_UINT64_PTR): Likewise.
109 * common.opt (flag_sanitize_coverage): New variable.
110 (fsanitize-coverage=trace-pc): Remove.
111 (fsanitize-coverage=): Add.
112 * flag-types.h (enum sanitize_coverage_code): New enum.
113 * fold-const.c (fold_range_test): Disable non-short-circuit
114 optimization if flag_sanitize_coverage.
115 (fold_truth_andor): Likewise.
116 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
117 * opts.c (COVERAGE_SANITIZER_OPT): Define.
118 (coverage_sanitizer_opts): New array.
119 (get_closest_sanitizer_option): Add OPTS argument, handle also
120 OPT_fsanitize_coverage_.
121 (parse_sanitizer_options): Adjusted to also handle
122 OPT_fsanitize_coverage_.
123 (common_handle_option): Add OPT_fsanitize_coverage_.
124 * sancov.c (instrument_comparison, instrument_switch): New function.
125 (sancov_pass): Add trace-cmp support.
126 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
127 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
128 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
129 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
130 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
131 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
132 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
133 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
134 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
135 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
137 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
139 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
140 error. Only quit immediately if parsing is complete.
141 (BEGIN): Initialize fatal_err and parse_done.
142 (begin fpu, end fpu): Check number of arguments.
143 (begin arch, end arch): Likewise.
144 (begin cpu, end cpu): Likewise.
145 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
146 (optalias): Likewise.
148 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
150 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
151 * config/arm/arm-isa.h: Delete. Move definitions to ...
152 * arm-cpus.in: ... here. Use new feature and fgroup values.
153 * config/arm/arm.c (arm_option_override): Use lower case for feature
155 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
156 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
157 * config/arm/parsecpu.awk (END): Add new command 'isa'.
159 (print_isa_bits_for): New function.
160 (gen_isa): New function.
161 (gen_comm_data): Use print_isa_bits_for.
162 (define feature): New keyword.
163 (define fgroup): New keyword.
164 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
165 (arm-isa.h): Add rule to generate file.
166 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
167 case for feature bit names.
169 2017-09-06 Richard Biener <rguenther@suse.de>
171 * tree-ssa-pre.c (NECESSARY): Remove.
172 (create_expression_by_pieces): Do not touch pass-local flags.
173 (insert_into_preds_of_block): Likewise.
174 (do_pre_regular_insertion): Likewise.
175 (eliminate_insert): Likewise.
176 (eliminate_dom_walker::before_dom_children): Likewise.
177 (fini_eliminate): Do not look at inserted_exprs.
178 (mark_operand_necessary): Remove.
179 (remove_dead_inserted_code): Replace with simple work-list
180 algorithm based on inserted_exprs and SSA uses.
181 (pass_pre::execute): Re-order fini_eliminate and
182 remove_dead_inserted_code.
184 2017-09-06 Olivier Hainque <hainque@adacore.com>
186 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
187 for VxWorks 7. Adjust surrounding comments.
189 2017-09-06 Richard Biener <rguenther@suse.de>
191 * gimple-ssa-strength-reduction.c
192 (find_candidates_dom_walker::before_dom_children): Also allow
195 2017-09-06 Richard Biener <rguenther@suse.de>
197 PR tree-optimization/82108
198 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
199 for gap in the non-permutation SLP case.
201 2017-09-06 Martin Jambor <mjambor@suse.cz>
203 PR tree-optimization/82078
204 * tree-sra.c (sort_and_splice_var_accesses): Move call to
205 add_access_to_work_queue...
206 (build_accesses_from_assign): ...here.
207 (propagate_all_subaccesses): Make sure racc is the group
208 representative, if there is one.
210 2017-09-06 Jakub Jelinek <jakub@redhat.com>
213 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
216 2017-09-06 Richard Biener <rguenther@suse.de>
218 * gimple-ssa-strength-reduction.c
219 (find_candidates_dom_walker::before_doom_children): Use a
220 type and not a mode check.
222 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
225 * config/arm/predicates.md (arm_general_adddi_operand): Create new
227 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
229 2017-09-05 Jeff Law <law@redhat.com>
231 PR tree-optimization/64910
232 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
233 cases where we have 3 or more operands.
235 2017-09-05 Jakub Jelinek <jakub@redhat.com>
238 * omp-low.c (lower_omp_for): Recompute tree invariant if
239 gimple_omp_for_initial/final is ADDR_EXPR.
242 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
243 into gimple val before gimplification fo the COND_EXPR.
245 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
247 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
248 REGION_COPY argument.
249 (thread_through_all_blocks): Remove unused argument to
250 duplicate_thread_path.
252 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
253 Alan Hayward <alan.hayward@arm.com>
254 David Sherwood <david.sherwood@arm.com>
256 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
257 Take a scalar_mode rather than a machine_mode.
258 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
259 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
260 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
261 (aarch64_gen_adjusted_ldpstp): Likewise.
262 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
264 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
265 Alan Hayward <alan.hayward@arm.com>
266 David Sherwood <david.sherwood@arm.com>
268 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
269 Take a scalar_int_mode instead of a machine_mode.
270 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
271 (aarch64_output_scalar_simd_mov_immediate): Likewise.
272 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
273 (aarch64_simd_attr_length_rglist): Delete.
274 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
275 a scalar_int_mode instead of a machine_mode.
276 (aarch64_add_offset): Likewise.
277 (aarch64_internal_mov_immediate): Likewise
278 (aarch64_add_constant_internal): Likewise.
279 (aarch64_add_constant): Likewise.
280 (aarch64_movw_imm): Likewise.
281 (aarch64_rtx_arith_op_extract_p): Likewise.
282 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
283 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
284 Remove assert that the mode isn't a vector.
285 (aarch64_output_scalar_simd_mov_immediate): Likewise.
286 (aarch64_expand_mov_immediate): Update calls after above changes.
287 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
288 (aarch64_and_bitmask_imm): Check for scalar integer modes.
289 (aarch64_move_imm): Likewise.
290 (aarch64_can_const_movi_rtx_p): Likewise.
291 (aarch64_strip_extend): Likewise.
292 (aarch64_extr_rtx_p): Likewise.
293 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
294 a CONST_INT when the mode parameter is VOIDmode.
295 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
297 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
299 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
300 * stor-layout.c (bitwise_mode_for_mode): Likewise.
301 (bitwise_type_for_mode): Update accordingly.
303 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
305 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
306 * stor-layout.c (mode_for_size_tree): Likewise.
307 (mode_for_array): Update accordingly.
308 (layout_decl): Likewise.
309 (compute_record_mode): Likewise. Only set the mode once.
311 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
313 * target.def (get_mask_mode): Change return type to opt_mode.
315 * doc/tm.texi: Regenerate.
316 * targhooks.h (default_get_mask_mode): Return an opt_mode.
317 * targhooks.c (default_get_mask_mode): Likewise.
318 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
319 * optabs-query.c (can_vec_mask_load_store_p): Update use of
320 targetm.get_mask_mode.
321 * tree.c (build_truth_vector_type): Likewise.
323 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
325 * machmode.h (mode_for_vector): Return an opt_mode.
326 * stor-layout.c (mode_for_vector): Likewise.
327 (mode_for_int_vector): Update accordingly.
328 (layout_type): Likewise.
329 * config/i386/i386.c (emit_memmov): Likewise.
330 (ix86_expand_set_or_movmem): Likewise.
331 (ix86_expand_vector_init): Likewise.
332 (ix86_get_mask_mode): Likewise.
333 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
335 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
336 * expmed.c (extract_bit_field_1): Likewise.
337 * expr.c (expand_expr_real_2): Likewise.
338 * optabs-query.c (can_vec_perm_p): Likewise.
339 (can_vec_mask_load_store_p): Likewise.
340 * optabs.c (expand_vec_perm): Likewise.
341 * targhooks.c (default_get_mask_mode): Likewise.
342 * tree-vect-stmts.c (vectorizable_store): Likewise.
343 (vectorizable_load): Likewise.
344 (get_vectype_for_scalar_type_and_size): Likewise.
346 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
348 * machmode.h (mode_for_int_vector): New function.
349 * stor-layout.c (mode_for_int_vector): Likewise.
350 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
351 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
352 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
353 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
354 (s390_expand_vcond): Likewise.
356 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
358 * machmode.h (opt_machine_mode): New type.
359 (opt_mode<T>): Allow construction from anything that can be
361 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
362 (mode_for_size): Return an opt_machine_mode.
363 * stor-layout.c (mode_for_size): Likewise.
364 (mode_for_size_tree): Update call accordingly.
365 (bitwise_mode_for_mode): Likewise.
366 (make_fract_type): Likewise.
367 (make_accum_type): Likewise.
368 * caller-save.c (replace_reg_with_saved_mem): Update call
370 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
371 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
372 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
373 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
374 * expmed.c (extract_bit_field_1): Likewise.
375 * reload.c (get_secondary_mem): Likewise.
376 * varasm.c (assemble_integer): Likewise.
377 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
380 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
382 * machmode.h (decimal_float_mode_for_size): New function.
383 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
384 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
385 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
386 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
387 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
388 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
390 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
392 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
393 (get_builtin_sync_mode): Likewise.
394 (expand_ifn_atomic_compare_exchange): Likewise.
395 (expand_builtin_atomic_clear): Likewise.
396 (expand_builtin_atomic_test_and_set): Likewise.
397 (fold_builtin_atomic_always_lock_free): Likewise.
398 * calls.c (compute_argument_addresses): Likewise.
399 (emit_library_call_value_1): Likewise.
400 (store_one_arg): Likewise.
401 * combine.c (combine_instructions): Likewise.
402 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
403 * config/arm/arm.c (arm_function_value): Likewise.
404 (aapcs_allocate_return_reg): Likewise.
405 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
406 * config/i386/i386.c (construct_container): Likewise.
407 (ix86_gimplify_va_arg): Likewise.
408 (ix86_expand_sse_cmp): Likewise.
409 (emit_memmov): Likewise.
410 (emit_memset): Likewise.
411 (expand_small_movmem_or_setmem): Likewise.
412 (ix86_expand_pextr): Likewise.
413 (ix86_expand_pinsr): Likewise.
414 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
415 * config/microblaze/microblaze.c (microblaze_block_move_straight):
417 * config/mips/mips.c (mips_function_value_1) Likewise.
418 (mips_block_move_straight): Likewise.
419 (mips_expand_ins_as_unaligned_store): Likewise.
420 * config/powerpcspe/powerpcspe.c
421 (rs6000_darwin64_record_arg_advance_flush): Likewise.
422 (rs6000_darwin64_record_arg_flush): Likewise.
423 * config/rs6000/rs6000.c
424 (rs6000_darwin64_record_arg_advance_flush): Likewise.
425 (rs6000_darwin64_record_arg_flush): Likewise.
426 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
427 (sparc_function_value_1): Likewise.
428 * config/spu/spu.c (adjust_operand): Likewise.
429 (spu_emit_branch_or_set): Likewise.
430 (arith_immediate_p): Likewise.
431 * emit-rtl.c (gen_lowpart_common): Likewise.
432 * expr.c (expand_expr_real_1): Likewise.
433 * function.c (assign_parm_setup_block): Likewise.
434 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
435 * reload1.c (alter_reg): Likewise.
436 * stor-layout.c (mode_for_vector): Likewise.
437 (layout_type): Likewise.
439 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
441 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
442 (spu_convert_move): Likewise.
443 * lower-subreg.c (resolve_simple_move): Likewise.
445 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
448 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
449 define_insn to a define_expand.
450 (altivec_vsum2sws_direct): New define_insn.
451 (altivec_vsumsws): Convert from a define_insn to a define_expand.
453 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
455 * config/arm/arm.c (arm_option_params_internal): Improve setting of
458 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
463 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
464 realign stack if stack alignment needed is less than incoming
467 2017-09-05 Marek Polacek <polacek@redhat.com>
470 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
473 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
475 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
477 2017-09-05 Richard Biener <rguenther@suse.de>
479 PR tree-optimization/82084
480 * fold-const.c (can_native_encode_string_p): Handle wide characters.
482 2017-09-05 Richard Biener <rguenther@suse.de>
484 PR tree-optimization/82102
485 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
487 2017-09-05 Martin Liska <mliska@suse.cz>
489 PR tree-optimization/82032
490 * tree-cfg.c (generate_range_test): New function.
491 * tree-cfg.h (generate_range_test): Declared here.
492 * tree-cfgcleanup.c (convert_single_case_switch): New function.
493 (cleanup_control_expr_graph): Use it.
494 * tree-switch-conversion.c (try_switch_expansion): Remove
496 (emit_case_nodes): Use generate_range_test.
498 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
501 * config/i386/i386.md (*<btsc><mode>_mask): Add
502 TARGET_USE_BT to insn constraint.
503 (*btr<mode>_mask): Ditto.
505 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
507 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
508 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
510 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
513 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
514 TARGET_NEON and TARGET_IWMMXT.
515 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
516 TARGET_NEON and TARGET_IWMMXT.
517 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
519 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
521 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
522 (ix86_rewrite_tls_address): Ditto.
523 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
524 (ix86_rewrite_tls_address_1): Ditto.
525 (ix86_rewrite_tls_address): Ditto.
526 * config/i386/predicates.md (tls_address_pattern): New predicate.
527 * config/i386/i386.md (TLS address splitter): New splitter.
529 2017-09-04 Richard Biener <rguenther@suse.de>
531 PR tree-optimization/82084
532 * fold-const.h (can_native_encode_string_p): Declare.
533 * fold-const.c (can_native_encode_string_p): Factor out from ...
534 (native_encode_string): ... here.
535 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
536 vectorizing stores from constants we later cannot handle.
538 2017-09-04 Marek Polacek <polacek@redhat.com>
541 * doc/invoke.texi: Update -Wtautological-compare documentation.
543 2017-09-04 Jeff Law <law@redhat.com>
545 PR tree-optimization/64910
546 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
547 swap the first and last operand if the last is a constant.
549 2017-09-04 Marek Polacek <polacek@redhat.com>
552 * convert.c (do_narrow): When sanitizing signed integer overflows,
553 bail out for signed types.
554 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
556 2017-09-04 Richard Biener <rguenther@suse.de>
558 PR tree-optimization/82060
559 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
560 Move devirtualization after stmt folding and before EH/AB/noreturn
561 cleanup to get the stmt refs canonicalized. Use a bool instead
562 of gimple_modified_p since that doesn't work for NOPs. Schedule
563 NOPs generated by folding for removal.
565 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
566 Alan Hayward <alan.hayward@arm.com>
567 David Sherwood <david.sherwood@arm.com>
569 * coretypes.h (pad_direction): New enum.
570 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
571 (FUNCTION_ARG_PADDING): Likewise.
572 * target.def (function_arg_padding): New hook.
573 * targhooks.h (default_function_arg_padding): Declare.
574 * targhooks.c (default_function_arg_padding): New function.
575 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
576 (TARGET_FUNCTION_ARG_PADDING): ...this.
577 * doc/tm.texi: Regenerate.
578 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
579 instead of direction.
580 (compute_argument_addresses): Likewise.
581 (load_register_parameters): Likewise.
582 (emit_library_call_value_1): Likewise.
583 (store_one_arg): Use targetm.calls.function_arg_padding instead
584 of FUNCTION_ARG_PADDING.
585 (must_pass_in_stack_var_size_or_pad): Likewise.
586 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
587 (emit_group_store): Likewise.
588 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
589 instead of FUNCTION_ARG_PADDING.
590 (emit_push_insn): Likewise, and propagate enum change throughout
592 * function.h (direction): Delete.
593 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
595 * function.c (assign_parm_find_stack_rtl): Likewise.
596 (assign_parm_setup_block_p): Likewise.
597 (assign_parm_setup_block): Likewise.
598 (gimplify_parameters): Likewise.
599 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
600 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
602 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
603 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
604 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
605 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
606 (aarch64_function_arg_padding): ...this new function.
607 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
608 (TARGET_FUNCTION_ARG_PADDING): Redefine.
609 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
610 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
611 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
612 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
613 (arm_pad_arg_upward): Replace with...
614 (arm_function_arg_padding): ...this new function.
615 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
617 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
618 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
619 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
620 (ia64_hpux_function_arg_padding): Replace with...
621 (ia64_function_arg_padding): ...this new function. Use pad_direction
622 instead of direction. Check for TARGET_HPUX.
623 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
624 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
625 (iq2000_function_arg_padding): New function.
626 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
627 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
628 (mips_function_arg_padding): ...this new function.
629 (mips_pad_reg_upward): Update accordingly.
630 (TARGET_FUNCTION_ARG_PADDING): Redefine.
631 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
632 targetm.calls.function_arg_padding.
633 (FUNCTION_ARG_PADDING): Delete.
634 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
635 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
636 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
637 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
638 (nios2_block_reg_padding): Return pad_direction instead of direction.
639 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
640 instead of direction.
641 (nios2_function_arg_padding): Likewise. Make static.
642 (TARGET_FUNCTION_ARG_PADDING): Redefine.
643 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
644 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
645 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
646 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
647 (pa_function_arg_padding): Make static. Return pad_direction instead
649 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
650 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
651 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
652 instead of direction. Use targetm.calls.function_arg_padding.
653 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
654 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
655 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
656 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
657 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
659 (function_arg_padding): Rename to...
660 (rs6000_function_arg_padding): ...this. Make static. Return
661 pad_direction instead of direction.
662 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
663 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
664 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
665 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
666 instead of direction. Use targetm.calls.function_arg_padding.
667 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
668 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
669 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
670 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
671 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
672 (function_arg_padding): Rename to...
673 (rs6000_function_arg_padding): ...this. Make static. Return
674 pad_direction instead of direction.
675 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
676 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
677 * config/s390/s390.c (s390_function_arg_padding): New function.
678 (TARGET_FUNCTION_ARG_PADDING): Redefine.
679 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
680 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
681 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
682 (function_arg_padding): Rename to...
683 (sparc_function_arg_padding): ...this. Make static. Return
684 pad_direction instead of direction.
685 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
686 * config/spu/spu.c (spu_function_arg_padding): New function.
687 (TARGET_FUNCTION_ARG_PADDING): Redefine.
688 * system.h (FUNCTION_ARG_PADDING): Poison.
690 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
691 Alan Hayward <alan.hayward@arm.com>
692 David Sherwood <david.sherwood@arm.com>
694 * target.def (modes_tieable_p): New hook.
695 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
696 (TARGET_MODES_TIEABLE_P): ...this.
697 * doc/tm.texi.in: Regenerate.
698 * hooks.h (hook_bool_mode_mode_true): Declare.
699 * hooks.c (hook_bool_mode_mode_true): New function.
700 * combine.c (subst): Use targetm.modes_tieable_p instead of
702 * dse.c (find_shift_sequence): Likewise.
703 * expmed.c (extract_low_bits): Likewise.
704 * lower-subreg.c: Include target.h.
705 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
707 * rtlanal.c (rtx_cost): Likewise.
708 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
709 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
710 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
711 (TARGET_MODES_TIEABLE_P): Redefine.
712 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
713 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
714 (TARGET_MODES_TIEABLE_P): Redefine.
715 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
716 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
717 (arc_modes_tieable_p): New function.
718 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
719 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
720 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
721 (arm_modes_tieable_p): Make static.
722 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
723 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
724 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
725 (TARGET_MODES_TIEABLE_P): Redefine.
726 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
727 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
728 (TARGET_MODES_TIEABLE_P): Redefine.
729 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
730 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
731 (cr16_modes_tieable_p): New function.
732 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
733 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
734 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
735 (TRULY_NOOP_TRUNCATION): Update comment.
736 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
737 (TRULY_NOOP_TRUNCATION): Update comment.
738 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
739 (frv_modes_tieable_p): New function.
740 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
741 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
742 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
743 (TARGET_MODES_TIEABLE_P): Redefine.
744 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
745 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
746 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
747 (TARGET_MODES_TIEABLE_P): Redefine.
748 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
749 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
750 (ia64_modes_tieable_p): New function.
751 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
752 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
753 (iq2000_modes_tieable_p): New function.
754 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
755 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
756 (lm32_modes_tieable_p): New function.
757 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
758 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
759 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
760 (TARGET_MODES_TIEABLE_P): Redefine.
761 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
762 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
763 (m32r_modes_tieable_p): New function.
764 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
765 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
766 (m68k_modes_tieable_p): New function.
767 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
768 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
769 (mcore_modes_tieable_p): New function.
770 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
771 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
773 (TARGET_MODES_TIEABLE_P): Redefine.
774 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
775 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
776 * config/mips/mips.c (mips_modes_tieable_p): Make static.
777 (TARGET_MODES_TIEABLE_P): Redefine.
778 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
779 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
780 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
781 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
782 (mn10300_modes_tieable_p): ...this and make static.
783 (TARGET_MODES_TIEABLE_P): Redefine.
784 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
785 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
786 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
787 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
788 (msp430_modes_tieable_p): Make static.
789 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
790 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
791 (TARGET_MODES_TIEABLE_P): Redefine.
792 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
793 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
794 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
795 (TARGET_MODES_TIEABLE_P): Redefine.
796 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
797 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
798 * config/pa/pa.c (pa_modes_tieable_p): Make static.
799 (TARGET_MODES_TIEABLE_P): Redefine.
800 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
801 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
802 (pdp11_modes_tieable_p): New function.
803 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
804 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
805 (rs6000_modes_tieable_p): New function.
806 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
807 * config/powerpcspe/powerpcspe.md: Update comment.
808 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
809 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
810 (TARGET_MODES_TIEABLE_P): Redefine.
811 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
812 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
813 (rl78_modes_tieable_p): New function.
814 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
815 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
816 (rs6000_modes_tieable_p): New function.
817 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
818 * config/rs6000/rs6000.md: Update comment.
819 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
820 * config/rx/rx.c (rx_modes_tieable_p): New function.
821 (TARGET_MODES_TIEABLE_P): Redefine.
822 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
823 * config/s390/s390.c (s390_modes_tieable_p): New function.
824 (TARGET_MODES_TIEABLE_P): Redefine.
825 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
826 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
827 (sh_modes_tieable_p): New function.
828 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
829 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
830 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
831 (sparc_modes_tieable_p): Make static.
832 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
833 * config/spu/spu.c (spu_modes_tieable_p): New function.
834 (TARGET_MODES_TIEABLE_P): Redefine.
835 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
836 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
837 (TARGET_MODES_TIEABLE_P): Redefine.
838 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
839 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
840 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
841 * config/v850/v850.c (v850_modes_tieable_p): New function.
842 (TARGET_MODES_TIEABLE_P): Redefine.
843 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
844 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
845 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
846 (visium_modes_tieable_p): New function.
847 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
848 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
849 (xtensa_modes_tieable_p): New function.
850 * system.h (MODES_TIEABLE_P): Poison.
852 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
853 Alan Hayward <alan.hayward@arm.com>
854 David Sherwood <david.sherwood@arm.com>
856 * target.def (hard_regno_mode_ok): New hook.
857 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
858 (TARGET_HARD_REGNO_MODE_OK): ...this.
859 * doc/tm.texi.in: Regenerate.
860 * hooks.h (hook_bool_uint_mode_true): Declare.
861 * hooks.c (hook_bool_uint_mode_true): New function.
862 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
864 * genpreds.c (write_insn_preds_c): Add an include of target.h.
865 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
866 instead of HARD_REGNO_MODE_OK.
867 * caller-save.c: Include target.h.
868 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
870 * combine.c (can_combine_p): Likewise.
871 (combinable_i3pat): Likewise.
872 (can_change_dest_mode): Likewise.
873 * expr.c (init_expr_target): Likewise.
874 (convert_move): Likewise.
875 (convert_modes): Likewise.
876 * ira.c (setup_prohibited_class_mode_regs): Likewise.
877 (setup_prohibited_mode_move_regs): Likewise.
878 * ira.h (target_ira): Likewise.
879 * lra-assigns.c (find_hard_regno_for_1): Likewise.
880 * lra-constraints.c (process_alt_operands): Likewise.
881 (split_reg): Likewise.
882 * recog.c (peep2_find_free_register): Likewise.
883 * ree.c (combine_reaching_defs): Likewise.
884 * regcprop.c (maybe_mode_change): Likewise.
885 * reginfo.c (init_reg_sets_1): Likewise.
886 (choose_hard_reg_mode): Likewise.
887 (simplifiable_subregs): Likewise.
888 * regrename.c (check_new_reg_p): Likewise.
889 * reload.c (find_valid_class): Likewise.
890 (find_valid_class_1): Likewise.
891 (reload_inner_reg_of_subreg): Likewise.
892 (push_reload): Likewise.
893 (combine_reloads): Likewise.
894 (find_dummy_reload): Likewise.
895 (find_reloads): Likewise.
896 * reload1.c (find_reg): Likewise.
897 (set_reload_reg): Likewise.
898 (allocate_reload_reg): Likewise.
899 (choose_reload_regs): Likewise.
900 (reload_adjust_reg_for_temp): Likewise.
901 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
902 (simplify_subreg_regno): Likewise.
903 * sel-sched.c (init_regs_for_mode): Likewise.
904 * varasm.c (make_decl_rtl): Likewise.
905 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
906 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
908 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
909 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
910 (TARGET_HARD_REGNO_MODE_OK): Redefine.
911 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
912 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
913 (TARGET_HARD_REGNO_MODE_OK): Redefine.
914 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
915 (arc_mode_class): Delete.
916 (HARD_REGNO_MODE_OK): Delete.
917 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
918 (arc_hard_regno_mode_ok): Rename old array to...
919 (arc_hard_regno_mode_ok_modes): ...this.
920 (arc_conditional_register_usage): Update accordingly.
921 (arc_mode_class): Make static.
922 (arc_hard_regno_mode_ok): New function.
923 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
924 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
925 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
926 (arm_hard_regno_mode_ok): Make static.
927 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
929 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
930 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
931 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
933 (TARGET_HARD_REGNO_MODE_OK): Redefine.
934 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
935 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
936 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
937 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
938 (TARGET_HARD_REGNO_MODE_OK): Redefine.
939 * config/bfin/predicates.md (valid_reg_operand): Use
940 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
941 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
942 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
943 (TARGET_HARD_REGNO_MODE_OK): Redefine.
944 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
945 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
946 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
947 (cr16_hard_regno_mode_ok): Make static and return a bool.
948 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
949 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
950 (cris_hard_regno_mode_ok): New function.
951 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
952 (epiphany_mode_class): Delete.
953 (HARD_REGNO_MODE_OK): Delete.
954 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
955 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
956 (hard_regno_mode_ok): Rename to...
957 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
958 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
959 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
961 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
962 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
963 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
964 (frv_hard_regno_mode_ok): Make static and return a bool.
965 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
967 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
968 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
969 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
970 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
972 (TARGET_HARD_REGNO_MODE_OK): Redefine.
973 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
974 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
975 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
977 (TARGET_HARD_REGNO_MODE_OK): Redefine.
978 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
979 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
980 (ia64_hard_regno_mode_ok): New function.
981 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
982 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
983 (iq2000_hard_regno_mode_ok): New function.
984 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
985 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
986 (lm32_hard_regno_mode_ok): New function.
987 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
988 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
989 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
990 instead of HARD_REGNO_MODE_OK.
991 (m32c_hard_regno_ok): Rename to...
992 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
993 (m32c_cannot_change_mode_class): Update accordingly.
994 (TARGET_HARD_REGNO_MODE_OK): Redefine.
995 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
996 (m32r_mode_class): Delete.
997 (HARD_REGNO_MODE_OK): Delete.
998 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
999 (m32r_hard_regno_mode_ok): Rename to...
1000 (m32r_hard_regno_modes): ...this.
1001 (m32r_mode_class): Make static.
1002 (m32r_hard_regno_mode_ok): New function.
1003 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
1004 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
1005 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1006 (m68k_hard_regno_mode_ok): Make static.
1007 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
1008 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1009 (mcore_hard_regno_mode_ok): New function.
1010 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
1011 (HARD_REGNO_MODE_OK): Delete.
1012 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
1014 (microblaze_hard_regno_mode_ok_p): ...this and make static.
1015 (microblaze_hard_regno_mode_ok): New function.
1016 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1017 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
1018 (mips_hard_regno_mode_ok): Delete.
1019 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
1020 (mips_hard_regno_mode_ok_p): ...this and make static.
1021 (mips_hard_regno_mode_ok_p): Rename to...
1022 (mips_hard_regno_mode_ok_uncached): ...this.
1023 (mips_hard_regno_mode_ok): New function.
1024 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
1025 of HARD_REGNO_MODE_OK.
1026 (mips_option_override): Update after above name changes.
1027 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1028 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
1029 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
1030 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
1031 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
1032 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1033 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
1034 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
1035 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
1036 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1037 (msp430_hard_regno_mode_ok): Make static and return a bool.
1038 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
1039 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
1040 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
1042 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1043 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
1044 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
1045 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
1046 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
1047 (PA_HARD_REGNO_MODE_OK): ...this
1048 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
1049 (PA_HARD_REGNO_MODE_OK): ...this.
1050 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1051 (pa_hard_regno_mode_ok): New function.
1052 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
1053 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1054 (pdp11_hard_regno_mode_ok): New function.
1055 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
1056 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
1058 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
1060 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1061 (rs6000_hard_regno_mode_ok): Rename to...
1062 (rs6000_hard_regno_mode_ok_uncached): ...this.
1063 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1064 (rs6000_hard_regno_mode_ok): New function.
1065 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
1066 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
1067 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
1068 (riscv_hard_regno_mode_ok): ...this and make static.
1069 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1070 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
1071 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
1072 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1073 (rl78_hard_regno_mode_ok): Make static and return bool.
1074 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
1075 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
1077 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
1078 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1079 (rs6000_hard_regno_mode_ok): Rename to...
1080 (rs6000_hard_regno_mode_ok_uncached): ...this.
1081 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1082 (rs6000_hard_regno_mode_ok): New function.
1083 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
1084 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
1085 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1086 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
1087 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
1088 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
1089 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1090 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
1091 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
1092 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1093 (sh_hard_regno_mode_ok): Make static.
1094 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
1095 instead of HARD_REGNO_MODE_OK.
1096 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
1097 (sparc_mode_class): Delete.
1098 (HARD_REGNO_MODE_OK): Delete.
1099 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1100 (hard_regno_mode_classes): Make static.
1101 (sparc_mode_class): Likewise.
1102 (sparc_hard_regno_mode_ok): New function.
1103 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
1104 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
1105 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
1107 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1108 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
1109 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
1110 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
1111 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
1112 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1113 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
1114 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
1115 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1116 (visium_hard_regno_mode_ok): New function.
1117 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
1118 instead of HARD_REGNO_MODE_OK.
1119 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
1120 (HARD_REGNO_MODE_OK): Delete.
1121 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
1122 (xtensa_hard_regno_mode_ok_p): ...this and make static.
1123 (xtensa_option_override): Update accordingly.
1124 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1125 (xtensa_hard_regno_mode_ok): New function.
1126 * system.h (HARD_REGNO_MODE_OK): Poison.
1128 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1129 Alan Hayward <alan.hayward@arm.com>
1130 David Sherwood <david.sherwood@arm.com>
1132 * target.def (hard_regno_call_part_clobbered): New hook.
1133 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
1134 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
1135 * doc/tm.texi: Regenerate.
1136 * hooks.h (hook_bool_uint_mode_false): Declare.
1137 * hooks.c (hook_bool_uint_mode_false): New function.
1138 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1139 * cselib.c (cselib_process_insn): Use
1140 targetm.hard_regno_call_part_clobbered instead of
1141 HARD_REGNO_CALL_PART_CLOBBERED.
1142 * ira-conflicts.c (ira_build_conflicts): Likewise.
1143 * ira-costs.c (ira_tune_allocno_costs): Likewise.
1144 * lra-constraints.c (need_for_call_save_p): Likewise.
1145 * lra-lives.c: Include target.h.
1146 (check_pseudos_live_through_calls): Use
1147 targetm.hard_regno_call_part_clobbered instead of
1148 HARD_REGNO_CALL_PART_CLOBBERED.
1149 * regcprop.c: Include target.h.
1150 (copyprop_hardreg_forward_1): Use
1151 targetm.hard_regno_call_part_clobbered instead of
1152 HARD_REGNO_CALL_PART_CLOBBERED.
1153 * reginfo.c (choose_hard_reg_mode): Likewise.
1154 * regrename.c (check_new_reg_p): Likewise.
1155 * reload.c (find_equiv_reg): Likewise.
1156 * reload1.c (emit_reload_insns): Likewise.
1157 * sched-deps.c (deps_analyze_insn): Likewise.
1158 * sel-sched.c (init_regs_for_mode): Likewise.
1159 (mark_unavailable_hard_regs): Likewise.
1160 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
1161 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1162 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
1164 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1165 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1166 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
1168 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
1170 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1171 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1172 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
1174 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1175 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1176 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
1178 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1179 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
1181 * config/powerpcspe/powerpcspe.c
1182 (rs6000_hard_regno_call_part_clobbered): New function.
1183 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1184 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1185 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
1187 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1188 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1189 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
1191 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1192 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1193 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
1195 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1196 Alan Hayward <alan.hayward@arm.com>
1197 David Sherwood <david.sherwood@arm.com>
1199 * rtl.h (subreg_memory_offset): Declare.
1200 * emit-rtl.c (subreg_memory_offset): New function.
1201 * expmed.c (store_bit_field_1): Use it.
1202 * expr.c (undefined_operand_subword_p): Likewise.
1203 * simplify-rtx.c (simplify_subreg): Likewise.
1205 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
1207 PR rtl-optimization/57448
1210 * optabs.c (expand_atomic_load): Place compiler memory barriers if
1211 using atomic_load pattern.
1212 (expand_atomic_store): Likewise.
1214 2017-09-04 Jakub Jelinek <jakub@redhat.com>
1217 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
1218 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
1219 handling. Use replace_call_with_value with NULL instead of
1220 gsi_replace, unlink_stmt_vdef and release_defs.
1222 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
1225 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
1227 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1230 * rtl.h (emit_library_call_value_1): Declare.
1231 (emit_library_call): Replace declaration with a series of overloads.
1232 Remove the parameter count argument.
1233 (emit_library_call_value): Likewise.
1234 * calls.c (emit_library_call_value_1): Make global. Replace varargs
1235 with an "rtx_mode_t *".
1236 (emit_library_call_value): Delete.
1237 (emit_library_call): Likewise.
1238 * asan.c (asan_emit_stack_protection): Update calls accordingly.
1239 (asan_emit_allocas_unpoison): Likewise.
1240 * builtins.c (expand_builtin_powi): Likewise.
1241 (expand_asan_emit_allocas_unpoison): Likewise.
1242 * cfgexpand.c (expand_main_function): Likewise.
1243 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
1244 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
1245 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
1246 * config/arm/arm.c (arm_trampoline_init): Likewise.
1247 (arm_call_tls_get_addr): Likewise.
1248 (arm_expand_divmod_libfunc): Likewise.
1249 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
1250 (smulsi3_highpart): Likewise.
1251 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
1252 (c6x_expand_compare): Likewise.
1253 (c6x_expand_movmem): Likewise.
1254 * config/frv/frv.c (frv_trampoline_init): Likewise.
1255 * config/i386/i386.c (ix86_trampoline_init): Likewise.
1256 (ix86_expand_divmod_libfunc): Likewise.
1257 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
1258 (ia64_expand_compare): Likewise.
1259 (ia64_profile_hook): Likewise.
1260 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
1261 (nonlocal_goto): Likewise.
1262 (restore_stack_nonlocal): Likewise.
1263 * config/m32r/m32r.c (block_move_call): Likewise.
1264 (m32r_trampoline_init): Likewise.
1265 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
1266 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
1267 (m68k_call_m68k_read_tp): Likewise.
1268 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
1269 (microblaze_expand_divide): Likewise.
1270 * config/mips/mips.h (mips_args): Likewise.
1271 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
1272 (MIPS_ICACHE_SYNC): Likewise.
1273 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
1274 (nios2_trampoline_init): Likewise.
1275 * config/pa/pa.c (hppa_tls_call): Likewise.
1276 (pa_trampoline_init): Likewise.
1277 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
1278 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
1279 (expand_strn_compare): Likewise.
1280 (rs6000_generate_compare): Likewise.
1281 (rs6000_expand_float128_convert): Likewise.
1282 (output_profile_hook): Likewise.
1283 (rs6000_trampoline_init): Likewise.
1284 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
1285 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
1286 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
1287 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
1288 (rs6000_generate_compare): Likewise.
1289 (rs6000_expand_float128_convert): Likewise.
1290 (output_profile_hook): Likewise.
1291 (rs6000_trampoline_init): Likewise.
1292 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
1293 * config/sh/sh.c (sh_trampoline_init): Likewise.
1294 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
1295 (sparc_emit_float_lib_cmp): Likewise.
1296 (sparc32_initialize_trampoline): Likewise.
1297 (sparc64_initialize_trampoline): Likewise.
1298 (sparc_profile_hook): Likewise.
1299 * config/spu/spu.c (ea_load_store): Likewise.
1300 * config/spu/spu.md (floatunssidf2): Likewise.
1301 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
1302 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
1303 * config/visium/visium.c (expand_block_move_4): Likewise.
1304 (expand_block_move_2): Likewise.
1305 (expand_block_move_1): Likewise.
1306 (expand_block_set_4): Likewise.
1307 (expand_block_set_2): Likewise.
1308 (expand_block_set_1): Likewise.
1309 (visium_trampoline_init): Likewise.
1310 (visium_profile_hook): Likewise.
1311 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
1312 (xtensa_setup_frame_addresses): Likewise.
1313 (xtensa_trampoline_init): Likewise.
1314 * except.c (sjlj_emit_function_enter): Likewise.
1315 (sjlj_emit_function_exit): Likewise.
1316 * explow.c (allocate_dynamic_stack_space): Likewise.
1317 (probe_stack_range): Likewise.
1318 * expr.c (convert_mode_scalar): Likewise.
1319 * optabs.c (expand_binop): Likewise.
1320 (expand_twoval_binop_libfunc): Likewise.
1321 (expand_unop): Likewise.
1322 (prepare_cmp_insn): Likewise.
1323 (prepare_float_lib_cmp): Likewise.
1324 (expand_float): Likewise.
1325 (expand_fix): Likewise.
1326 (expand_fixed_convert): Likewise.
1327 (maybe_emit_sync_lock_test_and_set): Likewise.
1328 (expand_atomic_compare_and_swap): Likewise.
1329 (expand_mem_thread_fence): Likewise.
1330 (expand_atomic_fetch_op): Likewise.
1332 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
1334 * doc/generic.texi (OpenACC): Adjust URL.
1335 * doc/invoke.texi (C Dialect Options): Ditto.
1337 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
1339 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
1340 predicate for operand 1. Add (m,<S>) constraint.
1341 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
1342 Prevent memory operand 1 with register operand 2.
1344 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
1346 PR rtl-optimization/82024
1347 * combine.c (try_combine): If the combination result is a PARALLEL,
1348 and we only need to retain the SET in there that would be placed
1349 at I2, check that we can place that at I3 instead, before doing so.
1351 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1354 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
1356 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
1357 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
1360 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
1361 Jeff Law <law@redhat.com>
1363 * varasm.c (bss_initializer_p): Do not put constants into .bss
1364 (categorize_decl_for_section): Handle bss_initializer_p returning
1365 false when DECL_INITIAL is NULL.
1367 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1370 * config/s390/s390.c (s390_can_inline_p): New function.
1372 2017-09-01 Jeff Law <law@redhat.com>
1374 PR tree-optimization/82052
1375 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
1376 Always initialize the returned slot after a hash table miss
1377 when INSERT is true.
1379 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
1381 * config/s390/s390.md (mem_signal_fence): Remove.
1382 * doc/md.texi (mem_signal_fence): Remove.
1383 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
1385 * target-insns.def (mem_signal_fence): Remove.
1387 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1390 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
1393 * asan.c (create_odr_indicator): Strip name encoding from assembler
1394 name before appending it after __odr_asan_.
1396 2017-09-01 Martin Liska <mliska@suse.cz>
1398 PR tree-optimization/82059
1399 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
1400 frequency only when an edge is redirected.
1402 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1404 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
1405 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
1406 (arc_conditional_register_usage): Remove ARC600 lp_count
1408 (arc_file_start): Emit Tag_ARC_CPU_variation.
1409 (arc_can_use_doloop_p): New conditions to use ZOLs.
1410 (hwloop_fail): New function.
1411 (hwloop_optimize): Likewise.
1412 (hwloop_pattern_reg): Likewise.
1413 (arc_doloop_hooks): New struct, to be used with reorg_loops.
1414 (arc_reorg_loops): New function, calls reorg_loops.
1415 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
1416 (arc600_corereg_hazard): Remove ZOL checking, case handled by
1418 (arc_loop_hazard): Remove function, functionality moved into
1420 (arc_hazard): Remove arc_loop_hazard call.
1421 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
1422 into hwloop_optimize.
1423 (arc_label_align): Remove ZOL handling.
1424 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
1425 * config/arc/arc.md (doloop_begin): Remove pattern.
1426 (doloop_begin_i): Likewise.
1427 (doloop_end_i): Likewise.
1428 (doloop_fallback): Likewise.
1429 (doloop_fallback_m): Likewise.
1430 (doloop_end): Reimplement expand.
1431 (arc_lp): New pattern for LP instruction.
1432 (loop_end): New pattern.
1433 (loop_fail): Likewise.
1434 (decrement_and_branch_until_zero): Likewise.
1435 * config/arc/arc.opt (mlpc-width): New option.
1436 * doc/invoke.texi (mlpc-width): Document option.
1438 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1440 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
1441 (arc_ccfsm_advance): Fix checking for delay slots.
1442 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
1444 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1446 * config/arc/arc.md (movqi_insn): Add stores to save constant long
1448 (movhi_insn): Update store instruction constraint which are saving
1449 6-bit short immediates.
1450 (movsi_insn): Consider also short scaled load operations.
1451 (zero_extendhisi2_i): Use Usd constraint instead of T.
1452 (extendhisi2_i): Add q constraint.
1453 (arc_clzsi2): Add type and length attributes.
1454 (arc_ctzsi2): Likewise.
1455 * config/arc/constraints.md (Usc): Update constraint, the
1456 assembler can parse two relocations for a single instruction.
1458 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1460 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
1461 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
1463 2017-08-31 Olivier Hainque <hainque@adacore.com>
1465 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
1466 match as powerpc-wrs-vxworks*.
1468 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
1470 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
1471 register constraint for by-element operand.
1472 (aarch64_mls_elt_merge<mode>): Likewise.
1474 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1476 * config/arc/arc.c (arc_can_follow_jump): Check for short
1479 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1481 * config.gcc: Use g.opt for arc.
1482 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
1483 functionality moved to ...
1484 (legitimate_scaled_address_p): New function, ...here.
1485 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
1486 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
1487 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
1489 (arc_override_options): Handle G option.
1490 (arc_output_pic_addr_const): Correct function definition.
1491 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
1492 (arc_decl_anon_ns_mem_p): Delete.
1493 (arc_in_small_data_p): Overhaul this function to take into
1494 consideration the value given via G option.
1495 (arc_rewrite_small_data_1): Renamed and corrected old
1496 arc_rewrite_small_data function.
1497 (arc_rewrite_small_data): New function.
1498 (small_data_pattern): Don't use pic_offset_table_rtx.
1499 * config/arc/arc.h (CC1_SPEC): Recognize G option.
1500 * config/arc/simdext.md (movmisalignv2hi): Use
1501 prepare_move_operands function.
1503 (movmisalign*): Likewise.
1504 * doc/invoke.texi (ARC options): Document -G option.
1506 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1508 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
1510 * config/arc/arc.c (arc_print_operand): Output scalled address for
1511 sdata whenever is possible.
1512 (arc_in_small_data_p): Allow sdata for 64bit datum when double
1513 load/stores are available.
1514 (compact_sda_memory_operand): Check for the alignment required by
1515 code density instructions.
1516 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
1518 * config/arc/constraints.md (Usd): Update constraint.
1519 (Us0): New constraint.
1520 (Usc): Update constraint.
1522 2017-08-31 Richard Biener <rguenther@suse.de>
1525 * dwarf2out.c (dwarf2out_early_global_decl): Process each
1528 2017-08-31 Tamar Christina <tamar.christina@arm.com>
1530 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
1531 Resize type_signature.
1533 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1534 Alan Hayward <alan.hayward@arm.com>
1535 David Sherwood <david.sherwood@arm.com>
1537 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
1538 subregs whose inner modes can be stored in GPRs.
1539 (aarch64_classify_index): Likewise.
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/iterators.md (V_cmp_result): Rename to...
1546 (V_INT_EQUIV): ...this.
1547 (v_cmp_result): Rename to...
1548 (v_int_equiv): ...this.
1549 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
1550 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
1551 (copysign<mode>3): Likewise.
1552 (aarch64_simd_bsl<mode>_internal): Likewise.
1553 (aarch64_simd_bsl<mode>): Likewise.
1554 (vec_cmp<mode><mode>): Likewise.
1555 (vcond<mode><mode>): Likewise.
1556 (vcond<v_cmp_mixed><mode>): Likewise.
1557 (vcondu<mode><v_cmp_mixed>): Likewise.
1558 (aarch64_cm<optab><mode>): Likewise.
1559 (aarch64_cmtst<mode>): Likewise.
1560 (aarch64_fac<optab><mode>): Likewise.
1561 (vec_perm_const<mode>): Likewise.
1562 (vcond_mask_<mode><v_cmp_result>): Rename to...
1563 (vcond_mask_<mode><v_int_equiv>): ...this.
1564 (vec_cmp<mode><v_cmp_result>): Rename to...
1565 (vec_cmp<mode><v_int_equiv>): ...this.
1567 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1568 Alan Hayward <alan.hayward@arm.com>
1569 David Sherwood <david.sherwood@arm.com>
1571 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
1573 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
1574 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
1575 (UNSPEC_LD4_DREG): New unspecs.
1576 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
1577 (aarch64_ld2<mode>_dreg_be): Replace with...
1578 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
1580 (aarch64_ld3<mode>_dreg_le)
1581 (aarch64_ld3<mode>_dreg_be): Replace with...
1582 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
1584 (aarch64_ld4<mode>_dreg_le)
1585 (aarch64_ld4<mode>_dreg_be): Replace with...
1586 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
1589 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1591 PR tree-optimization/81987
1592 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
1593 insert an initializer in a location not dominated by the stride
1596 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1598 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
1599 on the entire header of the finally block in the fallthru case.
1601 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1603 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
1605 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
1607 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
1608 rs6000_emit_move_from_cr and call renamed function.
1609 (rs6000_emit_prologue): Call renamed functions.
1610 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
1611 movesi_from_cr, remove volatile CRs.
1613 2017-08-30 Jon Beniston <jon@beniston.com>
1614 Richard Biener <rguenther@suse.de>
1616 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
1617 of VECTOR_MODE_P check.
1618 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
1619 element vector types.
1621 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1623 * df.h (df_read_modify_subreg_p): Remove in favor of...
1624 * rtl.h (read_modify_subreg_p): ...this new function. Take a
1625 const_rtx instead of an rtx.
1626 * cprop.c (local_cprop_find_used_regs): Update accordingly.
1627 * df-problems.c (df_word_lr_mark_ref): Likewise.
1628 * ira-lives.c (mark_pseudo_reg_live): Likewise.
1629 (mark_pseudo_reg_dead): Likewise.
1630 (mark_ref_dead): Likewise.
1631 * reginfo.c (init_subregs_of_mode): Likewise.
1632 * sched-deps.c (sched_analyze_1): Likewise.
1633 * df-scan.c (df_def_record_1): Likewise.
1634 (df_uses_record): Likewise.
1635 (df_read_modify_subreg_p): Remove in favor of...
1636 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
1637 const_rtx instead of an rtx.
1639 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1640 Alan Hayward <alan.hayward@arm.com>
1641 David Sherwood <david.sherwood@arm.com>
1643 * rtl.h (partial_subreg_p): New function.
1644 * caller-save.c (save_call_clobbered_regs): Use it.
1645 * calls.c (expand_call): Likewise.
1646 * combine.c (combinable_i3pat): Likewise.
1647 (simplify_set): Likewise.
1648 (make_extraction): Likewise.
1649 (make_compound_operation_int): Likewise.
1650 (gen_lowpart_or_truncate): Likewise.
1651 (force_to_mode): Likewise.
1652 (make_field_assignment): Likewise.
1653 (reg_truncated_to_mode): Likewise.
1654 (record_truncated_value): Likewise.
1655 (move_deaths): Likewise.
1656 * cse.c (record_jump_cond): Likewise.
1657 (cse_insn): Likewise.
1658 * cselib.c (cselib_lookup_1): Likewise.
1659 * expmed.c (extract_bit_field_using_extv): Likewise.
1660 * function.c (assign_parm_setup_reg): Likewise.
1661 * ifcvt.c (noce_convert_multiple_sets): Likewise.
1662 * ira-build.c (create_insn_allocnos): Likewise.
1663 * lra-coalesce.c (merge_pseudos): Likewise.
1664 * lra-constraints.c (match_reload): Likewise.
1665 (simplify_operand_subreg): Likewise.
1666 (curr_insn_transform): Likewise.
1667 * lra-lives.c (process_bb_lives): Likewise.
1668 * lra.c (new_insn_reg): Likewise.
1669 (lra_substitute_pseudo): Likewise.
1670 * regcprop.c (mode_change_ok): Likewise.
1671 (maybe_mode_change): Likewise.
1672 (copyprop_hardreg_forward_1): Likewise.
1673 * reload.c (push_reload): Likewise.
1674 (find_reloads): Likewise.
1675 (find_reloads_subreg_address): Likewise.
1676 * reload1.c (alter_reg): Likewise.
1677 (eliminate_regs_1): Likewise.
1678 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1680 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
1682 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
1683 back to if statements, including unpack.
1685 2017-08-30 Martin Liska <mliska@suse.cz>
1688 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
1690 (func_checker::compare_asm_inputs_outputs): ... this function.
1691 (func_checker::compare_gimple_asm): Use the function to compare
1692 also ASM constrains.
1693 * ipa-icf-gimple.h: Rename the function.
1695 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1696 Alan Hayward <alan.hayward@arm.com>
1697 David Sherwood <david.sherwood@arm.com>
1699 * coretypes.h (complex_mode): New type.
1700 * gdbhooks.py (build_pretty_printer): Handle it.
1701 * machmode.h (complex_mode): New class.
1702 (complex_mode::includes_p): New function.
1703 (is_complex_int_mode): Likewise.
1704 (is_complex_float_mode): Likewise.
1705 * genmodes.c (get_mode_class): Handle complex mode classes.
1706 * function.c (expand_function_end): Use is_complex_int_mode.
1708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1709 Alan Hayward <alan.hayward@arm.com>
1710 David Sherwood <david.sherwood@arm.com>
1712 * coretypes.h (scalar_mode_pod): New typedef.
1713 * gdbhooks.py (build_pretty_printer): Handle it.
1714 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
1715 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
1716 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
1717 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
1720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1721 Alan Hayward <alan.hayward@arm.com>
1722 David Sherwood <david.sherwood@arm.com>
1724 * machmode.h (mode_for_vector): Take a scalar_mode instead
1726 * stor-layout.c (mode_for_vector): Likewise.
1727 * explow.c (promote_mode): Use as_a <scalar_mode>.
1728 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
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 (preferred_simd_mode): Take a scalar_mode
1735 instead of a machine_mode.
1736 * targhooks.h (default_preferred_simd_mode): Likewise.
1737 * targhooks.c (default_preferred_simd_mode): Likewise.
1738 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
1739 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
1740 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
1741 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
1742 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
1743 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
1744 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
1745 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
1747 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
1748 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
1749 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
1750 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
1751 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
1752 * doc/tm.texi: Regenerate.
1753 * optabs-query.c (can_vec_mask_load_store_p): Return false for
1756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1757 Alan Hayward <alan.hayward@arm.com>
1758 David Sherwood <david.sherwood@arm.com>
1760 * target.def (scalar_mode_supported_p): Take a scalar_mode
1761 instead of a machine_mode.
1762 * targhooks.h (default_scalar_mode_supported_p): Likewise.
1763 * targhooks.c (default_scalar_mode_supported_p): Likewise.
1764 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
1765 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1766 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
1767 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
1768 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
1769 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
1770 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
1771 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
1772 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
1773 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
1774 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
1775 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
1777 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
1778 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
1779 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1780 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1781 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
1783 * doc/tm.texi: Regenerate.
1785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1786 Alan Hayward <alan.hayward@arm.com>
1787 David Sherwood <david.sherwood@arm.com>
1789 * coretypes.h (opt_scalar_mode): New typedef.
1790 * gdbhooks.py (build_pretty_printers): Handle it.
1791 * machmode.h (mode_iterator::get_2xwider): Add overload for
1793 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
1795 * expr.c (convert_mode_scalar): Likewise.
1796 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1797 * optabs.c (expand_float): Likewise.
1798 (expand_fix): Likewise.
1799 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1801 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1802 Alan Hayward <alan.hayward@arm.com>
1803 David Sherwood <david.sherwood@arm.com>
1805 * optabs.c (expand_float): Explicitly check for scalars before
1806 using a branching expansion.
1807 (expand_fix): Likewise.
1809 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1810 Alan Hayward <alan.hayward@arm.com>
1811 David Sherwood <david.sherwood@arm.com>
1813 * expr.c (convert_mode): Split scalar handling out into...
1814 (convert_mode_scalar): ...this new function. Treat the modes
1817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1818 Alan Hayward <alan.hayward@arm.com>
1819 David Sherwood <david.sherwood@arm.com>
1821 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
1823 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
1825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1826 Alan Hayward <alan.hayward@arm.com>
1827 David Sherwood <david.sherwood@arm.com>
1829 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
1830 rather than a machine_mode.
1831 (fixed_from_string): Likewise.
1832 (fixed_convert): Likewise.
1833 (fixed_convert_from_int): Likewise.
1834 (fixed_convert_from_real): Likewise.
1835 (real_convert_from_fixed): Likewise.
1836 * fixed-value.c (fixed_from_double_int): Likewise.
1837 (fixed_from_string): Likewise.
1838 (fixed_convert): Likewise.
1839 (fixed_convert_from_int): Likewise.
1840 (fixed_convert_from_real): Likewise.
1841 (real_convert_from_fixed): Likewise.
1842 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
1844 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1845 Alan Hayward <alan.hayward@arm.com>
1846 David Sherwood <david.sherwood@arm.com>
1848 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
1849 of separate mode class checks. Do not allow vector modes here.
1850 (immed_wide_int_const): Use as_a <scalar_mode>.
1851 * explow.c (trunc_int_for_mode): Likewise.
1852 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
1853 (wi::shwi): Likewise.
1854 (wi::min_value): Likewise.
1855 (wi::max_value): Likewise.
1856 * dwarf2out.c (loc_descriptor): Likewise.
1857 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
1860 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1861 Alan Hayward <alan.hayward@arm.com>
1862 David Sherwood <david.sherwood@arm.com>
1864 * tree.h (SCALAR_TYPE_MODE): New macro.
1865 * expr.c (expand_expr_addr_expr_1): Use it.
1866 (expand_expr_real_2): Likewise.
1867 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
1868 (fold_convert_const_fixed_from_int): Likewise.
1869 (fold_convert_const_fixed_from_real): Likewise.
1870 (native_encode_fixed): Likewise
1871 (native_encode_complex): Likewise
1872 (native_encode_vector): Likewise.
1873 (native_interpret_fixed): Likewise.
1874 (native_interpret_real): Likewise.
1875 (native_interpret_complex): Likewise.
1876 (native_interpret_vector): Likewise.
1877 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
1878 (simd_clone_adjust_argument_types): Likewise.
1879 (simd_clone_init_simd_arrays): Likewise.
1880 (simd_clone_adjust): Likewise.
1881 * stor-layout.c (layout_type): Likewise.
1882 * tree.c (build_minus_one_cst): Likewise.
1883 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
1884 * tree-inline.c (estimate_move_cost): Likewise.
1885 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
1886 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
1887 (vectorizable_reduction): Likewise.
1888 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
1889 (vect_recog_mixed_size_cond_pattern): Likewise.
1890 (check_bool_pattern): Likewise.
1891 (adjust_bool_pattern): Likewise.
1892 (search_type_for_mask_1): Likewise.
1893 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
1894 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1895 (vectorizable_load): Likewise.
1896 (vectorizable_store): Likewise.
1897 * ubsan.c (ubsan_encode_value): Likewise.
1898 * varasm.c (output_constant): Likewise.
1900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1901 Alan Hayward <alan.hayward@arm.com>
1902 David Sherwood <david.sherwood@arm.com>
1904 * coretypes.h (scalar_mode): New class.
1905 * machmode.h (scalar_mode): Likewise.
1906 (scalar_mode::includes_p): New function.
1907 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
1908 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
1909 * genmodes.c (get_mode_class): Handle remaining scalar modes.
1910 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
1911 * expmed.c (store_bit_field_1): Likewise.
1912 (extract_bit_field_1): Likewise.
1913 * expr.c (write_complex_part): Likewise.
1914 (read_complex_part): Likewise.
1915 (emit_move_complex_push): Likewise.
1916 (expand_expr_real_2): Likewise.
1917 * function.c (assign_parm_setup_reg): Likewise.
1918 (assign_parms_unsplit_complex): Likewise.
1919 * optabs.c (expand_binop): Likewise.
1920 * rtlanal.c (subreg_get_info): Likewise.
1921 * simplify-rtx.c (simplify_immed_subreg): Likewise.
1922 * varasm.c (output_constant_pool_2): Likewise.
1924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1925 Alan Hayward <alan.hayward@arm.com>
1926 David Sherwood <david.sherwood@arm.com>
1928 * expmed.c (extract_high_half): Use scalar_int_mode and remove
1930 (expmed_mult_highpart_optab): Likewise.
1931 (expmed_mult_highpart): Likewise.
1933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1934 Alan Hayward <alan.hayward@arm.com>
1935 David Sherwood <david.sherwood@arm.com>
1937 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
1938 instead of a machine_mode.
1939 (builtin_memset_read_str): Likewise.
1940 * builtins.c (c_readstr): Likewise.
1941 (builtin_memcpy_read_str): Likewise.
1942 (builtin_strncpy_read_str): Likewise.
1943 (builtin_memset_read_str): Likewise.
1944 (builtin_memset_gen_str): Likewise.
1945 (expand_builtin_signbit): Use scalar_int_mode for local variables.
1946 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
1947 instead of a machine_mode.
1948 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
1950 (make_extraction): Likewise.
1951 (try_widen_shift_mode): Take and return scalar_int_modes instead
1953 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
1954 a scalar_int_mode instead of a machine_mode.
1955 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
1956 (avr_addr_space_pointer_mode): Likewise.
1957 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
1958 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
1959 (msp430_unwind_word_mode): Likewise.
1960 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
1961 (spu_addr_space_pointer_mode): Likewise.
1962 (spu_addr_space_address_mode): Likewise.
1963 (spu_libgcc_cmp_return_mode): Likewise.
1964 (spu_libgcc_shift_count_mode): Likewise.
1965 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
1966 (rl78_addr_space_pointer_mode): Likewise.
1967 (fl78_unwind_word_mode): Likewise.
1968 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
1970 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
1971 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
1972 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
1973 (mips_valid_pointer_mode): Likewise.
1974 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
1975 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
1976 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
1978 (ft32_addr_space_address_mode): Likewise.
1979 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
1980 scalar_int_mode instead of a machine_mode.
1981 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
1983 (m32c_addr_space_address_mode): Likewise.
1984 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
1985 (rs6000_eh_return_filter_mode): Likewise.
1986 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
1987 (rs6000_eh_return_filter_mode): Likewise.
1988 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
1989 (s390_libgcc_shift_count_mode): Likewise.
1990 (s390_unwind_word_mode): Likewise.
1991 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
1993 * target.def (mode_rep_extended): Likewise.
1994 (valid_pointer_mode): Likewise.
1995 (addr_space.valid_pointer_mode): Likewise.
1996 (eh_return_filter_mode): Return a scalar_int_mode rather than
1998 (libgcc_cmp_return_mode): Likewise.
1999 (libgcc_shift_count_mode): Likewise.
2000 (unwind_word_mode): Likewise.
2001 (addr_space.pointer_mode): Likewise.
2002 (addr_space.address_mode): Likewise.
2003 * doc/tm.texi: Regenerate.
2004 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
2006 (do_jump): Use scalar_int_mode for local variables.
2007 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
2008 rather than a machine_mode.
2009 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
2010 (scompare_loc_descriptor_wide): Likewise.
2011 (scompare_loc_descriptor_narrow): Likewise.
2012 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
2014 * except.c (sjlj_emit_dispatch_table): Likewise.
2015 (expand_builtin_eh_copy_values): Likewise.
2016 * explow.c (convert_memory_address_addr_space_1): Likewise.
2017 Take a scalar_int_mode rather than a machine_mode.
2018 (convert_memory_address_addr_space): Take a scalar_int_mode rather
2019 than a machine_mode.
2020 (memory_address_addr_space): Use scalar_int_mode for local variables.
2021 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
2022 rather than a machine_mode.
2023 * expmed.c (mask_rtx): Likewise.
2024 (init_expmed_one_conv): Likewise.
2025 (expand_mult_highpart_adjust): Likewise.
2026 (extract_high_half): Likewise.
2027 (expmed_mult_highpart_optab): Likewise.
2028 (expmed_mult_highpart): Likewise.
2029 (expand_smod_pow2): Likewise.
2030 (expand_sdiv_pow2): Likewise.
2031 (emit_store_flag_int): Likewise.
2032 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
2034 (extract_low_bits): Likewise.
2035 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
2037 * expr.c (pieces_addr::adjust): Likewise.
2038 (can_store_by_pieces): Likewise.
2039 (store_by_pieces): Likewise.
2040 (clear_by_pieces_1): Likewise.
2041 (expand_expr_addr_expr_1): Likewise.
2042 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
2043 (expand_expr_real_1): Likewise.
2044 (try_casesi): Likewise.
2045 * final.c (shorten_branches): Likewise.
2046 * fold-const.c (fold_convert_const_int_from_fixed): Change the
2047 type of "mode" to machine_mode.
2048 * internal-fn.c (expand_arith_overflow_result_store): Take a
2049 scalar_int_mode rather than a machine_mode.
2050 (expand_mul_overflow): Use scalar_int_mode for local variables.
2051 * loop-doloop.c (doloop_modify): Likewise.
2052 (doloop_optimize): Likewise.
2053 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
2054 than a machine_mode.
2055 (expand_doubleword_shift_condmove): Likewise.
2056 (expand_doubleword_shift): Likewise.
2057 (expand_doubleword_clz): Likewise.
2058 (expand_doubleword_popcount): Likewise.
2059 (expand_doubleword_parity): Likewise.
2060 (expand_absneg_bit): Use scalar_int_mode for local variables.
2061 (prepare_float_lib_cmp): Likewise.
2062 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
2063 rather than a machine_mode.
2064 (convert_memory_address_addr_space): Likewise.
2065 (get_mode_bounds): Likewise.
2066 (get_address_mode): Return a scalar_int_mode rather than a
2068 * rtlanal.c (get_address_mode): Likewise.
2069 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
2070 than a machine_mode.
2071 * targhooks.c (default_mode_rep_extended): Likewise.
2072 (default_valid_pointer_mode): Likewise.
2073 (default_addr_space_valid_pointer_mode): Likewise.
2074 (default_eh_return_filter_mode): Return a scalar_int_mode rather
2075 than a machine_mode.
2076 (default_libgcc_cmp_return_mode): Likewise.
2077 (default_libgcc_shift_count_mode): Likewise.
2078 (default_unwind_word_mode): Likewise.
2079 (default_addr_space_pointer_mode): Likewise.
2080 (default_addr_space_address_mode): Likewise.
2081 * targhooks.h (default_eh_return_filter_mode): Likewise.
2082 (default_libgcc_cmp_return_mode): Likewise.
2083 (default_libgcc_shift_count_mode): Likewise.
2084 (default_unwind_word_mode): Likewise.
2085 (default_addr_space_pointer_mode): Likewise.
2086 (default_addr_space_address_mode): Likewise.
2087 (default_mode_rep_extended): Take a scalar_int_mode rather than
2089 (default_valid_pointer_mode): Likewise.
2090 (default_addr_space_valid_pointer_mode): Likewise.
2091 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
2093 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
2094 rather than a machine_mode.
2095 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
2096 for local variables.
2097 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2098 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
2099 than a machine_mode.
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 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
2106 the mode argument to scalar_int_mode.
2107 (do_jump_by_parts_zero_rtx): Likewise.
2108 (do_jump_by_parts_equality_rtx): Likewise.
2109 (do_jump_by_parts_greater): Take a mode argument.
2110 (do_jump_by_parts_equality): Likewise.
2111 (do_jump_1): Update calls accordingly.
2113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2114 Alan Hayward <alan.hayward@arm.com>
2115 David Sherwood <david.sherwood@arm.com>
2117 * is-a.h (safe_dyn_cast): New function.
2118 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
2119 (jump_table_for_label): Likewise.
2120 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
2121 instead of an rtx_insn *.
2122 (shorten_branches): Use dyn_cast instead of LABEL_P and
2123 JUMP_TABLE_DATA_P. Use jump_table_for_label and
2124 rtx_jump_table_data::get_data_mode.
2125 (final_scan_insn): Likewise.
2127 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2128 Alan Hayward <alan.hayward@arm.com>
2129 David Sherwood <david.sherwood@arm.com>
2131 * combine.c (try_combine): Use is_a <scalar_int_mode> when
2132 trying to combine a full-register integer set with a subreg
2135 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2136 Alan Hayward <alan.hayward@arm.com>
2137 David Sherwood <david.sherwood@arm.com>
2139 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
2140 that is always either address_mode or pointer_mode.
2142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2143 Alan Hayward <alan.hayward@arm.com>
2144 David Sherwood <david.sherwood@arm.com>
2146 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
2147 when the two are known to be equal.
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 * simplify-rtx.c (simplify_const_unary_operation): Use
2154 is_a <scalar_int_mode> instead of checking for a nonzero
2155 precision. Forcibly convert op_mode to a scalar_int_mode
2156 in that case. More clearly differentiate the operand and
2157 result modes and use the former when deciding what the value
2158 of a count-bits operation should be. Use is_int_mode instead
2159 of checking for a MODE_INT. Remove redundant check for whether
2160 this mode has a zero precision.
2162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2163 Alan Hayward <alan.hayward@arm.com>
2164 David Sherwood <david.sherwood@arm.com>
2166 * optabs.c (widen_leading): Change the type of the mode argument
2167 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
2168 (widen_bswap): Likewise.
2169 (expand_parity): Likewise.
2170 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
2171 (expand_ffs): Likewise.
2172 (epand_unop): Check for scalar integer modes before calling the
2175 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2176 Alan Hayward <alan.hayward@arm.com>
2177 David Sherwood <david.sherwood@arm.com>
2179 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
2181 (expand_expr_real_1): Update call accordingly.
2183 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2184 Alan Hayward <alan.hayward@arm.com>
2185 David Sherwood <david.sherwood@arm.com>
2187 * expmed.c (store_bit_field_using_insv): Add op0_mode and
2188 value_mode arguments. Use scalar_int_mode internally.
2189 (store_bit_field_1): Rename the new integer mode from imode
2190 to op0_mode and use it instead of GET_MODE (op0). Update calls
2191 to store_split_bit_field, store_bit_field_using_insv and
2192 store_fixed_bit_field.
2193 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
2194 Use scalar_int_mode internally. Use a bit count rather than a mode
2195 when calculating the largest bit size for get_best_mode.
2196 Update calls to store_split_bit_field and store_fixed_bit_field_1.
2197 (store_fixed_bit_field_1): Add mode and value_mode arguments.
2198 Remove assertion that OP0 has a scalar integer mode.
2199 (store_split_bit_field): Add op0_mode and value_mode arguments.
2200 Update calls to extract_fixed_bit_field.
2201 (extract_bit_field_using_extv): Add an op0_mode argument.
2202 Use scalar_int_mode internally.
2203 (extract_bit_field_1): Rename the new integer mode from imode to
2204 op0_mode and use it instead of GET_MODE (op0). Update calls to
2205 extract_split_bit_field, extract_bit_field_using_extv and
2206 extract_fixed_bit_field.
2207 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
2208 to extract_split_bit_field and extract_fixed_bit_field_1.
2209 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
2210 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
2212 (extract_split_bit_field): Add an op0_mode argument. Update call
2213 to extract_fixed_bit_field.
2215 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2216 Alan Hayward <alan.hayward@arm.com>
2217 David Sherwood <david.sherwood@arm.com>
2219 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
2220 * explow.c (hard_function_value): Likewise.
2221 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
2222 convert_to_mode call outside the loop.
2223 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
2224 for the mode iterator. Require the mode specified by max_pieces
2226 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
2228 (copy_blkmode_to_reg): Likewise.
2229 (set_storage_via_setmem): Likewise.
2230 * optabs.c (prepare_cmp_insn): Likewise.
2231 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2232 * stor-layout.c (finish_bitfield_representative): Likewise.
2234 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2235 Alan Hayward <alan.hayward@arm.com>
2236 David Sherwood <david.sherwood@arm.com>
2238 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
2239 * expr.c (convert_move): Use them.
2240 (convert_modes): Likewise.
2241 (store_expr_with_bounds): Likewise.
2243 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2244 Alan Hayward <alan.hayward@arm.com>
2245 David Sherwood <david.sherwood@arm.com>
2247 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
2248 parameter for the mode of "x". Remove the "known_x", "known_mode"
2249 and "known_ret" arguments. Change the type of the mode argument
2251 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
2252 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
2253 (reg_num_sign_bit_copies_for_combine): Likewise.
2254 * rtlanal.c (nonzero_bits1): Likewise.
2255 (num_sign_bit_copies1): Likewise.
2256 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
2257 (reg_num_sign_bit_copies_general): Likewise.
2258 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
2259 (reg_nonzero_bits_general): Likewise.
2261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2262 Alan Hayward <alan.hayward@arm.com>
2263 David Sherwood <david.sherwood@arm.com>
2265 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
2266 than in subroutines. Return 1 for non-integer modes.
2267 (cached_num_sign_bit_copies): Change the type of the mode parameter
2269 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
2270 classes. Handle CONST_INT_P first and then check whether X also
2271 has a scalar integer mode. Check the same thing for inner registers
2272 of a SUBREG and for values that are being extended or truncated.
2274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2275 Alan Hayward <alan.hayward@arm.com>
2276 David Sherwood <david.sherwood@arm.com>
2278 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
2279 in subroutines. Return the mode mask for non-integer modes.
2280 (cached_nonzero_bits): Change the type of the mode parameter
2282 (nonzero_bits1): Likewise. Remove early exit for other mode
2283 classes. Handle CONST_INT_P first and then check whether X
2284 also has a scalar integer mode.
2286 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2287 Alan Hayward <alan.hayward@arm.com>
2288 David Sherwood <david.sherwood@arm.com>
2290 * expr.c (widest_int_mode_for_size): Make the comment match the code.
2291 Return a scalar_int_mode and assert that the size is greater than
2293 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
2294 (op_by_pieces_d::op_by_pieces_d): Likewise.
2295 (op_by_pieces_d::run): Likewise.
2296 (can_store_by_pieces): Likewise.
2298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2299 Alan Hayward <alan.hayward@arm.com>
2300 David Sherwood <david.sherwood@arm.com>
2302 * combine.c (extract_left_shift): Add a mode argument and update
2304 (make_compound_operation_int): Change the type of the mode parameter
2305 to scalar_int_mode and update the call to extract_left_shift.
2307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2308 Alan Hayward <alan.hayward@arm.com>
2309 David Sherwood <david.sherwood@arm.com>
2311 * combine.c (simplify_and_const_int): Change the type of the mode
2312 parameter to scalar_int_mode.
2313 (simplify_and_const_int_1): Likewise. Update recursive call.
2315 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2316 Alan Hayward <alan.hayward@arm.com>
2317 David Sherwood <david.sherwood@arm.com>
2319 * combine.c (simplify_compare_const): Check that the mode is a
2320 scalar_int_mode (rather than VOIDmode) before testing its
2322 (simplify_comparison): Move COMPARISON_P handling out of the
2323 loop and restrict the latter part of the loop to scalar_int_modes.
2324 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
2325 and when considering SUBREG_REGs. Use is_int_mode instead of
2326 checking GET_MODE_CLASS against MODE_INT.
2328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2329 Alan Hayward <alan.hayward@arm.com>
2330 David Sherwood <david.sherwood@arm.com>
2332 * combine.c (try_widen_shift_mode): Move check for equal modes to...
2333 (simplify_shift_const_1): ...here. Use scalar_int_mode for
2334 shift_unit_mode and for modes involved in scalar shifts.
2336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2337 Alan Hayward <alan.hayward@arm.com>
2338 David Sherwood <david.sherwood@arm.com>
2340 * combine.c (force_int_to_mode): New function, split out from...
2341 (force_to_mode): ...here. Keep xmode up-to-date and use it
2342 instead of GET_MODE (x).
2344 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2345 Alan Hayward <alan.hayward@arm.com>
2346 David Sherwood <david.sherwood@arm.com>
2348 * optabs-query.h (extraction_insn::struct_mode): Change type to
2349 opt_scalar_int_mode and update comment.
2350 (extraction_insn::field_mode): Change type to scalar_int_mode.
2351 (extraction_insn::pos_mode): Likewise.
2352 * combine.c (make_extraction): Update accordingly.
2353 * optabs-query.c (get_traditional_extraction_insn): Likewise.
2354 (get_optab_extraction_insn): Likewise.
2355 * recog.c (simplify_while_replacing): Likewise.
2356 * expmed.c (narrow_bit_field_mem): Change the type of the mode
2357 parameter to opt_scalar_int_mode.
2359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2360 Alan Hayward <alan.hayward@arm.com>
2361 David Sherwood <david.sherwood@arm.com>
2363 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
2364 to a scalar_int_mode instead of a machine_mode.
2365 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
2366 (get_best_mode): Return a boolean and use a pointer argument to store
2367 the selected mode. Replace the limit mode parameter with a bit limit.
2368 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
2369 for the values returned by bit_field_mode_iterator::next_mode.
2370 (store_bit_field): Update call to get_best_mode.
2371 (store_fixed_bit_field): Likewise.
2372 (extract_fixed_bit_field): Likewise.
2373 * expr.c (optimize_bitfield_assignment_op): Likewise.
2374 * fold-const.c (optimize_bit_field_compare): Likewise.
2375 (fold_truth_andor_1): Likewise.
2376 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
2377 Update for new type of m_mode.
2378 (get_best_mode): As above.
2380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2381 Alan Hayward <alan.hayward@arm.com>
2382 David Sherwood <david.sherwood@arm.com>
2384 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
2385 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
2386 (store_bit_field): Check is_a <scalar_int_mode> before calling
2387 strict_volatile_bitfield_p.
2388 (extract_bit_field): Likewise.
2390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2391 Alan Hayward <alan.hayward@arm.com>
2392 David Sherwood <david.sherwood@arm.com>
2394 * target.def (cstore_mode): Return a scalar_int_mode.
2395 * doc/tm.texi: Regenerate.
2396 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
2397 * targhooks.h (default_cstore_mode): Likewise.
2398 * targhooks.c (default_cstore_mode): Likewise, using a forced
2400 * expmed.c (emit_cstore): Expect the target of the cstore to be
2403 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2404 Alan Hayward <alan.hayward@arm.com>
2405 David Sherwood <david.sherwood@arm.com>
2407 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
2409 (niter_desc): Likewise mode.
2410 (iv_analyze): Add a mode parameter.
2412 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
2413 and change its type to scalar_int_mode.
2414 * loop-iv.c: Update commentary at head of file.
2415 (iv_constant): Pass the mode paraeter before the rtx it describes
2416 and change its type to scalar_int_mode. Remove VOIDmode handling.
2417 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
2418 (iv_extend): Likewise.
2419 (shorten_into_mode): Likewise.
2420 (iv_add): Use scalar_int_mode.
2421 (iv_mult): Likewise.
2422 (iv_shift): Likewise.
2423 (canonicalize_iv_subregs): Likewise.
2424 (get_biv_step_1): Pass the outer_mode parameter before the rtx
2425 it describes and change its mode to scalar_int_mode. Also change
2426 the type of the returned inner_mode to scalar_int_mode.
2427 (get_biv_step): Likewise, turning outer_mode from a pointer
2428 into a direct parameter. Update call to get_biv_step_1.
2429 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
2430 iv_constant and get_biv_step.
2431 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
2432 and change its type to scalar_int_mode. Don't initialise iv->mode
2433 to VOIDmode and remove later checks for its still being VOIDmode.
2434 Update calls to iv_analyze_op and iv_analyze_expr. Check
2435 is_a <scalar_int_mode> when changing the mode under consideration.
2436 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
2437 Update call to iv_analyze_expr.
2438 (iv_analyze_op): Add a mode parameter. Reject subregs whose
2439 inner register is not also a scalar_int_mode. Update call to
2441 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
2442 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
2443 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
2444 separate mode class checks. Update calls to iv_analyze. Remove
2445 fix-up of VOIDmodes after iv_analyze_biv.
2446 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
2447 don't have a scalar_int_mode. Update call to biv_p.
2449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2450 Alan Hayward <alan.hayward@arm.com>
2451 David Sherwood <david.sherwood@arm.com>
2453 * cfgexpand.c (convert_debug_memory_address): Use
2454 as_a <scalar_int_mode>.
2455 * combine.c (expand_compound_operation): Likewise.
2456 (make_extraction): Likewise.
2457 (change_zero_ext): Likewise.
2458 (simplify_comparison): Likewise.
2459 * cse.c (cse_insn): Likewise.
2460 * dwarf2out.c (minmax_loc_descriptor): Likewise.
2461 (mem_loc_descriptor): Likewise.
2462 (loc_descriptor): Likewise.
2463 * expmed.c (init_expmed_one_mode): Likewise.
2464 (synth_mult): Likewise.
2465 (emit_store_flag_1): Likewise.
2466 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
2467 of a comparison with size.
2468 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
2469 (reduce_to_bit_field_precision): Likewise.
2470 * function.c (expand_function_end): Likewise.
2471 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
2472 * loop-doloop.c (doloop_modify): Likewise.
2473 * optabs.c (expand_binop): Likewise.
2474 (expand_unop): Likewise.
2475 (expand_copysign_absneg): Likewise.
2476 (prepare_cmp_insn): Likewise.
2477 (maybe_legitimize_operand): Likewise.
2478 * recog.c (const_scalar_int_operand): Likewise.
2479 * rtlanal.c (get_address_mode): Likewise.
2480 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2481 (simplify_cond_clz_ctz): Likewise.
2482 * tree-nested.c (get_nl_goto_field): Likewise.
2483 * tree.c (build_vector_type_for_mode): Likewise.
2484 * var-tracking.c (use_narrower_mode): Likewise.
2486 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2487 Alan Hayward <alan.hayward@arm.com>
2488 David Sherwood <david.sherwood@arm.com>
2490 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
2491 * builtins.c (expand_builtin_signbit): Use it.
2492 * cfgexpand.c (expand_debug_expr): Likewise.
2493 * dojump.c (do_jump): Likewise.
2494 (do_compare_and_jump): Likewise.
2495 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
2496 * expmed.c (make_tree): Likewise.
2497 * expr.c (expand_expr_real_2): Likewise.
2498 (expand_expr_real_1): Likewise.
2499 (try_casesi): Likewise.
2500 * fold-const-call.c (fold_const_call_ss): Likewise.
2501 * fold-const.c (unextend): Likewise.
2502 (extract_muldiv_1): Likewise.
2503 (fold_single_bit_test): Likewise.
2504 (native_encode_int): Likewise.
2505 (native_encode_string): Likewise.
2506 (native_interpret_int): Likewise.
2507 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
2508 * internal-fn.c (expand_addsub_overflow): Likewise.
2509 (expand_neg_overflow): Likewise.
2510 (expand_mul_overflow): Likewise.
2511 (expand_arith_overflow): Likewise.
2512 * match.pd: Likewise.
2513 * stor-layout.c (layout_type): Likewise.
2514 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2515 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
2516 * tree-ssanames.c (get_range_info): Likewise.
2517 * tree-switch-conversion.c (array_value_type) Likewise.
2518 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
2519 (vect_recog_divmod_pattern): Likewise.
2520 (vect_recog_mixed_size_cond_pattern): Likewise.
2521 * tree-vrp.c (extract_range_basic): Likewise.
2522 (simplify_float_conversion_using_ranges): Likewise.
2523 * tree.c (int_fits_type_p): Likewise.
2524 * ubsan.c (instrument_bool_enum_load): Likewise.
2525 * varasm.c (mergeable_string_section): Likewise.
2526 (narrowing_initializer_constant_valid_p): Likewise.
2527 (output_constant): Likewise.
2529 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2530 Alan Hayward <alan.hayward@arm.com>
2531 David Sherwood <david.sherwood@arm.com>
2533 * machmode.h (NARROWEST_INT_MODE): New macro.
2534 * expr.c (alignment_for_piecewise_move): Use it instead of
2535 GET_CLASS_NARROWEST_MODE (MODE_INT).
2536 (push_block): Likewise.
2537 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2539 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2541 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2542 Alan Hayward <alan.hayward@arm.com>
2543 David Sherwood <david.sherwood@arm.com>
2545 * postreload.c (move2add_valid_value_p): Change the type of the
2546 mode parameter to scalar_int_mode.
2547 (move2add_use_add2_insn): Add a mode parameter and use it instead
2549 (move2add_use_add3_insn): Likewise.
2550 (reload_cse_move2add): Update accordingly.
2552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2553 Alan Hayward <alan.hayward@arm.com>
2554 David Sherwood <david.sherwood@arm.com>
2556 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
2558 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
2559 * optabs.c (expand_unop): Likewise.
2561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2562 Alan Hayward <alan.hayward@arm.com>
2563 David Sherwood <david.sherwood@arm.com>
2565 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
2566 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
2567 (popcount_loc_descriptor): Likewise.
2568 (bswap_loc_descriptor): Likewise.
2569 (rotate_loc_descriptor): Likewise.
2570 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
2571 calling the functions above.
2573 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2574 Alan Hayward <alan.hayward@arm.com>
2575 David Sherwood <david.sherwood@arm.com>
2577 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
2579 (try_combine): Likewise.
2580 (simplify_if_then_else): Likewise.
2581 * cse.c (cse_insn): Likewise.
2582 * dwarf2out.c (mem_loc_descriptor): Likewise.
2583 * emit-rtl.c (gen_lowpart_common): Likewise.
2584 * simplify-rtx.c (simplify_truncation): Likewise.
2585 (simplify_binary_operation_1): Likewise.
2586 (simplify_const_relational_operation): Likewise.
2587 (simplify_ternary_operation): Likewise.
2588 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
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 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
2595 * reload.c (push_reload): Likewise.
2596 (find_reloads): Likewise.
2598 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2599 Alan Hayward <alan.hayward@arm.com>
2600 David Sherwood <david.sherwood@arm.com>
2602 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
2603 (make_compound_operation_int): Likewise.
2604 (change_zero_ext): Likewise.
2605 * expr.c (convert_move): Likewise.
2606 (convert_modes): Likewise.
2607 * fwprop.c (forward_propagate_subreg): Likewise.
2608 * loop-iv.c (get_biv_step_1): Likewise.
2609 * optabs.c (widen_operand): Likewise.
2610 * postreload.c (move2add_valid_value_p): Likewise.
2611 * recog.c (simplify_while_replacing): Likewise.
2612 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2613 (simplify_binary_operation_1): Likewise. Remove redundant
2614 mode equality check.
2616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2617 Alan Hayward <alan.hayward@arm.com>
2618 David Sherwood <david.sherwood@arm.com>
2620 * combine.c (combine_simplify_rtx): Add checks for
2621 is_a <scalar_int_mode>.
2622 (simplify_if_then_else): Likewise.
2623 (make_field_assignment): Likewise.
2624 (simplify_comparison): Likewise.
2625 * ifcvt.c (noce_try_bitop): Likewise.
2626 * loop-invariant.c (canonicalize_address_mult): Likewise.
2627 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2630 Alan Hayward <alan.hayward@arm.com>
2631 David Sherwood <david.sherwood@arm.com>
2633 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
2634 is_a <scalar_int_mode> instead of != BLKmode.
2636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2637 Alan Hayward <alan.hayward@arm.com>
2638 David Sherwood <david.sherwood@arm.com>
2640 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
2641 instead of != VOIDmode.
2642 * combine.c (if_then_else_cond): Likewise.
2643 (change_zero_ext): Likewise.
2644 * dwarf2out.c (mem_loc_descriptor): Likewise.
2645 (loc_descriptor): Likewise.
2646 * rtlanal.c (canonicalize_condition): Likewise.
2647 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
2649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2650 Alan Hayward <alan.hayward@arm.com>
2651 David Sherwood <david.sherwood@arm.com>
2653 * simplify-rtx.c (simplify_binary_operation_1): Use
2654 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
2656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2657 Alan Hayward <alan.hayward@arm.com>
2658 David Sherwood <david.sherwood@arm.com>
2660 * wide-int.h (int_traits<unsigned char>) New class.
2661 (int_traits<unsigned short>) Likewise.
2662 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
2663 Use GET_MODE_UNIT_PRECISION and remove redundant test for
2665 * combine.c (set_nonzero_bits_and_sign_copies): Use
2666 is_a <scalar_int_mode>.
2667 (find_split_point): Likewise.
2668 (combine_simplify_rtx): Likewise.
2669 (simplify_logical): Likewise.
2670 (expand_compound_operation): Likewise.
2671 (expand_field_assignment): Likewise.
2672 (make_compound_operation): Likewise.
2673 (extended_count): Likewise.
2674 (change_zero_ext): Likewise.
2675 (simplify_comparison): Likewise.
2676 * dwarf2out.c (scompare_loc_descriptor): Likewise.
2677 (ucompare_loc_descriptor): Likewise.
2678 (minmax_loc_descriptor): Likewise.
2679 (mem_loc_descriptor): Likewise.
2680 (loc_descriptor): Likewise.
2681 * expmed.c (init_expmed_one_mode): Likewise.
2682 * lra-constraints.c (lra_constraint_offset): Likewise.
2683 * optabs.c (prepare_libcall_arg): Likewise.
2684 * postreload.c (move2add_note_store): Likewise.
2685 * reload.c (operands_match_p): Likewise.
2686 * rtl.h (load_extend_op): Likewise.
2687 * rtlhooks.c (gen_lowpart_general): Likewise.
2688 * simplify-rtx.c (simplify_truncation): Likewise.
2689 (simplify_unary_operation_1): Likewise.
2690 (simplify_binary_operation_1): Likewise.
2691 (simplify_const_binary_operation): Likewise.
2692 (simplify_const_relational_operation): Likewise.
2693 (simplify_subreg): Likewise.
2694 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2695 * var-tracking.c (adjust_mems): Likewise.
2696 (prepare_call_arguments): Likewise.
2698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2699 Alan Hayward <alan.hayward@arm.com>
2700 David Sherwood <david.sherwood@arm.com>
2702 * machmode.h (is_int_mode): New fuction.
2703 * combine.c (find_split_point): Use it.
2704 (combine_simplify_rtx): Likewise.
2705 (simplify_if_then_else): Likewise.
2706 (simplify_set): Likewise.
2707 (simplify_shift_const_1): Likewise.
2708 (simplify_comparison): Likewise.
2709 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
2710 * cse.c (notreg_cost): Likewise.
2711 (cse_insn): Likewise.
2712 * cselib.c (cselib_lookup_1): Likewise.
2713 * dojump.c (do_jump_1): Likewise.
2714 (do_compare_rtx_and_jump): Likewise.
2715 * dse.c (get_call_args): Likewise.
2716 * dwarf2out.c (rtl_for_decl_init): Likewise.
2717 (native_encode_initializer): Likewise.
2718 * expmed.c (emit_store_flag_1): Likewise.
2719 (emit_store_flag): Likewise.
2720 * expr.c (convert_modes): Likewise.
2721 (store_field): Likewise.
2722 (expand_expr_real_1): Likewise.
2723 * fold-const.c (fold_read_from_constant_string): Likewise.
2724 * gimple-ssa-sprintf.c (get_format_string): Likewise.
2725 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
2726 * optabs.c (expand_binop): Likewise.
2727 (expand_unop): Likewise.
2728 (expand_abs_nojump): Likewise.
2729 (expand_one_cmpl_abs_nojump): Likewise.
2730 * simplify-rtx.c (mode_signbit_p): Likewise.
2731 (val_signbit_p): Likewise.
2732 (val_signbit_known_set_p): Likewise.
2733 (val_signbit_known_clear_p): Likewise.
2734 (simplify_relational_operation_1): Likewise.
2735 * tree.c (vector_type_mode): Likewise.
2737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2738 Alan Hayward <alan.hayward@arm.com>
2739 David Sherwood <david.sherwood@arm.com>
2741 * machmode.h (smallest_mode_for_size): Fix formatting.
2742 (smallest_int_mode_for_size): New function.
2743 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
2744 instead of smallest_mode_for_size.
2745 * combine.c (make_extraction): Likewise.
2746 * config/arc/arc.c (arc_expand_movmem): Likewise.
2747 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
2748 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2749 * config/s390/s390.c (s390_expand_insv): Likewise.
2750 * config/sparc/sparc.c (assign_int_registers): Likewise.
2751 * config/spu/spu.c (spu_function_value): Likewise.
2752 (spu_function_arg): Likewise.
2753 * coverage.c (get_gcov_type): Likewise.
2754 (get_gcov_unsigned_t): Likewise.
2755 * dse.c (find_shift_sequence): Likewise.
2756 * expmed.c (store_bit_field_1): Likewise.
2757 * expr.c (convert_move): Likewise.
2758 (store_field): Likewise.
2759 * internal-fn.c (expand_arith_overflow): Likewise.
2760 * optabs-query.c (get_best_extraction_insn): Likewise.
2761 * optabs.c (expand_twoval_binop_libfunc): Likewise.
2762 * stor-layout.c (layout_type): Likewise.
2763 (initialize_sizetypes): Likewise.
2764 * targhooks.c (default_get_mask_mode): Likewise.
2765 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
2767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2768 Alan Hayward <alan.hayward@arm.com>
2769 David Sherwood <david.sherwood@arm.com>
2771 * machmode.h (opt_mode::else_blk): New function.
2772 (int_mode_for_mode): Declare.
2773 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
2774 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
2776 * cfgexpand.c (expand_debug_expr): Likewise.
2777 * combine.c (gen_lowpart_or_truncate): Likewise.
2778 (gen_lowpart_for_combine): Likewise.
2779 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
2780 * config/avr/avr.c (avr_to_int_mode): Likewise.
2781 (avr_out_plus_1): Likewise.
2782 (avr_out_plus): Likewise.
2783 (avr_out_round): Likewise.
2784 * config/i386/i386.c (ix86_split_to_parts): Likewise.
2785 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2786 (s390_expand_vcond): Likewise.
2787 * config/spu/spu.c (spu_split_immediate): Likewise.
2788 (spu_expand_mov): Likewise.
2789 * dse.c (get_stored_val): Likewise.
2790 * expmed.c (store_bit_field_1): Likewise.
2791 (convert_extracted_bit_field): Use int_mode_for_mode instead of
2793 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
2794 (extract_low_bits): Likewise.
2795 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
2796 handling rather than repeating the check.
2797 (emit_group_store): Likewise.
2798 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
2799 * optabs.c (expand_absneg_bit): Likewise.
2800 (expand_copysign_absneg): Likewise.
2801 (expand_copysign_bit): Likewise.
2802 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
2803 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
2804 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
2805 * var-tracking.c (prepare_call_arguments): Likewise.
2806 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
2807 int_mode_for_mode instead of mode_for_size.
2808 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2810 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2811 Alan Hayward <alan.hayward@arm.com>
2812 David Sherwood <david.sherwood@arm.com>
2814 * machmode.h (int_mode_for_size): New function.
2815 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
2816 instead of mode_for_size.
2817 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
2819 * combine.c (expand_field_assignment): Use int_mode_for_size
2820 instead of mode_for_size.
2821 (make_extraction): Likewise.
2822 (simplify_shift_const_1): Likewise.
2823 (simplify_comparison): Likewise.
2824 * dojump.c (do_jump): Likewise.
2825 * dwarf2out.c (mem_loc_descriptor): Likewise.
2826 * emit-rtl.c (init_derived_machine_modes): Likewise.
2827 * expmed.c (flip_storage_order): Likewise.
2828 (convert_extracted_bit_field): Likewise.
2829 * expr.c (copy_blkmode_from_reg): Likewise.
2830 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
2831 * internal-fn.c (expand_mul_overflow): Likewise.
2832 * lower-subreg.c (simple_move): Likewise.
2833 * optabs-libfuncs.c (init_optabs): Likewise.
2834 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2835 * tree.c (vector_type_mode): Likewise.
2836 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
2837 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
2838 * tree-vect-generic.c (expand_vector_parallel): Likewise.
2839 * tree-vect-stmts.c (vectorizable_load): Likewise.
2840 (vectorizable_store): Likewise.
2842 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2843 Alan Hayward <alan.hayward@arm.com>
2844 David Sherwood <david.sherwood@arm.com>
2846 * coretypes.h (pod_mode): New type.
2847 (scalar_int_mode_pod): New typedef.
2848 * machmode.h (pod_mode): New class.
2849 (int_n_data_t::m): Change type to scalar_int_mode_pod.
2850 * genmodes.c (emit_mode_int_n): Update accordingly.
2851 * lower-subreg.h (target_lower_subreg): Change type to
2852 scalar_int_mode_pod.
2853 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
2854 scalar_int_mode_pod.
2856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2857 Alan Hayward <alan.hayward@arm.com>
2858 David Sherwood <david.sherwood@arm.com>
2860 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
2861 machine_mode to scalar_int_mode.
2862 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
2863 (rs6000_option_override_internal): Remove cast to int.
2864 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
2865 machine_mode to scalar_int_mode.
2866 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
2867 (rs6000_option_override_internal): Remove cast to int.
2868 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
2869 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
2871 * config/s390/s390.c (s390_expand_builtin): Likewise.
2872 * coretypes.h (scalar_int_mode): New type.
2873 (opt_scalar_int_mode): New typedef.
2874 * machmode.h (scalar_int_mode): New class.
2875 (scalar_int_mode::includes_p): New function.
2876 (byte_mode): Change type to scalar_int_mode.
2877 (word_mode): Likewise.
2878 (ptr_mode): Likewise.
2879 * emit-rtl.c (byte_mode): Likewise.
2880 (word_mode): Likewise.
2881 (ptr_mode): Likewise.
2882 (init_derived_machine_modes): Update accordingly.
2883 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
2884 and MODE_PARTIAL_INT.
2885 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
2886 opt_scalar_int_mode.
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 * target.def (libgcc_floating_mode_supported_p): Take a
2894 * doc/tm.texi: Regenerate.
2895 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
2897 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
2898 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
2901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2902 Alan Hayward <alan.hayward@arm.com>
2903 David Sherwood <david.sherwood@arm.com>
2905 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
2906 * doc/tm.texi: Regenerate.
2907 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
2908 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
2909 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
2910 * targhooks.h (default_floatn_mode): Likewise.
2911 * targhooks.c (default_floatn_mode): Likewise.
2912 * tree.c (build_common_tree_nodes): Update accordingly.
2914 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2915 Alan Hayward <alan.hayward@arm.com>
2916 David Sherwood <david.sherwood@arm.com>
2918 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
2919 (mode_iterator::iterate_p): Likewise.
2920 (mode_iterator::get_wider): Likewise.
2921 * expr.c (init_expr_target): Use opt_scalar_float_mode.
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 * coretypes.h (opt_scalar_float_mode): New typedef.
2928 * machmode.h (float_mode_for_size): New function.
2929 * emit-rtl.c (double_mode): Delete.
2930 (init_emit_once): Use float_mode_for_size.
2931 * stor-layout.c (layout_type): Likewise.
2932 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
2934 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2935 Alan Hayward <alan.hayward@arm.com>
2936 David Sherwood <david.sherwood@arm.com>
2938 * output.h (assemble_real): Take a scalar_float_mode.
2939 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
2940 * config/arm/arm.md (consttable_4): Likewise.
2941 (consttable_8): Likewise.
2942 (consttable_16): Likewise.
2943 * config/mips/mips.md (consttable_float): Likewise.
2944 * config/s390/s390.c (s390_output_pool_entry): Likewise.
2945 * varasm.c (assemble_real): Take a scalar_float_mode.
2946 (output_constant_pool_2): Update accordingly.
2947 (output_constant): Likewise.
2949 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2950 Alan Hayward <alan.hayward@arm.com>
2951 David Sherwood <david.sherwood@arm.com>
2953 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
2954 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
2955 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
2956 (native_encode_real): Likewise.
2957 (native_interpret_real): Likewise.
2958 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
2959 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2961 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2962 Alan Hayward <alan.hayward@arm.com>
2963 David Sherwood <david.sherwood@arm.com>
2965 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
2966 <scalar_float_mode>. Simplify.
2967 (gen_extend_conv_libfunc): Likewise.
2969 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2970 Alan Hayward <alan.hayward@arm.com>
2971 David Sherwood <david.sherwood@arm.com>
2973 * coretypes.h (scalar_float_mode): New type.
2974 * machmode.h (mode_traits::from_int): Use machine_mode if
2975 USE_ENUM_MODES is defined.
2976 (is_a): New function.
2978 (dyn_cast): Likewise.
2979 (scalar_float_mode): New class.
2980 (scalar_float_mode::includes_p): New function.
2981 (is_float_mode): Likewise.
2982 * gdbhooks.py (MachineModePrinter): New class.
2983 (build_pretty_printer): Use it for scalar_float_mode.
2984 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
2985 (format_helper::format_helper): Turn into a template.
2986 * genmodes.c (get_mode_class): New function.
2987 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
2988 or machine_mode if none.
2989 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
2990 as_a <scalar_float_mode>.
2991 * dwarf2out.c (mem_loc_descriptor): Likewise.
2992 (insert_float): Likewise.
2993 (add_const_value_attribute): Likewise.
2994 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2995 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
2996 (expand_unop): Update accordingly.
2997 (expand_abs_nojump): Likewise.
2998 (expand_copysign_absneg): Take a scalar_float_mode.
2999 (expand_copysign_bit): Likewise.
3000 (expand_copysign): Update accordingly.
3002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3003 Alan Hayward <alan.hayward@arm.com>
3004 David Sherwood <david.sherwood@arm.com>
3006 * coretypes.h (opt_mode): New class.
3007 * machmode.h (opt_mode): Likewise.
3008 (opt_mode::else_void): New function.
3009 (opt_mode::require): Likewise.
3010 (opt_mode::exists): Likewise.
3011 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
3012 (GET_MODE_2XWIDER_MODE): Likewise.
3013 (mode_iterator::get_wider): Update accordingly.
3014 (mode_iterator::get_2xwider): Likewise.
3015 (mode_iterator::get_known_wider): Likewise, turning into a template.
3016 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
3017 forcing a wider mode to exist.
3018 * config/cr16/cr16.h (LONG_REG_P): Likewise.
3019 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3020 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
3021 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
3022 * lower-subreg.c (init_lower_subreg): Likewise.
3023 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
3024 on the final iteration.
3025 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
3026 a wider mode exists before asking for a move pattern.
3027 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
3028 forcing a wider mode to exist.
3029 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
3030 returning false if no such mode exists.
3031 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
3032 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
3033 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
3034 Avoid checking for a MODE_INT if we already know the mode is not a
3036 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
3037 forcing a wider mode to exist.
3038 (expmed_mult_highpart_optab): Likewise.
3039 (expmed_mult_highpart): Likewise.
3040 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
3042 * lto-streamer-in.c (lto_input_mode_table): Likewise.
3043 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
3044 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3045 * internal-fn.c (expand_mul_overflow): Update use of
3046 GET_MODE_2XWIDER_MODE.
3047 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3048 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
3049 GET_MODE_WIDER_MODE.
3050 (convert_plusminus_to_widen): Likewise.
3051 * tree-switch-conversion.c (array_value_type): Likewise.
3052 * var-tracking.c (emit_note_insn_var_location): Likewise.
3053 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3054 Return false inside rather than outside the loop if no wider mode
3056 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
3057 and GET_MODE_2XWIDER_MODE
3058 (can_compare_p): Use else_void.
3059 * gdbhooks.py (OptMachineModePrinter): New class.
3060 (build_pretty_printer): Use it for opt_mode.
3062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3063 Alan Hayward <alan.hayward@arm.com>
3064 David Sherwood <david.sherwood@arm.com>
3066 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
3067 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
3069 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3070 Alan Hayward <alan.hayward@arm.com>
3071 David Sherwood <david.sherwood@arm.com>
3073 * machmode.h (mode_traits): New structure.
3074 (get_narrowest_mode): New function.
3075 (mode_iterator::start): Likewise.
3076 (mode_iterator::iterate_p): Likewise.
3077 (mode_iterator::get_wider): Likewise.
3078 (mode_iterator::get_known_wider): Likewise.
3079 (mode_iterator::get_2xwider): Likewise.
3080 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
3081 (FOR_EACH_MODE): Likewise.
3082 (FOR_EACH_MODE_FROM): Likewise.
3083 (FOR_EACH_MODE_UNTIL): Likewise.
3084 (FOR_EACH_WIDER_MODE): Likewise.
3085 (FOR_EACH_2XWIDER_MODE): Likewise.
3086 * builtins.c (expand_builtin_strlen): Use new mode iterators.
3087 * combine.c (simplify_comparison): Likewise
3088 * config/i386/i386.c (type_natural_mode): Likewise.
3089 * cse.c (cse_insn): Likewise.
3090 * dse.c (find_shift_sequence): Likewise.
3091 * emit-rtl.c (init_derived_machine_modes): Likewise.
3092 (init_emit_once): Likewise.
3093 * explow.c (hard_function_value): Likewise.
3094 * expmed.c (extract_fixed_bit_field_1): Likewise.
3095 (extract_bit_field_1): Likewise.
3096 (expand_divmod): Likewise.
3097 (emit_store_flag_1): Likewise.
3098 * expr.c (init_expr_target): Likewise.
3099 (convert_move): Likewise.
3100 (alignment_for_piecewise_move): Likewise.
3101 (widest_int_mode_for_size): Likewise.
3102 (emit_block_move_via_movmem): Likewise.
3103 (copy_blkmode_to_reg): Likewise.
3104 (set_storage_via_setmem): Likewise.
3105 (compress_float_constant): Likewise.
3106 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3107 * optabs-query.c (get_best_extraction_insn): Likewise.
3108 * optabs.c (expand_binop): Likewise.
3109 (expand_twoval_unop): Likewise.
3110 (expand_twoval_binop): Likewise.
3111 (widen_leading): Likewise.
3112 (widen_bswap): Likewise.
3113 (expand_parity): Likewise.
3114 (expand_unop): Likewise.
3115 (prepare_cmp_insn): Likewise.
3116 (prepare_float_lib_cmp): Likewise.
3117 (expand_float): Likewise.
3118 (expand_fix): Likewise.
3119 (expand_sfix_optab): Likewise.
3120 * postreload.c (move2add_use_add2_insn): Likewise.
3121 * reg-stack.c (reg_to_stack): Likewise.
3122 * reginfo.c (choose_hard_reg_mode): Likewise.
3123 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3124 * stor-layout.c (mode_for_size): Likewise.
3125 (smallest_mode_for_size): Likewise.
3126 (mode_for_vector): Likewise.
3127 (finish_bitfield_representative): Likewise.
3128 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
3129 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
3130 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3131 * var-tracking.c (prepare_call_arguments): Likewise.
3133 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3134 Alan Hayward <alan.hayward@arm.com>
3135 David Sherwood <david.sherwood@arm.com>
3137 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
3138 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
3139 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
3140 * machmode.h (mode_size): Move earlier in file.
3141 (mode_precision): Likewise.
3142 (mode_inner): Likewise.
3143 (mode_nunits): Likewise.
3144 (mode_unit_size): Likewise.
3145 (unit_unit_precision): Likewise.
3146 (mode_wider): Likewise.
3147 (mode_2xwider): Likewise.
3148 (machine_mode): New class.
3149 (mode_to_bytes): New function.
3150 (mode_to_bits): Likewise.
3151 (mode_to_precision): Likewise.
3152 (mode_to_inner): Likewise.
3153 (mode_to_unit_size): Likewise.
3154 (mode_to_unit_precision): Likewise.
3155 (mode_to_nunits): Likewise.
3156 (GET_MODE_SIZE): Use mode_to_bytes.
3157 (GET_MODE_BITSIZE): Use mode_to_bits.
3158 (GET_MODE_PRECISION): Use mode_to_precision.
3159 (GET_MODE_INNER): Use mode_to_inner.
3160 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
3161 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
3162 (GET_MODE_NUNITS): Use mode_to_nunits.
3163 * system.h (ALWAYS_INLINE): New macro.
3164 * config/powerpcspe/powerpcspe-c.c
3165 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
3166 int for arg1_mode and arg2_mode.
3168 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3169 Alan Hayward <alan.hayward@arm.com>
3170 David Sherwood <david.sherwood@arm.com>
3172 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
3173 Prefix mode names with E_ in case statements.
3174 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3175 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
3176 (aarch64_split_simd_move): Likewise.
3177 (aarch64_gen_storewb_pair): Likewise.
3178 (aarch64_gen_loadwb_pair): Likewise.
3179 (aarch64_gen_store_pair): Likewise.
3180 (aarch64_gen_load_pair): Likewise.
3181 (aarch64_get_condition_code_1): Likewise.
3182 (aarch64_constant_pool_reload_icode): Likewise.
3183 (get_rsqrte_type): Likewise.
3184 (get_rsqrts_type): Likewise.
3185 (get_recpe_type): Likewise.
3186 (get_recps_type): Likewise.
3187 (aarch64_gimplify_va_arg_expr): Likewise.
3188 (aarch64_simd_container_mode): Likewise.
3189 (aarch64_emit_load_exclusive): Likewise.
3190 (aarch64_emit_store_exclusive): Likewise.
3191 (aarch64_expand_compare_and_swap): Likewise.
3192 (aarch64_gen_atomic_cas): Likewise.
3193 (aarch64_emit_bic): Likewise.
3194 (aarch64_emit_atomic_swap): Likewise.
3195 (aarch64_emit_atomic_load_op): Likewise.
3196 (aarch64_evpc_trn): Likewise.
3197 (aarch64_evpc_uzp): Likewise.
3198 (aarch64_evpc_zip): Likewise.
3199 (aarch64_evpc_ext): Likewise.
3200 (aarch64_evpc_rev): Likewise.
3201 (aarch64_evpc_dup): Likewise.
3202 (aarch64_gen_ccmp_first): Likewise.
3203 (aarch64_gen_ccmp_next): Likewise.
3204 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3205 (alpha_emit_xfloating_libcall): Likewise.
3206 (emit_insxl): Likewise.
3207 (alpha_arg_type): Likewise.
3208 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
3209 (arc_preferred_simd_mode): Likewise.
3210 (arc_secondary_reload): Likewise.
3211 (get_arc_condition_code): Likewise.
3212 (arc_print_operand): Likewise.
3213 (arc_legitimate_constant_p): Likewise.
3214 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3215 * config/arc/arc.md (casesi_load): Likewise.
3216 (casesi_compact_jump): Likewise.
3217 * config/arc/predicates.md (proper_comparison_operator): Likewise.
3218 (cc_use_register): Likewise.
3219 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3220 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
3221 (arm_init_iwmmxt_builtins): Likewise.
3222 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
3223 (neon_expand_vector_init): Likewise.
3224 (arm_attr_length_move_neon): Likewise.
3225 (maybe_get_arm_condition_code): Likewise.
3226 (arm_emit_vector_const): Likewise.
3227 (arm_preferred_simd_mode): Likewise.
3228 (arm_output_iwmmxt_tinsr): Likewise.
3229 (thumb1_output_casesi): Likewise.
3230 (thumb2_output_casesi): Likewise.
3231 (arm_emit_load_exclusive): Likewise.
3232 (arm_emit_store_exclusive): Likewise.
3233 (arm_expand_compare_and_swap): Likewise.
3234 (arm_evpc_neon_vuzp): Likewise.
3235 (arm_evpc_neon_vzip): Likewise.
3236 (arm_evpc_neon_vrev): Likewise.
3237 (arm_evpc_neon_vtrn): Likewise.
3238 (arm_evpc_neon_vext): Likewise.
3239 (arm_validize_comparison): Likewise.
3240 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
3241 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
3242 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
3243 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
3244 (c6x_preferred_simd_mode): Likewise.
3245 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
3246 (epiphany_rtx_costs): Likewise.
3247 * config/epiphany/predicates.md (proper_comparison_operator):
3249 * config/frv/frv.c (condexec_memory_operand): Likewise.
3250 (frv_emit_move): Likewise.
3251 (output_move_single): Likewise.
3252 (output_condmove_single): Likewise.
3253 (frv_hard_regno_mode_ok): Likewise.
3254 (frv_matching_accg_mode): Likewise.
3255 * config/h8300/h8300.c (split_adds_subs): Likewise.
3256 (h8300_rtx_costs): Likewise.
3257 (h8300_print_operand): Likewise.
3258 (compute_mov_length): Likewise.
3259 (output_logical_op): Likewise.
3260 (compute_logical_op_length): Likewise.
3261 (compute_logical_op_cc): Likewise.
3262 (h8300_shift_needs_scratch_p): Likewise.
3263 (output_a_shift): Likewise.
3264 (compute_a_shift_length): Likewise.
3265 (compute_a_shift_cc): Likewise.
3266 (expand_a_rotate): Likewise.
3267 (output_a_rotate): Likewise.
3268 * config/i386/i386.c (classify_argument): Likewise.
3269 (function_arg_advance_32): Likewise.
3270 (function_arg_32): Likewise.
3271 (function_arg_64): Likewise.
3272 (function_value_64): Likewise.
3273 (ix86_gimplify_va_arg): Likewise.
3274 (ix86_legitimate_constant_p): Likewise.
3275 (put_condition_code): Likewise.
3276 (split_double_mode): Likewise.
3277 (ix86_avx256_split_vector_move_misalign): Likewise.
3278 (ix86_expand_vector_logical_operator): Likewise.
3279 (ix86_split_idivmod): Likewise.
3280 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
3281 (ix86_build_const_vector): Likewise.
3282 (ix86_build_signbit_mask): Likewise.
3283 (ix86_match_ccmode): Likewise.
3284 (ix86_cc_modes_compatible): Likewise.
3285 (ix86_expand_branch): Likewise.
3286 (ix86_expand_sse_cmp): Likewise.
3287 (ix86_expand_sse_movcc): Likewise.
3288 (ix86_expand_int_sse_cmp): Likewise.
3289 (ix86_expand_vec_perm_vpermi2): Likewise.
3290 (ix86_expand_vec_perm): Likewise.
3291 (ix86_expand_sse_unpack): Likewise.
3292 (ix86_expand_int_addcc): Likewise.
3293 (ix86_split_to_parts): Likewise.
3294 (ix86_vectorize_builtin_gather): Likewise.
3295 (ix86_vectorize_builtin_scatter): Likewise.
3296 (avx_vpermilp_parallel): Likewise.
3297 (inline_memory_move_cost): Likewise.
3298 (ix86_tieable_integer_mode_p): Likewise.
3299 (x86_maybe_negate_const_int): Likewise.
3300 (ix86_expand_vector_init_duplicate): Likewise.
3301 (ix86_expand_vector_init_one_nonzero): Likewise.
3302 (ix86_expand_vector_init_one_var): Likewise.
3303 (ix86_expand_vector_init_concat): Likewise.
3304 (ix86_expand_vector_init_interleave): Likewise.
3305 (ix86_expand_vector_init_general): Likewise.
3306 (ix86_expand_vector_set): Likewise.
3307 (ix86_expand_vector_extract): Likewise.
3308 (emit_reduc_half): Likewise.
3309 (ix86_emit_i387_round): Likewise.
3310 (ix86_mangle_type): Likewise.
3311 (ix86_expand_round_sse4): Likewise.
3312 (expand_vec_perm_blend): Likewise.
3313 (canonicalize_vector_int_perm): Likewise.
3314 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
3315 (expand_vec_perm_1): Likewise.
3316 (expand_vec_perm_interleave3): Likewise.
3317 (expand_vec_perm_even_odd_pack): Likewise.
3318 (expand_vec_perm_even_odd_1): Likewise.
3319 (expand_vec_perm_broadcast_1): Likewise.
3320 (ix86_vectorize_vec_perm_const_ok): Likewise.
3321 (ix86_expand_vecop_qihi): Likewise.
3322 (ix86_expand_mul_widen_hilo): Likewise.
3323 (ix86_expand_sse2_abs): Likewise.
3324 (ix86_expand_pextr): Likewise.
3325 (ix86_expand_pinsr): Likewise.
3326 (ix86_preferred_simd_mode): Likewise.
3327 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
3328 * config/i386/sse.md (*andnot<mode>3): Likewise.
3329 (<mask_codefor><code><mode>3<mask_name>): Likewise.
3330 (*<code><mode>3): Likewise.
3331 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
3332 (ia64_expand_atomic_op): Likewise.
3333 (ia64_arg_type): Likewise.
3334 (ia64_mode_to_int): Likewise.
3335 (ia64_scalar_mode_supported_p): Likewise.
3336 (ia64_vector_mode_supported_p): Likewise.
3337 (expand_vec_perm_broadcast): Likewise.
3338 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
3339 (iq2000_function_arg_advance): Likewise.
3340 (iq2000_function_arg): Likewise.
3341 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
3342 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
3343 (m68k_libcall_value): Likewise.
3344 (m68k_function_value): Likewise.
3345 (sched_attr_op_type): Likewise.
3346 * config/mcore/mcore.c (mcore_output_move): Likewise.
3347 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
3349 (microblaze_function_arg): Likewise.
3350 * config/mips/mips.c (mips16_build_call_stub): Likewise.
3351 (mips_print_operand): Likewise.
3352 (mips_mode_ok_for_mov_fmt_p): Likewise.
3353 (mips_vector_mode_supported_p): Likewise.
3354 (mips_preferred_simd_mode): Likewise.
3355 (mips_expand_vpc_loongson_even_odd): Likewise.
3356 (mips_expand_vec_unpack): Likewise.
3357 (mips_expand_vi_broadcast): Likewise.
3358 (mips_expand_vector_init): Likewise.
3359 (mips_expand_vec_reduc): Likewise.
3360 (mips_expand_msa_cmp): Likewise.
3361 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
3362 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
3363 (cc_flags_for_mode): Likewise.
3364 * config/msp430/msp430.c (msp430_print_operand): Likewise.
3365 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
3366 (nds32_output_casesi_pc_relative): Likewise.
3367 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3368 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
3369 (nvptx_gen_unpack): Likewise.
3370 (nvptx_gen_pack): Likewise.
3371 (nvptx_gen_shuffle): Likewise.
3372 (nvptx_gen_wcast): Likewise.
3373 (nvptx_preferred_simd_mode): Likewise.
3374 * config/pa/pa.c (pa_secondary_reload): Likewise.
3375 * config/pa/predicates.md (base14_operand): Likewise.
3376 * config/powerpcspe/powerpcspe-c.c
3377 (altivec_resolve_overloaded_builtin): Likewise.
3378 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
3380 (rs6000_preferred_simd_mode): Likewise.
3381 (output_vec_const_move): Likewise.
3382 (rs6000_expand_vector_extract): Likewise.
3383 (rs6000_split_vec_extract_var): Likewise.
3384 (reg_offset_addressing_ok_p): Likewise.
3385 (rs6000_legitimate_offset_address_p): Likewise.
3386 (rs6000_legitimize_address): Likewise.
3387 (rs6000_emit_set_const): Likewise.
3388 (rs6000_const_vec): Likewise.
3389 (rs6000_emit_move): Likewise.
3390 (spe_build_register_parallel): Likewise.
3391 (rs6000_darwin64_record_arg_recurse): Likewise.
3392 (swap_selector_for_mode): Likewise.
3393 (spe_init_builtins): Likewise.
3394 (paired_init_builtins): Likewise.
3395 (altivec_init_builtins): Likewise.
3396 (do_load_for_compare): Likewise.
3397 (rs6000_generate_compare): Likewise.
3398 (rs6000_expand_float128_convert): Likewise.
3399 (emit_load_locked): Likewise.
3400 (emit_store_conditional): Likewise.
3401 (rs6000_output_function_epilogue): Likewise.
3402 (rs6000_handle_altivec_attribute): Likewise.
3403 (rs6000_function_value): Likewise.
3404 (emit_fusion_gpr_load): Likewise.
3405 (emit_fusion_p9_load): Likewise.
3406 (emit_fusion_p9_store): Likewise.
3407 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
3408 (fusion_gpr_mem_load): Likewise.
3409 (fusion_addis_mem_combo_load): Likewise.
3410 (fusion_addis_mem_combo_store): Likewise.
3411 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
3412 (fusion_gpr_mem_load): Likewise.
3413 (fusion_addis_mem_combo_load): Likewise.
3414 (fusion_addis_mem_combo_store): Likewise.
3415 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3417 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
3418 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
3419 (rs6000_preferred_simd_mode): Likewise.
3420 (output_vec_const_move): Likewise.
3421 (rs6000_expand_vector_extract): Likewise.
3422 (rs6000_split_vec_extract_var): Likewise.
3423 (reg_offset_addressing_ok_p): Likewise.
3424 (rs6000_legitimate_offset_address_p): Likewise.
3425 (rs6000_legitimize_address): Likewise.
3426 (rs6000_emit_set_const): Likewise.
3427 (rs6000_const_vec): Likewise.
3428 (rs6000_emit_move): Likewise.
3429 (rs6000_darwin64_record_arg_recurse): Likewise.
3430 (swap_selector_for_mode): Likewise.
3431 (paired_init_builtins): Likewise.
3432 (altivec_init_builtins): Likewise.
3433 (rs6000_expand_float128_convert): Likewise.
3434 (emit_load_locked): Likewise.
3435 (emit_store_conditional): Likewise.
3436 (rs6000_output_function_epilogue): Likewise.
3437 (rs6000_handle_altivec_attribute): Likewise.
3438 (rs6000_function_value): Likewise.
3439 (emit_fusion_gpr_load): Likewise.
3440 (emit_fusion_p9_load): Likewise.
3441 (emit_fusion_p9_store): Likewise.
3442 * config/rx/rx.c (rx_gen_move_template): Likewise.
3443 (flags_from_mode): Likewise.
3444 * config/s390/predicates.md (s390_alc_comparison): Likewise.
3445 (s390_slb_comparison): Likewise.
3446 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
3447 (s390_vector_mode_supported_p): Likewise.
3448 (s390_cc_modes_compatible): Likewise.
3449 (s390_match_ccmode_set): Likewise.
3450 (s390_canonicalize_comparison): Likewise.
3451 (s390_emit_compare_and_swap): Likewise.
3452 (s390_branch_condition_mask): Likewise.
3453 (s390_rtx_costs): Likewise.
3454 (s390_secondary_reload): Likewise.
3455 (__SECONDARY_RELOAD_CASE): Likewise.
3456 (s390_expand_cs): Likewise.
3457 (s390_preferred_simd_mode): Likewise.
3458 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
3459 * config/sh/sh.c (sh_print_operand): Likewise.
3460 (dump_table): Likewise.
3461 (sh_secondary_reload): Likewise.
3462 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3463 * config/sh/sh.md (casesi_worker_1): Likewise.
3464 (casesi_worker_2): Likewise.
3465 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
3466 (fcc_comparison_operator): Likewise.
3467 * config/sparc/sparc.c (sparc_expand_move): Likewise.
3468 (emit_soft_tfmode_cvt): Likewise.
3469 (sparc_preferred_simd_mode): Likewise.
3470 (output_cbranch): Likewise.
3471 (sparc_print_operand): Likewise.
3472 (sparc_expand_vec_perm_bmask): Likewise.
3473 (vector_init_bshuffle): Likewise.
3474 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3475 (spu_vector_mode_supported_p): Likewise.
3476 (spu_expand_insv): Likewise.
3477 (spu_emit_branch_or_set): Likewise.
3478 (spu_handle_vector_attribute): Likewise.
3479 (spu_builtin_splats): Likewise.
3480 (spu_builtin_extract): Likewise.
3481 (spu_builtin_promote): Likewise.
3482 (spu_expand_sign_extend): Likewise.
3483 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3484 (tilegx_simd_int): Likewise.
3485 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
3486 (tilepro_simd_int): Likewise.
3487 * config/v850/v850.c (const_double_split): Likewise.
3488 (v850_print_operand): Likewise.
3489 (ep_memory_offset): Likewise.
3490 * config/vax/vax.c (vax_rtx_costs): Likewise.
3491 (vax_output_int_move): Likewise.
3492 (vax_output_int_add): Likewise.
3493 (vax_output_int_subtract): Likewise.
3494 * config/visium/predicates.md (visium_branch_operator): Likewise.
3495 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
3496 (visium_print_operand_address): Likewise.
3497 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3498 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
3499 (xtensa_expand_conditional_branch): Likewise.
3500 (xtensa_copy_incoming_a7): Likewise.
3501 (xtensa_output_literal): Likewise.
3502 * dfp.c (decimal_real_maxval): Likewise.
3503 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3505 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3506 Alan Hayward <alan.hayward@arm.com>
3507 David Sherwood <david.sherwood@arm.com>
3509 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
3510 (mode_nunits_inline): Likewise.
3511 (mode_inner_inline): Likewise.
3512 (mode_unit_size_inline): Likewise.
3513 (mode_unit_precision_inline): Likewise.
3514 (emit_insn_modes_h): Likewise. Also emit a #define of the
3516 (emit_mode_wider): Add an E_ prefix to mode names.
3517 (emit_mode_complex): Likewise.
3518 (emit_mode_inner): Likewise.
3519 (emit_mode_adjustments): Likewise.
3520 (emit_mode_int_n): Likewise.
3521 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
3522 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
3523 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
3524 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
3525 (CRC32_BUILTIN, ENTRY): Likewise.
3526 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
3527 (aarch64_pop_regs): Likewise.
3528 (aarch64_process_components): Likewise.
3529 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
3530 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
3531 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
3532 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
3533 * config/arm/arm.c (arm_init_libfuncs): Likewise.
3534 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
3536 * config/i386/i386-builtin.def (pcmpestr): Likewise.
3537 (pcmpistr): Likewise.
3538 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
3539 * config/mmix/mmix.c (mmix_output_condition): Likewise.
3540 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
3542 * config/rl78/rl78.c (mduc_regs): Likewise.
3543 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3544 (htm_expand_builtin): Likewise.
3545 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
3546 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
3547 * config/xtensa/xtensa.c (print_operand): Likewise.
3548 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
3549 (NUM_MODE_VECTOR_INT): Likewise.
3550 * genoutput.c (null_operand): Likewise.
3551 (output_operand_data): Likewise.
3552 * genrecog.c (print_parameter_value): Likewise.
3553 * lra.c (debug_operand_data): Likewise.
3555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3557 * dwarf2out.c (scompare_loc_descriptor_wide)
3558 (scompare_loc_descriptor_narrow): New functions, split out from...
3559 (scompare_loc_descriptor): ...here.
3560 * expmed.c (emit_store_flag_int): New function, split out from...
3561 (emit_store_flag): ...here.
3563 2017-08-30 Richard Biener <rguenther@suse.de>
3565 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
3566 (dwarf2out_early_finish): Move setting of AT_pubnames from
3567 early debug output to early finish.
3569 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3571 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
3572 and -mdata-region to the assembler.
3574 2017-08-30 Richard Biener <rguenther@suse.de>
3576 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
3578 (gen_subprogram_die): Add DW_AT_object_pointer only early.
3579 (dwarf2out_early_global_decl): Only generate a DIE for the
3580 abstract origin if it doesn't already exist or is a declaration DIE.
3581 (resolve_addr): Do not add the linkage name twice when
3582 generating a stub DIE for the DW_TAG_GNU_call_site target.
3584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3586 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3587 Use machine_mode rather than int for arg1_mode.
3589 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
3592 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
3593 that the second argument of the built-in functions to unpack
3594 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
3595 switch statement instead a lot of if statements.
3596 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
3597 Allow 64-bit values to be in Altivec registers as well as
3598 traditional floating point registers.
3599 (pack<mode>, FMOVE128_VSX iterator): Likewise.
3601 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
3603 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
3604 MAX_REGS_PER_ADDRESS == 1.
3606 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
3608 * config/i386/i386.opt (flag_fentry): Do not init to -1.
3609 * config/i386/i386.c (ix86_option_override_internal): Simplify
3610 setting of opts->x_flag_entry.
3612 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3613 Jakub Jelinek <jakub@redhat.com>
3614 Richard Biener <rguenther@suse.de>
3616 PR tree-optimization/81503
3617 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
3618 folded constant fits in the target type; reorder tests for clarity.
3620 2017-08-29 Martin Liska <mliska@suse.cz>
3622 * passes.def: Include pass_lower_switch.
3623 * stmt.c (dump_case_nodes): Remove and move to
3624 tree-switch-conversion.
3625 (case_values_threshold): Likewise.
3626 (expand_switch_as_decision_tree_p): Likewise.
3627 (emit_case_decision_tree): Likewise.
3628 (expand_case): Likewise.
3629 (balance_case_nodes): Likewise.
3630 (node_has_low_bound): Likewise.
3631 (node_has_high_bound): Likewise.
3632 (node_is_bounded): Likewise.
3633 (emit_case_nodes): Likewise.
3634 (struct simple_case_node): New struct.
3635 (add_case_node): Remove.
3636 (emit_case_dispatch_table): Use vector instead of case_list.
3637 (reset_out_edges_aux): Remove.
3638 (compute_cases_per_edge): Likewise.
3639 (expand_case): Build list of simple_case_node.
3640 (expand_sjlj_dispatch_table): Use it.
3641 * tree-switch-conversion.c (struct case_node): Moved from
3642 stmt.c and adjusted.
3643 (emit_case_nodes): Likewise.
3644 (node_has_low_bound): Likewise.
3645 (node_has_high_bound): Likewise.
3646 (node_is_bounded): Likewise.
3647 (case_values_threshold): Likewise.
3648 (reset_out_edges_aux): Likewise.
3649 (compute_cases_per_edge): Likewise.
3650 (add_case_node): Likewise.
3651 (dump_case_nodes): Likewise.
3652 (balance_case_nodes): Likewise.
3653 (expand_switch_as_decision_tree_p): Likewise.
3654 (emit_jump): Likewise.
3655 (emit_case_decision_tree): Likewise.
3656 (try_switch_expansion): Likewise.
3657 (do_jump_if_equal): Likewise.
3658 (emit_cmp_and_jump_insns): Likewise.
3659 (fix_phi_operands_for_edge): New function.
3660 (record_phi_operand_mapping): Likewise.
3661 (class pass_lower_switch): New pass.
3662 (pass_lower_switch::execute): New function.
3663 (make_pass_lower_switch): Likewise.
3664 (conditional_probability):
3665 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
3666 * tree-pass.h: Add make_pass_lower_switch.
3668 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3671 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
3674 2017-08-29 Richard Biener <rguenther@suse.de>
3676 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
3677 we do not add a DW_AT_inline attribute twice.
3678 (gen_subprogram_die): Remove code setting DW_AT_inline on
3679 DECL_ABSTRACT_P nodes.
3681 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
3683 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
3684 calls to internal functions.
3685 (gimplify_modify_expr): Likewise.
3686 * tree-call-cdce.c (use_internal_fn): Likewise.
3687 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
3688 (convert_to_divmod): Set the nothrow flag.
3689 * tree-if-conv.c (predicate_mem_writes): Likewise.
3690 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3691 (vectorizable_call): Likewise.
3692 (vectorizable_store): Likewise.
3693 (vectorizable_load): Likewise.
3694 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
3695 (vect_recog_mask_conversion_pattern): Likewise.
3697 2017-08-29 Martin Liska <mliska@suse.cz>
3700 * gcc.c (driver_handle_option): Add new argument.
3701 * opts-common.c (handle_option): Pass
3702 target_option_override_hook.
3703 * opts-global.c (lang_handle_option): Add new option.
3704 (set_default_handlers): Add new argument.
3705 (decode_options): Likewise.
3706 * opts.c (target_handle_option): Likewise.
3707 (common_handle_option): Call target_option_override_hook.
3708 * opts.h (struct cl_option_handler_func): Add hook for
3709 target option override.
3710 (struct cl_option_handlers): Likewise.
3711 (set_default_handlers): Add new argument.
3712 (decode_options): Likewise.
3713 (common_handle_option): Likewise.
3714 (target_handle_option): Likewise.
3715 * toplev.c (toplev::main): Pass targetm.target_option.override
3718 2017-08-29 Richard Biener <rguenther@suse.de>
3719 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3721 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
3722 life to the active subtree.
3724 2017-08-28 Jeff Law <law@redhat.com>
3726 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
3727 derive_equivalences.
3728 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
3730 (edge_info::derive_equivalences): New private member function
3732 * tree-ssa-dom.c (class edge_info): Changed from a struct
3733 to a class. Add ctor/dtor, methods and data members.
3734 (edge_info::edge_info): Renamed from allocate_edge_info.
3735 Initialize additional members.
3736 (edge_info::~edge_info): New.
3737 (free_dom_edge_info): Delete the edge info.
3738 (record_edge_info): Use new class & associated member functions.
3739 Tighten forms for testing for edge equivalences.
3740 (record_temporary_equivalences): Iterate over the simple
3741 equivalences rather than assuming there's only one per edge.
3742 (cprop_into_successor_phis): Iterate over the simple
3743 equivalences rather than assuming there's only one per edge.
3744 (optimize_stmt): Use operand_equal_p rather than pointer
3745 equality for mini-DSE code.
3747 2017-08-28 Nathan Sidwell <nathan@acm.org>
3749 * gcc.c (execute): Fold SIGPIPE handling into switch
3750 statement. Adjust internal error message.
3752 2017-08-28 Richard Biener <rguenther@suse.de>
3755 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
3756 Do nothing for removed DIEs.
3758 2017-08-28 Richard Biener <rguenther@suse.de>
3760 PR tree-optimization/81977
3761 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
3764 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
3767 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
3768 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
3769 using targetm.gen_mem_thread_fence.
3771 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
3774 * config/i386/i386.md (*<btsc><mode>): Change operand 2
3775 predicate to register_operand. Reorder operands.
3776 (*btr<mode>): Ditto.
3777 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
3778 (*btr<mode>_mask): Ditto.
3780 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
3782 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
3783 * config/rs6000/xmmintrin.h: New file.
3784 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
3786 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3789 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
3790 parameter and_insn and return it.
3791 (recombine_lvx_pattern): Insert a copy to ensure availability of
3792 the base register of the copied masking operation at the point of
3793 the instruction replacement.
3794 (recombine_stvx_pattern): Likewise.
3796 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3798 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
3799 undocumented switches.
3800 (-mpower9-dform-vector): Likewise.
3801 (-mpower9-dform): Likewise.
3802 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
3803 comments to delete references to -mpower9-dform* switches.
3804 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
3805 Delete reference to -mpower9-dform* switches, test for
3806 -mpower9-vector instead.
3807 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
3808 (OTHER_P9_VECTOR_MASKS): Likewise.
3809 (POWERPC_MASKS): Likewise.
3810 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
3811 tests against -mpower9-dform* to -mpower9-vector. Delete code
3812 that checked for -mpower9-dform* consistancy with other options.
3813 Add test for -mpower9-misc to enable other power9 switches.
3814 (rs6000_init_hard_regno_mode_ok): Likewise.
3815 (rs6000_option_override_internal): Likewise.
3816 (rs6000_emit_prologue): Likewise.
3817 (rs6000_emit_epilogue): Likewise.
3818 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
3819 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
3820 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
3822 (emit_fusion_p9_store): Likewise.
3823 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
3824 resetting these macros if the assembler does not support ISA 3.0
3826 (TARGET_P9_DFORM_VECTOR): Likewise.
3827 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
3828 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
3830 2017-08-25 Alan Modra <amodra@gmail.com>
3832 PR rtl-optimization/81747
3833 * cse.c (cse_extended_basic_block): Don't attempt to record
3834 equivalences for degenerate conditional branches that branch
3835 to their fall-through.
3837 2017-08-24 Martin Sebor <msebor@redhat.com>
3840 * gimple-fold.c (size_must_be_zero_p): New function.
3841 (gimple_fold_builtin_memory_op): Call it.
3843 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
3845 * config/rs6000/mm_malloc.h: New file.
3847 2017-08-24 Bin Cheng <bin.cheng@arm.com>
3849 PR tree-optimization/81913
3850 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
3851 analysis when either IVs in condition can wrap.
3853 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
3855 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
3856 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
3858 2017-08-24 Richard Biener <rguenther@suse.de>
3861 * targhooks.c (default_target_can_inline_p): Properly
3862 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
3863 is present and always compare.
3864 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
3865 infer -mfpmath=sse from TARGET_SSE_P.
3866 (ix86_can_inline_p): Properly use target_option_default_node when
3867 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
3869 2017-08-24 Richard Biener <rguenther@suse.de>
3872 * dwarf2out.c (output_die): Handle flag_generate_offload like
3874 (output_comp_unit): Likewise.
3875 (gen_array_type_die): Likewise.
3876 (dwarf2out_early_finish): Likewise.
3877 (note_variable_value_in_expr): Likewise.
3878 (dwarf2out_finish): Likewise. Adjust assert.
3879 * cgraphunit.c (symbol_table::compile): Move setting of
3880 flag_generate_offload earlier ...
3881 (symbol_table::finalize_compilation_unit): ... here, before
3882 early debug finalization.
3884 2017-08-24 Richard Biener <rguenther@suse.de>
3886 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
3887 and ipa-fnsummary.h.
3888 (ix86_can_inline_p): When ix86_fpmath flags do not match
3889 check whether the callee uses FP math at all.
3891 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
3894 * tree-ssanames.c (get_nonzero_bits): Use element_precision
3895 instead of TYPE_PRECISION.
3897 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
3898 Alan Hayward <alan.hayward@arm.com>
3899 David Sherwood <david.sherwood@arm.com>
3901 * combine.c (make_extraction): Use subreg_offset_from_lsb.
3903 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
3905 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
3907 (ix86_frame::stack_realign_allocate): New field.
3908 (struct machine_frame_state): Modify comments.
3909 (machine_frame_state::sp_realigned_fp_end): New field.
3910 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
3912 (sp_valid_at): Add assertion to assure no attempt to access invalid
3913 offset of a realigned stack.
3914 (fp_valid_at): Likewise.
3915 (choose_baseaddr): Modify comments.
3916 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
3917 ix86_expand_prologue.
3918 (ix86_expand_prologue): Modify stack realignment and allocation.
3919 (ix86_expand_epilogue): Modify comments.
3920 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
3921 avx2_runtime, avx512f, and avx512f_runtime.
3923 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
3925 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
3926 (mstackrealign): Do not init to -1.
3927 * config/i386/i386.c (ix86_option_override_internal):
3928 Check opts_set, not opts when setting default value of
3929 opts->x_ix86_force_align_arg_pointer.
3931 2017-08-23 Richard Biener <rguenther@suse.de>
3933 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
3934 lang_hooks.decl_printable_name.
3935 * print-rtl-function.c (print_rtx_function): Likewise.
3936 * tree-pretty-print.c (dump_function_header): Likewise.
3938 2017-08-23 Richard Biener <rguenther@suse.de>
3941 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
3942 -g0 at compile-time.
3944 2017-08-23 Tamar Christina <tamar.christina@arm.com>
3947 * doc/sourcebuild.texi (Other hardware attributes):
3950 2017-08-23 Tamar Christina <tamar.christina@arm.com>
3953 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
3954 Fix single-use check.
3956 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
3958 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
3960 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
3962 * doc/install.texi: Modify to add more details on running selected
3965 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
3967 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
3968 is combined with -mabi=ms.
3969 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
3972 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3974 PR tree-optimization/81488
3975 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
3976 and cached_basis fields.
3977 (MAX_SPREAD): New constant.
3978 (alloc_cand_and_find_basis): Initialize new fields.
3979 (clear_visited): New function.
3980 (create_phi_basis_1): Rename from create_phi_basis, set visited
3981 and cached_basis fields.
3982 (create_phi_basis): New wrapper function.
3983 (phi_add_costs_1): Rename from phi_add_costs, add spread
3984 parameter, set visited field, short-circuit when limits reached.
3985 (phi_add_costs): New wrapper function.
3986 (record_phi_increments_1): Rename from record_phi_increments, set
3988 (record_phi_increments): New wrapper function.
3989 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
3990 (phi_incr_cost): New wrapper function.
3991 (all_phi_incrs_profitable_1): Rename from
3992 all_phi_incrs_profitable, set visited field.
3993 (all_phi_incrs_profitable): New wrapper function.
3995 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
3996 Alan Hayward <alan.hayward@arm.com>
3997 David Sherwood <david.sherwood@arm.com>
3999 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
4000 that takes the outer and inner modes.
4001 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
4002 comparison as the canonical test for a paradoxical subreg.
4003 * combine.c (simplify_set): Use paradoxical_subreg_p.
4004 (make_extraction): Likewise.
4005 (force_to_mode): Likewise.
4006 (rtx_equal_for_field_assignment_p): Likewise.
4007 (gen_lowpart_for_combine): Likewise.
4008 (simplify_comparison): Likewise.
4009 * cse.c (equiv_constant): Likewise.
4010 * expmed.c (store_bit_field_1): Likewise.
4011 * final.c (alter_subreg): Likewise.
4012 * fwprop.c (propagate_rtx): Likewise.
4013 (forward_propagate_subreg): Likewise.
4014 * ira-conflicts.c (ira_build_conflicts): Likewise.
4015 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
4016 * lra-constraints.c (curr_insn_transform): Likewise.
4017 (split_reg): Likewise.
4018 * lra-eliminations.c (move_plus_up): Likewise.
4019 (lra_eliminate_regs_1): Likewise.
4020 * recog.c (general_operand): Likewise.
4021 * ree.c (combine_reaching_defs): Likewise.
4022 * reload.c (push_reload): Likewise.
4023 (find_reloads): Likewise.
4024 * reload1.c (elimination_effects): Likewise.
4025 (compute_reload_subreg_offset): Likewise.
4026 (choose_reload_regs): Likewise.
4027 * rtlanal.c (subreg_lsb_1): Likewise.
4028 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4029 (simplify_subreg): Likewise.
4030 * var-tracking.c (track_loc_p): Likewise.
4031 * emit-rtl.c (byte_lowpart_offset): Likewise.
4032 (paradoxical_subreg_p): Delete out-of-line definition.
4034 2017-08-22 Jeff Law <law@redhat.com>
4036 PR tree-optimization/81741
4037 PR tree-optimization/71947
4038 * tree-ssa-dom.c: Include tree-inline.h.
4039 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
4040 equivalences if one is more expensive to compute than the other.
4041 * tree-ssa-scopedtables.h (class const_or_copies): Make
4042 record_const_or_copy_raw method private.
4043 (class avail_exprs_stack): New method simplify_binary_operation.
4044 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
4045 avail_exprs_stack::simplify_binary_operation as needed.
4046 (avail_exprs_stack::simplify_binary_operation): New function.
4048 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
4050 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
4051 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
4052 (DOT_SYMBOLS): Likewise.
4053 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
4054 (RELOCATABLE_NEEDS_FIXUP): Likewise.
4055 (RS6000_ABI_NAME): Likewise.
4056 (TARGET_CMODEL): Likewise.
4057 (TOC_SECTION_ASM_OP): Likewise.
4058 (SET_CMODEL): New macro.
4059 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
4061 2017-08-22 Richard Biener <rguenther@suse.de>
4063 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
4064 to allow for free-lang-data replacements similar to verify_type_variant.
4066 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4067 Alan Hayward <alan.hayward@arm.com>
4068 David Sherwood <david.sherwood@arm.com>
4070 * config/aarch64/aarch64.md (casesi): Use DImode rather than
4071 VOIDmode for the LABEL_REF.
4073 2017-08-22 Richard Biener <rguenther@suse.de>
4075 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
4077 2017-08-22 Richard Biener <rguenther@suse.de>
4079 * common.opt (feliminate-dwarf2-dups): Ignore.
4080 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
4081 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
4082 same_die_p_wrap, compute_section_prefix,
4083 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
4084 (comdat_symbol_id, comdat_symbol_number): Likewise.
4085 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
4087 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
4088 (output_die): Mark unreachable path unreachable.
4089 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
4090 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
4091 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
4092 (dwarf2out_early_finish): Likewise.
4094 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
4096 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
4098 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
4101 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
4102 not VAR_P. Filter attribute warnings with OPT_Wattributes.
4103 (avr_attribute_table) <io, io_low, address>: Initialize
4104 .decl_required with true.
4106 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
4108 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
4109 undocumented debugging options.
4110 (-mvsx-scalar-double): Likewise.
4111 (-mallow-df-permute): Likewise.
4112 (-mvectorize-builtins): Likewise.
4113 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4114 (rs6000_builtin_vectorized_function): Likewise.
4115 (rs6000_builtin_md_vectorized_function): Likewise.
4116 (rs6000_opt_vars): Likewise.
4118 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
4121 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
4122 (*btrq_imm): Rename from *btrq.
4123 (*btcq_imm): Rename from *btcq.
4124 (btsc): New code attribute.
4125 (*<btsc><mode>): New insn pattern.
4126 (*btr<mode>): Ditto.
4127 (*<btsc><mode>_mask): New insn_and_split pattern.
4128 (*btr<mode>_mask): Ditto.
4130 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4131 Alan Hayward <alan.hayward@arm.com>
4132 David Sherwood <david.sherwood@arm.com>
4134 * function.c (pad_below): Simplify padding calculation.
4136 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4137 Alan Hayward <alan.hayward@arm.com>
4138 David Sherwood <david.sherwood@arm.com>
4140 * target.def (function_prologue): Remove frame size argument.
4141 (function_epilogue): Likewise.
4142 * doc/tm.texi: Regenerate.
4143 * final.c (final_start_function): Update call to function_prologue.
4144 (final_end_function): Update call to function_epilogue.
4145 (default_function_pro_epilogue): Remove frame size argument.
4146 * output.h (default_function_pro_epilogue): Likewise.
4147 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
4148 (arm_output_function_prologue): Likewise.
4149 * config/frv/frv.c (frv_function_prologue): Likewise.
4150 (frv_function_epilogue): Likewise.
4151 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
4152 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
4153 (ia64_output_function_epilogue): Likewise.
4154 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
4155 (m32r_output_function_epilogue): Likewise.
4156 * config/microblaze/microblaze.c (microblaze_function_prologue)
4157 (microblaze_function_epilogue): Likewise.
4158 * config/mips/mips.c (mips_output_function_prologue): Likewise.
4159 (mips_output_function_epilogue): Likewise.
4160 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
4161 (mmix_target_asm_function_epilogue): Likewise.
4162 * config/msp430/msp430.c (msp430_start_function): Likewise.
4163 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
4164 (nds32_asm_function_epilogue): Likewise.
4165 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
4166 * config/pa/pa.c (pa_output_function_prologue): Likewise.
4167 (pa_output_function_epilogue): Likewise.
4168 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
4169 (rs6000_output_function_epilogue): Likewise.
4170 * config/rl78/rl78.c (rl78_start_function): Likewise.
4171 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
4172 (rs6000_output_function_epilogue): Likewise.
4173 * config/rx/rx.c (rx_output_function_prologue): Likewise.
4174 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
4175 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
4176 (sparc_asm_function_epilogue): Likewise.
4178 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4180 * tree.h (type_has_mode_precision_p): New function.
4181 * convert.c (convert_to_integer_1): Use it.
4182 * expr.c (expand_expr_real_2): Likewise.
4183 (expand_expr_real_1): Likewise.
4184 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
4185 * match.pd: Likewise.
4186 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
4187 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
4188 * tree-tailcall.c (process_assignment): Likewise.
4189 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4190 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
4191 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
4192 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4193 (vectorizable_assignment): Likewise.
4194 (vectorizable_shift): Likewise.
4195 (vectorizable_operation): Likewise.
4196 * tree-vrp.c (register_edge_assert_for_2): Likewise.
4198 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
4200 * match.pd: Add pow (C, x) simplification.
4202 2017-08-21 Richard Biener <rguenther@suse.de>
4204 PR tree-optimization/81900
4205 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
4206 for blocks with abnormal predecessors.
4207 (compute_antic): Do not set visited flag prematurely.
4209 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
4212 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
4214 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4216 * stor-layout.h (vector_type_mode): Move to...
4217 * tree.h (vector_type_mode): ...here.
4218 * stor-layout.c (vector_type_mode): Move to...
4219 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
4221 2017-08-21 Richard Biener <rguenther@suse.de>
4223 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
4224 register_external_die hooks.
4225 (debug_false_tree_charstarstar_uhwistar): Declare.
4226 (debug_nothing_tree_charstar_uhwi): Likewise.
4227 * debug.c (do_nothing_debug_hooks): Adjust.
4228 (debug_false_tree_charstarstar_uhwistar): New do nothing.
4229 (debug_nothing_tree_charstar_uhwi): Likewise.
4230 * dbxout.c (dbx_debug_hooks): Adjust.
4231 (xcoff_debug_hooks): Likewise.
4232 * sdbout.c (sdb_debug_hooks): Likewise.
4233 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4234 * dwarf2out.c (macinfo_label_base): New global.
4235 (dwarf2out_register_external_die): New function for the
4236 register_external_die hook.
4237 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
4238 (dwarf2_debug_hooks): Use them.
4239 (dwarf2_lineno_debug_hooks): Adjust.
4240 (struct die_struct): Add with_offset flag.
4241 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
4242 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
4243 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
4244 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
4245 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
4246 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
4247 defining section names for the early LTO debug variants.
4248 (reset_indirect_string): New helper.
4249 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
4250 (print_dw_val): Add support for offsetted symbol references.
4251 (get_ultimate_context): Split out from is_cxx.
4252 (is_cxx): Use get_ultimate_context.
4253 (is_fortran): Add decl overload.
4254 (compute_comp_unit_symbol): Split out worker from
4255 compute_section_prefix.
4256 (compute_section_prefix): Call compute_comp_unit_symbol and
4257 set comdat_type_p here.
4258 (output_die): Skip DIE symbol output for the LTO added one.
4259 Handle DIE symbol references with offset.
4260 (output_comp_unit): Guard section name mangling properly.
4261 For LTO debug sections emit a symbol at the section beginning
4262 which we use to refer to its DIEs.
4263 (add_abstract_origin_attribute): For DIEs registered via
4264 dwarf2out_register_external_die directly refer to the early
4265 DIE rather than indirectly through the shadow one we created.
4266 Remove obsolete call to dwarf2out_abstract_function for
4267 non-function/block origins.
4268 (gen_array_type_die): When generating early LTO debug do
4269 not emit DW_AT_string_length.
4270 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
4271 late when in LTO. As suggested place a gcc_unreachable for
4272 the DECL_ABSTRACT_P case.
4273 (gen_subprogram_die): Avoid another specification DIE
4274 for early built declarations/definitions for the late LTO case.
4275 (gen_variable_die): Add type references for late duplicated VLA dies
4277 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
4278 we have the abstract instance already.
4279 (process_scope_var): Adjust decl DIE contexts in LTO which
4280 first puts them in limbo.
4281 (gen_decl_die): Do not generate type DIEs late apart from
4282 types for VLAs or for decls we do not yet have a DIE. Do not
4283 call dwarf2out_abstract_function late.
4284 (dwarf2out_early_global_decl): Make sure to create DIEs
4285 for abstract instances of a decl first.
4286 (dwarf2out_late_global_decl): Adjust comment.
4287 (output_macinfo_op): With multiple macro sections use
4288 macinfo_label_base to distinguish labels.
4289 (output_macinfo): Likewise. Update macinfo_label_base.
4290 Pass in the line info label.
4291 (note_variable_value_in_expr): When generating LTO resolve
4292 all variable values here by generating DIEs as needed.
4293 (init_sections_and_labels): Add early LTO debug flag parameter
4294 and generate different sections and names if set. Add generation
4295 counter for the labels so we can have multiple of them.
4296 (reset_dies): Helper to allow DIEs to be output multiple times.
4297 (dwarf2out_finish): When outputting DIEs to the fat part of an
4298 LTO object first reset DIEs.
4299 (dwarf2out_early_finish): Output early DIEs when generating LTO.
4300 (modified_type_die): Check for decl_ultimate_origin being self
4302 (gen_type_die_with_usage): Likewise.
4303 (gen_typedef_die): Allow decl_ultimate_origin being self.
4304 (set_decl_abstract_flags): Remove.
4305 (set_block_abstract_flags): Likewise.
4306 (dwarf2out_abstract_function): Treat the early generated DIEs
4307 as the abstract copy and only add DW_AT_inline and
4308 DW_AT_artificial here and call set_decl_origin_self.
4309 If the DIE has an abstract origin don't do anything.
4310 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
4311 if we have none yet (Go fails to build one, PR78628).
4312 (variably_modified_type_p): Prevent endless recursion for Ada
4313 cyclic pointer types.
4314 * lto-streamer-in.c: Include debug.h.
4315 (dref_queue): New global.
4316 (lto_read_tree_1): Stream in DIE references.
4317 (lto_input_tree): Register DIE references.
4318 (input_function): Stream DECL_DEBUG_ARGS.
4319 * lto-streamer-out.c: Include debug.h.
4320 (lto_write_tree_1): Output DIE references.
4321 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
4322 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
4323 (output_function): Stream DECL_DEBUG_ARGS.
4324 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
4325 Stream DECL_ABSTRACT_ORIGIN.
4326 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
4327 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
4328 DECL_CONTEXT for file-scope decls.
4329 * lto-streamer.h (struct dref_entry): Declare.
4330 (dref_queue): Likewise.
4331 * cfgexpand.c (pass_expand::execute): Do not call the
4332 outlining_inline_function hook here.
4333 * lto-wrapper.c (debug_obj): New global.
4334 (tool_cleanup): Unlink it if required.
4335 (debug_objcopy): New function.
4336 (run_gcc): Handle early debug sections in the IL files by
4337 extracting them to separate files, partially linkin them and
4338 feeding the result back as result to the linker.
4339 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
4340 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
4341 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
4342 sections into a separate segment.
4343 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
4345 (darwin_asm_dwarf_section): Likewise.
4346 (darwin_asm_output_dwarf_offset): Likewise.
4347 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
4349 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4350 Alan Hayward <alan.hayward@arm.com>
4351 David Sherwood <david.sherwood@arm.com>
4353 * read-md.h (md_reader::record_potential_iterator_use): Replace
4354 pointer argument with an rtx and an index.
4355 * read-rtl.c (iterator_group::apply_iterator): Likewise.
4356 (apply_mode_iterator): Likewise.
4357 (apply_code_iterator): Likewise.
4358 (apply_int_iterator): Likewise.
4359 (apply_subst_iterator): Likewise.
4360 (record_iterator_use): Likewise.
4361 (record_attribute_use): Likewise.
4362 (md_reader::record_potential_iterator_use): Likewise. Update calls
4363 to record_iterator_use and apply_iterator.
4364 (iterator_use): Replace ptr with x and index.
4365 (attribute_use): Likewise.
4366 (apply_attribute_uses): Update calls to apply_iterator.
4367 (apply_iterators): Likewise. Update initialization of iterator_use.
4368 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
4369 and record_potential_iterator_use.
4370 (rtx_reader::read_rtx_operand): Likewise.
4372 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4373 Alan Hayward <alan.hayward@arm.com>
4374 David Sherwood <david.sherwood@arm.com>
4376 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
4379 2017-08-21 Richard Biener <rguenther@suse.de>
4382 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
4383 at struct end conservatively when comparing common bases.
4385 2017-08-21 Richard Biener <rguenther@suse.de>
4387 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
4388 (mem_ref_in_stmt): Remove.
4389 (determine_max_movement): Use ref index to get at the reference.
4390 (invariantness_dom_walker::before_dom_children): Deal with
4391 lim data already initialized.
4392 (gather_mem_refs_stmt): Initialize lim data and record ref index.
4394 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4396 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
4397 (TARGET_ISA_ROUND): Ditto.
4398 (TARGET_ROUND): Ditto.
4399 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
4400 * config/i386/i386.md: Ditto.
4401 * config/i386/sse.md: Ditto.
4402 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
4403 with OPTION_MASK_ISA_SSE4_1.
4405 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4408 * doc/extend.texi (x86 Built-in Functions): Correct the name of
4409 __builtin_ia32_lzcnt_u16.
4411 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4414 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
4415 (rs6000_set_current_function): Rewrite function to use it.
4417 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
4420 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
4421 and TYPE_WARN_IF_NOT_ALIGN.
4422 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
4423 (handle_warn_if_not_align): New.
4424 (place_union_field): Call handle_warn_if_not_align.
4425 (place_field): Call handle_warn_if_not_align.
4426 Copy TYPE_WARN_IF_NOT_ALIGN.
4427 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
4428 (layout_type): Likewise.
4429 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
4431 (tree_decl_common): Add warn_if_not_align.
4432 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
4433 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
4434 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
4435 (DECL_WARN_IF_NOT_ALIGN): Likewise.
4436 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
4437 * doc/extend.texi: Document warn_if_not_aligned attribute.
4438 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
4440 2017-08-17 Martin Liska <mliska@suse.cz>
4443 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
4444 (get_data_dependence): Use it as pointer type.
4445 (distribute_loop): Likewise.
4447 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4449 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
4450 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
4451 (p8_vmrgow_<mode>_direct): New define_insn.
4452 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
4453 handle endianness for vmrgew and vmrgow permute patterns.
4455 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4457 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
4458 * config/rs6000/rs6000-cpus.def: Remove comment.
4459 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
4460 (POWERPC_MASKS): Likewise.
4461 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
4462 use of TARGET_VSX_TIMODE.
4463 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
4464 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
4465 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
4466 (rs6000_option_override_internal): Remove dead code.
4467 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
4468 (rs6000_legitimize_reload_address): Likewise.
4469 (rs6000_legitimate_address_p): Likewise.
4470 (rs6000_opt_masks): Delete "vsx-timode".
4471 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
4472 from function comment.
4473 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
4474 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
4475 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
4477 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
4478 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
4479 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
4482 (define_peephole2): Likewise.
4484 2017-08-17 Martin Sebor <msebor@redhat.com>
4487 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
4488 past the end of an array.
4489 (test_pp_format): Add test cases.
4491 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
4493 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
4494 missing ECF_NOTHROW flags.
4496 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4499 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
4500 operands residing in integer registers.
4501 (*vsx_le_perm_load_<mode>): Likewise.
4502 (*vsx_le_perm_store_<mode>): Likewise.
4503 (define_peephole2): Add peepholes to optimize the above.
4505 2017-08-17 Marek Polacek <polacek@redhat.com>
4508 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
4509 to mimic what shorten_compare did. Change the return type to bool.
4510 (fold_cond_expr_with_comparison): Update call to
4511 operand_equal_for_comparison_p.
4512 (fold_ternary_loc): Likewise.
4514 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
4516 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
4518 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
4520 2017-08-17 Richard Biener <rguenther@suse.de>
4522 * tree-ssa-structalias.c (solve_graph): When propagating
4523 to successors update the graphs succ edges and avoid duplicate work.
4525 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
4528 * config/i386/i386.c (ix86_option_override_internal): Save target
4529 specific options after ix86_stack_protector_guard_reg was changed.
4531 2017-08-17 Richard Biener <rguenther@suse.de>
4533 PR tree-optimization/81827
4534 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
4535 (new_var_info): Initialize it conservatively.
4536 (get_call_vi): Mark register vars.
4537 (new_scalar_tmp_constraint_exp): Likewise.
4538 (handle_rhs_call): Likewise.
4539 (handle_const_call): Likewise.
4540 (create_function_info_for): Likewise.
4541 (solve_constraints): Sort varinfos to separate register from
4542 non-register vars to pack points-to solution bitmaps during
4545 2017-08-17 Marek Polacek <polacek@redhat.com>
4547 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
4549 2017-08-17 Richard Biener <rguenther@suse.de>
4551 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
4552 to true when overflow is undefined and we saturated the result.
4554 2017-08-17 Alan Modra <amodra@gmail.com>
4557 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
4558 Don't use store multiple if only one reg needs saving.
4559 (interesting_frame_related_regno): New function.
4560 (rs6000_frame_related): Don't emit frame info for regs that
4562 (rs6000_emit_epilogue): Likewise.
4564 2017-08-16 Nathan Sidwell <nathan@acm.org>
4566 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
4567 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
4568 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
4569 * tree.c (free_lang_data_in_type): Use else-if chain. Always
4570 clear TYPE_LANG_1. Remove obsolete member-function stripping.
4571 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
4572 (verify_type): Adjust for TYPE_BINFO move.
4573 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
4574 process TYPE_BINFO directly.
4575 (hash_tree): Likewise.
4576 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
4578 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
4581 2017-08-16 David Malcolm <dmalcolm@redhat.com>
4583 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
4585 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
4588 * config/i386/i386.md (*anddi_1_btr): Change predicates of
4589 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
4590 Add ix86_binary_operator_ok to insn constraint.
4591 (*iordi_1_bts): Ditto.
4592 (*xordi_1_btc): Ditto.
4593 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
4594 Update corresponding peephole2 pattern.
4598 2017-08-16 Bin Cheng <bin.cheng@arm.com>
4600 PR tree-optimization/81832
4601 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
4602 copy loop header which has IFN_LOOP_DIST_ALIAS call.
4604 2017-08-16 Marek Polacek <polacek@redhat.com>
4607 * fold-const.c (fold_indirect_ref_1): Restore original behavior
4608 regarding size_zero_node.
4610 2017-08-16 Martin Liska <mliska@suse.cz>
4613 * config.gcc: Respect previously set extra_objs in case
4616 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
4618 PR tree-optimization/81835
4619 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
4620 the phi SSA_NAME. Check that the condition in a COND_EXPR does
4621 not depend on the phi.
4623 2017-08-16 Alan Modra <amodra@gmail.com>
4625 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
4628 2017-08-16 Alan Modra <amodra@gmail.com>
4630 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
4631 (save_reg_p): ..into this. Update all callers.
4632 (first_reg_to_save): Simplify.
4634 2017-08-16 Alan Modra <amodra@gmail.com>
4636 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
4639 2017-08-15 Joseph Myers <joseph@codesourcery.com>
4643 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
4644 constant count if that count is less than 32.
4646 2017-08-15 Nathan Sidwell <nathan@acm.org>
4648 * gcc.c (execute): Emit friendlier message if inferior is killed
4649 by an external cause.
4651 2017-08-15 Richard Biener <rguenther@suse.de>
4653 PR tree-optimization/81790
4654 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
4655 CONSTRUCTORs from simplifying and VN.
4657 2017-08-14 Martin Sebor <msebor@redhat.com>
4659 * builtin-attrs.def: Add comments.
4661 2017-08-14 Martin Sebor <msebor@redhat.com>
4664 * doc/extend.texi (attribute nonstring): Document new attribute.
4666 2017-08-14 Martin Sebor <msebor@redhat.com>
4669 * tree-diagnostic.c (default_tree_printer): Handle %G.
4670 * gimple-pretty-print.h (percent_G_format): Declare new function.
4671 * gimple-pretty-print.c (percent_G_format): Define.
4672 * tree-pretty-print.c (percent_K_format): Add argument.
4674 2017-08-14 Martin Sebor <msebor@redhat.com>
4676 PR translation/79998
4677 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
4678 Remove a stray space.
4680 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
4683 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
4684 (*iordi_1_bts): Ditto.
4685 (*xordi_1_btc): Ditto.
4687 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4690 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
4691 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4693 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
4694 quoted strings, and make more translator-friendly.
4695 (darwin_rs6000_override_options): Likewise.
4696 (rs6000_option_override_internal): Likewise.
4697 (rs6000_return_in_memory): Fix overlong line.
4698 (init_cmulative_args): Use quoted strings, and make more
4699 translator-friendly.
4700 (rs6000_pass_by_reference): Fix overlong line.
4701 (def_builtin): Use quoted strings.
4702 (altivec_expand_predicate_builtin): Use quoted strings, and make
4703 more translator-friendly.
4704 (htm_expand_builtin): Use quoted strings.
4705 (cpu_expand_builtin): Use quoted strings, and make more
4706 translator-friendly.
4707 (altivec_expand_builtin): Likewise.
4708 (paired_expand_predicate_builtin): Likewise.
4709 (rs6000_invalid_builtin): Likewise.
4710 (builtin_function_type): Use quoted strings.
4711 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
4712 more translator-friendly.
4713 (rs6000_trampoline_init): Likewise.
4714 (rs6000_handle_altivec_attribute): Likewise.
4715 (rs6000_inner_target_options): Use quoted strings.
4716 (rs6000_disable_incompatible_switches): Likewise.
4717 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
4718 strings, and make more translator-friendly.
4719 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
4721 2017-08-14 Bin Cheng <bin.cheng@arm.com>
4723 PR tree-optimization/81799
4724 * tree-loop-distribution.c (version_loop_by_alias_check): Force
4725 cond_expr to simple gimple operand.
4727 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
4730 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
4732 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
4736 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
4737 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
4738 TARGET_GASISR_PROLOGUES.
4739 * config/avr/avr.c (avr_option_override): Same.
4740 (avr_pass_pre_proep::execute): Same.
4742 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
4745 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
4746 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
4748 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
4750 * config/i386/i386.md (*load_tp_<mode>): Redefine as
4751 define_insn_and_split. Split to a memory load from 0 in
4752 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
4753 using PTR mode iterator.
4754 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
4755 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
4756 (*add_tp_<mode>): Redefine as define_insn_and_split.
4757 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
4758 address space. Merge with *add_tp_x32 using PTR mode iterator.
4759 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
4760 Split to an add with a memory load from 0 in
4761 DEFAULT_TLS_SEG_REG address space.
4763 2017-08-12 Andrew Pinski <apinski@cavium.com>
4765 * config/aarch64/aarch64-option-extensions.def (rdma):
4766 Fix feature string to what Linux prints out in /proc/cpuinfo.
4768 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
4771 * dwarf2out.c (modified_type_die): For C typedef types that have
4772 an ultimate origin, process the ultimate origin instead of the
4774 (gen_typedef_die): Assert that input DECLs have no ultimate
4776 (gen_type_die_with_usage): For typedef variants that have an
4777 ultimate origin, just call gen_decl_die on the original DECL.
4778 (process_scope_var): Avoid creating DIEs for local typedefs and
4779 concrete static variables.
4781 2017-08-12 Alan Modra <amodra@gmail.com>
4785 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
4786 match gnu-user.h startfile.
4787 (ENDFILE_LINUX_SPEC): Similarly.
4789 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
4792 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
4794 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
4796 2017-08-11 Tamar Christina <tamar.christina@arm.com>
4797 * config/aarch64/aarch64.md (mov<mode>): Change.
4798 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4799 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
4800 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
4802 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
4804 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
4805 for storage order barriers.
4807 2017-08-11 Martin Liska <mliska@suse.cz>
4810 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
4811 variables of void type.
4813 2017-08-11 Martin Liska <mliska@suse.cz>
4815 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
4816 TARGET_SUPPORTS_ALIASES.
4817 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
4818 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
4819 (optimize_weakref): Likewise.
4820 * symtab.c (symtab_node::noninterposable_alias): Likewise.
4821 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
4822 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
4824 2017-08-11 Martin Liska <mliska@suse.cz>
4827 * config/i386/i386.c (make_resolver_func): Do complete
4828 refactoring of the function.
4830 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4833 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
4834 * config/i386/i386.c (ix86_stack_protect_guard): Use
4835 ix86_stack_protect_guard_symbol_str to generate varible declaration.
4836 * doc/invoke.texi (x86 Options): Document
4837 -mstack-protector-guard-symbol= option.
4839 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4841 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
4842 * config/i386/i386.c (ix86_split_stack_guard): New function.
4843 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
4844 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
4845 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
4846 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
4847 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
4848 (split_stack_space_check): Call ix86_split_stack_guard.
4850 2017-08-10 Martin Sebor <msebor@redhat.com>
4852 * print-tree.c (print_node): Print location using the established
4854 Replace spaces with colons.
4855 (debug_raw, debug): Ditto.
4857 2017-08-10 Martin Sebor <msebor@redhat.com>
4860 * pretty-print.c (pp_format): Correct the handling of %s precision.
4862 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
4865 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
4867 (ix86_finalize_stack_frame_flags): This. Also clear
4868 frame_pointer_needed if -fno-omit-frame-pointer is used without
4870 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
4871 with ix86_finalize_stack_frame_flags.
4872 (ix86_expand_epilogue): Likewise.
4873 (ix86_expand_split_stack_prologue): Likewise.
4874 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
4876 2017-08-10 Martin Liska <mliska@suse.cz>
4879 * c-attribs.c (handle_target_attribute):
4880 Report warning for an empty string argument of target attribute.
4882 2017-08-09 Jakub Jelinek <jakub@redhat.com>
4885 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
4887 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
4888 or DECL_NONLOCAL labels.
4889 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
4890 or DECL_NONLOCAL labels here.
4892 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
4894 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
4895 to indicate when early gimple folding has been disabled.
4896 (rs6000_gimple_fold_builtin): Add debug content.
4897 (rs6000_invalid_builtin): Fix whitespace.
4898 (rs6000_expand_builtin): Fix whitespace.
4899 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
4901 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
4904 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
4905 SAVE_MULTIPLE if not all the registers that saves, should be saved.
4907 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
4909 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
4910 (qdf24xx): Likewise.
4911 * config/aarch64/aarch64.md: Include falkor.md.
4912 * config/aarch64/falkor.md: New.
4914 2017-08-09 Marek Polacek <polacek@redhat.com>
4917 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
4918 * diagnostic.c (emit_diagnostic): Add a comment.
4919 (emit_diagnostic_valist): New function.
4921 2017-08-09 Marek Polacek <polacek@redhat.com>
4924 * input.c (make_location): New overload.
4925 * input.h (make_location): Declare.
4927 2017-08-08 Alan Modra <amodra@gmail.com>
4928 H.J. Lu <hongjiu.lu@intel.com>
4931 * gcc.c (NO_PIE_SPEC): Delete.
4932 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
4934 (LINK_PIE_SPEC): ..to here.
4935 (LINK_COMMAND_SPEC): Support -no-pie.
4936 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
4937 chain of crtbegin*.o selection, update for PIE_SPEC changes and
4939 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
4940 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
4941 (ENDFILE_CRTEND_SPEC): Similarly.
4943 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
4946 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
4947 (mstack-protector-guard-offset=): Ditto.
4948 * config/i386/i386.c (ix86_option_override): Handle
4949 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
4951 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
4952 ix86_stack_protect_guard_offset variables.
4953 (TARGET_STACK_PROTECT_GUARD): Always define.
4954 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
4955 and -mstack-protector-guard-offset= options.
4957 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4959 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
4960 boundary case for the last candidate.
4962 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4964 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
4965 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
4967 2017-08-08 Tamar Christina <tamar.christina@arm.com>
4970 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
4971 * config/aarch64/aarch64-builtins.c
4972 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
4973 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
4974 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
4976 2017-08-08 Tamar Christina <tamar.christina@arm.com>
4977 Andrew Pinski <pinskia@gmail.com>
4980 * internal-fn.def (XORSIGN): New.
4981 * optabs.def (xorsign_optab): New.
4982 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
4983 (convert_expand_mult_copysign): New.
4984 (pass_optimize_widening_mul::execute): Call
4985 convert_expand_mult_copysign.
4987 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4989 PR tree-optimization/81354
4990 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
4991 Insert on edges rather than explicitly creating landing pads.
4992 (analyze_candidates_and_replace): Commit edge inserts.
4994 2017-08-08 Richard Biener <rguenther@suse.de>
4997 * tree-ssa-loop-niter.c: Include tree-dfa.h.
4998 (expand_simple_operations): Also look through ADDR_EXPRs with
4999 MEM_REF bases treating them as POINTER_PLUS_EXPR.
5001 2017-08-08 Richard Biener <rguenther@suse.de>
5003 PR tree-optimization/81723
5004 * tree-vect-slp.c (struct bst_traits): New hash traits.
5005 (bst_fail): New global.
5006 (vect_build_slp_tree_2): New worker, split out from ...
5007 (vect_build_slp_tree): ... this now wrapping it with using
5008 bst_fail set to cache SLP tree build fails. Properly handle
5010 (vect_analyze_slp_instance): Allocate and free bst_fail.
5012 2017-08-08 Martin Liska <mliska@suse.cz>
5015 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
5016 LABEL_DECLs that can be from a different function.
5018 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5020 PR tree-optimization/81744
5021 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
5022 loop's number of iterations.
5024 2017-08-08 Martin Liska <mliska@suse.cz>
5026 * asan.c: Include header files.
5027 * attribs.c (build_decl_attribute_variant): New function moved
5029 (build_type_attribute_qual_variant): Likewise.
5030 (cmp_attrib_identifiers): Likewise.
5031 (simple_cst_list_equal): Likewise.
5032 (omp_declare_simd_clauses_equal): Likewise.
5033 (attribute_value_equal): Likewise.
5034 (comp_type_attributes): Likewise.
5035 (build_type_attribute_variant): Likewise.
5036 (lookup_ident_attribute): Likewise.
5037 (remove_attribute): Likewise.
5038 (merge_attributes): Likewise.
5039 (merge_type_attributes): Likewise.
5040 (merge_decl_attributes): Likewise.
5041 (merge_dllimport_decl_attributes): Likewise.
5042 (handle_dll_attribute): Likewise.
5043 (attribute_list_equal): Likewise.
5044 (attribute_list_contained): Likewise.
5045 * attribs.h (lookup_attribute): New function moved from tree.[ch].
5046 (lookup_attribute_by_prefix): Likewise.
5047 * bb-reorder.c: Include header files.
5048 * builtins.c: Likewise.
5049 * calls.c: Likewise.
5050 * cfgexpand.c: Likewise.
5051 * cgraph.c: Likewise.
5052 * cgraphunit.c: Likewise.
5053 * convert.c: Likewise.
5054 * dwarf2out.c: Likewise.
5055 * final.c: Likewise.
5056 * fold-const.c: Likewise.
5057 * function.c: Likewise.
5058 * gimple-expr.c: Likewise.
5059 * gimple-fold.c: Likewise.
5060 * gimple-pretty-print.c: Likewise.
5061 * gimple.c: Likewise.
5062 * gimplify.c: Likewise.
5063 * hsa-common.c: Likewise.
5064 * hsa-gen.c: Likewise.
5065 * internal-fn.c: Likewise.
5066 * ipa-chkp.c: Likewise.
5067 * ipa-cp.c: Likewise.
5068 * ipa-devirt.c: Likewise.
5069 * ipa-fnsummary.c: Likewise.
5070 * ipa-inline.c: Likewise.
5071 * ipa-visibility.c: Likewise.
5073 * lto-cgraph.c: Likewise.
5074 * omp-expand.c: Likewise.
5075 * omp-general.c: Likewise.
5076 * omp-low.c: Likewise.
5077 * omp-offload.c: Likewise.
5078 * omp-simd-clone.c: Likewise.
5079 * opts-global.c: Likewise.
5080 * passes.c: Likewise.
5081 * predict.c: Likewise.
5082 * sancov.c: Likewise.
5083 * sanopt.c: Likewise.
5084 * symtab.c: Likewise.
5085 * toplev.c: Likewise.
5086 * trans-mem.c: Likewise.
5087 * tree-chkp.c: Likewise.
5088 * tree-eh.c: Likewise.
5089 * tree-into-ssa.c: Likewise.
5090 * tree-object-size.c: Likewise.
5091 * tree-parloops.c: Likewise.
5092 * tree-profile.c: Likewise.
5093 * tree-ssa-ccp.c: Likewise.
5094 * tree-ssa-live.c: Likewise.
5095 * tree-ssa-loop.c: Likewise.
5096 * tree-ssa-sccvn.c: Likewise.
5097 * tree-ssa-structalias.c: Likewise.
5098 * tree-ssa.c: Likewise.
5099 * tree-streamer-in.c: Likewise.
5100 * tree-vectorizer.c: Likewise.
5101 * tree-vrp.c: Likewise.
5103 * ubsan.c: Likewise.
5104 * varasm.c: Likewise.
5105 * varpool.c: Likewise.
5106 * tree.c: Remove functions moved to attribs.[ch].
5108 * config/aarch64/aarch64.c: Add attrs.h header file.
5109 * config/alpha/alpha.c: Likewise.
5110 * config/arc/arc.c: Likewise.
5111 * config/arm/arm.c: Likewise.
5112 * config/avr/avr.c: Likewise.
5113 * config/bfin/bfin.c: Likewise.
5114 * config/c6x/c6x.c: Likewise.
5115 * config/cr16/cr16.c: Likewise.
5116 * config/cris/cris.c: Likewise.
5117 * config/darwin.c: Likewise.
5118 * config/epiphany/epiphany.c: Likewise.
5119 * config/fr30/fr30.c: Likewise.
5120 * config/frv/frv.c: Likewise.
5121 * config/ft32/ft32.c: Likewise.
5122 * config/h8300/h8300.c: Likewise.
5123 * config/i386/winnt.c: Likewise.
5124 * config/ia64/ia64.c: Likewise.
5125 * config/iq2000/iq2000.c: Likewise.
5126 * config/lm32/lm32.c: Likewise.
5127 * config/m32c/m32c.c: Likewise.
5128 * config/m32r/m32r.c: Likewise.
5129 * config/m68k/m68k.c: Likewise.
5130 * config/mcore/mcore.c: Likewise.
5131 * config/microblaze/microblaze.c: Likewise.
5132 * config/mips/mips.c: Likewise.
5133 * config/mmix/mmix.c: Likewise.
5134 * config/mn10300/mn10300.c: Likewise.
5135 * config/moxie/moxie.c: Likewise.
5136 * config/msp430/msp430.c: Likewise.
5137 * config/nds32/nds32-isr.c: Likewise.
5138 * config/nds32/nds32.c: Likewise.
5139 * config/nios2/nios2.c: Likewise.
5140 * config/nvptx/nvptx.c: Likewise.
5141 * config/pa/pa.c: Likewise.
5142 * config/pdp11/pdp11.c: Likewise.
5143 * config/powerpcspe/powerpcspe.c: Likewise.
5144 * config/riscv/riscv.c: Likewise.
5145 * config/rl78/rl78.c: Likewise.
5146 * config/rx/rx.c: Likewise.
5147 * config/s390/s390.c: Likewise.
5148 * config/sh/sh.c: Likewise.
5149 * config/sol2.c: Likewise.
5150 * config/sparc/sparc.c: Likewise.
5151 * config/spu/spu.c: Likewise.
5152 * config/stormy16/stormy16.c: Likewise.
5153 * config/tilegx/tilegx.c: Likewise.
5154 * config/tilepro/tilepro.c: Likewise.
5155 * config/v850/v850.c: Likewise.
5156 * config/vax/vax.c: Likewise.
5157 * config/visium/visium.c: Likewise.
5158 * config/xtensa/xtensa.c: Likewise.
5160 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5163 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
5164 constraints since the -mupper-regs-* switches have been
5166 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
5167 into a vector from a double word element that was extracted from
5168 another vector, and eliminate extra XXPERMDI instructions.
5169 (vsx_concat_<mode>_2): Likewise.
5170 (vsx_concat_<mode>_3): Likewise.
5171 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
5172 concat to allow optimizing inserts from previous extracts.
5174 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
5176 * config/i386/i386.c (ix86_stack_protect_guard): Generate
5177 memory reference to a SSP offset in TLS address space.
5178 (ix86_print_operand) <case '@'>: Remove.
5179 (ix86_print_operand_punct_valid_p): Remove '@' code.
5180 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
5182 (stack_tls_protect_set_<mode>): Remove.
5183 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
5184 (stack_tls_protect_test_<mode>): Remove.
5185 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
5187 2017-08-07 Olivier Hainque <hainque@adacore.com>
5190 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
5192 2017-08-07 Douglas Rupp <rupp@adacore.com>
5194 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
5195 variable was referenced as multidir in command.
5197 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5200 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
5203 2017-08-07 Martin Liska <mliska@suse.cz>
5205 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
5206 * config/rl78/rl78.c: Add include of attribs.h.
5207 * config/sh/sh.c: Likewise.
5208 * config/v850/v850.c: Likewise.
5210 2017-08-07 Tom de Vries <tom@codesourcery.com>
5213 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
5215 2017-08-07 Martin Liska <mliska@suse.cz>
5217 * config/mips/mips.c: Include attribs.h.
5219 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
5222 * doc/invoke.texi: Document change in behvaior for -Ofast for
5225 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
5227 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
5229 (aarch64_pop_regs): Likewise.
5230 (aarch64_gen_load_pair): Likewise.
5231 (aarch64_save_callee_saves): Likewise.
5232 (aarch64_restore_callee_saves): Likewise.
5234 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5236 * config/i386/i386.c: Revert the last change.
5238 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5241 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5243 (ix86_finalize_stack_frame_flags): This. Also clear
5244 frame_pointer_needed if -fno-omit-frame-pointer is used without
5246 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5247 with ix86_finalize_stack_frame_flags.
5248 (ix86_expand_epilogue): Likewise.
5249 (ix86_expand_split_stack_prologue): Likewise.
5251 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5254 * config/i386/i386.c (get_builtin_code_for_version): Set priority
5255 to P_AES for Westmere.
5257 2017-08-07 Jonathan Yong <10walls@gmail.com>
5259 * config/i386/mingw.opt (fset-stack-executable): Removed.
5260 * config/i386/cygming.opt (fset-stack-executable): Moved
5262 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
5264 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
5266 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
5268 2017-08-07 Marek Polacek <polacek@redhat.com>
5271 * fold-const.c (fold_indirect_ref_1): Check type_domain.
5273 2017-08-07 Martin Liska <mliska@suse.cz>
5275 * attribs.h (canonicalize_attr_name): New function.
5276 (cmp_attribs): Move from c-format.c and adjusted.
5277 (is_attribute_p): Moved from tree.h.
5278 * tree-inline.c: Add new includes.
5279 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
5280 (private_is_attribute_p): Remove.
5281 (private_lookup_attribute): Likewise.
5282 (private_lookup_attribute_by_prefix): Simplify.
5283 (remove_attribute): Use is_attribute_p.
5284 * tree.h: Remove removed declarations.
5286 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5289 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
5290 instead of computing it in the function. Formatting fix.
5291 (expand_case): Don't rely on default_edge being the first edge,
5292 clear it if removing it, pass default_edge to
5293 emit_case_dispatch_table.
5294 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
5297 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
5299 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
5300 insn in the function, emit NOP after the insn.
5302 2017-08-06 Tom de Vries <tom@codesourcery.com>
5304 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
5307 2017-08-06 Tom de Vries <tom@codesourcery.com>
5309 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
5312 2017-08-04 Yury Gribov <tetra2005@gmail.com>
5314 PR tree-optimization/57371
5315 * match.pd: New pattern.
5317 2017-08-04 Marek Polacek <polacek@redhat.com>
5320 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
5321 perform the computation in offset_int.
5323 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5325 PR tree-optimization/81136
5326 * tree-vectorizer.h: Include tree-hash-traits.h.
5327 (vec_base_alignments): New typedef.
5328 (vec_info): Add a base_alignments field.
5329 (vect_record_base_alignments): Declare.
5330 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
5332 (DR_IS_CONDITIONAL_IN_STMT): New macro.
5333 (create_data_ref): Add an is_conditional_in_stmt argument.
5334 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
5335 the is_conditional_in_stmt field.
5336 (data_ref_loc): Add an is_conditional_in_stmt field.
5337 (get_references_in_stmt): Set the is_conditional_in_stmt field.
5338 (find_data_references_in_stmt): Update call to create_data_ref.
5339 (graphite_find_data_references_in_stmt): Likewise.
5340 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
5341 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
5342 (vect_record_base_alignment): New function.
5343 (vect_record_base_alignments): Likewise.
5344 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
5345 for nested statements even if we fail to compute a misalignment.
5346 Use pooled base alignments for unconditional references.
5347 (vect_find_same_alignment_drs): Compare base addresses instead
5349 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
5350 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
5352 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5354 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
5355 Add an explicit name for the enum. Use auto_vec for slp_instances
5357 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
5359 (_bb_vec_info): Add a constructor and destructor.
5360 (vinfo_for_stmt): Return NULL for uids of -1 as well.
5361 (destroy_loop_vec_info): Delete.
5362 (vect_destroy_datarefs): Likewise.
5363 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
5364 (vec_info::vec_info): New function.
5365 (vec_info::~vec_info): Likewise.
5366 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
5367 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
5368 destroy_loop_vec_info.
5369 * tree-vect-loop.c (new_loop_vec_info): Replace with...
5370 (_loop_vec_info::_loop_vec_info): ...this.
5371 (destroy_loop_vec_info): Replace with...
5372 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
5373 the stmt_vec_infos. Leave handling of vec_info information to its
5374 destructor. Remove explicit vector releases.
5375 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
5376 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
5377 * tree-vect-slp.c (new_bb_vec_info): Replace with...
5378 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
5379 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
5380 (destroy_bb_vec_info): Replace with...
5381 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
5382 information to its destructor.
5383 (vect_slp_analyze_bb_1): Use new and delete instead of
5384 new_bb_vec_info and destroy_bb_vec_info.
5385 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
5388 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5390 * tree-data-ref.h (subscript): Add access_fn field.
5391 (data_dependence_relation): Add could_be_independent_p.
5392 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
5393 (same_access_functions): Move to tree-data-ref.c.
5394 * tree-data-ref.c (ref_contains_union_access_p): New function.
5395 (access_fn_component_p): Likewise.
5396 (access_fn_components_comparable_p): Likewise.
5397 (dr_analyze_indices): Add a reference to access_fn_component_p.
5398 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
5400 (constant_access_functions): Likewise.
5401 (add_other_self_distances): Likewise.
5402 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
5403 (initialize_data_dependence_relation): Use XCNEW and remove
5404 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
5405 of access functions that have the same type. Allow the
5406 subsequence to end with different bases in some circumstances.
5407 Record the chosen access functions in SUB_ACCESS_FN.
5408 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
5409 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
5410 (subscript_dependence_tester_1): Likewise dra and drb.
5411 (build_classic_dist_vector): Update calls accordingly.
5412 (subscript_dependence_tester): Likewise.
5413 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
5414 DDR_COULD_BE_INDEPENDENT_P.
5415 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
5416 comp_alias_ddrs instead of may_alias_ddrs.
5417 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
5419 (vect_analyze_data_ref_dependence): Use it if
5420 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
5421 distance vectors if that fails.
5422 (dependence_distance_ge_vf): New function.
5423 (vect_prune_runtime_alias_test_list): Use it. Don't clear
5424 LOOP_VINFO_MAY_ALIAS_DDRS.
5426 2017-08-04 Richard Biener <rguenther@suse.de>
5429 * fold-const.c (fold_binary_loc): Properly restrict
5430 minus_var0 && minus_var1 case when associating undefined overflow
5433 2017-08-04 Tom de Vries <tom@codesourcery.com>
5435 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
5437 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5439 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5440 Don't start diagnostic messages with a capital letter.
5441 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5443 (rs6000_invalid_builtin): Likewise.
5444 (rs6000_trampoline_init): Likewise.
5446 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5449 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
5450 after setting changeable df flags.
5452 2017-08-03 Richard Biener <rguenther@suse.de>
5454 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
5455 up if the use is in USE - X.
5457 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
5459 * toplev.c (dumpfile.h): New include.
5460 (internal_error_reentered): New static function. Use it...
5461 (internal_error_function): ...here to handle reentered internal_error.
5463 2017-08-03 Richard Biener <rguenther@suse.de>
5466 * fold-const.c (split_tree): Add minus_var and minus_con
5467 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
5468 here but always use minus_*.
5469 (associate_trees): Assert we never associate with MINUS_EXPR
5470 and NULL first operand. Do not recurse for PLUS_EXPR operands
5471 when associating as MINUS_EXPR either.
5472 (fold_binary_loc): Track minus_var and minus_con.
5474 2017-08-03 Tom de Vries <tom@codesourcery.com>
5477 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
5478 ACCEL_COMPILER, apply finish_options on
5479 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
5481 2017-08-03 Tom de Vries <tom@codesourcery.com>
5484 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
5485 function_entry_patch_area_size > 0.
5487 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5490 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
5491 instead of 10??LU, perform unit multiplication in wide_int,
5492 don't change alloc_object_size_limit if the limit is larger
5495 PR tree-optimization/81655
5496 PR tree-optimization/81588
5497 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
5498 the case when ranges[i].low and high are 1 for unsigned type with
5502 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
5503 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
5505 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5507 * tree-vrp.h: Add include guard.
5509 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
5512 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
5513 (ud2): New insn pattern.
5514 * config/i386/i386.c (ix86_expand_epilogue):
5515 For naked functions, generate ud2 instead of trap insn.
5517 2017-08-02 Marek Polacek <polacek@redhat.com>
5520 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
5522 2017-08-02 Tom de Vries <tom@codesourcery.com>
5523 Cesar Philippidis <cesar@codesourcery.com>
5525 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
5526 Add missing edge probabilities.
5528 2017-08-02 Tamar Christina <tamar.christina@arm.com>
5530 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
5533 2017-08-02 Jakub Jelinek <jakub@redhat.com>
5536 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
5537 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
5538 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
5539 prologue_seq sequences - if any.
5541 2017-08-02 Richard Biener <rguenther@suse.de>
5543 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
5544 via vectors if supported, integer extracts via punning if supported
5545 or otherwise vector extracts.
5547 2017-08-02 Richard Biener <rguenther@suse.de>
5549 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
5551 (bitmap_insert_into_set): ... this.
5553 2017-08-02 Richard Biener <rguenther@suse.de>
5555 PR tree-optimization/81633
5557 2015-08-17 Alan Hayward <alan.hayward@arm.com>
5559 PR tree-optimization/71752
5560 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
5562 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
5564 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
5565 (machine_function::call_ms2sysv_pad_out): Remove field.
5566 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
5567 (ix86_compute_frame_layout): Likewise.
5569 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
5572 * config/i386/i386.c (ix86_set_func_type): Disallow naked
5573 attribute with interrupt attribute.
5575 2017-08-01 Andrew Pinski <apinski@cavium.com>
5577 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
5578 BIT_INSERT_EXPR's operand 1
5579 to see if the types precision matches.
5581 2017-08-01 Martin Liska <mliska@suse.cz>
5584 * builtins.c (expand_builtin_memcpy_args): Remove.
5585 (expand_builtin_memcpy): Call newly added function
5586 expand_builtin_memory_copy_args.
5587 (expand_builtin_memcpy_with_bounds): Likewise.
5588 (expand_builtin_mempcpy): Remove last argument.
5589 (expand_builtin_mempcpy_with_bounds): Likewise.
5590 (expand_builtin_memory_copy_args): New function created from
5591 expand_builtin_mempcpy_args with small modifications.
5592 (expand_builtin_mempcpy_args): Remove.
5593 (expand_builtin_stpcpy): Remove unused argument.
5594 (expand_builtin): Likewise.
5595 (expand_builtin_with_bounds): Likewise.
5597 2017-08-01 Martin Liska <mliska@suse.cz>
5600 Make mempcpy more optimal (PR middle-end/70140).
5602 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5605 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5606 __builtin_vec_cmpne verify both arguments are compatible vectors
5607 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
5608 verify arg1_type is a pointer or array type. For __builtin_vec_st,
5609 move computation of aligned to after checking the argument types.
5613 * config/rs6000/vsx.md (vextract_fp_from_shorth,
5614 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
5617 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
5618 Doug Rupp <rupp@adacore.com>
5619 Olivier Hainque <hainque@adacore.com>
5621 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
5622 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
5624 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
5625 for TARGET_OS_CPP_BUILTIN.
5626 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
5627 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
5629 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
5630 passing required abi options to the assembler for EABI configurations.
5631 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
5632 of .text.hot and .text.unlikely sections for kernel modules when
5633 using ARM style exceptions.
5634 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
5635 options. Add EXTRA_CC1_SPEC.
5636 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
5638 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
5640 (ARM_TARGET2_DWARF_FORMAT): Define.
5641 * config/arm/t-vxworks: Adjust multilib control to removal of the
5642 Diab command line options.
5644 2017-08-01 Martin Liska <mliska@suse.cz>
5646 PR gcov-profile/81561
5647 * gcov.c (unblock): Make unblocking safe as we need to preserve
5648 index correspondence of blocks and block_lists.
5650 2017-08-01 Richard Biener <rguenther@suse.de>
5652 PR tree-optimization/81181
5653 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5654 (compute_antic): ... end of iteration here.
5656 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
5658 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
5659 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
5660 (ftree-slp-vectorize): Likewise.
5661 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
5662 can no longer be set independent of flag_tree_loop_vectorize.
5663 * omp-general.c (emp_max_vf): Likewise.
5664 * opts.c (enable_fdo_optimizations): Remove references to
5665 flag_tree_vectorize, these are now implicit.
5666 (common_handle_option): Remove handling for OPT_ftree_vectorize,
5667 and leave it for the options machinery.
5669 2017-08-01 Martin Liska <mliska@suse.cz>
5672 * builtins.c (expand_builtin_memcpy_args): Remove.
5673 (expand_builtin_memcpy): Call newly added function
5674 expand_builtin_memory_copy_args.
5675 (expand_builtin_memcpy_with_bounds): Likewise.
5676 (expand_builtin_mempcpy): Remove last argument.
5677 (expand_builtin_mempcpy_with_bounds): Likewise.
5678 (expand_builtin_memory_copy_args): New function created from
5679 expand_builtin_mempcpy_args with small modifications.
5680 (expand_builtin_mempcpy_args): Remove.
5681 (expand_builtin_stpcpy): Remove unused argument.
5682 (expand_builtin): Likewise.
5683 (expand_builtin_with_bounds): Likewise.
5685 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5688 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
5689 print "ds:" only for immediates in generic address space.
5691 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5694 * config/i386/i386.c (ix86_funciton_naked): New prototype.
5695 (ix86_function_ok_for_sibcall): Return false for naked functions.
5697 2017-08-01 Richard Biener <rguenther@suse.de>
5699 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
5700 (compute_antic): Seed worklist with exit block predecessors.
5701 * cfganal.c (dfs_find_deadend): For a cycle return the source
5702 of the edge closing it.
5704 2017-08-01 Tamar Christina <tamar.christina@arm.com>
5706 * config/aarch64/aarch64.c
5707 (aarch64_can_const_movi_rtx_p): Move 0 check.
5709 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5711 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
5712 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
5714 * match.pd: Ditto in address comparison pattern.
5716 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5718 PR tree-optimization/81627
5719 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
5720 closed ssa form for store-store chain.
5722 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5724 PR tree-optimization/81620
5725 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
5726 for store-store chain.
5728 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5730 PR tree-optimization/81588
5731 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
5732 ranges[i].in_p, invert comparison code ccode. For >/>=,
5733 swap rhs1 and rhs2 and comparison code unconditionally,
5734 for </<= don't do that. Don't swap rhs1/rhs2 again if
5735 ranges[i].in_p, instead invert comparison code ccode if
5736 opcode or oe->rank is BIT_IOR_EXPR.
5739 * optabs.def (vec_extract_optab, vec_init_optab): Change from
5740 a direct optab to conversion optab.
5741 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
5742 with GET_MODE_INNER as last argument instead of optab_handler.
5743 * expmed.c (extract_bit_field_1): Likewise. Use vector from
5744 vector extraction if possible and optab is available.
5745 * expr.c (store_constructor): Use convert_optab_handler instead
5746 of optab_handler. Use vector initialization from smaller
5747 vectors if possible and optab is available.
5748 * tree-vect-stmts.c (vectorizable_load): Likewise.
5749 * doc/md.texi (vec_extract, vec_init): Document that the optabs
5751 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
5752 of vec_init from half-sized vectors with the same element mode.
5753 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
5754 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
5755 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
5756 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
5757 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
5758 after mode in gen_vec_extract* calls.
5759 (vec_extract<mode>): Renamed to ...
5760 (vec_extract<mode><ssescalarmodelower>): ... this.
5761 (vec_extract<mode><ssehalfvecmodelower>): New expander.
5762 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
5763 element mode after mode in gen_vec_init* calls.
5764 (VEC_INIT_HALF_MODE): New mode iterator.
5765 (vec_init<mode>): Renamed to ...
5766 (vec_init<mode><ssescalarmodelower>): ... this.
5767 (vec_init<mode><ssehalfvecmodelower>): New expander.
5768 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
5769 (vec_extractv2sfsf): ... this.
5770 (vec_initv2sf): Renamed to ...
5771 (vec_initv2sfsf): ... this.
5772 (vec_extractv2si): Renamed to ...
5773 (vec_extractv2sisi): ... this.
5774 (vec_initv2si): Renamed to ...
5775 (vec_initv2sisi): ... this.
5776 (vec_extractv4hi): Renamed to ...
5777 (vec_extractv4hihi): ... this.
5778 (vec_initv4hi): Renamed to ...
5779 (vec_initv4hihi): ... this.
5780 (vec_extractv8qi): Renamed to ...
5781 (vec_extractv8qiqi): ... this.
5782 (vec_initv8qi): Renamed to ...
5783 (vec_initv8qiqi): ... this.
5784 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
5785 (vec_init<mode>): Renamed to ...
5786 (vec_init<mode><VEC_base_l>): ... this.
5787 (vec_extract<mode>): Renamed to ...
5788 (vec_extract<mode><VEC_base_l>): ... this.
5789 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
5790 (vec_initv2sfsf): ... this.
5791 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
5792 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5793 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
5794 element mode after mode in gen_vec_init* calls.
5795 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
5796 (vec_init<mode><Vel>): ... this.
5797 (vec_extract<mode>): Renamed to ...
5798 (vec_extract<mode><Vel>): ... this.
5799 * config/aarch64/iterators.md (Vel): New mode attribute.
5800 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
5801 Add element mode after mode in gen_vec_extract* calls.
5802 * config/s390/vector.md (non_vec_l): New mode attribute.
5803 (vec_extract<mode>): Renamed to ...
5804 (vec_extract<mode><non_vec_l>): ... this.
5805 (vec_init<mode>): Renamed to ...
5806 (vec_init<mode><non_vec_l>): ... this.
5807 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
5808 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
5810 * config/arm/iterators.md (V_elem_l): New mode attribute.
5811 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
5812 (vec_extract<mode><V_elem_l>): ... this.
5813 (vec_extractv2di): Renamed to ...
5814 (vec_extractv2didi): ... this.
5815 (vec_init<mode>): Renamed to ...
5816 (vec_init<mode><V_elem_l>): ... this.
5817 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
5818 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
5819 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
5820 Add element mode after gen_vec_extract* calls.
5821 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
5822 (vec_init<mode><unitmode>): ... this.
5823 (vec_extract<mode>): Renamed to ...
5824 (vec_extract<mode><unitmode>): ... this.
5825 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
5826 (vec_init<mode><unitmode>): ... this.
5827 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
5828 (vec_initv2sfsf): ... this.
5829 (vec_extractv2sf): Renamed to ...
5830 (vec_extractv2sfsf): ... this.
5831 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
5832 Add element mode after gen_vec_extract* calls.
5833 * config/mips/mips.md (unitmode): New mode iterator.
5834 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
5835 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
5836 * config/spu/spu.md (inner_l): New mode attribute.
5837 (vec_init<mode>): Renamed to ...
5838 (vec_init<mode><inner_l>): ... this.
5839 (vec_extract<mode>): Renamed to ...
5840 (vec_extract<mode><inner_l>): ... this.
5841 * config/sparc/sparc.md (veltmode): New mode iterator.
5842 (vec_init<VMALL:mode>): Renamed to ...
5843 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
5844 * config/ia64/vect.md (vec_initv2si): Renamed to ...
5845 (vec_initv2sisi): ... this.
5846 (vec_initv2sf): Renamed to ...
5847 (vec_initv2sfsf): ... this.
5848 (vec_extractv2sf): Renamed to ...
5849 (vec_extractv2sfsf): ... this.
5850 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
5851 (vec_init<mode>): Renamed to ...
5852 (vec_init<mode><VEC_base_l>): ... this.
5853 (vec_extract<mode>): Renamed to ...
5854 (vec_extract<mode><VEC_base_l>): ... this.
5855 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
5856 (vec_initv2sfsf): ... this.
5857 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
5858 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5859 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
5860 gen_vec_init* calls.
5862 2017-08-01 Richard Biener <rguenther@suse.de>
5864 PR tree-optimization/81297
5865 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
5866 TREE_OVERFLOW from INTEGER_CSTs.
5868 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
5870 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
5872 2017-07-31 Carl Love <cel@us.ibm.com>
5874 * config/rs6000/rs6000-c: Add support for built-in functions
5875 vector signed char vec_xl_be (signed long long, signed char *);
5876 vector unsigned char vec_xl_be (signed long long, unsigned char *);
5877 vector signed int vec_xl_be (signed long long, signed int *);
5878 vector unsigned int vec_xl_be (signed long long, unsigned int *);
5879 vector signed long long vec_xl_be (signed long long, signed long long *);
5880 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
5881 vector signed short vec_xl_be (signed long long, signed short *);
5882 vector unsigned short vec_xl_be (signed long long, unsigned short *);
5883 vector double vec_xl_be (signed long long, double *);
5884 vector float vec_xl_be (signed long long, float *);
5885 * config/rs6000/altivec.h (vec_xl_be): Add #define.
5886 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
5887 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
5889 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
5890 (altivec_expand_builtin): Add switch statement to call
5891 altivec_expand_xl_be for each builtin.
5892 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
5893 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
5894 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
5895 __builtin_vsx_le_be_v16qi.
5896 * doc/extend.texi: Update the built-in documentation file for the
5897 new built-in functions.
5899 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5902 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
5904 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5906 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5908 * config.gcc: Add z14.
5909 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
5910 CPU model numbers for z13s and z14.
5911 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
5913 * config/s390/s390-opts.h (enum processor_type): Rename
5914 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5915 * config/s390/s390.c (processor_table): Add field for CPU name to
5916 be passed to Binutils.
5917 (s390_asm_output_machine_for_arch): Use the new field in
5918 processor_table for Binutils.
5919 (s390_expand_builtin): Replace arch12 with z14.
5920 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5921 (s390_get_sched_attrmask): Likewise.
5922 (s390_get_unit_mask): Likewise.
5923 * config/s390/s390.opt: Add z14 to processor_type enum.
5925 2017-07-31 Martin Jambor <mjambor@suse.cz>
5928 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
5929 regardless of optimization level.
5931 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
5932 Martin Liska <mliska@suse.cz>
5934 * predict.def: Remove old comment and adjust probability.
5935 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
5938 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5941 * config/i386/i386.c (ix86_function_naked): New function.
5942 (ix86_can_use_return_insn_p): Return false for naked functions.
5943 (ix86_expand_prologue): Skip prologue for naked functions.
5944 (ix86_expand_epilogue): Skip epilogue for naked functions
5945 and emit trap instruction.
5946 (ix86_warn_func_return): New function.
5947 (ix86_attribute_table): Add "naked" attribute specification.
5948 (TARGET_WARN_FUNC_RETURN): Define.
5949 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
5951 2017-07-31 Martin Liska <mliska@suse.cz>
5953 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
5954 (dump_gimple_bb_header): Always dump BB info.
5955 (pp_cfg_jump): Do not append info about BB when dumping a jump.
5957 2017-07-31 Martin Liska <mliska@suse.cz>
5960 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
5961 also with current_function_decl non-null equality.
5963 2017-07-31 Jakub Jelinek <jakub@redhat.com>
5966 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
5967 change type to the element type, instead add eltype variable and
5968 use it where we are interested in the element type.
5970 PR tree-optimization/81603
5971 * ipa-polymorphic-call.c
5972 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
5973 offset arithmetic in offset_int, bail out if the resulting bit offset
5974 doesn't fit into shwi.
5976 2017-07-31 Martin Liska <mliska@suse.cz>
5978 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
5979 (gimplify_save_expr): Fix comment.
5981 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
5984 * config/i386/i386.c (ix86_function_arg): Update arguments for
5986 (ix86_compute_frame_layout): Set the initial stack offset to
5987 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
5988 INCOMING_FRAME_SP_OFFSET.
5989 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
5990 stack before exception handler returns.
5991 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
5992 the 'ERROR_CODE' for exception handler.
5994 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
5996 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
5997 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
5998 (ASM_OUTPUT_REG_POP): Ditto.
5999 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
6000 instead of asm_fprintf to output pure string.
6002 2017-07-29 Jakub Jelinek <jakub@redhat.com>
6004 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
6005 to imported_module_or_decl hook.
6006 (debug_nothing_tree_tree_tree_bool): Remove.
6007 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
6008 * debug.c (do_nothing_debug_hooks): Use
6009 debug_nothing_tree_tree_tree_bool_bool instead of
6010 debug_nothing_tree_tree_tree_bool.
6011 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6012 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
6013 * sdbout.c (sdb_debug_hooks): Likewise.
6014 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
6015 (gen_namespace_die): Add DW_AT_export_symbols attribute if
6017 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
6018 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
6019 attribute, don't add anything.
6021 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6023 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
6024 (fold_build2_stat_loc): Likewise.
6025 (fold_build3_stat_loc): Likewise.
6026 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
6027 (fold_build1_loc): Remove macro.
6028 (fold_build2_loc): Likewise.
6029 (fold_build3_loc): Likewise.
6031 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6033 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
6034 (gimple_build_debug_bind_source_stat): Likewise.
6035 * gimple.h (gimple_build_debug_bind): Remove macro.
6036 (gimple_build_debug_bind_source): Likewise.
6038 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6040 * bitmap.c (bitmap_alloc): Adjust.
6041 (bitmap_gc_alloc): Likewise.
6042 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
6044 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6046 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
6047 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
6048 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
6049 (bitmap_gc_alloc_stat): Likewise.
6050 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
6052 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6054 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
6055 * rtl.h (shallow_copy_rtx): Remove macro.
6057 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6059 * emit-rtl.c (gen_raw_REG): Adjust.
6060 * gengenrtl.c (gendef): Likewise.
6061 * rtl.c (rtx_alloc_stat): Remove _stat from name.
6062 * rtl.h (rtx_alloc): Remove macro.
6064 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6066 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
6067 (build_tree_list_stat): Likewise.
6068 * tree.h (build_tree_list): Remove macro.
6069 (build_tree_list_vec): Likewise.
6071 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6073 * tree.c (make_vector_stat): Remove _stat from name.
6074 (build_vector_stat): Likewise.
6075 * tree.h (make_vector_stat): Remove macro.
6076 (build_vector_stat): Likewise.
6078 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6080 * tree.h (build_var_debug_value): Remove prototype.
6082 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6084 * tree.c (tree_cons_stat): Remove _stat from name.
6085 * tree.h (tree_cons): Remove macro.
6087 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6089 * tree.c (build_vl_exp_stat): Remove _stat from name.
6090 * tree.h (build_vl_exp): Remove macro.
6092 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6094 * tree.c (build_decl_stat): Remove _stat from name.
6095 * tree.h (build_decl): Remove macro.
6097 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6099 * gimple.c (gimple_build_with_ops_stat): Adjust.
6100 (gimple_alloc_stat): Remove _stat from name.
6101 * gimple.h (gimple_alloc): Remove macro.
6103 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6105 * tree.c (make_tree_vec_stat): Remove _stat from name.
6106 (grow_tree_vec_stat): Likewise.
6107 * tree.h (make_tree_vec_stat): Adjust prototype.
6108 (grow_tree_vec_stat): Likewise.
6109 (make_tree_vec): Remove macro.
6110 (grow_tree_vec): Likewise.
6112 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6114 * fold-const.c (fold_build1_stat_loc): Adjust.
6115 (fold_build2_stat_loc): Likewise.
6116 (fold_build3_stat_loc): Likewise.
6117 * tree.c (build0_stat): Remove _stat from name.
6118 (build1_stat): Likewise.
6119 (build2_stat): Likewise.
6120 (build3_stat): Likewise.
6121 (build4_stat): Likewise.
6122 (build5_stat): Likewise.
6123 * tree.h (build1_loc): Remove macro, and rename _stat function
6125 (build2_loc): Likewise.
6126 (build3_loc): Likewise.
6127 (build4_loc): Likewise.
6128 (build5_loc): Likewise.
6130 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6132 * tree.c (make_int_cst_stat): Remove _stat from name.
6133 * tree.h (make_int_cst_stat): Adjust prototype.
6134 (make_int_cst): Remove macro.
6136 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6138 * tree.c (make_tre_binfo_stat): Remove _stat from name.
6139 * tree.h (make_tree_binfo_stat): Adjust prototype.
6140 (make_tree_binfo): Remove.
6142 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6144 * tree.c (copy_node_stat): Rename to copy_node.
6145 (build_distinct_type_copy): Adjust.
6146 * tree.h (copy_node_stat): Adjust prototype.
6147 (copy_node): Remove macro.
6149 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6151 * tree.c (make_node_stat): rename to make_node.
6152 (build_tree_list_stat): Adjust.
6153 (build0_stat): Likewise.
6154 (build2_stat): Likewise.
6155 (build3_stat): Likewise.
6156 (build4_stat): Likewise.
6157 (build5_stat): Likewise.
6158 (build_decl_stat): Likewise.
6159 * tree.h (make_node_stat): Adjust prototype.
6160 (make_node): remove macro.
6162 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
6164 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
6165 (PPC_FEATURE2_SCV): Likewise.
6166 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
6168 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6170 * config/aarch64/aarch64.c
6171 (aarch64_internal_mov_immediate): Add new special pattern.
6172 * config/aarch64/aarch64.md (*movdi_aarch64):
6173 Add reg/32bit const mov case.
6175 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6176 Richard Sandiford <richard.sandiford@linaro.org>
6178 * config/aarch64/aarch64.md (mov<mode>): Generalize.
6179 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6180 Add integer and movi cases.
6181 (movi-split-hf-df-sf split, fp16): New.
6182 (enabled): Added TARGET_FP_F16INST.
6183 * config/aarch64/iterators.md (GPF_HF): New.
6184 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
6186 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6188 * config/aarch64/aarch64.c
6189 (aarch64_simd_container_mode): Add prototype.
6190 (aarch64_expand_mov_immediate): Add HI support.
6191 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
6192 (aarch64_can_const_movi_rtx_p): New.
6193 (aarch64_preferred_reload_class):
6194 Remove restrictions of using FP registers for certain SIMD operations.
6195 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
6196 (aarch64_valid_floating_const): Add integer move validation.
6197 (aarch64_simd_imm_scalar_p): Remove.
6198 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
6199 (aarch64_legitimate_constant_p): Expand list of supported cases.
6200 * config/aarch64/aarch64-protos.h
6201 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
6202 (aarch64_reinterpret_float_as_int): New.
6203 (aarch64_simd_imm_scalar_p): Remove.
6204 * config/aarch64/constraints.md (Uvi): New.
6205 (Dd): Split into Ds and new Dd.
6206 * config/aarch64/aarch64.md (*movsi_aarch64):
6208 (*movdi_aarch64): Add SIMD mov case.
6210 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6212 * tree-predcom.c: (struct chain): Handle store-store chain in which
6213 stores for elimination only store loop invariant values.
6214 (execute_pred_commoning_chain): Ditto.
6215 (prepare_initializers_chain_store_elim): Ditto.
6216 (prepare_finalizers): Ditto.
6217 (is_inv_store_elimination_chain): New function.
6218 (initialize_root_vars_store_elim_1): New function.
6220 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6222 * tree-predcom.c: Revise general description of the pass.
6223 (enum chain_type): New enum type for store elimination.
6224 (struct chain): New field supporting store elimination.
6225 (struct component): Ditto.
6226 (dump_chain): Dump store-stores chain.
6227 (release_chain): Release resources.
6228 (split_data_refs_to_components): Compute and create component
6229 contains only stores for elimination.
6230 (get_chain_last_ref_at): New function.
6231 (make_invariant_chain): Initialization.
6232 (make_rooted_chain): Specify chain type in parameter and record it.
6233 (add_looparound_copies): Skip for store-stores chain.
6234 (determine_roots_comp): Compute type of chain and pass it to
6236 (initialize_root_vars_store_elim_2): New function.
6237 (finalize_eliminated_stores): New function.
6238 (remove_stmt): Handle store for elimination.
6239 (execute_pred_commoning_chain): Execute predictive commoning on
6241 (determine_unroll_factor): Skip unroll for store-stores chain.
6242 (prepare_initializers_chain_store_elim): New function.
6243 (prepare_initializers_chain): Hanlde store-store chain.
6244 (prepare_finalizers_chain, prepare_finalizers): New function.
6245 (tree_predictive_commoning_loop): Return integer value indicating
6246 if loop is unrolled or lcssa form is corrupted.
6247 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
6249 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6251 * tree-predcom.c (initialize_root): Delete.
6252 (execute_pred_commoning_chain): Initialize root vars and replace
6253 reference of non-combined chain directly, rather than call above
6256 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6258 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
6259 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
6261 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6263 * tree-predcom.c (struct chain): New field init_seq.
6264 (release_chain): Release init_seq.
6265 (prepare_initializers_chain): Record intialization stmts in above
6267 (insert_init_seqs): New function.
6268 (tree_predictive_commoning_loop): Call insert_init_seqs.
6270 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6272 * tree-predcom.c (determine_roots_comp): Skip trivial components.
6274 2017-07-28 Richard Biener <rguenther@suse.de>
6276 * match.pd: Remove superfluous :c.
6277 * genmatch.c (simplify::id): Add member.
6278 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
6280 (current_id): New global.
6281 (dt_node::parent): Move from ...
6282 (dt_operand::parent): ... here. Add for_id member.
6283 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
6284 (decision_tree::find_node): Relax order requirement when
6285 merging DT_TRUE nodes to ones inbetween the current simplify
6286 and the one we try to merge with. Add diagnostic whenever
6287 we need to enforce pattern order by not merging.
6288 (decision_tree::insert): Set current_id.
6289 (decision_tree::print_node): Dump parent node and for_id.
6290 (parser::last_id): Add member.
6291 (parser::push_simplify): Assign unique id.
6292 (parser::parser): Initialize last_id.
6294 2017-07-28 Martin Liska <mliska@suse.cz>
6297 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
6298 gimple_build_debug_bind.
6300 2017-07-28 Richard Biener <rguenther@suse.de>
6302 PR tree-optimization/81502
6303 * match.pd: Add pattern combining BIT_INSERT_EXPR with
6305 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
6307 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
6308 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
6309 for BIT_FIELD_REF args.
6310 * fold-const.c (make_bit_field_ref): Likewise.
6311 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6313 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6316 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
6317 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
6318 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
6319 Or it into SANITIZER_UNDEFINED.
6320 * ubsan.c: Include gimple-fold.h and varasm.h.
6321 (ubsan_expand_ptr_ifn): New function.
6322 (instrument_pointer_overflow): New function.
6323 (maybe_instrument_pointer_overflow): New function.
6324 (instrument_object_size): Formatting fix.
6325 (pass_ubsan::execute): Call instrument_pointer_overflow
6326 and maybe_instrument_pointer_overflow.
6327 * internal-fn.c (expand_UBSAN_PTR): New function.
6328 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
6329 * sanitizer.def (__ubsan_handle_pointer_overflow,
6330 __ubsan_handle_pointer_overflow_abort): New builtins.
6331 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
6332 * internal-fn.def (UBSAN_PTR): New internal function.
6333 * opts.c (sanitizer_opts): Add pointer-overflow.
6334 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
6335 * fold-const.c (build_range_check): Compute pointer range check in
6336 integral type if pointer arithmetics would be needed. Formatting
6339 2017-07-28 Martin Liska <mliska@suse.cz>
6342 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
6343 parameters that are of a variable-length.
6345 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
6347 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
6349 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
6350 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
6351 (CRT_CALL_STATIC_FUNCTION): Likewise.
6352 (ASM_DEFAULT_SPEC): New define.
6353 (ASM_SPEC32): Likewise.
6354 (ASM_SPEC64): Likewise.
6355 (ASM_SPEC_COMMON): Likewise.
6356 (ASM_SPEC): Likewise.
6357 (INVALID_64BIT): Likewise.
6358 (LINK_OS_DEFAULT_SPEC): Likewise.
6359 (LINK_OS_SPEC32): Likewise.
6360 (LINK_OS_SPEC64): Likewise.
6361 (POWERPC_LINUX): Likewise.
6362 (PTRDIFF_TYPE): Likewise.
6363 (RESTORE_FP_PREFIX): Likewise.
6364 (RESTORE_FP_SUFFIX): Likewise.
6365 (SAVE_FP_PREFIX): Likewise.
6366 (SAVE_FP_SUFFIX): Likewise.
6367 (SIZE_TYPE): Likewise.
6368 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
6369 (TARGET_64BIT): Likewise.
6370 (TARGET_64BIT): Likewise.
6371 (TARGET_AIX): Likewise.
6372 (WCHAR_TYPE_SIZE): Likewise.
6373 (WCHAR_TYPE): Undefine.
6374 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
6375 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
6376 (CPP_OS_RTEMS_SPEC): Delete.
6377 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
6378 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
6380 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
6382 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6384 PR tree-optimization/81578
6385 * tree-parloops.c (build_new_reduction): Bail out if
6386 reduction_code isn't one of the standard OpenMP reductions.
6387 Move the details printing after that decision.
6389 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6391 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
6392 related to reload_in_progress.
6393 (splat_input_operand): Likewise.
6394 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
6396 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
6398 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
6399 (TARGET_INSTANTIATE_DECLS): Likewise.
6400 (legitimate_indexed_address_p): Delete reload_in_progress code.
6401 (rs6000_debug_legitimate_address_p): Likewise.
6402 (rs6000_eliminate_indexed_memrefs): Likewise.
6403 (rs6000_emit_le_vsx_store): Likewise.
6404 (rs6000_emit_move_si_sf_subreg): Likewise.
6405 (rs6000_emit_move): Likewise.
6406 (register_to_reg_type): Likewise.
6407 (rs6000_pre_atomic_barrier): Likewise.
6408 (rs6000_machopic_legitimize_pic_address): Likewise.
6409 (rs6000_allocate_stack_temp): Likewise.
6410 (rs6000_address_for_fpconvert): Likewise.
6411 (rs6000_address_for_altivec): Likewise.
6412 (rs6000_secondary_memory_needed_rtx): Delete function.
6413 (rs6000_check_sdmode): Likewise.
6414 (rs6000_alloc_sdmode_stack_slot): Likewise.
6415 (rs6000_instantiate_decls): Likewise.
6416 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
6417 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
6418 Delete reload_in_progress.
6419 (*vec_reload_and_plus_<mptrsize>): Likewise.
6420 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
6421 (vsx_div_v2di): Likewise.
6422 (vsx_udiv_v2di): Likewise.
6424 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6426 * config/rs6000/rs6000.opt (mlra): Replace with stub.
6427 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
6428 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
6429 (rs6000_debug_reg_global): Delete print of LRA status.
6430 (rs6000_option_override_internal): Delete dead LRA related code.
6431 (rs6000_lra_p): Delete function.
6432 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
6434 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
6436 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
6437 * config/riscv/rtems.h: New file.
6439 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6440 Sudakshina Das <sudi.das@arm.com>
6442 * config/aarch64/aarch64.md
6443 (define_split for and<mode>3nr_compare): Move
6444 non aarch64_logical_operand to a register.
6445 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
6446 register immediate operand to a register.
6447 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
6449 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6452 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
6454 2017-07-27 Richard Biener <rguenther@suse.de>
6456 PR tree-optimization/81573
6457 PR tree-optimization/81494
6458 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
6459 multi defuse cycle case.
6461 2017-07-27 Richard Biener <rguenther@suse.de>
6463 PR tree-optimization/81571
6464 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
6467 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
6469 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
6470 earlier and only if MASK_FPU is set. Adjust formatting.
6472 2017-07-27 Martin Liska <mliska@suse.cz>
6474 * opt-functions.awk: Add validation of value of Init.
6475 * optc-gen.awk: Pass new argument.
6477 2017-07-27 Martin Liska <mliska@suse.cz>
6479 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
6480 Fix wrong condition.
6482 2017-07-27 Martin Liska <mliska@suse.cz>
6484 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
6485 BBs and edges seen by autoFDO.
6487 2017-07-27 Richard Biener <rguenther@suse.de>
6489 PR tree-optimization/81502
6490 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
6491 with incompatible but same sized type.
6492 (execute_update_addresses_taken): Likewise.
6494 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
6496 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
6497 flag_tree_loop_vectorize rather than flag_tree_vectorize.
6499 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6502 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
6503 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
6504 Change s_operand to memory_operand.
6506 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
6508 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
6509 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
6510 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
6511 Emit instructions rather than returning an expression. Handle TFmode
6512 and KFmode by casting to TImode.
6513 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
6514 (rs6000_emit_le_vsx_store): Likewise.
6515 * config/rs6000/vsx.md (VSX_TI): New iterator.
6516 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
6517 (*vsx_le_undo_permute_<mode>): Likewise.
6518 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
6519 emit the split sequence.
6520 (*vsx_le_perm_store_<mode>): Likewise.
6522 2017-07-27 Jakub Jelinek <jakub@redhat.com>
6524 PR tree-optimization/81555
6525 PR tree-optimization/81556
6526 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
6527 if true, force CHANGED for the recursive invocation.
6528 (reassociate_bb): Remember original length of ops array, pass
6529 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
6531 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
6532 attributes for noipa attribute. For naked attribute use
6533 lookup_attribute first before lookup_attribute_spec.
6534 * final.c (rest_of_handle_final): Disable IPA RA for functions with
6536 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
6537 for functions with noipa attribute.
6538 (cgraph_externally_visible_p): Return true for functions with noipa
6540 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
6541 for functions with noipa attribute.
6542 * doc/extend.texi: Document noipa function attribute.
6543 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
6544 also for functions with noipa attribute.
6545 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
6547 2017-07-26 Andrew Pinski <apinski@cavium.com>
6549 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
6550 vec_unalign_load_cost and vec_unalign_store_cost.
6552 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6554 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
6555 -mvsx-small-integer option.
6556 (ISA_3_0_MASKS_IEEE): Likewise.
6557 (OTHER_VSX_VECTOR_MASKS): Likewise.
6558 (POWERPC_MASKS): Likewise.
6559 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
6560 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
6561 code, only testing for DImode being allowed in non-VSX floating
6563 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
6564 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
6566 (rs6000_option_override_internal): Delete -mvsx-small-integer.
6567 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
6568 TARGET_P8_VECTOR test.
6569 (rs6000_secondary_reload_simple_move): Likewise.
6570 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
6571 since TARGET_P9_VECTOR was already tested.
6572 (rs6000_opt_masks): Remove -mvsx-small-integer.
6573 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
6574 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6576 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
6577 test for TARGET_VEXTRACTUB was used, and that uses
6579 (p9 extract splitter): Likewise.
6580 (vsx_extract_<mode>_di_p9): Likewise.
6581 (vsx_extract_<mode>_store_p9): Likewise.
6582 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
6583 for TARGET_P9_VECTOR was used. Delete code that is now dead with
6584 the elimination of TARGET_VSX_SMALL_INTEGER.
6585 (vsx_extract_<mode>_p8): Likewise.
6586 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
6587 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
6588 (vsx_set_<mode>_p9): Likewise.
6589 (vsx_set_v4sf_p9): Likewise.
6590 (vsx_set_v4sf_p9_zero): Likewise.
6591 (vsx_insert_extract_v4sf_p9): Likewise.
6592 (vsx_insert_extract_v4sf_p9_2): Likewise.
6593 * config/rs6000/rs6000.md (sign extend splitter): Change
6594 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
6595 (floatsi<mode>2_lfiwax_mem): Likewise.
6596 (floatunssi<mode>2_lfiwzx_mem): Likewise.
6597 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
6598 since a test for TARGET_P9_VECTOR was used.
6599 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6600 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
6601 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6602 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6603 TARGET_P8_VECTOR test.
6604 (fix_trunc<mode>si2_stfiwx): Likewise.
6605 (fix_trunc<mode>si2_internal): Likewise.
6606 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
6607 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6609 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6610 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6611 TARGET_P8_VECTOR test.
6612 (fixuns_trunc<mode>si2_stfiwx): Likewise.
6613 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
6614 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6616 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6617 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
6618 since a test for TARGET_P9_VECTOR was used.
6619 (splitter for loading small constants): Likewise.
6621 2017-07-26 Andrew Pinski <apinski@cavium.com>
6623 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
6626 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
6629 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
6630 (fp_valid_at): Likewise.
6632 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6634 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
6635 (qdf24xx_addrcost_table): Likewise.
6636 (cortexa57_tunings): Update to use generic_branch_cost.
6637 (cortexa72_tunings): Likewise.
6638 (cortexa73_tunings): Likewise.
6639 (qdf24xx_tunings): Likewise.
6641 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6643 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
6644 (thunderx2t99_branch_cost): Likewise.
6645 (cortexa35_tunings): Update to use generic_branch_cost.
6646 (cortexa53_tunings): Likewise.
6647 (cortexa57_tunings): Likewise.
6648 (cortexa72_tunings): Likewise.
6649 (cortexa73_tunings): Likewise.
6650 (thunderx2t99_tunings): Likewise.
6652 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6654 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
6655 (sparc_option_override): Honour MASK_FSMULD.
6656 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
6657 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
6658 * config/sparc/sparc.opt (mfsmuld): New option.
6659 * doc/invoke.texi (mfsmuld): Document option.
6661 2017-07-26 Marek Polacek <polacek@redhat.com>
6664 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
6666 2017-07-26 Richard Biener <rguenther@suse.de>
6668 * gimple-match-head.c (do_valueize): Return OP if valueize
6670 (get_def): New helper to get at the def stmt of a SSA name
6672 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
6673 do_valueize to get at the def stmt.
6674 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
6676 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
6679 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
6681 2017-07-26 Martin Liska <mliska@suse.cz>
6684 * function.c (expand_function_start): Make expansion of
6685 nonlocal_goto_save_area after parm_birth_insn.
6687 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6689 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
6690 from all CPU target flags enable members.
6692 2017-07-26 Richard Biener <rguenther@suse.de>
6694 * genmatch.c (dt_simplify::gen): Make iterator vars const.
6695 (decision_tree::gen): Make 'type' const.
6696 (write_predicate): Likewise.
6698 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6700 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
6701 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
6702 (rs6000_option_override_internal): Likewise.
6703 (rs6000_expand_vector_set): Likewise.
6704 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
6705 (TARGET_UPPER_REGS_SF): Likewise.
6706 (TARGET_UPPER_REGS_DI): Likewise.
6707 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
6708 (TARGET_DIRECT_MOVE_64BIT): Likewise.
6709 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
6710 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6711 (Splitters for DI constants in Altivec registers): Likewise.
6712 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
6713 (vsx_set_v4sf_p9): Likewise.
6714 (vsx_set_v4sf_p9_zero): Likewise.
6715 (vsx_insert_extract_v4sf_p9): Likewise.
6716 (vsx_insert_extract_v4sf_p9_2): Likewise.
6718 2017-07-25 Carl Love <cel@us.ibm.com>
6720 * doc/extend.texi: Update the built-in documentation file for the
6721 existing built-in functions
6722 vector signed char vec_cnttz (vector signed char);
6723 vector unsigned char vec_cnttz (vector unsigned char);
6724 vector signed short vec_cnttz (vector signed short);
6725 vector unsigned short vec_cnttz (vector unsigned short);
6726 vector signed int vec_cnttz (vector signed int);
6727 vector unsigned int vec_cnttz (vector unsigned int);
6728 vector signed long long vec_cnttz (vector signed long long);
6729 vector unsigned long long vec_cnttz (vector unsigned long long);
6731 2017-07-25 Andrew Pinski <apinski@cavium.com>
6733 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
6734 accesses where the use is for the first operand of a BIT_INSERT.
6736 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6739 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6740 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6742 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6744 * config/i386/gstabs.h: Delete.
6745 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
6747 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
6749 * config/i386/i386.c (ix86_decompose_address): Do not check for
6750 register RTX when looking at index_reg or base_reg.
6751 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
6753 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
6755 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
6756 to update EH info here.
6758 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6760 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
6762 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6764 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
6766 2017-07-25 Torsten Duwe <duwe@suse.de>
6768 * common.opt: Introduce -fpatchable-function-entry
6769 command line option, and its variables function_entry_patch_area_size
6770 and function_entry_patch_area_start.
6771 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
6772 including a two-value parser.
6773 * target.def (print_patchable_function_entry): New target hook.
6774 * targhooks.h (default_print_patchable_function_entry): New function.
6775 * targhooks.c (default_print_patchable_function_entry): Likewise.
6776 * toplev.c (process_options): Switch off IPA-RA if
6777 patchable function entries are being generated.
6778 * varasm.c (assemble_start_function): Look at the
6779 patchable-function-entry command line switch and current
6780 function attributes and maybe generate NOP instructions by
6781 calling the print_patchable_function_entry hook.
6782 * doc/extend.texi: Document patchable_function_entry attribute.
6783 * doc/invoke.texi: Document -fpatchable_function_entry
6784 command line option.
6785 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
6787 * doc/tm.texi: Re-generate.
6789 2017-07-25 Jakub Jelinek <jakub@redhat.com>
6792 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
6793 TARGET_AVX512DQ rather than TARGET_AVX512BW.
6795 2017-07-25 Tamar Christina <tamar.christina@arm.com>
6797 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
6799 2017-07-25 Richard Biener <rguenther@suse.de>
6801 PR tree-optimization/81455
6802 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
6803 not walk in cycles when looking for guards.
6805 2017-07-25 Richard Biener <rguenther@suse.de>
6807 PR tree-optimization/81529
6808 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
6809 when optimizing backedge uses.
6811 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
6813 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
6815 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
6816 to dl_section_ref. On AIX, append an expression to subtract
6817 the size of the section length to dl_section_ref.
6819 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
6821 * configure.ac: If any of the config.* scripts fail, exit 1.
6822 * configure: Regenerate.
6824 2017-07-25 Richard Biener <rguenther@suse.de>
6827 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
6828 of immediate uses, be more verbose on errors.
6830 2017-07-25 Richard Biener <rguenther@suse.de>
6832 PR tree-optimization/81510
6833 * tree-vect-loop.c (vect_is_simple_reduction): When the
6834 reduction stmt is not inside the loop bail out.
6836 2017-07-25 Richard Biener <rguenther@suse.de>
6838 PR tree-optimization/81303
6839 * tree-vect-loop-manip.c (vect_loop_versioning): Build
6840 profitability check against LOOP_VINFO_NITERSM1.
6842 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6844 * domwalk.c (cmp_bb_postorder): Simplify.
6845 (sort_bbs_postorder): New function. Use it...
6846 (dom_walker::walk): ...here to optimize common cases.
6848 2017-07-25 Martin Liska <mliska@suse.cz>
6851 * ipa-visibility.c (function_and_variable_visibility): Make the
6852 redirection just on target that supports aliasing.
6853 Fix GNU coding style.
6855 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
6858 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
6860 * config/arm/rtems.h: Likewise.
6861 * config/bfin/rtems.h: Likewise.
6862 * config/i386/rtemself.h: Likewise.
6863 * config/lm32/rtems.h: Likewise.
6864 * config/m32c/rtems.h: Likewise.
6865 * config/m68k/rtemself.h: Likewise.
6866 * config/microblaze/rtems.h: Likewise.
6867 * config/mips/rtems.h: Likewise.
6868 * config/moxie/rtems.h: Likewise.
6869 * config/nios2/rtems.h: Likewise.
6870 * config/powerpcspe/rtems.h: Likewise.
6871 * config/rs6000/rtems.h: Likewise.
6872 * config/rtems.h: Likewise.
6873 * config/sh/rtems.h: Likewise.
6874 * config/sh/rtemself.h: Likewise.
6875 * config/sparc/rtemself.h: Likewise.
6877 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6880 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
6881 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
6882 * tree-ssa-structalias.c (alias_get_name): Same.
6884 2017-07-25 Bin Cheng <bin.cheng@arm.com>
6887 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
6888 instructions if no du chain is found.
6890 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6892 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
6894 2017-07-25 Richard Biener <rguenther@suse.de>
6897 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
6900 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6902 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
6904 (ISA_2_7_MASKS_SERVER): Likewise.
6905 (ISA_3_0_MASKS_IEEE): Likewise.
6906 (OTHER_P8_VECTOR_MASKS): Likewise.
6907 (OTHER_VSX_VECTOR_MASKS): Likewise.
6908 (POWERPC_MASKS): Likewise.
6909 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
6910 duplicate list of options.
6911 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
6912 explicit -mupper-regs options.
6913 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
6914 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
6915 alias for -mupper-regs-df.
6916 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6917 (rs6000_init_hard_regno_mode_ok): Likewise.
6918 (rs6000_option_override_internal): Likewise.
6919 (rs6000_opt_masks): Likewise.
6920 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
6921 options in terms of whether -mvsx or -mpower8-vector was used.
6922 (TARGET_UPPER_REGS_DI): Likewise.
6923 (TARGET_UPPER_REGS_SF): Likewise.
6924 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
6925 -mupper-regs-* options.
6927 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6929 * passes.c (emergency_dump_function): Print some empty lines and a
6930 header before the RTL dump.
6932 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6934 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
6936 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
6939 * config/aarch64/aarch64.c (aarch64_classify_symbol):
6940 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
6942 2017-07-24 Carl Love <cel@us.ibm.com>
6944 * config/rs6000/rs6000-c.c: Add support for built-in functions
6945 vector float vec_extract_fp32_from_shorth (vector unsigned short);
6946 vector float vec_extract_fp32_from_shortl (vector unsigned short);
6947 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
6948 vec_extract_fp_from_shortl): Add defines for the two builtins.
6949 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
6950 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
6952 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
6953 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
6954 * doc/extend.texi: Update the built-in documentation file for the
6955 new built-in function.
6957 2017-07-24 Jakub Jelinek <jakub@redhat.com>
6960 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
6962 * doc/generic.texi: Likewise.
6963 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6964 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6966 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
6968 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
6969 (aarch64_mls_elt_merge<mode>): Likewise.
6971 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
6973 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
6974 having __cxa_atexit.
6976 2017-07-23 Michael Collison <michael.collison@arm.com>
6978 * config/arm/arm.c (arm_option_override): Deprecate
6979 use of -mstructure-size-boundary.
6980 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
6981 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
6983 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6986 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
6987 Reduce cost estimate for direct moves.
6989 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
6992 * config/i386/i386.c (ix86_option_override_internal): Disable
6993 BMI, BMI2 and TBM instructions for -m16.
6995 2017-07-21 Carl Love <cel@us.ibm.com>
6997 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6998 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6999 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7000 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7001 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7002 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7003 VMULOSW): New enum "unspec" values.
7004 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7005 altivec_vmulosw): New patterns.
7006 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7007 VMULOSW): Add definitions.
7009 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
7011 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
7012 (qdf24xx): Likewise.
7013 * config/aarch64/aarch64-options-extensions.def (rdma); New.
7014 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
7015 (AARCH64_FL_V8_1): Renumber.
7016 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
7017 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
7018 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
7019 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
7020 rdma to feature modifiers list.
7022 2017-07-21 Yury Gribov <tetra2005@gmail.com>
7025 * ipa-visibility (function_and_variable_visibility): Convert
7026 recursive PLT call to direct call if appropriate.
7028 2017-07-21 Andrew Pinski <apinski@cavium.com>
7030 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
7031 operand 1 to see if the types precision matches.
7032 * fold-const.c (operand_equal_p): Likewise.
7034 2017-07-21 Richard Biener <rguenther@suse.de>
7036 PR tree-optimization/81303
7037 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
7038 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
7039 (vect_peeling_hash_get_lowest_cost): Adjust.
7040 (vect_enhance_data_refs_alignment): Likewise. Use
7041 vect_get_peeling_costs_all_drs to compute the penalty for no
7042 peeling to match up costs.
7044 2017-07-21 Richard Biener <rguenther@suse.de>
7046 PR tree-optimization/81500
7047 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
7048 we didn't identify a reduction path.
7050 2017-07-21 Tom de Vries <tom@codesourcery.com>
7051 Cesar Philippidis <cesar@codesourcery.com>
7053 PR gcov-profile/81442
7054 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
7057 2017-07-21 Tom de Vries <tom@codesourcery.com>
7060 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
7062 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
7063 nvptx_override_options_after_change.
7065 2017-07-21 Ulrich Drepper <drepper@redhat.com>
7067 * dwarf2out.c (output_file_names): Avoid double testing for
7070 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
7072 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
7074 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7076 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
7078 (try_crossjump_to_edge): Do not punt on partitioned functions.
7080 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7082 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
7083 Put all BBs reachable only via paths crossing cold region to cold
7085 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
7087 2016-07-21 Richard Biener <rguenther@suse.de>
7089 PR tree-optimization/81303
7090 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
7091 into account prologue and epilogue iterations when raising
7092 min_profitable_iters to sth at least covering one vector iteration.
7094 2017-07-21 Tamar Christina <tamar.christina@arm.com>
7096 * config/arm/arm.c (arm_test_cpu_arch_dat):
7099 2017-07-20 Nathan Sidwell <nathan@acm.org>
7101 Remove TYPE_METHODS.
7102 * tree.h (TYPE_METHODS): Delete.
7103 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
7104 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
7105 (dbxout_type_methods): Scan TYPE_FIELDS.
7106 (dbxout_type): Don't check TYPE_METHODS here.
7107 * function.c (use_register_for_decl): Always ignore register for
7108 class types when not optimizing.
7109 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
7110 * tree.c (free_lang_data_in_type): Stitch out member functions and
7111 templates from TYPE_FIELDS.
7112 (build_distinct_type_copy, verify_type_variant,
7113 verify_type): Member fns are on TYPE_FIELDS.
7114 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
7115 * tree-pretty-print.c (dump_generic_node): Likewise.
7117 2017-07-20 Jakub Jelinek <jakub@redhat.com>
7120 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
7121 V2TImode and V4TImode.
7122 (ix86_expand_vector_extract): Likewise.
7123 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
7124 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
7125 (ssescalarmode): Handle V4TImode and V2TImode.
7126 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
7127 (*vec_extractv2ti, *vec_extractv4ti): New insns.
7128 (VEXTRACTI128_MODE): New mode iterator.
7129 (splitter for *vec_extractv?ti first element): New.
7130 (VEC_INIT_MODE): New mode iterator.
7131 (vec_init<mode>): Consolidate 3 expanders into one using
7132 VEC_INIT_MODE mode iterator.
7134 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7136 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
7137 non_spilled_static_chain_regno_p.
7139 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7141 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
7143 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
7145 * bb-reorder.c (connect_traces): Allow copying of blocks within
7148 2017-07-20 Richard Biener <rguenther@suse.de>
7150 * gimple.h (gimple_phi_result): Add gphi * overload.
7151 (gimple_phi_result_ptr): Likewise.
7152 (gimple_phi_arg): Likewise. Adjust index assert to only
7153 allow actual argument accesses rather than all slots available
7155 (gimple_phi_arg_def): Add gphi * overload.
7156 * tree-phinodes.c (make_phi_node): Initialize only actual
7158 (resize_phi_node): Clear memory not covered by old node,
7159 do not initialize excess argument slots.
7160 (reserve_phi_args_for_new_edge): Initialize new argument slot
7163 2017-07-20 Bin Cheng <bin.cheng@arm.com>
7165 PR tree-optimization/81388
7167 2016-07-21 Bin Cheng <bin.cheng@arm.com>
7169 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
7170 by removing computation of may_be_zero.
7172 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7173 Tom de Vries <tom@codesourcery.com>
7176 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
7177 when gimple level profile disagrees with what RTL expander did.
7179 2017-07-20 Richard Biener <rguenther@suse.de>
7181 PR tree-optimization/61171
7182 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
7183 (vect_analyze_stmt): Add slp instance parameter.
7184 (vectorizable_reduction): Likewise.
7185 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
7186 (vect_is_simple_reduction): Deal with chains not detected
7187 as SLP reduction chain, specifically not properly associated
7188 chains containing a mix of plus/minus.
7189 (get_reduction_op): Remove.
7190 (get_initial_defs_for_reduction): Simplify, pass in whether
7191 this is a reduction chain, pass in the SLP node for the PHIs.
7192 (vect_create_epilog_for_reduction): Get the SLP instance as
7194 (vectorizable_reduction): Get the SLP instance as arg.
7195 During analysis remember the SLP node with the PHIs in the
7196 instance. Simplify getting at the vectorized reduction PHIs.
7197 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
7198 through SLP instance.
7199 (vect_slp_analyze_operations): Likewise.
7200 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
7201 (vect_transform_stmt): Likewise.
7203 2017-07-20 Tom de Vries <tom@codesourcery.com>
7205 PR tree-optimization/81489
7206 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
7207 read of phi arg location to before loop that modifies phi.
7209 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7211 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
7214 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7217 * except.c (execute): Fix ordering issue.
7219 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7221 PR rtl-optimization/81423
7222 * combine.c (make_compound_operation_int): Don't try to optimize
7223 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
7225 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7227 PR rtl-optimization/81423
7228 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
7229 with a constant that is -1 in the truncated to mode.
7231 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7233 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
7234 (determine_unlikely_bbs): ... here.
7235 * predict.h (propagate_unlikely_bbs_forward): Declare.
7236 * cfgexpand.c (pass_expand::execute): Use it.
7237 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
7239 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
7240 propagate_unlikely_bbs_forward.
7242 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7245 * except.c (maybe_add_nop_after_section_switch): New function.
7248 2017-07-19 Tom de Vries <tom@codesourcery.com>
7250 * gimple.h (gimple_phi_set_arg): Make assert more strict.
7252 2017-07-19 Tom de Vries <tom@codesourcery.com>
7254 * gimple.h (gimple_phi_arg): Make assert more strict.
7256 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
7258 * config.gcc (powerpc*-*-*): Add mmintrin.h.
7259 * config/rs6000/mmintrin.h: New file.
7260 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
7262 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7264 PR tree-optimization/81346
7265 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
7267 2017-07-19 Tom de Vries <tom@codesourcery.com>
7269 * config/nvptx/nvptx.md (VECIM): Add V2DI.
7271 2017-07-19 Tom de Vries <tom@codesourcery.com>
7273 * config/nvptx/nvptx-modes.def: Add V2DImode.
7274 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
7275 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
7276 (nvptx_output_mov_insn): Handle lack of mov.b128.
7277 (nvptx_print_operand): Handle 'H' and 'L' codes.
7278 (nvptx_vector_mode_supported): Allow V2DImode.
7279 (nvptx_preferred_simd_mode): New function.
7280 (nvptx_data_alignment): New function.
7281 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
7282 nvptx_preferred_simd_mode.
7283 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
7285 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
7287 2017-07-19 Tom de Vries <tom@codesourcery.com>
7289 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
7290 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
7291 (nvptx_vector_mode_supported): New function. Allow V2SImode.
7292 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
7293 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
7294 (mov<VECIM>_insn): New define_insn.
7295 (define_expand "mov<VECIM>): New define_expand.
7297 2017-07-19 Tom de Vries <tom@codesourcery.com>
7299 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
7301 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7303 PR tree-optimization/81346
7304 * fold-const.h (fold_div_compare, range_check_type): Declare.
7305 * fold-const.c (range_check_type): New function.
7306 (build_range_check): Use range_check_type.
7307 (fold_div_compare): No longer static, rewritten into
7308 a match.pd helper function.
7309 (fold_comparison): Don't call fold_div_compare here.
7310 * match.pd (X / C1 op C2): New optimization using fold_div_compare
7313 2017-07-19 Nathan Sidwell <nathan@acm.org>
7315 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
7316 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
7317 * tree.c (find_decls_types_r, verify_type): Use
7318 TYPE_{MIN,MAX}_VALUE_RAW.
7319 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
7320 (hash_tree): Likewise.
7321 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7323 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7326 2017-07-18 Tom de Vries <tom@codesourcery.com>
7329 * omp-expand.c (expand_omp_for_static_chunk): Handle
7330 equal-argument loop exit phi.
7332 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
7335 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
7336 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
7337 operand 2 predicate.
7338 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
7339 operand 2 predicate.
7340 (ror,rol -> rorx splitters): Use const_int_operand as
7341 operand 2 predicate.
7343 2017-06-18 Richard Biener <rguenther@suse.de>
7345 PR tree-optimization/81410
7346 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
7347 the gap in the ! slp_perm SLP case after each group.
7349 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7352 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
7355 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7358 * predict.c (set_even_probabilities): Cleanup; do not affect
7359 probabilities that are already known.
7360 (combine_predictions_for_bb): Call even when count is set.
7362 2017-07-18 Nathan Sidwell <nathan@acm.org>
7364 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
7367 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7370 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
7371 optimization for loop niter analysis.
7373 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
7376 * config/avr/avr.c (avr_optimize_casesi): Don't use
7377 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
7379 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7381 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
7382 body_cost_vec from _vect_peel_extended_info.
7383 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
7384 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
7387 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7389 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
7391 2017-07-18 Richard Biener <rguenther@suse.de>
7393 PR tree-optimization/80620
7394 PR tree-optimization/81403
7395 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
7396 info when re-using a VN table entry.
7398 2017-07-18 Richard Biener <rguenther@suse.de>
7400 PR tree-optimization/81418
7401 * tree-vect-loop.c (vectorizable_reduction): Properly compute
7402 vectype_in. Verify that with lane-reducing reduction operations
7403 we have a single def-use cycle.
7405 2017-07-17 Carl Love <cel@us.ibm.com>
7407 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
7409 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7410 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7411 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7412 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7413 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7414 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7415 VMULOSW): New enum "unspec" values.
7416 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
7417 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
7418 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7419 altivec_vmulosw): New patterns.
7420 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7421 VMULOSW): Add definitions.
7422 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
7424 * config/alpha/alpha.c: Include predict.h.
7426 2017-07-17 Yury Gribov <tetra2005@gmail.com>
7428 * tree-vrp.c (compare_assert_loc): Fix comparison function
7429 to return predictable results.
7431 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7433 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
7436 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
7437 * doc/invoke.texi (mexpand-adddi): Update text.
7439 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7441 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
7442 that also clobbers the CC register. The old expand code is moved
7444 (*arc_clzsi2): ... here.
7445 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
7446 the CC register. The old expand code is moved to ...
7447 (arc_ctzsi2): ... here.
7449 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7451 * config/arc/arc.opt (mindexed-loads): Use initial value
7452 TARGET_INDEXED_LOADS_DEFAULT.
7453 (mauto-modify-reg): Use initial value
7454 TARGET_AUTO_MODIFY_REG_DEFAULT.
7455 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7456 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7457 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7458 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7460 2017-07-17 Martin Liska <mliska@suse.cz>
7463 * opts.c (finish_options): Do not allow -fgnu-tm
7464 w/ -fsanitize={kernel-,}address. Say sorry.
7466 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7469 * tree-loop-distribution.c (classify_partition): Only assert on
7470 numer of iterations.
7471 (merge_dep_scc_partitions): Delete prameter. Update function call.
7472 (distribute_loop): Remove code handling loop with unknown niters.
7473 (pass_loop_distribution::execute): Skip loop with unknown niters.
7475 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7478 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
7479 function sort_partitions_by_post_order.
7481 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7483 PR tree-optimization/81374
7484 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
7485 the max index of basic blocks, rather than number of basic blocks.
7487 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7489 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
7491 (arc_legitimate_pic_operand_p): Likewise.
7492 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
7494 (arc_needs_pcl_p): Likewise.
7495 (arc_legitimate_pc_offset_p): Likewise.
7496 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
7497 function is also used in constrains.md.
7498 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
7499 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
7500 PLUS. Only return true/false in known cases, otherwise assert.
7501 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
7502 is already called in arc_legitimate_constant_p.
7503 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
7505 (LEGITIMATE_PIC_OPERAND_P): Use
7506 arc_raw_symbolic_reference_mentioned_p function.
7507 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
7512 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7513 Andrew Burgess <andrew.burgess@embecosm.com>
7515 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
7516 (arc_return_address_register): New function.
7517 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
7518 (arc_handle_fndecl_attribute): Add naked attribute.
7519 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7520 (TARGET_WARN_FUNC_RETURN): Likewise.
7521 (arc_allocate_stack_slots_for_args): New function.
7522 (arc_warn_func_return): Likewise.
7523 (machine_function): Change type fn_type.
7524 (arc_compute_function_type): Consider new naked function type,
7525 change function return type.
7526 (arc_must_save_register): Adapt to handle new
7527 arc_compute_function_type's return type.
7528 (arc_expand_prologue): Likewise.
7529 (arc_expand_epilogue): Likewise.
7530 (arc_return_address_regs): Delete.
7531 (arc_return_address_register): New function.
7532 (arc_epilogue_uses): Use above function.
7533 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
7534 (arc_function_type): Change encoding, add naked type.
7535 (ARC_INTERRUPT_P): Change to handle the new encoding.
7536 (ARC_FAST_INTERRUPT_P): Likewise.
7537 (ARC_NORMAL_P): Define.
7538 (ARC_NAKED_P): Likewise.
7539 (arc_compute_function_type): Delete prototype.
7540 * config/arc/arc.md (in_ret_delay_slot): Use
7541 arc_return_address_register function.
7542 (simple_return): Likewise.
7543 (p_return_i): Likewise.
7545 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7547 PR tree-optimization/81428
7548 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
7549 can't be built for those types.
7551 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7553 Remove stuff dead since r239246.
7555 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
7556 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
7557 (avr_inform_devices): Remove dead stuff.
7559 2017-07-17 Tamar Christina <tamar.christina@arm.com>
7561 * config/arm/arm_neon.h: Fix softp typo.
7563 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7565 PR tree-optimization/81365
7566 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
7567 aggregate moves onto bb predecessor edges, make sure there are no
7568 loads that could alias the lhs in between the start of bb and the
7571 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7574 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
7575 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
7576 [LSHIFTRT, outer_code = TRUNCATE]: Same.
7578 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7580 PR tree-optimization/81396
7581 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
7582 (init_symbolic_number): Initialize it to 1.
7583 (perform_symbolic_merge): Add n_ops from both operands into the new
7585 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
7586 without base_addr as useless if they need more than one operation.
7587 (bswap_replace): Handle !bswap case for NULL base_addr.
7589 2017-07-17 Tom de Vries <tom@codesourcery.com>
7592 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
7595 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
7597 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
7598 conditional builtin define __FIX_LEON3FT_B2BST.
7600 2017-07-17 Daniel Cederman <cederman@gaisler.com>
7602 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
7603 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
7606 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7608 PR rtl-optimization/81424
7609 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
7610 to remove potential trapping from operands if -fnon-call-exceptions.
7612 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7614 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
7615 profile_proability for scalling.
7616 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
7618 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7620 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
7622 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7624 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
7625 fixpoint arithmetics.
7627 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7629 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
7630 fixpoint arithmetics.
7632 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7634 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
7635 fixpoint arithmetics.
7637 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7639 * profile-count.h (profile_probability::from_reg_br_prob_note,
7640 profile_probability::to_reg_br_prob_note): New functions.
7641 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
7642 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
7643 * predict.c (probability_reliable_p): Update.
7644 (edge_probability_reliable_p): Update.
7645 (br_prob_note_reliable_p): Update.
7646 (invert_br_probabilities): Update.
7647 (add_reg_br_prob_note): New function.
7648 (combine_predictions_for_insn): Update.
7649 * asan.c (asan_clear_shadow): Update.
7650 * cfgbuild.c (compute_outgoing_frequencies): Update.
7651 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
7652 (update_br_prob_note): Update.
7653 (rtl_verify_edges): Update.
7654 (purge_dead_edges): Update.
7655 (fixup_reorder_chain): Update.
7656 * emit-rtl.c (try_split): Update.
7657 * ifcvt.c (cond_exec_process_insns): Update.
7658 (cond_exec_process_if_block): Update.
7659 (dead_or_predicable): Update.
7660 * internal-fn.c (expand_addsub_overflow): Update.
7661 (expand_neg_overflow): Update.
7662 (expand_mul_overflow): Update.
7663 * loop-doloop.c (doloop_modify): Update.
7664 * loop-unroll.c (compare_and_jump_seq): Update.
7665 * optabs.c (emit_cmp_and_jump_insn_1): Update.
7666 * predict.h: Update.
7667 * reorg.c (mostly_true_jump): Update.
7669 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
7670 * config/alpha/alpha.c (emit_unlikely_jump): Update.
7671 * config/arc/arc.c: (emit_unlikely_jump): Update.
7672 * config/arm/arm.c: (emit_unlikely_jump): Update.
7673 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
7674 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
7675 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
7676 (ix86_print_operand): Update.
7677 (ix86_split_fp_branch): Update.
7678 (predict_jump): Update.
7679 * config/ia64/ia64.c (ia64_print_operand): Update.
7680 * config/mmix/mmix.c (mmix_print_operand): Update.
7681 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
7682 (rs6000_expand_split_stack_prologue): Update.
7683 * config/rs6000/rs6000.c: Update.
7684 * config/s390/s390.c (s390_expand_vec_strlen): Update.
7685 (s390_expand_vec_movstr): Update.
7686 (s390_expand_cs_tdsi): Update.
7687 (s390_expand_split_stack_prologue): Update.
7688 * config/sh/sh.c (sh_print_operand): Update.
7689 (expand_cbranchsi4): Update.
7690 (expand_cbranchdi4): Update.
7691 * config/sparc/sparc.c (output_v9branch): Update.
7692 * config/spu/spu.c (get_branch_target): Update.
7693 (ea_load_store_inline): Update.
7694 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
7695 * config/tilepro/tilepro.c: Update.
7697 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7699 * gimplify.c (mostly_copy_tree_r): Revert latest change.
7700 (gimplify_save_expr): Likewise.
7702 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7704 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
7706 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7708 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
7710 * timevar.def (TV_IPA_FNSUMMARY): Define.
7712 2017-07-16 Daniel Cederman <cederman@gaisler.com>
7714 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
7715 to back store errata sensitive sequence from being generated.
7716 (sqrtdf2_fix): Likewise.
7718 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7720 * tree-ssa-threadupdate.c (compute_path_counts,
7721 update_joiner_offpath_counts): Use profile_probability.
7723 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
7726 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7728 * config/arm/arm-c.c (arm_cpu_builtins): Define
7729 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7731 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
7733 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7734 array entries to represent __ieee128 versions of the
7735 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
7736 scalar_extract_sig, and scalar_insert_exp built-in functions.
7737 (altivec_resolve_overloaded_builtin): Add special case handling
7738 for the __builtin_scalar_insert_exp function, as represented by
7739 the P9V_BUILTIN_VEC_VSIEDP constant.
7740 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
7741 exponent support for __ieee128 argument.
7742 (VSESQP): Add scalar extract signature support for __ieee128
7744 (VSTDCNQP): Add scalar test negative support for __ieee128
7746 (VSIEQP): Add scalar insert exponent support for __int128 argument
7747 with __ieee128 result.
7748 (VSIEQPF): Add scalar insert exponent support for __ieee128
7749 argument with __ieee128 result.
7750 (VSTDCQP): Add scalar test data class support for __ieee128
7752 (VSTDCNQP): Add overload support for scalar test negative with
7754 (VSTDCQP): Add overload support for scalar test data class
7756 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
7758 (UNSPEC_VSX_SIEXPQP): New constant.
7759 (xsxexpqp): New insn for VSX scalar extract exponent quad
7761 (xsxsigqp): New insn for VSX scalar extract significand quad
7763 (xsiexpqpf): New insn for VSX scalar insert exponent quad
7764 precision with floating point argument.
7765 (xststdcqp): New expand for VSX scalar test data class quad
7767 (xststdcnegqp): New expand for VSX scalar test negative quad
7769 (xststdcqp): New insn to match expansions for VSX scalar test data
7770 class quad precision and VSX scalar test negative quad precision.
7771 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7772 special case operand checking to enforce that second operand of
7773 VSX scalar test data class with quad precision argument is a 7-bit
7775 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
7776 prototypes and descriptions of __ieee128 versions of
7777 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
7778 scalar_test_data_class, and scalar_test_neg built-in functions.
7780 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7782 PR tree-optimization/81162
7783 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
7784 replace a negate with an add.
7786 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
7788 * doc/invoke.texi (arm/-mcpu): Document +crypto.
7790 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7792 * config/arm/arm-c.c (arm_cpu_builtins): Define
7793 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7795 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7797 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
7798 (armv8-r): Set ARM Cortex-R52 as default CPU.
7799 * config/arm/arm-tables.opt: Regenerate.
7800 * config/arm/arm-tune.md: Regenerate.
7801 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7803 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
7804 extension for -mcpu=cortex-r52.
7806 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7808 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
7809 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
7810 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
7811 (fp-armv8): Define it as FP_ARMv8 only.
7812 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
7813 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
7815 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
7816 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
7817 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
7818 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
7819 than TARGET_FPU_ARMV8.
7820 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
7821 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
7822 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
7824 * config/arm/neon.md (neon_vrint): Likewise.
7825 (neon_vcvt): Likewise.
7826 (neon_<fmaxmin_op><mode>): Likewise.
7827 (<fmaxmin><mode>3): Likewise.
7828 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
7829 * config/arm/predicates.md (arm_cond_move_operator): Check against
7830 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
7832 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
7834 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
7837 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7839 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
7840 loop in comment with memset.
7842 2017-07-14 Martin Liska <mliska@suse.cz>
7844 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
7845 * dwarf2out.c (is_java): Remove the function.
7846 (output_pubname): Remove usage of the function.
7847 (lower_bound_default): Remove usage of DW_LANG_Java.
7848 (gen_compile_unit_die): Likewise.
7849 * gcc.c: Remove compiler defaults for .java and .zip files.
7850 * gimple-expr.c (remove_suffix): Change as there's no longer
7851 extension than 4-letter one.
7852 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
7853 (gimplify_save_expr): Likewise.
7854 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
7855 as it's possible even for other languages than Java.
7856 * langhooks.h (struct lang_hooks): Remove Java from a comment.
7857 * lto-opts.c (lto_write_options): Remove reference to Java.
7858 * opts.c (strip_off_ending): Update file extension handling.
7859 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
7860 * tree-eh.c (lower_resx): Likewise.
7861 * tree.c (free_lang_data_in_type): Remove dead code.
7862 (find_decls_types_r): Likewise.
7863 (build_common_builtin_nodes): Remove Java from a comment.
7864 (verify_type): Remove dead code.
7865 * varasm.c (assemble_external): Remove Java from a comment.
7867 2017-07-14 Martin Liska <mliska@suse.cz>
7869 * opts.c (finish_options): Add quotes.
7870 (common_handle_option): Likewise.
7872 2017-07-14 Martin Liska <mliska@suse.cz>
7874 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
7875 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
7877 * dwarf2out.c (lower_bound_default): Do not handle
7879 (gen_compile_unit_die): Likewise.
7880 * gcc.c: Remove default extension binding for GNU Pascal.
7881 * stmt.c: Remove Pascal language from a comment.
7882 * xcoffout.c: Likewise.
7884 2017-07-13 David Malcolm <dmalcolm@redhat.com>
7887 * diagnostic-show-locus.c (fixit_cmp): New function.
7888 (layout::layout): Sort m_fixit_hints.
7889 (column_range::column_range): Assert that the values are valid.
7890 (struct char_span): New struct.
7891 (correction::overwrite): New method.
7892 (struct source_line): New struct.
7893 (line_corrections::add_hint): Add assertions. Reimplement memcpy
7894 calls in terms of classes source_line and char_span, and
7895 correction::overwrite.
7896 (selftest::test_overlapped_fixit_printing_2): New function.
7897 (selftest::diagnostic_show_locus_c_tests): Call it.
7899 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7901 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
7902 early if there is no lhs.
7904 2017-07-13 Martin Liska <mliska@suse.cz>
7906 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
7907 (gen_reference_type_die): Likewise.
7908 * stor-layout.c: Remove Pascal-related comment.
7910 2017-07-13 Martin Liska <mliska@suse.cz>
7912 * opts.c (finish_options): Add quotes to error messages.
7913 (parse_sanitizer_options): Likewise.
7915 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
7917 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
7919 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
7921 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
7923 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
7925 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
7927 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
7929 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7932 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
7933 provides the hardware capability bits, define the macro
7934 __BUILTIN_CPU_SUPPORTS__.
7935 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
7936 if GLIBC does not provide the hardware capability bits. Add a
7937 gcc_unreachable call if the built-in cpu function is neither
7938 __builtin_cpu_is nor __builtin_cpu_supports.
7939 (rs6000_get_function_versions_dispatcher): Change the warning
7940 that an old GLIBC is used which does not export the capability
7941 bits to be an error.
7942 * doc/extend.texi (target_clones attribute): Document the
7943 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
7944 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
7945 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
7946 the macros defined by GCC if the newer GLIBC is available.
7948 2017-07-12 Jeff Law <law@redhat.com>
7950 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
7951 remaining includes slightly.
7952 * config/riscv/riscv-builtins.c: Include profile-count.h.
7954 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
7957 * config/avr/avr.c (avr_set_current_function): In diagnostic
7958 messages: Quote keywords and (parts of) identifiers.
7959 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
7962 2017-07-12 Carl Love <cel@us.ibm.com>
7964 * config/rs6000/rs6000-c.c: Add support for built-in functions
7965 vector bool char vec_revb (vector bool char);
7966 vector bool short vec_revb (vector short char);
7967 vector bool int vec_revb (vector bool int);
7968 vector bool long long vec_revb (vector bool long long);
7969 * doc/extend.texi: Update the built-in documentation file for the
7970 new built-in functions.
7972 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7974 * config/s390/s390.md: Remove movcc splitter.
7976 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7978 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
7979 load/store on condition.
7981 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
7984 * config/avr/avr.c (avr_encode_section_info)
7985 [progmem && !TREE_READONLY]: Error if progmem object needs
7988 2017-07-11 Michael Collison <michael.collison@arm.com>
7990 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
7993 2017-07-11 Carl Love <cel@us.ibm.com>
7995 * config/rs6000/rs6000-c.c: Add support for builtins
7996 vector unsigned int vec_parity_lsbb (vector signed int);
7997 vector unsigned int vec_parity_lsbb (vector unsigned int);
7998 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
7999 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
8000 vector unsigned long long vec_parity_lsbb (vector signed long long);
8001 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
8002 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
8003 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
8004 * doc/extend.texi: Update the built-in documentation file for the
8005 new built-in functions.
8007 2017-07-11 David Malcolm <dmalcolm@redhat.com>
8009 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
8010 (layout::m_primary_loc): New field.
8011 (layout::layout): Initialize new field. Move location filtering
8012 logic from here to...
8013 (layout::maybe_add_location_range): ...this new method. Add
8014 support for filtering to just the lines already specified by other
8016 (layout::will_show_line_p): New method.
8017 (gcc_rich_location::add_location_if_nearby): New method.
8018 (selftest::test_add_location_if_nearby): New test function.
8019 (selftest::diagnostic_show_locus_c_tests): Call it.
8020 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
8023 2017-07-11 Tom de Vries <tom@codesourcery.com>
8025 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
8026 (bb_first_real_insn): New function.
8027 (nvptx_single): Add extra initialization of broadcasted condition
8030 2017-07-11 Nathan Sidwell <nathan@acm.org>
8032 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
8034 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
8036 * doc/extend.texi (AVR Function Attributes): Remove weblink to
8037 Binutils doc as TEXI will mess them up.
8038 * doc/invoke.texi (AVR Options): Same here.
8040 2017-07-11 Daniel Cederman <cederman@gaisler.com>
8042 * config/sparc/sparc.opt (mfix-ut700): New option.
8043 (mfix-gr712rc): Likewise.
8044 (sparc_fix_b2bst): New variable.
8045 * doc/invoke.texi (SPARC options): Document them.
8046 (ARM options): Fix warnings.
8047 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
8048 instructions to prevent sequences that can trigger the store-store
8049 errata for certain LEON3FT processors.
8050 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
8051 (sparc_option_override): Set sparc_fix_b2bst appropriately.
8052 * config/sparc/sparc.md (fix_b2bst): New attribute.
8053 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
8055 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
8058 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
8060 (*rsqrtsf2_sse): Ditto.
8062 (div<mode>3): Macroize insn from divdf3 and divsf3
8063 using MODEF mode iterator.
8065 2017-07-10 Martin Sebor <msebor@redhat.com>
8067 PR tree-optimization/80397
8068 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
8069 instead of testing for equality to INTEGER_TYPE.
8071 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
8073 * config.gcc: Remove uclibc from arc target spec.
8075 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
8077 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
8079 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8082 * lto-wrapper.c (remove_option): New function.
8083 (merge_and_complain): Merge PIC/PIE options more realistically.
8085 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8087 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
8091 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
8092 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
8093 * config.in: Regenerate.
8094 * configure: Regenerate.
8095 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
8096 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
8097 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
8098 (TARGET_GASISR_PROLOGUES): ...target mask.
8099 * common/config/avr/avr-common.c
8100 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
8101 Set -mgas-isr-prologues.
8102 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
8103 INSERT_PASS_BEFORE for it.
8104 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
8105 * config/avr/avr.c (avr_option_override)
8106 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
8107 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
8108 (avr_attribute_table) <no_gccisr>: Add new function attribute.
8109 (avr_set_current_function) <is_no_gccisr>: Init machine field.
8110 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
8112 (make_avr_pass_pre_proep): New function.
8113 (emit_push_sfr) <treg>: Add argument to function and use it
8115 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
8116 and set machine->gasisr.yes.
8117 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
8118 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
8119 __gcc_isr.n_pushed to .L__stack_usage.
8120 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
8121 (avr_asm_final_postscan_insn): ...this new static function.
8122 * config/avr/avr.h (machine_function)
8123 <is_no_gccisr, use_L__stack_usage>: New fields.
8124 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
8125 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
8126 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
8127 (gasisr, *gasisr): New expander and insn.
8128 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
8129 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
8130 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
8132 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
8134 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
8137 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8139 Move jump-tables out of .text again.
8142 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
8143 (ASM_OUTPUT_ADDR_VEC): New function.
8144 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
8145 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
8146 INSN_ADDRESSes as asm comment.
8147 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
8148 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
8149 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
8150 * config/avr/avr.md (*tablejump): Adjust comment.
8151 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
8152 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
8154 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
8155 (avr_output_addr_vec): New proto.
8156 (avr_log_t) <insn_addresses>: New field.
8158 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
8161 * config/i386/i386.c (ix86_function_arg_advance): Set
8162 outgoing_args_on_stack to true if there are outgoing arguments
8164 (ix86_function_arg): Likewise.
8165 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
8166 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
8167 * config/i386/i386.h (machine_function): Add
8168 outgoing_args_on_stack.
8170 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
8172 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8174 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
8176 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
8178 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
8179 (REAL_H): Remove $(MACHMODE_H).
8180 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
8182 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
8183 $(MACHMODE_H) and double-int.h.
8184 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
8186 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
8189 2017-07-07 Andrew Pinski <apinski@cavium.com>
8191 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
8192 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
8194 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8196 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
8197 Add warning if GCC was not configured to link against a GLIBC that
8198 exports the hardware capability bits.
8199 (make_resolver_func): Make resolver function private and not a
8200 COMDAT function. Create the name with clone_function_name instead
8201 of make_unique_name.
8204 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
8205 correct operand in doing the split.
8207 2017-07-07 Carl Love <cel@us.ibm.com>
8209 * config/rs6000/rs6000-c: Add support for built-in function
8210 vector unsigned short vec_pack_to_short_fp32 (vector float,
8212 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
8213 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
8214 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
8215 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
8216 (convert_4f32_8i16): Add define_expand.
8217 * doc/extend.texi: Update the built-in documentation file for the
8218 new built-in function.
8220 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8222 * config/sparc/m8.md: New file.
8223 * config/sparc/sparc.md: Include m8.md.
8225 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8227 * config/sparc/sparc.opt: New option -mvis4b.
8228 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
8229 (sparc_option_override): Handle VIS4B.
8230 (enum sparc_builtins): Define
8231 SPARC_BUILTIN_DICTUNPACK{8,16,32},
8232 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
8233 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
8234 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
8235 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
8236 (check_constant_argument): New function.
8237 (sparc_vis_init_builtins): Define builtins
8238 __builtin_vis_dictunpack{8,16,32},
8239 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
8240 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
8241 __builtin_vis_fpcmpde{8,16,32}shl and
8242 __builtin_vis_fpcmpur{8,16,32}shl.
8243 (sparc_expand_builtin): Check that the constant operands to
8244 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
8245 constant and in range.
8246 * config/sparc/sparc-c.c (sparc_target_macros): Handle
8248 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
8249 (SPARC_IMM5_P): Likewise.
8250 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
8251 (enabled): Handle vis4b.
8252 (UNSPEC_DICTUNPACK): New unspec.
8253 (UNSPEC_FPCMPSHL): Likewise.
8254 (UNSPEC_FPUCMPSHL): Likewise.
8255 (UNSPEC_FPCMPDESHL): Likewise.
8256 (UNSPEC_FPCMPURSHL): Likewise.
8257 (cpu_feature): New CPU feature `vis4b'.
8258 (dictunpack{8,16,32}): New insns.
8259 (FPCSMODE): New mode iterator.
8260 (fpcscond): New code iterator.
8261 (fpcsucond): Likewise.
8262 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
8263 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
8264 (fpcmpde{8,16,32}{si,di}shl): Likewise.
8265 (fpcmpur{8,16,32}{si,di}shl): Likewise.
8266 * config/sparc/constraints.md: Define constraints `q' for unsigned
8267 2-bit integer constants and `t' for unsigned 5-bit integer
8269 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
8271 (imm5_operand_dictunpack16): Likewise.
8272 (imm5_operand_dictunpack32): Likewise.
8273 (imm2_operand): Likewise.
8274 * doc/invoke.texi (SPARC Options): Document -mvis4b.
8275 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
8276 ditunpack* and fpcmp*shl builtins.
8278 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8280 * config.gcc: Handle m8 in --with-{cpu,tune} options.
8281 * config.in: Add HAVE_AS_SPARC6 define.
8282 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
8284 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
8286 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
8287 (CPP_CPU_SPEC): Handle m8.
8288 (ASM_CPU_SPEC): Likewise.
8289 * config/sparc/sparc-opts.h (enum processor_type): Add
8291 * config/sparc/sparc.c (m8_costs): New struct.
8292 (sparc_option_override): Handle TARGET_CPU_m8.
8293 (sparc32_initialize_trampoline): Likewise.
8294 (sparc64_initialize_trampoline): Likewise.
8295 (sparc_issue_rate): Likewise.
8296 (sparc_register_move_cost): Likewise.
8297 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
8298 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
8299 (ASM_CPU64_DEFAULT_SPEC): Likewise.
8300 (CPP_CPU_SPEC): Handle M8.
8301 (ASM_CPU_SPEC): Likewise.
8302 (AS_M8_FLAG): Define.
8303 * config/sparc/sparc.md: Add m8 to the cpu attribute.
8304 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
8305 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
8307 * configure: Regenerate.
8308 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
8311 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8313 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
8315 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
8316 ("*movdi_insn_sp32"): Do not set v3pipe.
8317 ("*movsi_insn"): Likewise.
8318 ("*movdi_insn_sp64"): Likewise.
8319 ("*movsf_insn"): Likewise.
8320 ("*movdf_insn_sp32"): Likewise.
8321 ("*movdf_insn_sp64"): Likewise.
8322 ("*zero_extendsidi2_insn_sp64"): Likewise.
8323 ("*sign_extendsidi2_insn"): Likewise.
8324 ("*mov<VM32:mode>_insn"): Likewise.
8325 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8326 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8327 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8328 ("<vlop:code><VL:mode>3"): Likewise.
8329 ("*not_<vlop:code><VL:mode>3"): Likewise.
8330 ("*nand<VL:mode>_vis"): Likewise.
8331 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
8332 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
8333 ("one_cmpl<VL:mode>2"): Likewise.
8334 ("faligndata<VM64:mode>_vis"): Likewise.
8335 ("alignaddrsi_vis"): Likewise.
8336 ("alignaddrdi_vis"): Likweise.
8337 ("alignaddrlsi_vis"): Likewise.
8338 ("alignaddrldi_vis"): Likewise.
8339 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8340 ("bmaskdi_vis"): Likewise.
8341 ("bmasksi_vis"): Likewise.
8342 ("bshuffle<VM64:mode>_vis"): Likewise.
8343 ("cmask8<P:mode>_vis"): Likewise.
8344 ("cmask16<P:mode>_vis"): Likewise.
8345 ("cmask32<P:mode>_vis"): Likewise.
8346 ("pdistn<P:mode>_vis"): Likewise.
8347 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8349 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8351 * config/sparc/sparc.md ("subtype"): New insn attribute.
8352 ("*wrgsr_sp64"): Set insn subtype.
8353 ("*rdgsr_sp64"): Likewise.
8354 ("alignaddrsi_vis"): Likewise.
8355 ("alignaddrdi_vis"): Likewise.
8356 ("alignaddrlsi_vis"): Likewise.
8357 ("alignaddrldi_vis"): Likewise.
8358 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8359 ("fexpand_vis"): Likewise.
8360 ("fpmerge_vis"): Likewise.
8361 ("faligndata<VM64:mode>_vis"): Likewise.
8362 ("bshuffle<VM64:mode>_vis"): Likewise.
8363 ("cmask8<P:mode>_vis"): Likewise.
8364 ("cmask16<P:mode>_vis"): Likewise.
8365 ("cmask32<P:mode>_vis"): Likewise.
8366 ("fchksm16_vis"): Likewise.
8367 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
8368 ("fmean16_vis"): Likewise.
8369 ("fp<plusminus_insn>64_vis"): Likewise.
8370 ("<plusminus_insn>v8qi3"): Likewise.
8371 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8372 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
8373 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
8374 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
8375 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
8376 ("*movqi_insn"): Likewise.
8377 ("*movhi_insn"): Likewise.
8378 ("*movsi_insn"): Likewise.
8379 ("movsi_pic_gotdata_op"): Likewise.
8380 ("*movdi_insn_sp32"): Likewise.
8381 ("*movdi_insn_sp64"): Likewise.
8382 ("movdi_pic_gotdata_op"): Likewise.
8383 ("*movsf_insn"): Likewise.
8384 ("*movdf_insn_sp32"): Likewise.
8385 ("*movdf_insn_sp64"): Likewise.
8386 ("*zero_extendhisi2_insn"): Likewise.
8387 ("*zero_extendqihi2_insn"): Likewise.
8388 ("*zero_extendqisi2_insn"): Likewise.
8389 ("*zero_extendqidi2_insn"): Likewise.
8390 ("*zero_extendhidi2_insn"): Likewise.
8391 ("*zero_extendsidi2_insn_sp64"): Likewise.
8392 ("ldfsr"): Likewise.
8393 ("prefetch_64"): Likewise.
8394 ("prefetch_32"): Likewise.
8395 ("tie_ld32"): Likewise.
8396 ("tie_ld64"): Likewise.
8397 ("*tldo_ldub_sp32"): Likewise.
8398 ("*tldo_ldub1_sp32"): Likewise.
8399 ("*tldo_ldub2_sp32"): Likewise.
8400 ("*tldo_ldub_sp64"): Likewise.
8401 ("*tldo_ldub1_sp64"): Likewise.
8402 ("*tldo_ldub2_sp64"): Likewise.
8403 ("*tldo_ldub3_sp64"): Likewise.
8404 ("*tldo_lduh_sp32"): Likewise.
8405 ("*tldo_lduh1_sp32"): Likewise.
8406 ("*tldo_lduh_sp64"): Likewise.
8407 ("*tldo_lduh1_sp64"): Likewise.
8408 ("*tldo_lduh2_sp64"): Likewise.
8409 ("*tldo_lduw_sp32"): Likewise.
8410 ("*tldo_lduw_sp64"): Likewise.
8411 ("*tldo_lduw1_sp64"): Likewise.
8412 ("*tldo_ldx_sp64"): Likewise.
8413 ("*mov<VM32:mode>_insn"): Likewise.
8414 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8415 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8417 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8419 * config/sparc/sparc.md ("type"): New insn type viscmp.
8420 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
8422 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
8423 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
8424 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8425 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
8427 ("n7_vis_logical_11cycle"): Likewise.
8428 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
8429 * config/sparc/niagara2.md ("niag3_vis": Likewise.
8430 * config/sparc/niagara.md ("niag_vis"): Likewise.
8431 * config/sparc/ultra3.md ("us3_fga"): Likewise.
8432 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
8434 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8436 * config/sparc/sparc.md: New instruction type `bmask'.
8437 (bmaskdi_vis): Use the `bmask' type.
8438 (bmasksi_vis): Likewise.
8439 * config/sparc/ultra3.md (us3_array): Likewise.
8440 * config/sparc/niagara7.md (n7_array): Likewise.
8441 * config/sparc/niagara4.md (n4_array): Likewise.
8442 * config/sparc/niagara2.md (niag2_vis): Likewise.
8443 (niag3_vis): Likewise.
8444 * config/sparc/niagara.md (niag_vis): Likewise.
8446 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8448 * ipa-comdats.c: Remove optimize check from gate.
8449 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
8450 for functions not optimized.
8451 (ipa_fn_summary_read): Skip optimize check.
8452 (ipa_fn_summary_write): Likewise.
8453 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
8455 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
8457 (can_inline_edge_p): Check flag_pcc_struct_return for match.
8458 (check_callers): Give up on caller which is not optimized.
8459 (inline_small_functions): Likewise.
8460 (ipa_inline): Do not give up when not optimizing.
8461 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
8462 away unoptimizes cdtors.
8463 (whole_program_function_and_variable_visibility): Do
8464 ipa_discover_readonly_nonaddressable_vars in LTO mode.
8465 * ipa.c (process_references): Do not check optimize.
8466 (symbol_table::remove_unreachable_nodes): Update optimize check.
8467 (set_writeonly_bit): Update optimize check.
8468 (pass_ipa_cdtor_merge::gate): Do not check optimize.
8469 (pass_ipa_single_use::gate): Remove.
8471 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8473 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
8474 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
8475 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
8476 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
8477 permute_load, permute_store, adjust_extract, adjust_splat,
8478 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
8479 replace_swap_with_copy, dump_swap_insn_table,
8480 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
8481 recombine_lvx_pattern, recombine_stvx_pattern,
8482 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
8483 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
8484 to file rs6000-p8swap.c.
8485 * config/rs6000/rs6000-p8swap.c: New file.
8486 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
8487 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
8488 and rs6000*-*-* targets.
8490 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8492 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
8494 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8496 * lto-wrapper.c (merge_and_complain): Do not merge
8497 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
8498 fsigned_zeros, ftrapping_math, fwrapv.
8499 (append_compiler_options): Do not track these options.
8500 (append_linker_options): Likewie
8502 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8504 * cgraphunit.c (cgraph_node::finalize_function): When
8505 !flag_toplevel_reorde set no_reorder flag.
8506 (varpool_node::finalize_decl): Likewise.
8507 (symbol_table::compile): Drop no toplevel reorder path.
8509 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8511 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
8512 edges; zero probability is not better than uninitialized.
8514 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8516 * asan.h (asan_sanitize_allocas_p): Declare.
8517 * asan.c (asan_sanitize_allocas_p): New function.
8518 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
8519 (handle_builtin_alloca): Likewise.
8520 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
8521 if !asan_sanitize_allocas_p.
8522 * params.def (asan-instrument-allocas): Add new option.
8523 * params.h (ASAN_PROTECT_ALLOCAS): Define.
8524 * opts.c (common_handle_option): Disable allocas sanitization for
8527 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8529 * asan.c: Include gimple-fold.h.
8530 (get_last_alloca_addr): New function.
8531 (handle_builtin_stackrestore): Likewise.
8532 (handle_builtin_alloca): Likewise.
8533 (asan_emit_allocas_unpoison): Likewise.
8534 (get_mem_refs_of_builtin_call): Add new parameter, remove const
8535 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
8536 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
8537 (instrument_builtin_call): Pass gimple iterator to
8538 get_mem_refs_of_builtin_call.
8539 (last_alloca_addr): New global.
8540 * asan.h (asan_emit_allocas_unpoison): Declare.
8541 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
8542 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
8543 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
8544 if function calls alloca.
8545 * gimple-fold.c (replace_call_with_value): Remove static keyword.
8546 * gimple-fold.h (replace_call_with_value): Declare.
8547 * internal-fn.c: Include asan.h.
8548 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
8549 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
8551 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8553 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
8554 (C_SELFTEST_FLAGS): New.
8555 (CPP_SELFTEST_FLAGS): New.
8556 (SELFTEST_DEPS): New, from deps of s-selftest.
8557 (C_SELFTEST_DEPS): New, from deps of s-selftest.
8558 (CPP_SELFTEST_DEPS): New.
8559 (selftest): Add dependency on s-selftest-c++.
8560 (s-selftest): Rename to...
8561 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
8562 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
8563 than SELFTEST_FLAGS.
8564 (selftest-gdb): Rename to...
8565 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
8567 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
8568 (selftest-valgrind): Rename to...
8569 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
8571 (selftest-valgrind): Reintroduce as an alias for
8572 selftest-c-valgrind.
8573 (s-selftest-c++): New.
8574 (selftest-c++-gdb): New.
8575 (selftest-c++-valgrind): New.
8577 2017-07-06 Olivier Hainque <hainque@adacore.com>
8579 * gcc.c (process_command): When deciding if undefined variables
8580 should be ignored when processing specs, accept "gcc -v" as well.
8582 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8584 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
8585 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
8587 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
8589 * config/arm/arm-cpus.in (armv8-r): Add new entry.
8590 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
8591 * config/arm/arm-tables.opt: Regenerate.
8592 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
8594 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
8596 2017-07-06 Carl Love <cel@us.ibm.com>
8598 * ChangeLog: Clean up from mid air collision
8600 2017-07-06 Carl Love <cel@us.ibm.com>
8602 * config/rs6000/rs6000-c.c: Add support for built-in functions
8603 vector signed int vec_subc (vector signed int, vector signed int);
8604 vector signed __int128 vec_subc (vector signed __int128,
8605 vector signed __int128);
8606 vector unsigned __int128 vec_subc (vector unsigned __int128,
8607 vector unsigned __int128);
8608 vector signed int vec_sube (vector signed int, vector signed int,
8610 vector unsigned int vec_sube (vector unsigned int,
8611 vector unsigned int,
8612 vector unsigned int);
8613 vector signed __int128 vec_sube (vector signed __int128,
8614 vector signed __int128,
8615 vector signed__int128);
8616 vector unsigned __int128 vec_sube (vector unsigned __int128,
8617 vector unsigned __int128,
8618 vector unsigned __int128);
8619 vector signed int vec_subec (vector signed int, vector signed int,
8621 vector unsigned int vec_subec (vector unsigned int,
8622 vector unsigned int,
8623 vector unsigned int);
8624 vector signed __int128 vec_subec (vector signed __int128,
8625 vector signed __int128,
8626 vector signed__int128);
8627 vector unsigned __int128 vec_subec (vector unsigned __int128,
8628 vector unsigned __int128,
8629 vector unsigned __int128);
8630 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
8631 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
8632 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
8633 BU_ALTIVEC_OVERLOAD_X definitions.
8634 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
8635 * doc/extend.texi: Update the built-in documentation file for the new
8638 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8641 * diagnostic-show-locus.c (layout::layout): Use start and finish
8642 spelling location for the start and finish of each range.
8643 * genmatch.c (linemap_client_expand_location_to_spelling_point):
8644 Add unused aspect param.
8645 * input.c (expand_location_1): Add "aspect" param, and use it
8646 to access the correct part of the location.
8647 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
8649 (expand_location_to_spelling_point): Likewise.
8650 (linemap_client_expand_location_to_spelling_point): Add "aspect"
8651 param, and pass it to expand_location_1.
8653 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
8655 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
8656 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
8657 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
8658 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
8659 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
8660 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
8661 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
8662 _mm_maskz_getmant_ss): New intrinsics.
8663 (__builtin_ia32_getexpss128_mask): Changed to ...
8664 __builtin_ia32_getexpss128_round ... this.
8665 (__builtin_ia32_getexpsd128_mask): Changed to ...
8666 __builtin_ia32_getexpsd128_round ... this.
8667 * config/i386/i386-builtin-types.def
8668 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
8669 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
8670 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
8671 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
8672 __builtin_ia32_getmantss_mask_round): New builtins.
8673 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
8674 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
8675 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
8676 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
8677 * config/i386/sse.md
8678 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
8679 avx512f_sgetexp<mode><mask_scalar_name>
8680 <round_saeonly_scalar_name> ... this.
8681 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
8682 %0, %1, %2<round_saeonly_op3>}): Changed to ...
8683 vgetexp<ssescalarmodesuffix>
8684 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8685 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
8686 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
8687 avx512f_vgetmant<mode><mask_scalar_name>
8688 <round_saeonly_scalar_name> ... this.
8689 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
8690 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
8691 vgetmant<ssescalarmodesuffix>
8692 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
8693 %0<mask_scalar_operand4>, %1, %2
8694 <round_saeonly_scalar_mask_op4>, %3} ... this.
8695 * config/i386/subst.md (mask_scalar_operand4,
8696 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
8697 round_saeonly_scalar_nimm_predicate): New subst attributes.
8699 2017-07-06 Julia Koval <julia.koval@intel.com>
8701 * config/i386/i386.c (ix86_erase_embedded_rounding):
8702 Remove code for old rounding pattern.
8704 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
8706 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
8708 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
8710 * doc/sourcebuild.texi (Test Directives, Variants of
8711 dg-require-support): Add documentation for dg-require-stack-check.
8713 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
8715 * config/i386/subst.md (mask_scalar, round_scalar,
8716 round_saeonly_scalar): New meta-templates.
8717 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
8718 round_scalar_mask_operand3, round_scalar_mask_op3,
8719 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
8720 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
8721 round_saeonly_scalar_constraint,
8722 round_saeonly_scalar_prefix): New subst attribute.
8723 * config/i386/sse.md
8724 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
8725 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
8726 <round_scalar_name> ... this.
8727 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
8728 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
8729 <round_scalar_name> ... this.
8730 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
8731 <sse>_vm<code><mode>3<mask_scalar_name>
8732 <round_saeonly_scalar_name> ... this.
8733 (v<plusminus_mnemonic><ssescalarmodesuffix>
8734 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8735 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8736 v<plusminus_mnemonic><ssescalarmodesuffix>
8737 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8738 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8739 (v<multdiv_mnemonic><ssescalarmodesuffix>
8740 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8741 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8742 v<multdiv_mnemonic><ssescalarmodesuffix>
8743 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8744 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8745 (v<maxmin_float><ssescalarmodesuffix>
8746 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
8747 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
8748 v<maxmin_float><ssescalarmodesuffix>
8749 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8750 %0<mask_scalar_operand3>, %1, %<iptr>2
8751 <round_saeonly_scalar_mask_op3>} ... this.
8753 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
8755 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
8756 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
8758 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
8759 Alan Hayward <alan.hayward@arm.com>
8760 David Sherwood <david.sherwood@arm.com>
8762 * combine.c (simplify_if_then_else): Remove "enum" before
8764 * compare-elim.c (can_eliminate_compare): Likewise.
8765 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8767 (aarch64_lookup_simd_builtin_type): Likewise.
8768 (aarch64_simd_builtin_type): Likewise.
8769 (aarch64_init_simd_builtin_types): Likewise.
8770 (aarch64_simd_expand_args): Likewise.
8771 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
8773 (aarch64_reverse_mask): Likewise.
8774 (aarch64_simd_emit_reg_reg_move): Likewise.
8775 (aarch64_gen_adjusted_ldpstp): Likewise.
8776 (aarch64_ccmp_mode_to_code): Likewise.
8777 (aarch64_operands_ok_for_ldpstp): Likewise.
8778 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8779 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
8781 (aarch64_min_divisions_for_recip_mul): Likewise.
8782 (aarch64_reassociation_width): Likewise.
8783 (aarch64_get_condition_code_1): Likewise.
8784 (aarch64_simd_emit_reg_reg_move): Likewise.
8785 (aarch64_simd_attr_length_rglist): Likewise.
8786 (aarch64_reverse_mask): Likewise.
8787 (aarch64_operands_ok_for_ldpstp): Likewise.
8788 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8789 (aarch64_gen_adjusted_ldpstp): Likewise.
8790 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
8792 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
8793 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8794 (arm_lookup_simd_builtin_type): Likewise.
8795 (arm_simd_builtin_type): Likewise.
8796 (arm_init_simd_builtin_types): Likewise.
8797 (arm_expand_builtin_args): Likewise.
8798 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
8799 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
8800 (ft32_setup_incoming_varargs): Likewise.
8801 (ft32_function_arg): Likewise.
8802 (ft32_function_arg_advance): Likewise.
8803 (ft32_pass_by_reference): Likewise.
8804 (ft32_arg_partial_bytes): Likewise.
8805 (ft32_valid_pointer_mode): Likewise.
8806 (ft32_addr_space_pointer_mode): Likewise.
8807 (ft32_addr_space_legitimate_address_p): Likewise.
8808 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
8810 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
8811 (ix86_emit_outlined_ms2sysv_restore): Likewise.
8812 (iamcu_alignment): Likewise.
8813 (canonicalize_vector_int_perm): Likewise.
8814 (ix86_noce_conversion_profitable_p): Likewise.
8815 (ix86_mpx_bound_mode): Likewise.
8816 (ix86_operands_ok_for_move_multiple): Likewise.
8817 * config/microblaze/microblaze-protos.h
8818 (microblaze_expand_conditional_branch_reg): Likewise.
8819 * config/microblaze/microblaze.c
8820 (microblaze_expand_conditional_branch_reg): Likewise.
8821 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8823 (rs6000_reassociation_width): Likewise.
8824 (rs6000_invalid_binary_op): Likewise.
8825 (fusion_p9_p): Likewise.
8826 (emit_fusion_p9_load): Likewise.
8827 (emit_fusion_p9_store): Likewise.
8828 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
8830 (riscv_hard_regno_mode_ok_p): Likewise.
8831 (riscv_address_insns): Likewise.
8832 (riscv_split_symbol): Likewise.
8833 (riscv_legitimize_move): Likewise.
8834 (riscv_function_value): Likewise.
8835 (riscv_hard_regno_nregs): Likewise.
8836 (riscv_expand_builtin): Likewise.
8837 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
8838 (riscv_build_integer): Likewise.
8839 (riscv_split_integer): Likewise.
8840 (riscv_legitimate_constant_p): Likewise.
8841 (riscv_cannot_force_const_mem): Likewise.
8842 (riscv_regno_mode_ok_for_base_p): Likewise.
8843 (riscv_valid_base_register_p): Likewise.
8844 (riscv_valid_offset_p): Likewise.
8845 (riscv_valid_lo_sum_p): Likewise.
8846 (riscv_classify_address): Likewise.
8847 (riscv_legitimate_address_p): Likewise.
8848 (riscv_address_insns): Likewise.
8849 (riscv_load_store_insns): Likewise.
8850 (riscv_force_binary): Likewise.
8851 (riscv_split_symbol): Likewise.
8852 (riscv_force_address): Likewise.
8853 (riscv_legitimize_address): Likewise.
8854 (riscv_move_integer): Likewise.
8855 (riscv_legitimize_const_move): Likewise.
8856 (riscv_legitimize_move): Likewise.
8857 (riscv_address_cost): Likewise.
8858 (riscv_subword): Likewise.
8859 (riscv_output_move): Likewise.
8860 (riscv_canonicalize_int_order_test): Likewise.
8861 (riscv_emit_int_order_test): Likewise.
8862 (riscv_function_arg_boundary): Likewise.
8863 (riscv_pass_mode_in_fpr_p): Likewise.
8864 (riscv_pass_fpr_single): Likewise.
8865 (riscv_pass_fpr_pair): Likewise.
8866 (riscv_get_arg_info): Likewise.
8867 (riscv_function_arg): Likewise.
8868 (riscv_function_arg_advance): Likewise.
8869 (riscv_arg_partial_bytes): Likewise.
8870 (riscv_function_value): Likewise.
8871 (riscv_pass_by_reference): Likewise.
8872 (riscv_setup_incoming_varargs): Likewise.
8873 (riscv_print_operand): Likewise.
8874 (riscv_elf_select_rtx_section): Likewise.
8875 (riscv_save_restore_reg): Likewise.
8876 (riscv_for_each_saved_reg): Likewise.
8877 (riscv_register_move_cost): Likewise.
8878 (riscv_hard_regno_mode_ok_p): Likewise.
8879 (riscv_hard_regno_nregs): Likewise.
8880 (riscv_class_max_nregs): Likewise.
8881 (riscv_memory_move_cost): Likewise.
8882 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
8883 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
8884 (rl78_addr_space_address_mode): Likewise.
8885 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8887 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8888 (rs6000_reassociation_width): Likewise.
8889 (rs6000_invalid_binary_op): Likewise.
8890 (fusion_p9_p): Likewise.
8891 (emit_fusion_p9_load): Likewise.
8892 (emit_fusion_p9_store): Likewise.
8893 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
8894 (ok_for_simple_move_operands): Likewise.
8895 (ok_for_simple_move_strict_operands): Likewise.
8896 (ok_for_simple_arith_logic_operands): Likewise.
8897 (visium_legitimize_reload_address): Likewise.
8898 (visium_select_cc_mode): Likewise.
8899 (output_cbranch): Likewise.
8900 (visium_split_double_move): Likewise.
8901 (visium_expand_copysign): Likewise.
8902 (visium_expand_int_cstore): Likewise.
8903 (visium_expand_fp_cstore): Likewise.
8904 * config/visium/visium.c (visium_pass_by_reference): Likewise.
8905 (visium_function_arg): Likewise.
8906 (visium_function_arg_advance): Likewise.
8907 (visium_libcall_value): Likewise.
8908 (visium_setup_incoming_varargs): Likewise.
8909 (visium_legitimate_constant_p): Likewise.
8910 (visium_legitimate_address_p): Likewise.
8911 (visium_legitimize_address): Likewise.
8912 (visium_secondary_reload): Likewise.
8913 (visium_register_move_cost): Likewise.
8914 (visium_memory_move_cost): Likewise.
8915 (prepare_move_operands): Likewise.
8916 (ok_for_simple_move_operands): Likewise.
8917 (ok_for_simple_move_strict_operands): Likewise.
8918 (ok_for_simple_arith_logic_operands): Likewise.
8919 (visium_function_value_1): Likewise.
8920 (rtx_ok_for_offset_p): Likewise.
8921 (visium_legitimize_reload_address): Likewise.
8922 (visium_split_double_move): Likewise.
8923 (visium_expand_copysign): Likewise.
8924 (visium_expand_int_cstore): Likewise.
8925 (visium_expand_fp_cstore): Likewise.
8926 (visium_split_cstore): Likewise.
8927 (visium_select_cc_mode): Likewise.
8928 (visium_split_cbranch): Likewise.
8929 (output_cbranch): Likewise.
8930 (visium_print_operand_address): Likewise.
8931 * expmed.c (flip_storage_order): Likewise.
8932 * expmed.h (emit_cstore): Likewise.
8933 (flip_storage_order): Likewise.
8934 * genrecog.c (validate_pattern): Likewise.
8935 * hsa-gen.c (gen_hsa_addr): Likewise.
8936 * internal-fn.c (expand_arith_overflow): Likewise.
8937 * ira-color.c (allocno_copy_cost_saving): Likewise.
8938 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8939 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
8940 (process_invariant_for_inheritance): Likewise.
8941 * lra-eliminations.c (move_plus_up): Likewise.
8942 * omp-low.c (lower_oacc_reductions): Likewise.
8943 * simplify-rtx.c (simplify_subreg): Likewise.
8944 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
8945 (TARGET_CHKP_BOUND_MODE): Likewise..
8946 * targhooks.c (default_chkp_bound_mode): Likewise.
8947 (default_setup_incoming_vararg_bounds): Likewise.
8948 * targhooks.h (default_chkp_bound_mode): Likewise.
8949 (default_setup_incoming_vararg_bounds): Likewise.
8950 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
8951 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
8952 (have_whole_vector_shift): Likewise.
8953 * tree-vect-stmts.c (vectorizable_load): Likewise.
8954 * doc/tm.texi: Regenerate.
8956 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
8958 Graceful degrade if Binutils PR21472 is not available.
8961 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
8962 .rodata in flash test fails.
8963 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
8964 * confgure: Regenerate.
8965 * config.in: Regenerate.
8966 * config/avr/avr.c (avr_asm_named_section)
8967 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
8968 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
8969 (avr_asm_init_sections): Same.
8971 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8973 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
8974 (fma<VH:mode>4_intrinsic): Likewise.
8975 (*fmsub<VCVTF:mode>4): Likewise.
8976 (*fmsub<VH:mode>4_intrinsic): Likewise.
8978 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
8981 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
8982 Don't depend on "optimize > 0".
8983 (out_movhi_r_mr, out_movqi_mr_r): Same.
8984 (out_movhi_mr_r, out_movqi_r_mr): Same.
8985 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
8986 io_address_operand on "optimize > 0".
8988 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8990 * tree-loop-distribution.c: Add general explanantion on the pass.
8991 (generate_loops_for_partition): Mark distributed loop.
8992 (pg_add_dependence_edges): New parameter. Handle alias data
8993 dependence specially and record it in the parameter if asked.
8994 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
8995 (init_partition_graph_vertices, add_partition_graph_edge): New.
8996 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
8997 (free_partition_graph_vdata, build_partition_graph): New.
8998 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
8999 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
9000 (data_ref_segment_size, latch_dominated_by_data_ref): New.
9001 (compute_alias_check_pairs, version_loop_by_alias_check): New.
9002 (version_for_distribution_p, finalize_partitions): New.
9003 (distribute_loop): Handle alias data dependence specially. Factor
9004 out loop fusion code as functions and call these functions.
9006 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9008 * tree-loop-distribution.c (classify_partition): New parameter and
9009 better handle reduction statement.
9010 (rdg_build_partitions): Revise comment.
9011 (distribute_loop): Compute statements in all partitions and pass it
9012 to classify_partition.
9014 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9016 * tree-loop-distribution.c (enum partition_type): New.
9017 (struct partition): New field type.
9018 (partition_merge_into): Add parameter. Update partition type.
9019 (data_dep_in_cycle_p, update_type_for_merge): New functions.
9020 (build_rdg_partition_for_vertex): Compute partition type.
9021 (rdg_build_partitions): Dump partition type.
9022 (distribute_loop): Update calls to partition_merge_into.
9024 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9026 * tree-loop-distribution.c (struct ddr_hasher): New.
9027 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
9029 (classify_partition): Call get_data_dependence.
9030 (pg_add_dependence_edges): Ditto.
9031 (distribute_loop): Release data dependence hash table.
9033 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9035 * tree-loop-distribution.c (ref_base_address): Delete.
9036 (similar_memory_accesses): Rename ...
9037 (share_memory_accesses): ... to this. Check if partitions access
9038 the same memory reference.
9039 (distribute_loop): Call share_memory_accesses.
9041 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9043 * tree-loop-distribution.c (struct partition): New field recording
9045 (partition_alloc, partition_free): Init and release data refs.
9046 (partition_merge_into): Merge data refs.
9047 (build_rdg_partition_for_vertex): Collect data refs for partition.
9048 (pg_add_dependence_edges): Change parameters from vector to bitmap.
9050 (distribute_loop): Remve data refs from vertice data of partition
9053 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9055 * tree-loop-distribution.c (params.h): Include header file.
9056 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
9057 (datarefs_vec): New global var.
9058 (create_rdg_vertices): Use datarefs_vec directly.
9059 (free_rdg): Don't free data references.
9060 (build_rdg): Update use. Don't free data references.
9061 (distribute_loop): Compute global variable for data references.
9062 Bail out if there are too many data references.
9064 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9066 * tree-loop-distribution.c (loop_nest): New global var.
9067 (build_rdg): Use loop directly, rather than loop nest.
9068 (pg_add_dependence_edges): Remove loop nest parameter. Use global
9070 (distribute_loop): Compute global variable loop nest. Update use.
9072 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9074 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
9075 (partition_merge_into): New parameter. Dump reason for fusion.
9076 (distribute_loop): Update use of partition_merge_into.
9078 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9080 * tree-loop-distribution.c (bb_top_order_index): New.
9081 (bb_top_order_index_size, bb_top_order_cmp): New.
9082 (stmts_from_loop): Use topological order.
9083 (pass_loop_distribution::execute): Compute and release topological
9084 order for basic blocks.
9086 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9088 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
9091 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9093 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
9094 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
9095 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
9096 * internal-fn.def (LOOP_DIST_ALIAS): New.
9097 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
9098 (fold_loop_internal_call): ... this.
9099 (vect_loop_dist_alias_call): New function.
9100 (set_uid_loop_bbs): Call fold_loop_internal_call.
9101 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
9104 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9107 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
9108 Require dead FLAGS_REG at the beginning of a peephole.
9110 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9113 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
9114 arguments in the call to __builtin_ia32_sbb_u32.
9115 (_subborrow_u64): Swap _X and _Y arguments in the call to
9116 __builtin_ia32_sbb_u64.
9118 2017-07-04 Jakub Jelinek <jakub@redhat.com>
9121 * tree-vrp.c (compare_assert_loc): Turn into a function template
9122 with stable template parameter. Only test if a->e is NULL,
9123 !a->e == !b->e has been verified already. Use e == NULL or
9124 e != NULL instead of e or ! e tests. If stable is true, don't use
9125 iterative_hash_expr, on the other side allow a or b or both NULL
9126 and sort the NULLs last.
9127 (process_assert_insertions): Sort using compare_assert_loc<false>
9128 instead of compare_assert_loc, later sort using
9129 compare_assert_loc<true> before calling process_assert_insertions_for
9130 in a loop. Use break instead of continue once seen NULL pointer.
9132 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9134 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9135 Cortex-R7 and Cortex-R8 processors.
9137 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9139 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
9140 uninitialized while src is not.
9142 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
9144 * common/config/arm/arm-common.c: Adjust include path for
9146 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
9147 (arm-cpu.h): Create in build directory. Adjust dependency rules.
9148 (arm-cpu-data.h): Likewise.
9149 (arm-cpu-cdata.h): Likewise.
9150 * config/arm/arm-cpu.h: Delete.
9151 * config/arm/arm-cpu-cdata.h: Delete.
9152 * config/arm/arm-cpu-data.h: Delete.
9154 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
9156 * config/arm/arm-cpus.in (cortex-a55): New.
9157 (cortex-a75): Likewise.
9158 (cortex-a75.cortex-a55): Likewise.
9159 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
9161 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
9162 * config/arm/arm-cpu-cdata.h: Regenerate.
9163 * config/arm/arm-cpu-data.h: Regenerate.
9164 * config/arm/arm-cpu.h: Regenerate.
9165 * config/arm/arm-tables.opt: Regenerate.
9166 * config/arm/arm-tune.md: Regenerate.
9168 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9170 * haifa-sched.c (sched_create_recovery_edges): Update profile.
9172 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9174 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
9177 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
9179 PR tree-optimization/81292
9180 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
9181 full_string_p, also call adjust_related_strinfos if the adjustment
9182 is simple, otherwise invalidate related strinfos.
9184 2017-07-04 Martin Liska <mliska@suse.cz>
9187 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
9188 newly created variable as DECL_IGNORED_P.
9190 2017-07-04 Martin Liska <mliska@suse.cz>
9193 * ipa-inline.c (inline_small_functions):
9194 Use xstrdup_for_dump.
9196 2017-07-04 Tom de Vries <tom@codesourcery.com>
9198 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
9200 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9203 * config/darwin.c (darwin_function_switched_text_sections):
9206 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
9208 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
9210 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9212 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
9214 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9216 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
9217 min_profitable_iters, and th as inclusive lower bounds.
9218 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
9219 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
9220 for min_profitable_iters and min_profitable_estimate.
9221 (vect_transform_loop): Treat th as an inclusive lower bound.
9222 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
9224 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9227 * config/darwin.c (darwin_function_switched_text_sections):
9228 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
9229 in two pieces, and suppress the use of buf.
9231 2017-07-03 Nathan Sidwell <nathan@acm.org>
9233 * hash-table.h (hash_table_mod1): Fix indentation.
9235 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9238 * predict.c (force_edge_cold): Be more careful about propagation
9240 * profile-count.h (profile_probability::guessed,
9241 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
9243 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
9245 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
9247 * doc/invoke.texi (rcpc architecture extension): Document it.
9249 2017-07-03 Richard Biener <rguenther@suse.de>
9251 PR tree-optimization/60510
9252 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
9253 the scalar reduction PHI and use it.
9254 (vectorizable_reduction): Properly guard the single_defuse_cycle
9255 path for non-SLP reduction chains where we cannot use it.
9256 Rework reduc_def/index and vector type deduction. Rework
9257 vector operand gathering during reduction op code-gen.
9258 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
9259 chains dissolve the chain and leave it to non-SLP reduction
9262 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9264 * tree-data-ref.h (dr_alignment): Declare.
9265 * tree-data-ref.c (dr_alignment): New function.
9266 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
9267 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
9269 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
9271 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9273 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
9274 and base_misalignment fields.
9275 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
9276 * tree-data-ref.c: Include builtins.h.
9277 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
9278 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
9279 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
9280 * tree-vect-data-refs.c: Include tree-cfg.h.
9281 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
9282 fields instead of calculating an alignment here.
9283 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
9284 innermost_loop_behavior fields.
9286 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9288 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
9290 (DR_STEP_ALIGNMENT): New macro.
9291 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
9292 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
9293 (create_data_ref): Print it.
9294 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
9295 to tell whether the step preserves vector (mis)alignment.
9296 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9297 Move the check for an integer step and generalise to all INTEGER_CST.
9298 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
9299 Print the outer step alignment.
9301 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9303 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
9304 with offset_alignment.
9305 (DR_ALIGNED_TO): Delete.
9306 (DR_OFFSET_ALIGNMENT): New macro.
9307 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
9308 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
9309 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
9310 (create_data_ref): Likewise.
9311 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9312 (vect_analyze_data_refs): Likewise.
9313 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
9314 creating dummy innermost behavior.
9316 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9318 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
9319 with a "innermost_loop_behavior *" and refeence tree.
9320 * tree-data-ref.c (dr_analyze_innermost): Likewise.
9321 (create_data_ref): Update call accordingly.
9322 * tree-predcom.c (find_looparound_phi): Likewise.
9324 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9326 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
9327 fields with dr_wrt_vec_loop.
9328 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
9329 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
9330 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
9331 (vect_dr_behavior): New function.
9332 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9333 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
9334 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
9335 track whether the step preserves the misalignment.
9336 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9337 Use vect_dr_behavior.
9338 (vect_setup_realignment): Update call accordingly.
9339 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
9340 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
9341 call to vect_create_addr_base_for_vector_ref.
9342 (vect_create_cond_for_align_checks): Likewise.
9343 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
9344 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
9345 (vect_recog_mask_conversion_pattern): Likewise.
9346 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
9347 (new_stmt_vec_info): Remove redundant zeroing.
9349 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9351 * common/config/arm/arm-common.c (arm_be8_option): New function.
9352 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
9353 (ISA_ARMv6): Add isa_bit_be8.
9354 * config/arm/arm.h (arm_be8_option): Add prototype.
9355 (BE8_SPEC_FUNCTION): New define.
9356 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
9357 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
9358 (mlittle-endian): Similarly.
9359 (mbe8, mbe32): New options.
9360 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
9361 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
9363 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9365 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
9367 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9369 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
9370 (cleanup_tree_cfg_bb): Use it.
9371 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
9373 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
9375 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9378 * loop-doloop.c (add_test): Update profile.
9380 2017-07-03 Martin Liska <mliska@suse.cz>
9383 * sanopt.c (rewrite_usage_of_param): New function.
9384 (sanitize_rewrite_addressable_params): Likewise.
9385 (pass_sanopt::execute): Call rewrite_usage_of_param.
9387 2017-07-03 Richard Biener <rguenther@suse.de>
9389 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
9390 back to using VIEW_CONVERT_EXPR.
9392 2017-07-03 Martin Liska <mliska@suse.cz>
9395 * doc/extend.texi: Document when a resolver function is
9396 generated for target_clones.
9398 2017-07-03 Martin Liska <mliska@suse.cz>
9400 * asan.c (asan_emit_stack_protection): Unpoison just red zones
9401 and shadow memory of auto variables which are subject of
9402 use-after-scope sanitization.
9403 (asan_expand_mark_ifn): Add do set only when is_poison.
9405 2016-07-03 Richard Biener <rguenther@suse.de>
9407 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
9409 (vect_force_simple_reduction): Record reduction def -> phi mapping.
9410 (vectorizable_reduction): Perform reduction PHI creation when
9411 visiting a reduction PHI and adjust and simplify code generation
9412 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
9413 (vect_transform_loop): Visit reduction PHIs.
9414 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
9415 defs into the SLP tree.
9416 (vect_build_slp_tree): Reduction defs terminate the recursion.
9417 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
9419 (vect_get_vec_defs_for_stmt_copy): Export.
9420 (vect_get_vec_defs): Likewise.
9421 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
9423 (vect_get_vec_defs_for_stmt_copy): Declare.
9424 (vect_get_vec_defs): Likewise.
9426 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9428 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
9429 parameter with a "loop" parameter and use it instead of the
9430 loop containing DR_STMT. Don't check simple_iv when doing
9431 BB analysis. Describe the two analysis modes in the comment.
9433 2017-07-03 Tom de Vries <tom@codesourcery.com>
9435 PR tree-optimization/69468
9436 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
9437 (find_same_succ_bb): Handle ignore_edge_flags.
9439 2017-07-03 Tom de Vries <tom@codesourcery.com>
9441 PR tree-optimization/81192
9442 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
9444 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
9446 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
9448 2017-07-03 Tom de Vries <tom@codesourcery.com>
9450 PR tree-optimization/81192
9451 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
9452 BB_SAME_SUCC (bb) == NULL.
9454 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9456 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
9459 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9461 * dumpfile.c: Include profile-count.h
9462 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
9464 (insert_cond_bb): Update profile.
9465 * tree-cfg.h (insert_cond_bb): Update prototype.
9466 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
9467 * tree-dump.c: Do not include tree-cfg.
9469 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9471 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
9473 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9475 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
9478 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9480 * tree-complex.c (expand_complex_div_wide): update profile.
9482 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9483 Alan Hayward <alan.hayward@arm.com>
9484 David Sherwood <david.sherwood@arm.com>
9486 * Makefile.in (MACHMODE_H): Remove insn-modes.h
9487 (CORETYPES_H): New define.
9488 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
9489 (insn-modes-inline.h, s-modes-inline-h): New rules.
9490 (generated_files): Add insn-modes-inline.h.
9491 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
9492 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
9493 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
9494 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
9495 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
9496 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
9497 (build/gencodes.o, build/genconditions.o): Likewise.
9498 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
9499 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
9500 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
9501 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
9502 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
9503 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
9504 * coretypes.h: Include everything up to real.h for generators.
9505 Include insn-modes.h first. Include wide-int-print.h after
9506 wide-int.h. Include insn-modes-inline.h and then machmode.h.
9507 * machmode.h: Don't include insn-modes.h here.
9508 * function-tests.c: Remove includes of signop.h, machmode.h,
9509 double-int.h and wide-int.h.
9511 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
9513 * optc-save-gen.awk: Likewise.
9514 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
9515 * godump.c: Remove include of wide-int-print.h.
9516 * pretty-print.h: Likewise.
9517 * wide-int-print.cc: Likewise.
9518 * wide-int.cc: Likewise.
9519 * hash-map-tests.c: Remove include of signop.h.
9520 * hash-set-tests.c: Likewise.
9521 * rtl-tests.c: Likewise.
9522 * mkconfig.sh: Remove include of machmode.h.
9523 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
9525 (emit_insn_modes_inline_h): ...this new function. Emit the code
9526 into an insn-modes-inline.h header file, adding appropriate
9527 include guards and end comments.
9528 (emit_insn_modes_c_header): Remove include of machmode.h.
9529 (emit_min_insn_modes_c_header): Include coretypes.h rather than
9531 (main): Handle -i flag and call emit_insn_modes_inline_h when
9534 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9536 * tree-ssa-strlen.c (strinfo): Rename the length field to
9537 nonzero_chars. Add a full_string_p field.
9538 (compare_nonzero_chars, zero_length_string_p): New functions.
9539 (get_addr_stridx): Add an offset_out parameter.
9540 Use compare_nonzero_chars.
9541 (get_stridx): Update accordingly. Use compare_nonzero_chars.
9542 (new_strinfo): Update after above changes to strinfo.
9543 (set_endptr_and_length): Set full_string_p.
9544 (get_string_length): Update after above changes to strinfo.
9545 (unshare_strinfo): Update call to new_strinfo.
9546 (maybe_invalidate): Likewise.
9547 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
9548 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
9549 as a uhwi instead of an shwi. Update after above changes to
9550 strinfo and new_strinfo.
9551 (zero_length_string): Assert that chainsi contains full strings.
9552 Use zero_length_string_p. Update call to new_strinfo.
9553 (adjust_related_strinfos): Update after above changes to strinfo.
9554 Copy full_string_p from origsi.
9555 (adjust_last_stmt): Use zero_length_string_p.
9556 (handle_builtin_strlen): Update after above changes to strinfo and
9557 new_strinfo. Install the lhs as the string length if the previous
9558 entry didn't describe a full string.
9559 (handle_builtin_strchr): Update after above changes to strinfo
9561 (handle_builtin_strcpy): Likewise.
9562 (handle_builtin_strcat): Likewise.
9563 (handle_builtin_malloc): Likewise.
9564 (handle_pointer_plus): Likewise.
9565 (handle_builtin_memcpy): Likewise. Track nonzero characters
9566 that aren't necessarily followed by a nul terminator.
9567 (handle_char_store): Likewise.
9569 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9571 PR tree-optimization/80769
9572 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
9573 for malloc and calloc. Document the new invariant that all related
9574 strinfos have delayed lengths or none do.
9575 (verify_related_strinfos): Move earlier in file.
9576 (set_endptr_and_length): New function, split out from...
9577 (get_string_length): ...here. Also set the lengths of related
9579 (zero_length_string): Assert that chainsi has known (rather than
9581 (adjust_related_strinfos): Likewise.
9583 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9585 PR tree-optimization/81136
9586 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
9587 assert that two references with the same misalignment have the same
9588 compile-time misalignment if those compile-time misalignments
9591 2017-07-01 Andi Kleen <ak@linux.intel.com>
9593 * print-tree.c (print_node): Print all attributes.
9595 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9597 * cfg.c (scale_bbs_frequencies): New function.
9598 * cfg.h (scale_bbs_frequencies): Declare it.
9599 * cfgloopanal.c (single_likely_exit): Cleanup.
9600 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
9602 (scale_loop_profile): Likewise.
9603 (loop_version): Likewise.
9604 (create_empty_loop_on_edge): Update.
9605 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
9606 scale_loop_frequencies, scale_loop_profile, loopify,
9607 loop_version): Update prototypes.
9608 * modulo-sched.c (sms_schedule): Update.
9609 * predict.c (unlikely_executed_edge_p): Also check probability.
9610 (probably_never_executed_edge_p): Fix typo.
9611 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9612 * tree-parloops.c (gen_parallel_loop): Update.
9613 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
9614 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9615 * tree-ssa-loop-split.c (split_loop): Update.
9616 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9617 * tree-vect-loop-manip.c (vect_do_peeling): Update.
9618 (vect_loop_versioning): Update.
9619 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9621 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9623 * trans-mem.c (split_bb_make_tm_edge): Update profile.
9625 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9627 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
9628 to keep profile consistent.
9630 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9632 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
9633 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
9634 * profile-count.h (max_safe_multiplier): Make unsigned.
9635 (profile_count::guessed_zero): New.
9637 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9639 * bb-reorder.c (fix_up_crossing_landing_pad,
9640 fix_crossing_conditional_branches): Use make_single_succ_edge
9641 to keep profile consistent.
9643 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9645 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
9648 2017-07-01 Jakub Jelinek <jakub@redhat.com>
9651 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
9652 the right scopes, make sure cond_jump isn't preserved between multiple
9653 iterations. Search for fallthru edge whenever there are 3+ edges and
9654 use find_fallthru_edge for it.
9656 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9658 Patch by Alexander Monakov <amonakov@ispras.ru>
9659 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
9660 probabilities consistently.
9662 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9664 * pa.c (pa_expand_compare_and_swap_loop): Update call of
9665 emit_cmp_and_jump_insns.
9667 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9670 * tree-inline.c (expand_call_inline): Combine profile statuses.
9672 2017-06-30 Andrew Pinski <apinski@cavium.com>
9674 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
9675 fold_stmt returned true.
9677 2017-06-30 Nathan Sidwell <nathan@acm.org>
9679 * ggc.h (empty_string): Delete.
9680 * cfgexpand.c (expand_asm_stmt): Use plain "".
9681 * optabs.c (expand_asm_memory_barrier): Likewise.
9682 * stringpool.c (empty_string): Delete.
9683 (digit_vector, digit_string): Delete.
9684 (ggc_alloc_string): Use plain "", don't optimize single digit
9685 strings. Use ggc_alloc_atomic.
9687 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
9689 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
9690 comparison set and one other set, use the cost of the non-comparison
9693 2017-06-30 Nathan Sidwell <nathan@acm.org>
9695 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
9698 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
9700 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
9701 loops. Remove now unneeded calls to gimple_switch_set_label() that
9702 just set removed labels to NULL_TREE.
9704 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
9706 * tree-ssanames.c (set_range_info_raw): Abstract from ...
9707 (set_range_info): ...here. Only call set_range_info_raw if domain
9709 (set_nonzero_bits): Call set_range_info_raw.
9710 * tree-ssanames.h (set_range_info_raw): New.
9712 2017-06-30 Jakub Jelinek <jakub@redhat.com>
9715 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
9716 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
9717 of nonimmediate_operand and <store_mask_constraint> instead of m
9718 for the input operand. For V8FI iterator, always split if input
9719 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
9720 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
9721 <store_mask_predicate> instead of register_operand and
9722 <store_mask_constraint> instead of v for the input operand. Make
9723 sure both operands aren't MEMs for if not <mask_applied>.
9725 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
9727 * lto-wrapper.c (copy_file) Close both file descriptors before
9730 2017-06-30 Martin Liska <mliska@suse.cz>
9733 * multiple_target.c (create_dispatcher_calls): Make ifunc
9734 also for function that don't have calls or are not referenced.
9736 2017-06-30 Richard Biener <rguenther@suse.de>
9738 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
9739 analyze the first scalar stmt. Move vector type computation
9740 for the BB case here from ...
9741 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
9742 live operation processing in the SLP case properly.
9744 2017-06-30 Richard Biener <rguenther@suse.de>
9746 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
9748 2017-06-30 Martin Liska <mliska@suse.cz>
9751 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
9752 before BUILT_IN_UNWIND_RESUME when ASAN is used.
9754 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
9756 * doc/invoke.texi (AArch64): Add missing options and remove redundant
9759 2017-06-30 Richard Biener <rguenther@suse.de>
9761 PR tree-optimization/81249
9762 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
9763 condition reduction result to original scalar type.
9765 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9767 * profile-count.h (enum profile_quality): Fix typos and whitespace
9770 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9772 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
9773 type for branch probabilities.
9775 2017-06-29 Julian Brown <julian@codesourcery.com>
9776 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9778 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
9779 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
9780 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
9781 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
9783 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9785 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
9786 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
9787 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
9788 CC usage from generic code to here.
9789 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
9790 CC usage into the target macros.
9792 2017-06-29 Maya Rashish <coypu@sdf.org>
9794 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
9797 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9799 * arm/arm-builtins.c: Include profile-count.h
9800 * except.c (sjlj_emit_function_enter): Use
9801 profile_probability::unlikely.
9803 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9805 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
9806 and tocrel_offset be pointer args rather than implicitly using
9808 (legitimate_constant_pool_address_p, rs6000_emit_move,
9809 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
9810 tocrel_offset and use in toc_relative_expr_p call.
9811 (print_operand, print_operand_address): Use static tocrel_base_oac
9812 and tocrel_offset_oac.
9813 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
9816 2017-06-29 Maya Rashish <coypu@sdf.org>
9818 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
9820 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
9822 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
9823 objects, take into account only the alignment of 'op0' and 'mode1' if
9826 2017-06-29 Steve Ellcey <sellcey@cavium.com>
9828 * ccmp.c (ccmp_tree_comparison_p): New function.
9829 (ccmp_candidate_p): Update to use above function.
9830 (get_compare_parts): New function.
9831 (expand_ccmp_next): Update to use new functions.
9832 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
9834 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
9835 take mode as argument.
9836 * ccmp.h (expand_ccmp_expr): Add mode as argument.
9837 * expr.c (expand_expr_real_1): Pass mode as argument.
9839 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
9841 * combine.c (combine_instructions): Print insns to dump_file, together
9844 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9846 * asan.c (asan_emit_stack_protection): Update.
9847 (create_cond_insert_point): Update.
9848 * auto-profile.c (afdo_propagate_circuit): Update.
9849 * basic-block.h (struct edge_def): Turn probability to
9850 profile_probability.
9851 (EDGE_FREQUENCY): Update.
9852 * bb-reorder.c (find_traces_1_round): Update.
9853 (better_edge_p): Update.
9854 (sanitize_hot_paths): Update.
9855 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
9856 (make_single_succ_edge): Update.
9857 (check_bb_profile): Update.
9858 (dump_edge_info): Update.
9859 (update_bb_profile_for_threading): Update.
9860 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
9861 probabilitycount to 0.
9862 * cfgbuild.c (compute_outgoing_frequencies): Update.
9863 * cfgcleanup.c (try_forward_edges): Update.
9864 (outgoing_edges_match): Update.
9865 (try_crossjump_to_edge): Update.
9866 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
9867 (expand_gimple_tailcall): Update.
9868 (construct_init_block): Use make_single_succ_edge.
9869 (construct_exit_block): Use make_single_succ_edge.
9870 * cfghooks.c (verify_flow_info): Update.
9871 (redirect_edge_succ_nodup): Update.
9872 (split_edge): Update.
9873 (account_profile_record): Update.
9874 * cfgloopanal.c (single_likely_exit): Update.
9875 * cfgloopmanip.c (scale_loop_profile): Update.
9876 (set_zero_probability): Remove.
9877 (duplicate_loop_to_header_edge): Update.
9878 * cfgloopmanip.h (loop_version): Update prototype.
9879 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9880 (force_nonfallthru_and_redirect): Update.
9881 (update_br_prob_note): Update.
9882 (rtl_verify_edges): Update.
9883 (purge_dead_edges): Update.
9884 (rtl_lv_add_condition_to_bb): Update.
9885 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
9886 * cgraphunit.c (init_lowered_empty_function): Update.
9887 (cgraph_node::expand_thunk): Update.
9888 * cilk-common.c: Include profile-count.h
9889 * dojump.c (inv): Remove.
9890 (jumpifnot): Update.
9891 (jumpifnot_1): Update.
9892 (do_jump_1): Update.
9894 (do_jump_by_parts_greater_rtx): Update.
9895 (do_compare_rtx_and_jump): Update.
9896 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
9897 do_jump_1. do_compare_rtx_and_jump): Update prototype.
9898 * dwarf2cfi.c: Include profile-count.h
9899 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
9900 (sjlj_emit_dispatch_table): Likewise.
9901 * explow.c: Include profile-count.h
9902 * expmed.c (emit_store_flag_force): Update.
9903 (do_cmp_and_jump): Update.
9904 * expr.c (compare_by_pieces_d::generate): Update.
9905 (compare_by_pieces_d::finish_mode): Update.
9906 (emit_block_move_via_loop): Update.
9907 (store_expr_with_bounds): Update.
9908 (store_constructor): Update.
9909 (expand_expr_real_2): Update.
9910 (expand_expr_real_1): Update.
9911 * expr.h (try_casesi, try_tablejump): Update prototypes.
9912 * gimple-pretty-print.c (dump_probability): Update.
9913 (dump_profile): New.
9914 (dump_gimple_label): Update.
9915 (dump_gimple_bb_header): Update.
9916 * graph.c (draw_cfg_node_succ_edges): Update.
9917 * hsa-gen.c (convert_switch_statements): Update.
9918 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
9919 (find_if_case_1): Update.
9920 (find_if_case_2): Update.
9921 * internal-fn.c (expand_arith_overflow_result_store): Update.
9922 (expand_addsub_overflow): Update.
9923 (expand_neg_overflow): Update.
9924 (expand_mul_overflow): Update.
9925 (expand_vector_ubsan_overflow): Update.
9926 * ipa-cp.c (good_cloning_opportunity_p): Update.
9927 * ipa-split.c (split_function): Use make_single_succ_edge.
9928 * ipa-utils.c (ipa_merge_profiles): Update.
9929 * loop-doloop.c (add_test): Update.
9930 (doloop_modify): Update.
9931 * loop-unroll.c (compare_and_jump_seq): Update.
9932 (unroll_loop_runtime_iterations): Update.
9933 * lra-constraints.c (lra_inheritance): Update.
9934 * lto-streamer-in.c (input_cfg): Update.
9935 * lto-streamer-out.c (output_cfg): Update.
9936 * mcf.c (adjust_cfg_counts): Update.
9937 * modulo-sched.c (sms_schedule): Update.
9938 * omp-expand.c (expand_omp_for_init_counts): Update.
9939 (extract_omp_for_update_vars): Update.
9940 (expand_omp_ordered_sink): Update.
9941 (expand_omp_for_ordered_loops): Update.
9942 (expand_omp_for_generic): Update.
9943 (expand_omp_for_static_nochunk): Update.
9944 (expand_omp_for_static_chunk): Update.
9945 (expand_cilk_for): Update.
9946 (expand_omp_simd): Update.
9947 (expand_omp_taskloop_for_outer): Update.
9948 (expand_omp_taskloop_for_inner): Update.
9949 * omp-simd-clone.c (simd_clone_adjust): Update.
9950 * optabs.c (expand_doubleword_shift): Update.
9951 (expand_abs): Update.
9952 (emit_cmp_and_jump_insn_1): Update.
9953 (expand_compare_and_swap_loop): Update.
9954 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
9955 * predict.c (predictable_edge_p): Update.
9956 (edge_probability_reliable_p): Update.
9957 (set_even_probabilities): Update.
9958 (combine_predictions_for_insn): Update.
9959 (combine_predictions_for_bb): Update.
9960 (propagate_freq): Update.
9961 (estimate_bb_frequencies): Update.
9962 (force_edge_cold): Update.
9963 * profile-count.c (profile_count::dump): Add missing space into dump.
9964 (profile_count::debug): Add newline.
9965 (profile_count::differs_from_p): Explicitly convert to unsigned.
9966 (profile_count::stream_in): Update.
9967 (profile_probability::dump): New member function.
9968 (profile_probability::debug): New member function.
9969 (profile_probability::differs_from_p): New member function.
9970 (profile_probability::differs_lot_from_p): New member function.
9971 (profile_probability::stream_in): New member function.
9972 (profile_probability::stream_out): New member function.
9973 * profile-count.h (profile_count_quality): Rename to ...
9974 (profile_quality): ... this one.
9975 (profile_probability): New.
9976 (profile_count): Update.
9977 * profile.c (compute_branch_probabilities): Update.
9978 * recog.c (peep2_attempt): Update.
9979 * sched-ebb.c (schedule_ebbs): Update.
9980 * sched-rgn.c (find_single_block_region): Update.
9981 (compute_dom_prob_ps): Update.
9982 (schedule_region): Update.
9983 * sel-sched-ir.c (compute_succs_info): Update.
9984 * stmt.c (struct case_node): Update.
9985 (do_jump_if_equal): Update.
9986 (get_outgoing_edge_probs): Update.
9987 (conditional_probability): Update.
9988 (emit_case_dispatch_table): Update.
9989 (expand_case): Update.
9990 (expand_sjlj_dispatch_table): Update.
9991 (emit_case_nodes): Update.
9992 * targhooks.c: Update.
9993 * tracer.c (better_p): Update.
9994 (find_best_successor): Update.
9995 * trans-mem.c (expand_transaction): Update.
9996 * tree-call-cdce.c: Update.
9997 * tree-cfg.c (gimple_split_edge): Upate.
9998 (move_sese_region_to_fn): Upate.
9999 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
10000 * tree-eh.c (lower_resx): Upate.
10001 (cleanup_empty_eh_move_lp): Upate.
10002 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10003 * tree-inline.c (copy_edges_for_bb): Update.
10004 (copy_cfg_body): Update.
10005 * tree-parloops.c (gen_parallel_loop): Update.
10006 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
10007 (gimple_gen_time_profiler): Update.
10008 * tree-ssa-dce.c (remove_dead_stmt): Update.
10009 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10010 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10011 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10012 (unloop_loops): Update.
10013 (try_peel_loop): Update.
10014 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10015 * tree-ssa-loop-split.c (connect_loops): Update.
10016 (split_loop): Update.
10017 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10018 (hoist_guard): Update.
10019 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10020 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10021 (value_replacement): Update.
10022 * tree-ssa-reassoc.c (branch_fixup): Update.
10023 * tree-ssa-tail-merge.c (replace_block_by): Update.
10024 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10025 (create_edge_and_update_destination_phis): Update.
10026 (compute_path_counts): Update.
10027 (recompute_probabilities): Update.
10028 (update_joiner_offpath_counts): Update.
10029 (freqs_to_counts_path): Update.
10030 (duplicate_thread_path): Update.
10031 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10032 (struct switch_conv_info): Update.
10033 (gen_inbound_check): Update.
10034 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10035 (vect_do_peeling): Update.
10036 (vect_loop_versioning): Update.
10037 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10038 (optimize_mask_stores): Update.
10039 * ubsan.c (ubsan_expand_null_ifn): Update.
10040 * value-prof.c (gimple_divmod_fixed_value): Update.
10041 (gimple_divmod_fixed_value_transform): Update.
10042 (gimple_mod_pow2): Update.
10043 (gimple_mod_pow2_value_transform): Update.
10044 (gimple_mod_subtract): Update.
10045 (gimple_mod_subtract_transform): Update.
10046 (gimple_ic): Update.
10047 (gimple_stringop_fixed_value): Update.
10048 (gimple_stringops_transform): Update.
10049 * value-prof.h: Update.
10051 2017-06-29 Carl Love <cel@us.ibm.com>
10053 * config/rs6000/rs6000-c.c: Add support for built-in functions
10054 vector signed int vec_signed (vector float);
10055 vector signed long long vec_signed (vector double);
10056 vector signed int vec_signed2 (vector double, vector double);
10057 vector signed int vec_signede (vector double);
10058 vector signed int vec_signedo (vector double);
10059 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
10060 instruction generator.
10061 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10062 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
10063 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
10065 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
10066 vunsignede_v2df): Add define_expands.
10067 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
10068 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
10069 VEC_UNSIGNEDO): Add definitions.
10070 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10071 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
10072 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
10073 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
10074 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
10075 * config/rs6000/altivec.h (vec_signed, vec_signed2,
10076 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
10077 vec_unsignede, vec_unsignedo): Add builtin defines.
10078 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
10080 * doc/extend.texi: Update the built-in documentation file for the
10081 new built-in functions.
10083 2017-06-29 Richard Biener <rguenther@suse.de>
10085 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
10086 reduction chains to LOOP_VINFO_REDUCTIONS.
10087 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
10088 SLP reductions after processing reduction chains.
10090 2017-06-29 Nathan Sidwell <nathan@acm.org>
10092 * builtins.c (fold_builtin_FUNCTION): Use
10093 lang_hooks.decl_printable_name.
10095 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
10097 PR middle-end/81194
10098 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
10099 with only one label.
10100 * stmt.c (expand_case): Assert NCASES is greater than one.
10102 2017-06-29 Richard Biener <rguenther@suse.de>
10104 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
10106 (group_case_labels): Likewise.
10107 (find_taken_edge): Push sanity checking on val to workers...
10108 (find_taken_edge_cond_expr): ... here
10109 (find_taken_edge_switch_expr): ... and here, handle cases
10110 with just a default label.
10111 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
10112 (group_case_labels): Likewise.
10113 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
10114 group_case_labels does anything cleanup the CFG again.
10116 2017-06-29 Bin Cheng <bin.cheng@arm.com>
10118 PR tree-optimization/81196
10119 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
10120 exit condition comparing two IVs.
10122 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
10124 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
10125 profile to the dummy entry at the end of the list of architectures.
10126 * config/arm/arm-cpu-cdata.h: Regenerated.
10128 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10129 Michael Collison <michael.collison@arm.com>
10132 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
10134 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
10135 (*aarch64_reg_<mode>3_minus_mask): New pattern.
10136 (*aarch64_<optab>_reg_di3_mask2): New pattern.
10137 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
10138 of shift when the shift amount is masked with constant equal to
10139 the size of the mode.
10140 * config/aarch64/predicates.md (subreg_lowpart_operator): New
10143 2017-06-29 Martin Liska <mliska@suse.cz>
10145 * config/i386/i386.opt: Change range from [1,5] to [0,5].
10147 2017-06-29 Yury Gribov <tetra2005@gmail.com>
10150 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
10153 (edge_growth_cache_entry::edge_growth_cache_entry): New
10155 (reset_edge_growth_cache): Update to use constructor.
10157 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10159 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
10160 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10161 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10163 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
10165 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
10166 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
10168 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
10170 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
10171 (*-linux-uclibc*): Add t-uclibc tmake_file.
10172 * config/t-musl: New.
10173 * config/t-uclibc: New.
10175 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
10177 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
10179 (gen_comm_data): Emit architectural setting of arch_prof.
10180 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
10182 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
10183 (armv8-m.base, armv8-m.main): Likewise.
10184 * arm-protos.h (arm_build_target): Add profile field.
10185 (arch_option): Likewise.
10186 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
10188 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
10189 arm_active_target.profile.
10191 2017-06-28 Richard Biener <rguenther@suse.de>
10193 PR middle-end/81227
10194 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
10195 TYPE_OVERFLOW_WRAPS.
10196 * match.pd (negate_expr_p): Likewise.
10197 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
10198 fold_build2, not fold_binary.
10200 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10202 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
10203 Convert memory address to Pmode.
10204 (aarch64_print_operand): Assert MEM operands are always Pmode.
10206 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10209 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
10210 Remove redundant if.
10211 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
10212 * config/arm/aarch-common-protos.h
10213 (aarch_forward_to_shift_is_not_shifted_re): Remove.
10214 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
10216 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
10219 * multiple_target.c (create_dispatcher_calls): Set the default
10220 clone to be static, not public.
10222 2017-06-28 Richard Biener <rguenther@suse.de>
10224 * tree-vect-loop.c (vectorizable_reduction): Move special
10225 cond reduction IV var creation ...
10226 (vect_create_epilog_for_reduction): ... here. Remove induction_index
10227 parameter. Use STMT_VINFO_VECTYPE.
10228 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
10231 2017-06-28 Martin Liska <mliska@suse.cz>
10234 * ipa-visibility.c (non_local_p): Handle visibility.
10236 2017-06-28 Martin Liska <mliska@suse.cz>
10239 * common.opt: Add IntegerRange to various options.
10240 * opt-functions.awk (integer_range_info): New function.
10241 * optc-gen.awk: Add integer_range_info to cl_options struct.
10242 * opts-common.c (decode_cmdline_option): Handle
10243 CL_ERR_INT_RANGE_ARG.
10244 (cmdline_handle_error): Likewise.
10245 * opts.c (print_filtered_help): Show valid interval in
10246 when --help is provided.
10247 * opts.h (struct cl_option): Add range_min and range_max fields.
10248 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
10250 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
10252 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
10253 (x * C EQ/NE y * C): New transformation.
10255 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
10257 * genmultilib (combination_space): Accept '+' in option names.
10259 2017-06-28 Martin Liska <mliska@suse.cz>
10262 * asan.c (instrument_derefs): Bail out inner references
10263 that are hard register variables.
10265 2017-06-28 Jakub Jelinek <jakub@redhat.com>
10268 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
10269 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
10271 2017-06-28 Richard Biener <rguenther@suse.de>
10273 * tree-vectorizer.h (vect_get_vec_defs): Remove.
10274 (vect_get_slp_defs): Adjust.
10275 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
10277 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
10279 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10280 get_initial_defs_for_reduction instead of vect_get_vec_defs.
10281 (vectorizable_reduction): Adjust.
10282 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
10284 (vect_get_slp_defs): Likewise.
10285 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
10286 (vectorizable_bswap): Adjust.
10287 (vectorizable_call): Likewise.
10288 (vectorizable_conversion): Likewise.
10289 (vectorizable_assignment): Likewise.
10290 (vectorizable_shift): Likewise.
10291 (vectorizable_operation): Likewise.
10292 (vectorizable_store): Likewise.
10293 (vectorizable_condition): Likewise.
10294 (vectorizable_comparison): Likewise.
10296 2017-06-28 Michael Collison <michael.collison@arm.com>
10299 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
10301 (arm_gen_movmemqi): Removed unused variable 'i'.
10302 Convert 'for' loop into 'while' loop.
10303 (arm_expand_prologue): Remove last unnecessary set of insn.
10304 (thumb_pop): Remove unused variable 'pushed_words'.
10305 (thumb_exit): Remove last unnecessary set of regs_to_pop.
10307 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10309 * config/s390/predicates.md: Use s390_rel_address_ok_p.
10310 * config/s390/s390-protos.h: Add prototype of
10311 s390_rel_address_ok_p.
10312 * config/s390/s390.c (s390_got_symbol): New function.
10313 (s390_rel_address_ok_p): New function.
10314 (legitimize_pic_address): Use s390_rel_address_ok_p.
10315 (s390_load_got): Use s390_got_symbol.
10316 (s390_option_override): Issue error if
10317 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
10318 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
10320 * config/s390/s390.opt: New option mpic-data-is-text-relative.
10322 2017-06-27 Andrew Pinski <apinski@cavium.com>
10324 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
10325 (X * copysign (1.0, X)): New pattern.
10326 (X * copysign (1.0, -X)): New pattern.
10327 (copysign (-1.0, CST)): New pattern.
10329 2017-06-27 Joseph Myers <joseph@codesourcery.com>
10331 * genmultilib (combination_space): Remove variable.
10332 Validate reuse rules against regular expression for any sequence
10333 of multilib options in any order.
10335 2017-06-27 Michael Collison <michael.collison@arm.com>
10337 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
10338 call aarch64_split_simd_combine.
10339 * (aarch64_combine_internal<mode>): Delete pattern.
10340 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
10341 Allow register and subreg operands.
10343 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10345 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
10346 specific need, just fallback on defaults.
10347 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
10349 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10350 Olivier Hainque <hainque@adacore.com>
10352 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
10354 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
10355 targets. Pick a default if no particular attempt applied.
10356 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
10359 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10361 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
10362 (x86_64-wrs-vxworks7): Likewise.
10364 2017-06-27 Marek Polacek <polacek@redhat.com>
10367 * ubsan.c (instrument_null): Check get_base_address's result for null.
10369 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10371 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
10373 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10375 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
10376 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
10377 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
10378 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
10379 New function types.
10380 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
10381 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
10382 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
10383 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
10384 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
10385 BUILT_IN_FEUPDATEENV): New builtins.
10386 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
10387 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
10388 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
10389 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
10391 (builtin_structptr_types): Adjust size.
10392 * tree.c (builtin_structptr_types): Add four entries.
10394 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10395 Olivier Hainque <hainque@adacore.com>
10397 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
10398 (TLS_SYM): New local macro, forcing reference to __tls__ on
10399 link command lines for VxWorks 7 RTPs, triggering initialization
10401 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
10402 OS features TLS support, true for RTPs on VxWorks 7.
10403 * config/vxworks.c (vxworks_override_options): Setup emutls
10406 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10408 * predict.c (test_prediction_value_range): Use -1U instead of -1
10409 to avoid narrowing conversion warning.
10410 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
10411 to avoid narrowing conversion warning.
10412 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
10414 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
10416 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10418 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
10419 64bit configurations.
10420 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
10421 (SIZE_TYPE): Likewise.
10422 * config/vxworks.c (vxworks_emutls_var_fields): Use
10423 long_unsigned_type_node instead of unsigned_type_node as the offset
10424 field type, which is "pointer" mode in emutls.c.
10426 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10429 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
10431 PR middle-end/81207
10432 * gimple-fold.c (replace_call_with_call_and_fold): Handle
10433 gimple_vuse copying separately from gimple_vdef copying.
10435 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10437 * value-prof.c (free_hist): Remove call to memset and the enclosing if
10440 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10441 Olivier Hainque <hainque@adacore.com>
10443 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
10444 for all vxworks7 targets.
10445 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
10446 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
10447 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
10448 variations for VX6/VX7 and 32/64bits later on in ...
10449 (VXWORKS_LIB_SPEC): Leverage new macros.
10450 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
10451 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
10453 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10455 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
10456 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
10458 2017-06-26 Carl Love <cel@us.ibm.com>
10460 * config/rs6000/rs6000-c.c: Add support for built-in functions
10461 vector bool char vec_reve (vector bool char);
10462 vector signed char vec_reve (vector signed char);
10463 vector unsigned char vec_reve (vector unsigned char);
10464 vector bool int vec_reve (vector bool int);
10465 vector signed int vec_reve (vector signed int);
10466 vector unsigned int vec_reve (vector unsigned int);
10467 vector bool long long vec_reve (vector bool long long);
10468 vector signed long long vec_reve (vector signed long long);
10469 vector unsigned long long vec_reve (vector unsigned long long);
10470 vector bool short vec_reve (vector bool short);
10471 vector signed short vec_reve (vector signed short);
10472 vector double vec_reve (vector double);
10473 vector float vec_reve (vector float);
10474 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
10475 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
10476 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
10477 (altivec_vreve): New pattern.
10478 * config/rs6000/altivec.h (vec_reve): New define.
10479 * doc/extend.texi (vec_rev): Update the built-in documentation file
10480 for the new built-in functions.
10482 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10484 PR tree-optimization/71815
10485 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
10487 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
10489 (slsr_process_phi): Likewise.
10490 (replace_uncond_cands_and_profitable_phis): Don't replace a
10491 multiply candidate with a stride of 1 (copy or cast).
10492 (phi_incr_cost): Call uses_consumed_by_stmt rather than
10494 (lowest_cost_path): Likewise.
10495 (total_savings): Likewise.
10497 2017-06-26 Richard Biener <rguenther@suse.de>
10500 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
10501 Use def_builtin_pure for all gather builtins.
10503 2017-06-26 Richard Biener <rguenther@suse.de>
10505 PR tree-optimization/81203
10506 * tree-tailcall.c (find_tail_calls): Do not move stmts into
10507 non-dominating BBs.
10509 2017-06-26 Marek Polacek <polacek@redhat.com>
10512 * doc/invoke.texi: Document -Wmultistatement-macros.
10514 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
10516 * doc/sourcebuild.texi (ARM-specific attributes): Document new
10517 arm_neon_ok_no_float_abi effective target.
10519 2017-06-26 Richard Biener <rguenther@suse.de>
10521 PR tree-optimization/80928
10522 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
10523 (copy_bbs): Set BB_DUPLICATED flag early.
10524 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
10526 (execute_on_shrinking_pred): Likewise.
10527 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
10528 BB_DUPLICATED blocks.
10529 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
10530 iterate over all PHIs considering removal of *gsi.
10532 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
10534 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
10537 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10539 * config/rs6000/rs6000-string.c: (expand_block_clear,
10540 do_load_for_compare, select_block_compare_mode,
10541 compute_current_alignment, expand_block_compare,
10542 expand_strncmp_align_check, expand_strn_compare,
10543 expand_block_move, rs6000_output_load_multiple)
10544 Move functions related to string/block move/compare
10545 to a separate file.
10546 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
10547 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
10548 for this function which is now used in two files.
10549 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
10550 * config.gcc: Add rs6000-string.o to extra_objs for
10551 targets powerpc*-*-* and rs6000*-*-*.
10553 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
10556 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
10557 32-bit, since indexed is not valid for DImode.
10558 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
10559 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
10560 (define_peephole2 for Altivec d-form load): Add 32-bit support.
10561 (define_peephole2 for Altivec d-form store): Likewise.
10564 * multiple_target.c (create_dispatcher_calls): Only create the
10565 dispatcher call if the function is the default clone of a
10566 versioned function.
10568 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
10570 PR middle-end/80902
10571 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
10572 a call, force the call to not be a tail call.
10574 2017-06-23 Jeff Law <law@redhat.com>
10576 * doc/contrib.texi: Add entry for Steven Pemberton's work on
10579 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
10581 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
10582 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
10583 handling for early expansion of vector shifts (sl,sr,sra,rl).
10584 (builtin_function_type): Add vector shift right instructions
10585 to the unsigned argument list.
10587 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
10589 rtl-optimizatoin/79286
10590 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
10591 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
10592 trap. PIC register plus a const unspec without offset can never trap.
10594 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
10596 * tree.h (builtin_structptr_type): New type.
10597 (builtin_structptr_types): Declare new array.
10598 * tree.c (builtin_structptr_types): New array.
10599 (free_lang_data, build_common_tree_nodes): Use it.
10601 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
10604 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
10607 2017-06-22 Matt Turner <mattst88@gmail.com>
10609 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
10610 Lake models to skylake case. Assume skylake for unknown
10611 models with clflushopt.
10613 2017-06-22 Jeff Law <law@redhat.com>
10615 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
10616 frame sizes that do not satisfy aarch64_uimm12_shift.
10618 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
10620 * profile-count.h (apply_probability,
10621 apply_scale, probability_in): Fix checks for zero.
10623 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10625 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
10626 * doc/cppdiropts.texi (-I @var{dir}): Document it.
10628 2016-06-22 Richard Biener <rguenther@suse.de>
10630 * tree-vect-loop.c (vect_model_reduction_cost): Handle
10631 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
10632 REDUC_MAX_EXPR support.
10633 (vectorizable_reduction): Likewise.
10634 (vect_create_epilog_for_reduction): Likewise.
10636 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
10638 * match.pd (A / (1 << B) -> A >> B): New.
10639 * generic-match-head.c: Include optabs-tree.h.
10640 * gimple-match-head.c: Likewise.
10641 * optabs-tree.h (target_supports_op_p): New.
10642 * optabs-tree.c (target_supports_op_p): New.
10644 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10646 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
10647 $gcc_cv_ld --help output.
10648 (gcc_cv_ld_demangle): Likewise.
10649 (gcc_cv_ld_eh_frame_hdr): Likewise.
10650 (gcc_cv_ld_pie): Likewise.
10651 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
10652 (gcc_cv_ld_buildid): Likewise.
10653 (gcc_cv_ld_sysroot): Likewise.
10654 (ld_bndplt_support): Likewise.
10655 (ld_pushpopstate_support): Likewise.
10656 * configure: Regenerate.
10657 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
10659 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10662 * config/i386/sse.md (round<mode>2): Renumber match_dup and
10663 operands indexes to avoid gap between operands and match_dups.
10665 2017-06-21 Andrew Pinski <apinski@cavium.com>
10667 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
10668 Increment Arith_shift and Arith_shift_reg by 1.
10669 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
10671 * config/aarch64/aarch64.c (thunderx_tunings): Enable
10672 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
10673 (aarch64_strip_extend): Add new argument and test for it.
10674 (aarch64_cheap_mult_shift_p): New function.
10675 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
10676 add a cost if it is true.
10677 Update calls to aarch64_strip_extend.
10678 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
10680 2017-06-21 Andrew Pinski <apinski@cavium.com>
10682 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
10684 (thunderxt88): Likewise.
10685 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
10686 (thunderx_prefetch_tune): New variable.
10687 (thunderx2t99_prefetch_tune): Update for the correct values.
10688 (thunderxt88_tunings): New variable.
10689 (thunderx_tunings): Use thunderx_prefetch_tune instead of
10690 generic_prefetch_tune.
10691 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
10693 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10695 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
10696 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
10697 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
10698 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
10699 (aarch64_atomic_cas<mode>, GPI): Likewise.
10701 2017-06-21 Martin Liska <mliska@suse.cz>
10703 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
10704 statements on cold and hot labels.
10705 * predict.c (tree_estimate_probability_bb): Remove the
10706 prediction from this place.
10708 2017-06-21 Martin Liska <mliska@suse.cz>
10710 PR tree-optimization/79489
10711 * gimplify.c (maybe_add_early_return_predict_stmt): New
10713 (gimplify_return_expr): Call the function.
10714 * predict.c (tree_estimate_probability_bb): Remove handling
10716 * predict.def: Update comment about early return predictor.
10717 * gimple-predict.h (is_gimple_predict): New function.
10718 * predict.def: Change default value of early return to 66.
10719 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
10721 * passes.def: Put pass_strip_predict_hints to the beginning of
10724 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
10726 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
10727 FUNCTION_DECL declarations.
10728 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
10730 (dwaf2out_decl): Likewise.
10731 * godump.c (go_early_global_decl): Skip call to the real debug hook
10732 for FUNCTION_DECL declarations.
10733 * passes.c (rest_of_decl_compilation): Skip call to the
10734 early_global_decl debug hook for FUNCTION_DECL declarations, unless
10735 -fdump-go-spec is passed.
10737 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10739 * config/i386/i386.c (struct builtin_isa): New field pure_p.
10740 Reorder for compactness.
10741 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
10742 (def_builtin_pure, def_builtin_pure2): New functions.
10743 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
10745 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10747 * match.pd (nop_convert): New predicate.
10748 ((A +- CST1) +- CST2): Allow some NOP conversions.
10750 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10753 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
10754 with ctors/dtors if GOVD_SHARED is set.
10756 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10758 * config/aarch64/aarch64.md (movti_aarch64):
10759 Emit mov rather than orr.
10760 (movtf_aarch64): Likewise.
10761 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
10762 Emit mov rather than orr.
10764 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10766 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
10767 Swap alternatives, make integer dup more expensive.
10769 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10771 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
10772 Return true for non-tls symbols.
10774 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
10776 * config/aarch64/aarch64-cores.def (cortex-a55): New.
10777 (cortex-a75): Likewise.
10778 (cortex-a75.cortex-a55): Likewise.
10779 * config/aarch64/aarch64-tune.md: Regenerate.
10780 * doc/invoke.texi (-mtune): Document new values for -mtune.
10782 2017-06-21 Tom de Vries <tom@codesourcery.com>
10784 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
10785 stack_size feature.
10786 (Effective-Target Keywords, Other attributes): Suggest using
10787 dg-add-options stack_size feature to get stack limit in stack_size
10788 effective target documentation.
10790 2017-06-21 Julian Brown <julian@codesourcery.com>
10791 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10793 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
10794 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
10795 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
10797 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
10798 attribute type list for neon_multiply.
10799 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
10800 attribute type list for neon_multiply.
10801 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
10802 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
10803 attribute type list for neon_multiply.
10804 * config/arm/types.md (crypto_pmull): Add.
10805 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
10806 attribute type list.
10808 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
10810 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
10813 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10815 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
10816 to make sure not to dereference a NULL cost_classes_ptr pointer.
10818 2017-06-20 Carl Love <cel@us.ibm.com>
10820 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10821 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10822 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10823 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10824 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10825 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10826 VMULOSW): New enum "unspec" values.
10827 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10828 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10829 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10830 altivec_vmulosw): New patterns.
10831 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10832 VMULOSW): Add definitions.
10834 2017-06-20 Julia Koval <julia.koval@intel.com>
10836 * config/i386/i386.c: Fix rounding expand for new pattern.
10837 * config/i386/subst.md: Fix pattern (parallel -> unspec).
10839 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10841 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
10842 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
10844 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10846 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
10849 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10851 * config/aarch64/aarch64-cores.def: Rearrange to sort by
10852 architecture, then by implementer ID.
10853 * config/aarch64/aarch64-tune.md: Regenerate.
10855 2017-06-20 Richard Biener <rguenther@suse.de>
10857 PR middle-end/81097
10858 * fold-const.c (split_tree): Fold to type before negating.
10860 2017-06-20 David Malcolm <dmalcolm@redhat.com>
10862 * diagnostic-show-locus.c
10863 (selftest::test_fixit_deletion_affecting_newline): New function.
10864 (selftest::diagnostic_show_locus_c_tests): Call it.
10866 2017-06-20 Andreas Schwab <schwab@suse.de>
10869 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
10872 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10874 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
10875 __ARM_FEATURE_COPROC according to support.
10877 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10879 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10880 Rewritten to avoid overflow for > 32-bit pointers.
10883 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
10884 by removing enum keyword.
10885 (ubsan_type_descriptor): Likewise. Formatting fix.
10888 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
10889 splitter): Require TARGET_SSE2 in the condition.
10891 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
10894 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
10895 for doing vector set of SFmode on ISA 3.0.
10896 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
10897 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
10899 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
10900 SFmode value into a V4SF variable that was extracted from another
10901 V4SF variable without converting the element to double precision
10902 and back to single precision vector format.
10903 (vsx_insert_extract_v4sf_p9_2): Likewise.
10905 2017-06-19 Jakub Jelinek <jakub@redhat.com>
10907 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
10908 in UWHI to avoid undefined overflow.
10911 * ubsan.h (enum ubsan_encode_value_phase): New.
10912 (ubsan_encode_value): Change second argument to
10913 enum ubsan_encode_value_phase with default value of
10914 UBSAN_ENCODE_VALUE_GENERIC.
10915 * ubsan.c (ubsan_encode_value): Change second argument to
10916 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
10917 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
10918 create_tmp_var_raw instead of create_tmp_var and use a
10920 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
10921 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
10922 ubsan_encode_value callers.
10925 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
10926 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
10927 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
10929 2017-06-19 Richard Biener <rguenther@suse.de>
10931 PR middle-end/81118
10932 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
10933 estimates if we changed anything.
10935 2017-06-19 Richard Biener <rguenther@suse.de>
10937 PR tree-optimization/80887
10938 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
10939 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
10940 simplified lookups, then reset mprts_hook.
10941 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
10943 (try_to_simplify): Likewise.
10945 2017-06-19 Martin Liska <mliska@suse.cz>
10948 * gimplify.c (gimplify_switch_expr):
10949 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
10951 2017-06-19 Martin Liska <mliska@suse.cz>
10953 * doc/install.texi: Document that PGO runs in 4 stages.
10955 2017-06-19 Martin Liska <mliska@suse.cz>
10958 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
10959 to dispatcher function name.
10960 * multiple_target.c (replace_function_decl): New function.
10961 (create_dispatcher_calls): Redirect both edges and references.
10963 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
10965 * profile-count.c (profile_count::dump): Dump quality.
10966 (profile_count::differs_from_p): Update for unsigned val.
10967 * profile-count.h (profile_count_quality): New enum.
10968 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
10970 2017-06-19 Richard Biener <rguenther@suse.de>
10972 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
10973 struct function as arg.
10974 (estimate_numbers_of_iterations): Export overload with loop arg.
10975 (free_numbers_of_iterations_estimates_loop): Use an overload of
10976 free_numbers_of_iterations_estimates instead.
10977 * tree-cfg.c (remove_bb): Adjust.
10978 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
10979 * tree-parloops.c (gen_parallel_loop): Likewise.
10980 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
10982 (tree_unroll_loops_completely): Likewise.
10983 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
10984 Use an overload instead and export.
10985 (estimated_loop_iterations): Adjust.
10986 (max_loop_iterations): Likewise.
10987 (likely_max_loop_iterations): Likewise.
10988 (estimate_numbers_of_iterations): Take struct function as arg
10990 (loop_exits_before_overflow): Adjust.
10991 (free_numbers_of_iterations_estimates_loop): Use an overload.
10992 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
10993 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
10995 2017-06-19 Richard Biener <rguenther@suse.de>
10998 * ipa-prop.c (find_constructor_constant_at_offset): Handle
10999 RANGE_EXPR conservatively.
11001 2017-06-16 Carl Love <cel@us.ibm.com>
11003 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11004 definitions for vec_float, vec_float2, vec_floato,
11005 vec_floate built-ins.
11006 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
11007 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
11009 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
11010 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
11011 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
11012 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
11013 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
11014 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
11015 vec_floato): Add builtin defines.
11016 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
11017 Update the built-in documentation file for the new built-in
11020 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11022 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
11023 (mthumb): Mark as the negative of -marm.
11025 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11027 * doc/invoke.texi (ARM Options, -mcpu): Document supported
11029 (ARM Options, -mtune): Document that this accepts the same
11030 extension options as -mcpu.
11031 (ARM Options, -mfpu): Document addition of -mfpu=auto.
11033 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11035 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
11036 permitted extensions.
11038 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11040 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
11041 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
11042 (armv8-m.main): Add option +nodsp.
11043 * config/arm/arm-cpu-cdata.h: Regenerated.
11045 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11047 * config/arm/t-fuchsia: New file.
11048 * config.gcc (arm*-*-fuchsia*): Use it.
11050 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11052 * config/arm/t-symbian: Rewrite for new option infrastructure.
11054 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11056 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
11057 (MULTILIB_REQUIRED): Likewise.
11059 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11061 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
11062 (MULTILIB_RESUE): Likewise.
11063 (MULTILIB_MATCHES): Likewise.
11064 (MULTLIB_REQUIRED): Likewise.
11066 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11068 * config/arm/t-rtems: Rewrite for new option framework.
11070 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11072 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
11073 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
11074 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
11075 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
11076 * config/arm/t-multilib: ... here.
11077 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
11078 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
11079 armv7-a and armv8*-a when A-profile libraries have not been built.
11080 * config/arm/t-rmprofile: Rewrite.
11082 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11084 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
11085 with a backslash. Remove the backslash after substituting unescaped
11087 * doc/fragments.texi (MULTILIB_REUSE): Document it.
11089 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11091 * config.gcc: (arm*-*-*): When building a-profile libraries, force
11092 the driver to pass through the default setting of -mfloat-abi.
11093 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
11095 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
11096 (all_feat_combs): New rule.
11097 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
11099 * config/arm/t-aprofile: Rewrite.
11101 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11103 * config/arm/arm.h (FPUTYPE_AUTO): Define.
11104 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
11105 fpu is not specified by the user/command-line.
11106 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
11107 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
11108 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
11109 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
11110 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
11111 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
11113 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11115 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
11116 * config/arm/t-arm-elf: Rewritten.
11118 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11120 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
11121 have some floating-point instructions.
11122 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
11123 (TARGET_MAYBE_HARD_FLOAT): New macro.
11124 * config/arm/arm-builtins.c (arm_init_builtins): Use
11125 TARGET_MAYBE_HARD_FLOAT.
11126 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
11128 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11130 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
11131 (configargs.h): Include it.
11132 (arm_print_hint_for_fpu_option): New function.
11133 (arm_parse_fpu_option): New function.
11134 (candidate_extension): New class.
11135 (arm_canon_for_multilib): New function.
11136 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
11137 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
11138 (ARCH_CANONICAL_SPECS): New macro.
11139 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
11141 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11143 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
11144 are set after handling multilib fragments. Set target_cpu_default2
11147 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11149 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
11151 (arm*-*-*): Set target_cpu_default2 to a quoted string.
11152 * config/arm/parsecpu.awk (check_cpu): Validate any extension
11154 (check_arch): Likewise.
11155 * config/arm/arm.c (arm_configure_build_target): Handle
11156 TARGET_CPU_DEFAULT being a string constant. Scan any feature
11157 options in the default.
11159 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11161 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
11162 when an option is an alias of another.
11163 * config/arm/parsecpu.awk (optalias): New parser token.
11164 (gen_comm_data): Mark non-alias options as such. Emit entries
11165 for extension aliases.
11166 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
11167 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
11168 (armv6kz, armv6zk, armv6t2): Likewise.
11169 (armv7): Make vfpv3-d16 an alias.
11170 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
11172 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
11173 Sort in canonical order.
11174 (armv8-a): Sort in canonical order.
11175 (armv8.1-a, armv8.2-a): Likewise.
11176 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
11178 (cortex-a9): Sort in canonical order.
11179 * config/arm/arm.c (selftests.h): Include it.
11180 (arm_test_cpu_arch_data): New function.
11181 (arm_run_self_tests): New function.
11182 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
11183 (targetm): Move declaration to the end of the file.
11184 * arm-cpu-cdata.h: Regenerated.
11186 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11188 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
11189 call to target_mode_check describing the type of option passed.
11190 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
11191 (arm_target_thumb_only): Use arm_parse_arch_option_name or
11192 arm_parse_cpu_option_name to match parameters against list of
11194 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
11195 arm_arch_core_flags data structure.
11196 * config/arm/arm-cpu_cdata.h: Regenerated.
11198 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11200 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
11202 (arm_print_hint_for_cpu_option): Likewise.
11203 (arm_print_hint_for_arch_option): Likewise.
11204 (arm_parse_cpu_option_name): Likewise.
11205 (arm_parse_arch_option_name): Likewise.
11206 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
11207 of entries in the all_fpus list.
11208 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
11209 (arm_parse_cpu_option_name): Declare.
11210 (arm_parse_arch_option_name): Declare.
11211 (arm_parse_option_features): Declare.
11212 (arm_intialize_isa): Declare.
11213 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
11215 (gen_comm_data): ... here. Make definitions non-static.
11216 * config/arm/arm-cpu-data.h: Regenerated.
11217 * config/arm/arm-cpu-cdata.h: Regenerated.
11219 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11221 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
11222 (cpu_arch_extension): New structure.
11223 (cpu_arch_option, arch_option, cpu_option): New structures.
11224 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
11225 architecture types.
11226 (gen_data): Generate new format data tables.
11227 * config/arm/arm.c (cpu_tune): New structure.
11228 (cpu_option, processors): Delete.
11229 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
11230 (arm_print_hint_for_cpu_option): ... this and ...
11231 (arm_print_hint_for_arch_option): ... this.
11232 (arm_parse_arch_cpu_name): Delete. Replace with ...
11233 (arm_parse_cpu_option_name): ... this and ...
11234 (arm_parse_arch_option_name): ... this.
11235 (arm_unrecognized_feature): Change type of target parameter to
11237 (arm_parse_arch_cpu_features): Delete. Replace with ...
11238 (arm_parse_option_features): ... this.
11239 (arm_configure_build_target): Rework to use new configuration data
11241 (arm_print_tune_info): Rework for new configuration data tables.
11242 * config/arm/arm-cpu-data.h: Regenerated.
11243 * config/arm/arm-cpu.h: Regenerated.
11245 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11247 * Makefile.in (OBJS): Move sbitmap.o from here ...
11248 (OBJS-libcommon): ... to here.
11250 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11252 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
11253 (ISA_ALL_CRYPTO): New macro.
11254 (ISA_ALL_SIMD): New macro
11255 (ISA_ALL_FP): New macro.
11256 * config/arm/arm.c (fpu_bitlist): Update initializer.
11257 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
11259 (arm9e): Add fpu. Add option for nofp
11260 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
11261 (arm926ej-s, arm1026ej-s): Likewise.
11262 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
11263 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
11264 neon-fp16, neon-vfpv4, nofp and nosimd.
11265 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
11266 (cortex-a8): Add fpu. Add option for nofp.
11267 (cortex-a9): Add fpu. Add options for nosimd and nofp.
11268 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
11269 (cortex-r4f): Add fpu.
11270 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
11271 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
11273 (cortex-r8): Likewise.
11274 (cortex-m4): Add fpu. Add option for nofp.
11275 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
11276 (cortex-a17.cortex-a7): Likewise.
11277 (cortex-a32): Add fpu. Add options for crypto and nofp.
11278 (cortex-a35, cortex-a53): Likewise.
11279 (cortex-a57): Add fpu. Add option for crypto.
11280 (cortex-a72, cortex-a73): Likewise.
11281 (exynos-m1): Likewise.
11282 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
11283 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
11284 (cortex-m33): Add fpu. Add option for nofp.
11285 * config/arm/arm-cpu-cdata.h: Regenerated
11286 * config/arm/arm-cpu-data.h: Regenerated.
11288 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11290 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
11291 (armv5te, armv5tej): Likewise.
11292 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
11293 (armv7): Add options fp and vfpv3-d16.
11294 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
11295 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
11297 (armv7ve): Likewise.
11298 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
11299 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
11300 (armv8-a): Add nocrypto option.
11301 (armv8.1-a, armv8.2-a): Likewise.
11302 (armv8-m.main): add options fp, fp.dp and nofp.
11304 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11306 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
11308 (armv8-a+crc): Delete.
11309 (armv8.1-a): Add options simd, crypto and nofp.
11310 (armv8.2-a): Add options fp16, simd, crypto and nofp.
11311 (armv8.2-a+fp16): Delete.
11312 (armv8-m.main): Add option dsp.
11313 (armv8-m.main+dsp): Delete.
11314 (cortex-a8): Add fpu. Add nofp option.
11315 (cortex-a9): Add fpu. Add nofp and nosimd options.
11316 * config/arm/parsecpu.awk (gen_data): Generate option tables and
11317 link to main cpu and architecture data structures.
11318 (gen_comm_data): Only put isa attributes from the main architecture
11320 (option): New statement for architecture and CPU entries.
11321 * arm.c (struct cpu_option): New structure.
11322 (struct processors): Add entry for options.
11323 (arm_unrecognized_feature): New function.
11324 (arm_parse_arch_cpu_name): Ignore any characters after the first
11326 (arm_parse_arch_cpu_feature): New function.
11327 (arm_configure_build_target): Separate out any CPU and architecture
11328 features and parse separately. Don't error out if -mfpu=auto is
11329 used with only an architecture string.
11330 (arm_print_asm_arch_directives): New function.
11331 (arm_file_start): Call it.
11332 * config/arm/arm-cpu-cdata.h: Regenerated.
11333 * config/arm/arm-cpu-data.h: Likewise.
11334 * config/arm/arm-tables.opt: Likewise.
11336 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11338 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
11339 assembler when it is not -mfpu=auto.
11341 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11343 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
11344 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
11345 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
11346 (ASM_CPU_SPEC): Rewrite.
11347 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
11348 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
11349 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
11350 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
11351 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
11352 copied string is NUL-terminated. Also strip any characters prefixed
11354 (arm_rewrite_selected_arch): New function.
11355 (arm_rewrite_march): New function.
11357 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11359 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
11360 (x_arm_cpu_string, x_arm_tune_string): Likewise.
11361 (march, mcpu, mtune): Convert to string-based options.
11362 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
11363 (arm_parse_arch_cpu_name): New function.
11364 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
11365 identify selected architecture or CPU.
11366 (arm_option_save): New function.
11367 (TARGET_OPTION_SAVE): Redefine.
11368 (arm_option_restore): Restore string options.
11369 (arm_option_print): Print string options.
11371 2017-06-16 Martin Sebor <msebor@redhat.com>
11373 PR tree-optimization/80933
11374 PR tree-optimization/80934
11375 * builtins.c (fold_builtin_3): Do not handle bcmp here.
11376 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
11377 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
11378 (gimple_fold_builtin): Call them.
11380 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11382 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
11383 as unlikely; update profile.
11385 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11387 * predict.c (force_edge_cold): Handle declaring edges impossible
11390 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11392 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
11394 (try_unroll_loop_completely): Fix reporting.
11396 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11398 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
11400 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
11403 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
11404 if given a non-constant argument for an intrinsic which requires a
11407 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11409 * profile.c (compare_freqs): New function.
11410 (branch_prob): Sort edge list.
11411 (find_spanning_tree): Assume that the list is priority sorted.
11413 2017-06-16 Richard Biener <rguenther@suse.de>
11415 PR tree-optimization/81090
11416 * passes.def (pass_record_bounds): Remove.
11417 * tree-pass.h (make_pass_record_bounds): Likewise.
11418 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
11419 make_pass_record_bounds): Likewise.
11420 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
11421 not free niter estimates at the beginning but at the end.
11422 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
11424 2017-06-16 Richard Biener <rguenther@suse.de>
11426 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
11427 initializer to workaround ICE in host GCC 4.8.
11429 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11431 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
11433 (clone_inlined_nodes): Update.
11435 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11437 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
11438 prefetch settings, and enable prefetching by default at -O3.
11440 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11442 * config/aarch64/aarch64.c (aarch64_override_options_internal):
11443 Set flag_prefetch_loop_arrays according to tuning data.
11445 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11447 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
11448 New tune structure.
11449 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
11450 [Unrelated to main purpose of the patch] Place the pointer field last
11451 to enable type checking errors when tune structure are wrongly merged.
11452 * config/aarch64/aarch64.c (generic_prefetch_tune,)
11453 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
11454 (thunderx2t99_prefetch_tune): New tune constants.
11455 (tune_params *_tunings): Update all tunings (no functional change).
11456 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
11457 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
11458 from tunings structures.
11460 2017-06-16 Jakub Jelinek <jakub@redhat.com>
11463 * ubsan.c (instrument_null): Add T argument, use it instead
11464 of computing it based on IS_LHS.
11465 (instrument_object_size): Likewise.
11466 (pass_ubsan::execute): Adjust instrument_null and
11467 instrument_object_size callers to pass gimple_get_lhs or
11468 gimple_assign_rhs1 result to it. Use instrument_null instead of
11469 calling get_base_address and instrument_mem_ref. Handle
11470 aggregate call arguments for object-size sanitization.
11472 2017-06-16 Yury Gribov <tetra2005@gmail.com>
11474 PR tree-optimization/81089
11475 * tree-vrp.c (is_masked_range_test): Validate operands of
11478 2017-06-15 Martin Sebor <msebor@redhat.com>
11481 * dumpfile.c (dump_register): Avoid calling memset to initialize
11482 a class with a default ctor.
11483 * gcc.c (struct compiler): Remove const qualification.
11484 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
11485 * hash-table.h: Ditto.
11486 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
11488 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
11489 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
11491 * params.h (struct param_info): Make struct members non-const.
11492 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
11493 with default initialization.
11494 * vec.h (vec_copy_construct, vec_default_construct): New helper
11496 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
11497 with vec_copy_construct.
11498 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
11499 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
11500 * doc/invoke.texi (-Wclass-memaccess): Document.
11502 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11504 * emit-rtl.h (is_leaf): Update comment about local
11505 register allocator.
11507 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11510 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
11511 for a variable to have a section before checking if the section has a
11513 Set section to.persistent if persistent attribute is set.
11514 Warn if .persistent attribute is used on an automatic variable.
11516 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
11518 PR rtl-optimization/80474
11519 * reorg.c (update_block): Do not ignore instructions in a delay slot.
11521 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
11523 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
11526 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
11528 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
11529 (casesi): Emit bounds checking as RTL.
11530 (casesi_internal_mips16_<mode>): Remove bounds checking.
11532 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
11534 * config/xtensa/xtensa.c (xtensa_option_override): Append
11535 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
11536 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
11537 xtensa_doloop_hooks): Define unconditionally.
11538 (xtensa_reorg_loops): Only call reorg_loops in the presence of
11540 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
11541 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
11542 for it in xtensa_option_override.
11543 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
11544 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
11546 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
11548 * doc/cppopts.texi: Document '-' special value to -MF.
11550 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
11552 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
11553 (cortex_a53_fconst): Likewise.
11554 (cortex_a53_fpmul): Likewise.
11555 (cortex_a53_f_load_64): Likewise.
11556 (cortex_a53_f_load_many): Likewise.
11557 (cortex_a53_advsimd_alu): Likewise.
11558 (cortex_a53_advsimd_alu_q): Likewise.
11559 (cortex_a53_advsimd_mul): Likewise.
11560 (cortex_a53_advsimd_mul_q): Likewise.
11561 (fpmac bypass): Add new bypass for fpmac-fpmac case.
11562 Add missing fmul, r2f_cvt and fconst cases.
11564 2017-06-14 Richard Biener <rguenther@suse.de>
11566 PR middle-end/81088
11567 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
11569 (fold_binary_loc): When associating do not treat pre-existing
11570 TREE_OVERFLOW on literal constants as a reason to allow
11571 TREE_OVERFLOW on associated literal constants.
11573 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11575 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
11576 (MASK_FEATURES): New macro.
11577 * config/sparc/sparc.c (sparc_option_override): Remove the special
11578 handling of -mfpu and generalize it to all MASK_FEATURES switches.
11580 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11582 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
11583 a division of 0 if non-call exceptions are enabled.
11585 2017-06-14 Andrew Pinski <apinski@cavium.com>
11586 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11589 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
11590 Improve vector initialization code gen for only variable case.
11592 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11594 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
11596 2017-06-14 Richard Biener <rguenther@suse.de>
11598 PR tree-optimization/81083
11599 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
11602 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11604 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
11605 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
11606 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
11607 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
11608 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
11609 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
11611 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11613 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
11614 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
11616 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11618 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
11620 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11622 * config/rs6000/t-rtems: Don't handle SPE.
11624 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11626 * config/rs6000/t-linux: Don't handle SPE.
11628 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11630 * config/rs6000/eabispe.h: Delete file.
11632 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11634 * config/rs6000/t-spe: Delete file.
11636 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11638 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
11639 (rs6000_legitimate_offset_address_p): Return false for anything in
11640 V2SImode or V2SFmode.
11642 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11644 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
11645 except V2SF and V2SI. Rearrange the vector modes, and add comments.
11646 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
11648 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
11649 (rs6000_legitimate_offset_address_p): Ditto.
11650 (rs6000_emit_move): Ditto.
11651 (rs6000_init_builtins): Remove V4HI_type_node.
11653 2017-06-13 Martin Liska <mliska@suse.cz>
11656 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
11657 (gate_asan): Likewise.
11658 * asan.h (asan_no_sanitize_address_p): Remove the function.
11659 (sanitize_flags_p): New function.
11660 * builtins.def: Fix coding style.
11661 * common.opt: Use renamed enum value.
11662 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
11663 * doc/extend.texi: Document no_sanitize attribute.
11664 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
11665 to SANITIZE_UNDEFINED_NONDEFAULT.
11666 * gcc.c (sanitize_spec_function): Use the renamed enum value.
11667 * gimple-fold.c (optimize_atomic_compare_exchange_p):
11668 Use sanitize_flags_p.
11669 * gimplify.c (gimplify_function_tree): Likewise.
11670 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
11671 * opts.c (parse_no_sanitize_attribute): New function.
11672 (common_handle_option): Use renamed enum value.
11673 * opts.h (parse_no_sanitize_attribute): Declare.
11674 * tree.c (sanitize_flags_p): New function.
11675 * tree.h: Declared here.
11676 * tsan.c: Use sanitize_flags_p.
11677 * ubsan.c (ubsan_expand_null_ifn): Likewise.
11678 (instrument_mem_ref): Likewise.
11679 (instrument_bool_enum_load): Likewise.
11680 (do_ubsan_in_current_function): Remove the function.
11681 (pass_ubsan::execute): Use sanitize_flags_p.
11682 * ubsan.h: Remove do_ubsan_in_current_function
11683 * tree-cfg.c (print_no_sanitize_attr_value): New function.
11684 (dump_function_to_file): Use it here.
11686 2017-06-13 Martin Jambor <mjambor@suse.cz>
11688 PR tree-optimization/80803
11689 PR tree-optimization/81063
11690 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
11691 (propagate_subaccesses_across_link): Enqueue subtree whenever
11692 necessary instead of relying on the caller.
11694 2017-06-13 Martin Jambor <mjambor@suse.cz>
11696 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
11697 that have a first_link.
11698 (sort_and_splice_var_accesses): Do not check first_link before
11700 (subtree_mark_written_and_enqueue): Likewise.
11701 (propagate_all_subaccesses): Likewise and do not stop at first
11702 parent with a first_link.
11704 2017-06-13 Martin Jambor <mjambor@suse.cz>
11706 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
11709 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11711 * match.pd: New pattern.
11713 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11715 * tree-vrp.c (is_masked_range_test): New function.
11716 (register_edge_assert_for): Determine ranges for
11719 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11721 PR tree-optimization/67328
11722 * fold-const.c (maskable_range_p): New function.
11723 (build_range_check): Generate bittests if possible.
11725 2017-06-13 Martin Liska <mliska@suse.cz>
11727 * gimple-pretty-print.c (dump_probability): Add new argument.
11728 (dump_edge_probability): Dump both probability and count.
11729 (dump_gimple_label): Likewise.
11730 (dump_gimple_bb_header): Likewise.
11732 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
11735 * config/avr/avr-devices.c: Fix indentation.
11736 * config/avr/gen-avr-mmcu-specs.c: Dito.
11738 2017-06-13 Richard Biener <rguenther@suse.de>
11740 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
11741 instead get vector type from stmt_info.
11742 (vectorizable_reduction): Adjust. Remove dead code.
11744 2017-06-13 Richard Biener <rguenther@suse.de>
11746 PR middle-end/81065
11747 * fold-const.c (extract_muldiv_1): Remove bogus distribution
11748 case of C * (x * C2 + C3).
11749 (fold_addr_of_array_ref_difference): Properly fold index difference.
11751 2017-06-12 David S. Miller <davem@davemloft.net>
11754 * config/sparc/sparc.md (return expander): Emit frame blockage if
11755 function uses alloca.
11757 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
11759 * combine.c (make_field_assignment): Check len rather than the mode
11760 precision when calling force_to_mode.
11762 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
11764 Support multilibs and devices that see flash in RAM address range.
11767 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
11768 (avr_mcu_t) <flash_pm_offset>: New field.
11769 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
11770 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
11771 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
11772 (AVR_TINY_PM_OFFSET): Remove macro.
11773 * config/avr/avr.opt (-mshort-calls): New option.
11774 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11775 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
11776 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
11777 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
11778 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
11779 instead of avr_arch->have_jmp_call.
11780 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
11781 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
11782 avr_arch->flash_pm_offset to define.
11783 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
11784 new field flash_pm_offset. Add entry for avrxmega3.
11785 (avr_texinfo): Add entry for avrxmega3.
11786 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
11787 attiny212, attiny214,
11788 attiny412, attiny414, attiny416, attiny417,
11789 attiny814, attiny816, attiny817,
11790 attiny1614, attiny1616, attiny1617,
11791 attiny3214, attiny3216, attiny3217.
11792 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
11793 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
11794 (avr_print_operand_address) [AVR_TINY]: Same.
11795 (avr_asm_init_sections) <readonly_data_section>: Only patch
11796 callback if avr_arch->flash_pm_offset = 0.
11797 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
11798 for rodata if avr_arch->flash_pm_offset != 0.
11799 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
11800 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
11801 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
11802 (m_options): Append opt_rcall.
11803 (m_dirnames): Append dir_rcall.
11804 * config/avr/t-multilib: Regenerate.
11806 * configure.ac [target=avr]: Check whether avrxmega3 default
11807 linker description file works as needed.
11808 * configure: Regenerate.
11809 * doc/avr-mmcu.texi: Regenerate.
11810 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
11811 <__AVR_ARCH__>: Document avrxmega3 and 103.
11812 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
11813 <__AVR_SHORT_CALLS__>: Document it.
11814 <__AVR_PM_BASE_ADDRESS__>: Document it.
11815 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
11816 (AVR Variable Attributes) <progmem>: Document this is
11817 not needed for avrxmega3.
11818 (AVR Named Address Spaces) <__flash>: Dito.
11820 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
11822 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
11824 2017-06-12 Doug Rupp <rupp@adacore.com>
11826 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
11827 Append vxworks-stdint.h to the tm_file list.
11828 * config/vxworks-stdint.h: New file.
11830 2017-06-12 Martin Liska <mliska@suse.cz>
11832 PR tree-optimization/81041
11833 * tree-profile.c (gimple_gen_ic_func_profiler):
11834 Create an extra BB in profile-generate
11835 (gimple_gen_time_profiler): Likewise.
11837 2017-06-12 Jakub Jelinek <jakub@redhat.com>
11839 PR tree-optimization/81003
11840 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
11841 (update_range_test): Use it instead of force_gimple_operand_gsi.
11843 2017-06-12 Richard Biener <rguenther@suse.de>
11845 PR tree-optimization/81053
11846 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
11847 with backedge value not defined in loop. Simplify def stmt
11850 2017-06-11 Tom de Vries <tom@codesourcery.com>
11853 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
11855 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
11856 nvptx_cannot_force_const_mem.
11858 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11860 * opts.c (finish_options): Move test for flag_split_stack after
11861 it has been initialized.
11863 2017-06-11 Jason Merrill <jason@redhat.com>
11865 * tree.h (id_equal): New.
11866 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
11867 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
11868 instead of strcmp of IDENTIFIER_POINTER.
11870 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11872 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
11873 (mark_all_inlined_calls_cdtor): Fix formating.
11874 (inline_transform): Rescale profile before inlining.
11876 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11878 * cgraph.h (cgraph_edge::clone): Update prototype.
11879 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
11880 (cgraph_node::create_clone): Update.
11881 (cgraph_node::create_version_clone): Update.
11882 * tree-inline.c (copy_bb): Update.
11883 (expand_call_inline): Update.
11885 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11887 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
11888 factored out from ...
11889 (rs6000_emit_prologue): ... here.
11891 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11893 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
11894 factored out from ...
11895 (rs6000_emit_prologue): ... here.
11897 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11899 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
11901 (handle_missing_profiles): Fix computation of tp_first_run.
11902 (counts_to_freqs): Do not touch freqs when count is 0.
11904 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11906 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
11909 2017-06-10 Tom de Vries <tom@codesourcery.com>
11911 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
11912 attributes): Document signal effective target.
11914 2017-06-10 Tom de Vries <tom@codesourcery.com>
11916 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
11917 Document effective target stack_size.
11919 2017-06-09 David Malcolm <dmalcolm@redhat.com>
11921 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
11922 to the edit_context if they can be auto-applied.
11924 2017-06-9 Ian Lance Taylor <iant@golang.org>
11926 * opts.c (finish_options): If -fsplit-stack, disable implicit
11927 -forder-blocks-and-partition.
11928 * doc/invoke.texi (Optimize Options): Document that when using
11929 -fsplit-stack -forder-blocks-and-partition is not implicitly
11932 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11934 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
11935 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
11936 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
11937 * builtins.def (abort, trap, unreachable): Declare cold.
11938 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
11939 * tree-core.h (ECF_COLD): New.
11940 * tree.c (set_call_expr_flags): Handle ECF_COLD.
11941 (build_common_builtin_nodes): Mark unreachable and abort as cold.
11943 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11945 * predict.c (unlikely_executed_stmt_p): Cleanup.
11947 2017-06-09 Richard Biener <rguenther@suse.de>
11949 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
11950 model if the ref is always written to.
11952 2017-06-09 Tamar Christina <tamar.christina@arm.com>
11954 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
11956 2017-06-09 Tamar Christina <tamar.christina@arm.com>
11958 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
11961 2017-06-09 Tom de Vries <tom@codesourcery.com>
11964 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
11965 "target cannot support label values" when encountering LABEL_REF.
11967 2017-06-09 Martin Liska <mliska@suse.cz>
11969 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
11970 (gimple_gen_ic_func_profiler): Emit direct comparison
11971 of __gcov_indirect_call_callee with NULL.
11972 (gimple_gen_time_profiler): Change probability from
11973 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
11975 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11977 * profile.c (edge_gcov_counts): Turn to pointer.
11978 (compute_branch_probabilities, compute_branch_probabilities): Update.
11979 (branch_prob): Do not clear edge_gcov_count.
11980 * profile.h (edge_gcov_counts): Turn to pointer.
11981 (edge_gcov_count): Update.
11983 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11985 * gimple.h (gimple_check_failed): Mark cold.
11987 2017-06-09 Richard Biener <rguenther@suse.de>
11989 PR tree-optimization/66623
11990 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
11991 refactor check_reduction into two parts, properly computing
11992 whether we have to check reduction validity for outer loop
11995 2017-06-09 Richard Biener <rguenther@suse.de>
11997 PR tree-optimization/79483
11998 * graphite-scop-detection.c (order): New global.
11999 (get_order): Compute bb to order mapping that satisfies code
12000 generation constraints.
12001 (cmp_pbbs): New helper.
12002 (build_scops): Start domwalk at entry block, sort generated
12005 2017-06-09 Richard Biener <rguenther@suse.de>
12007 PR middle-end/81007
12008 * ipa-polymorphic-call.c
12009 (ipa_polymorphic_call_context::restrict_to_inner_class):
12010 Skip FIELD_DECLs with error_mark_node type.
12011 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
12014 2017-06-09 Martin Liska <mliska@suse.cz>
12016 * predict.c (struct branch_predictor): New struct.
12017 (test_prediction_value_range): New test.
12018 (predict_c_tests): New function.
12019 * selftest-run-tests.c (selftest::run_tests): Run the function.
12020 * selftest.h: Declare new tests.
12022 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
12025 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
12026 gen_add3_insn did not fail.
12027 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
12028 r0, construct that number in a temporary reg and add that reg to r0.
12029 If asked to put the result in r0 as well, fail.
12031 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
12033 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12034 for early expansion of vec_eqv.
12036 2017-06-08 Jakub Jelinek <jakub@redhat.com>
12038 PR middle-end/81005
12039 * ubsan.c (instrument_null): Avoid pointless code temporary.
12040 (pass_ubsan::execute): Instrument aggregate arguments of calls.
12042 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
12046 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
12049 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
12052 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12054 * predict.c (unlikely_executed_edge_p): Move ahead.
12055 (probably_never_executed_edge_p): Use it.
12057 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12059 PR middle-end/79988
12060 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
12061 gimple_call_builtin_p call.
12063 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12065 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
12066 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
12067 rtl_check_failed_type2, rtl_check_failed_code1,
12068 rtl_check_failed_code2, rtl_check_failed_code_mode,
12069 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
12070 rtvec_check_failed_bounds, rtl_check_failed_flag,
12071 _fatal_insn_not_found, _fatal_insn): Likewise.
12072 * tree.h (tree_contains_struct_check_failed,
12073 tree_check_failed, tree_not_check_failed,
12074 tree_class_check_failed, tree_range_check_failed,
12075 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
12076 tree_vec_elt_check_failed, phi_node_elt_check_failed,
12077 tree_operand_check_failed, omp_clause_check_failed,
12078 omp_clause_operand_check_failed, omp_clause_range_check_failed):
12081 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12083 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
12084 flag_branch_probabilities.
12085 * ipa-inline.c (edge_badness): Likewise.
12086 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12087 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
12088 * predict.c (maybe_hot_frequency_p): Likewise.
12089 (probably_never_executed): Likewise.
12090 * sched-ebb.c (schedule_ebbs): Likewise.
12091 * sched-rgn.c (find_single_block_region): Likewise.
12092 * tracer.c (tail_duplicate): Likewise.
12094 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12096 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
12097 longer requires x_flag_profile_use.
12099 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12101 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
12102 instead of flag_reorder_blocks_and_partition.
12103 * dbxout.c (dbxout_function_end): Likewise.
12104 * dwarf2out.c (gen_subprogram_die): Likewise.
12105 * haifa-sched.c (sched_create_recovery_edges): Likewise.
12106 * hw-doloop.c (reorg_loops): Likewise.
12107 * varasm.c (assemble_start_function,
12108 assemble_end_function): Likewise.
12109 (decide_function_section): Do not check for
12110 flag_reorder_blocks_and_partition.
12112 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12114 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12116 (chkp_get_hard_register_fake_addr_expr): Ditto.
12117 (chkp_build_addr_expr): Add check for hard reg case.
12118 (chkp_parse_array_and_component_ref): Ditto.
12119 (chkp_find_bounds_1): Ditto.
12120 (chkp_process_stmt): Don't generate bounds store for
12123 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12125 * predict.c (maybe_hot_bb_p): Do not check profile status.
12126 (maybe_hot_edge_p): Likewise.
12127 (probably_never_executed): Check for zero counts even if profile
12129 (unlikely_executed_edge_p): New function.
12130 (unlikely_executed_stmt_p): New function.
12131 (unlikely_executed_bb_p): New function.
12132 (set_even_probabilities): Use unlikely predicates.
12133 (combine_predictions_for_bb): Likewise.
12134 (predict_paths_for_bb): Likewise.
12135 (predict_paths_leading_to_edge): Likewise.
12136 (determine_unlikely_bbs): New function.
12137 (estimate_bb_frequencies): Use it.
12138 (compute_function_frequency): Use zero counts even if profile is
12140 * profile-count.h: Fix typo.
12142 2017-08-08 Julia Koval <julia.koval@intel.com>
12144 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
12145 _mm512_mask_cvtsepi16_storeu_epi8,
12146 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
12147 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
12148 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
12149 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
12150 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
12151 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
12152 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
12153 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
12154 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
12155 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
12156 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
12157 __builtin_ia32_pmovuswb256mem_mask,
12158 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
12159 __builtin_ia32_pmovwb512mem_mask): New builtins.
12161 2017-08-08 Julia Koval <julia.koval@intel.com>
12163 PR target/73350,80862
12164 * config/i386/subst.md (round): Fix round pattern.
12165 * config/i386/i386.c (ix86_erase_embedded_rounding):
12166 Fix erasing rounding for the fixed pattern.
12168 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12170 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
12172 2017-06-08 Martin Liska <mliska@suse.cz>
12174 PR gcov-profile/80911
12175 * gcov.c (block_info::block_info): New constructor.
12177 2017-06-07 Carl Love <cel@us.ibm.com>
12179 * config/rs6000/rs6000-c: The return type of the following
12180 built-in functions was implemented as int not long long. Fix sign
12181 of return value for the unsigned version of vec_mulo and vec_mule.
12182 vector unsigned long long vec_bperm (vector unsigned long long,
12183 vector unsigned char)
12184 vector signed long long vec_mule (vector signed int,
12186 vector unsigned long long vec_mule (vector unsigned int,
12187 vector unsigned int)
12188 vector signed long long vec_mulo (vector signed int,
12190 vector unsigned long long vec_mulo (vector unsigned int,
12191 vector unsigned int)
12192 * doc/extend.texi: Fix the documentation for the built-in
12195 2017-06-07 Carl Love <cel@us.ibm.com>
12198 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
12201 2017-06-07 Carl Love <cel@us.ibm.com>
12203 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
12204 support, Generate doublehv for signed int/float for BE case only.
12206 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
12208 * doc/invoke.texi (mcx16): Rewrite.
12210 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12212 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
12213 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
12214 *mov<mode>_softfloat, and an anonymous splitter): Use
12215 nonimmediate_operand instead of rs6000_nonimmediate_operand.
12217 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12219 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
12221 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12222 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
12223 (rs6000_debug_reg_global): Adjust.
12224 (rs6000_init_hard_regno_mode_ok): Adjust.
12225 (rs6000_dbx_register_number): Adjust.
12226 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
12227 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12228 Remove SPE_ACC and SPEFSCR.
12229 (REG_ALLOC_ORDER): Ditto.
12230 (FRAME_POINTER_REGNUM): Change to 111.
12231 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
12232 (REG_CLASS_NAMES): Ditto.
12233 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
12234 (REGISTER_NAMES): Ditto.
12235 (ADDITIONAL_REG_NAMES): Ditto.
12236 (rs6000_reg_names): Ditto.
12237 * config/rs6000/rs6000.md: Renumber some register number
12240 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12242 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
12244 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12245 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
12247 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
12248 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12249 Delete the SPE high registers.
12250 (REG_ALLOC_ORDER): Ditto.
12251 (enum reg_class): Remove SPE_HIGH_REGS.
12252 (REG_CLASS_NAMES): Ditto.
12253 (REG_CLASS_CONTENTS): Delete the SPE high registers.
12254 (REGISTER_NAMES): Ditto.
12255 (rs6000_reg_names): Ditto.
12256 * doc/tm.texi.in: Remove SPE as example.
12257 * doc/tm.texi: Regenerate.
12259 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12261 * config/rs6000/8540.md (ppc8540_brinc): Delete.
12262 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
12263 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
12264 * config/rs6000/rs6000.md (type): Remove "brinc".
12266 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12268 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
12269 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
12270 * config/rs6000/linuxspe.h: Delete file.
12271 * config/rs6000/rs6000.md: Don't include spe.md.
12272 * config/rs6000/spe.h: Delete file.
12273 * config/rs6000/spe.md: Delete file.
12274 * config/rs6000/t-rs6000: Remove spe.md.
12276 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12278 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
12279 (reg_or_none500mem_operand): Delete.
12280 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
12281 instead of reg_or_none500mem_operand.
12283 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12285 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
12286 handling of SPE flags.
12287 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
12289 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12291 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
12293 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
12294 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
12295 paired_divv2sf3): Similar.
12296 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
12297 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
12298 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
12300 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
12301 Rename the paired_* instruction patterns.
12302 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
12304 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
12305 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
12306 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
12307 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
12308 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
12309 PAIRED_VECTOR_MODE.
12310 (struct machine_function): Delete field spe_insn_chain_scanned_p.
12311 (spe_func_has_64bit_regs_p): Delete.
12312 (spe_expand_predicate_builtin): Delete.
12313 (spe_expand_evsel_builtin): Delete.
12314 (TARGET_DWARF_REGISTER_SPAN): Do not define.
12315 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
12316 (invalid_e500_subreg): Delete.
12317 (rs6000_legitimize_address): Always force_reg op2 as well, for
12318 paired single memory accesses.
12319 (rs6000_member_type_forces_blk): Delete.
12320 (rs6000_spe_function_arg): Delete.
12321 (rs6000_expand_unop_builtin): Delete SPE handling.
12322 (rs6000_expand_binop_builtin): Ditto.
12323 (spe_expand_stv_builtin): Delete.
12324 (bdesc_2arg_spe): Delete.
12325 (spe_expand_builtin): Delete.
12326 (spe_expand_predicate_builtin): Delete.
12327 (spe_expand_evsel_builtin): Delete.
12328 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
12329 (spe_init_builtins): Delete.
12330 (spe_func_has_64bit_regs_p): Delete.
12331 (savres_routine_name): Delete "info" parameter. Adjust callers.
12332 (rs6000_emit_stack_reset): Ditto.
12333 (rs6000_dwarf_register_span): Delete.
12334 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
12335 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
12336 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
12338 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
12340 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
12341 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
12342 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
12343 mulv2sf3, divv2sf3): Delete expanders.
12345 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12347 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
12349 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12351 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
12352 * config/rs6000/rs6000.c: Ditto.
12354 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12356 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
12357 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
12358 comparison_operator.
12360 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12362 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
12363 * config/rs6000/rs6000.opt: Ditto.
12364 * config/rs6000/t-rtems: Ditto.
12366 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12368 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
12369 TARGET_E500_SINGLE by 0, simplify.
12370 * config/rs6000/rs6000.c: Ditto.
12371 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
12372 (spe_build_register_parallel): Delete.
12373 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
12374 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
12375 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
12376 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
12377 (E500_CONVERT): Delete.
12378 * config/rs6000/spe.md: Remove many patterns and all define_constants.
12380 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12382 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
12383 * config/rs6000/dfp.md: Ditto.
12384 (negdd2, *negdd2_fpr): Merge.
12385 (absdd2, *absdd2_fpr): Merge.
12386 (negtd2, *negtd2_fpr): Merge.
12387 (abstd2, *abstd2_fpr): Merge.
12388 * config/rs6000/e500.h: Delete file.
12389 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
12390 TARGET_FPRS by 1 and simplify.
12391 * config/rs6000/rs6000-c.c: Ditto.
12392 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
12393 TARGET_DF_SPE by 0.
12394 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
12396 * config/rs6000/rs6000.md: Ditto.
12397 (floatdidf2, *floatdidf2_fpr): Merge.
12398 (move_from_CR_gt_bit): Delete.
12399 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
12400 (E500_CR_IOR_COMPARE): Delete.
12401 (All patterns that require !TARGET_FPRS): Delete.
12402 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
12404 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12406 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
12408 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12410 * graphds.c (add_edge): Intitialize edge's attached data.
12411 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
12412 pointer parameter. Call pointed function on each edge during
12413 graph traversing. Skip traversing the edge when the function
12415 (graphds_dfs, graphds_scc): Ditto.
12416 (for_each_edge): New parameter. Pass the new parameter to callback
12418 * graphds.h (skip_edge_callback): New function pointer type.
12419 (graphds_dfs, graphds_scc): New function pointer parameter.
12420 (graphds_edge_callback, for_each_edge): New parameter.
12422 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12424 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
12425 out code checking if runtime alias check is possible to below ...
12426 Call the new function.
12427 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
12428 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
12430 2017-06-07 Marek Polacek <polacek@redhat.com>
12433 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
12434 TYPE_OVERFLOW_WRAPS check.
12436 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12438 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
12439 if versioning is required.
12440 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
12441 peeling with the check for versioning.
12443 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12445 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
12446 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
12447 Set true to new parameter if new ssa variable is defined.
12448 (vect_gen_vector_loop_niters): Refactor. Set range information
12449 for the new vector loop bound variable.
12450 (vect_do_peeling): Ditto.
12452 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12454 * tree-affine.c (ssa.h): Include header file.
12455 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
12456 has wrapping overflow behavior.
12458 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12460 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
12462 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12464 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
12465 (tree_to_aff_combination): ... here.
12467 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12469 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
12470 reg_pressure model function.
12471 (ivopts_global_cost_for_size): Delete.
12472 (determine_set_costs, iv_ca_recount_cost): Call new model function
12473 ivopts_estimate_reg_pressure.
12475 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12477 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
12478 expensive than udiv. Remove floating point cases from mod.
12480 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12482 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
12483 Increase idiv cost.
12485 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12487 * config/aarch64/aarch64.md
12488 (copysignsf3): Fix mask generation.
12490 2017-06-07 Jakub Jelinek <jakub@redhat.com>
12492 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
12494 (class dump_manager): Add register_dumps method.
12495 * dumpfile.c: Include langhooks.h.
12496 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
12497 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
12498 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
12499 (dump_manager::dump_register): Start with 512 entries instead of 32.
12500 (dump_manager::register_dumps): New method.
12501 * toplev.c (general_init): Instead of invoking register_dumps
12502 langhook, invoke register_dumps method on the dump manager.
12503 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
12506 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
12508 * doc/md.texi: Clarify the restrictions on a define_insn condition.
12509 Say that # requires an associated define_split to exist, and that
12510 the define_split must be suitable for use after register allocation.
12512 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12514 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
12515 (compute_outgoing_frequencies): Also initialize zero counts.
12516 (find_many_sub_basic_blocks): Do not produce uninitialized profile
12517 around loops; preserve more of profile when nothing changes.
12519 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
12521 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
12523 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
12524 * config/arm/arm-cpu-cdata.h: Regenerate.
12525 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
12526 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
12527 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
12528 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
12529 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
12531 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
12533 * config/arm/t-rmprofile: Likewise.
12534 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
12536 2017-06-06 David S. Miller <davem@davemloft.net>
12539 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
12540 blockage if function uses alloca.
12542 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12544 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
12545 New "uid" fields to hold pretty-print IDs of group and ref.
12546 Memory references are now identified as <group_id>:<ref_id>
12547 instead of using [random] addresses.
12548 (dump_mem_details): Simplify, no functional change.
12549 (dump_mem_ref): Simplify and make output more concise.
12550 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
12551 (find_or_create_group): Initialize group uid.
12552 (record_ref): Initialize ref uid. Improve debug output.
12553 (prune_group_by_reuse, should_issue_prefetch_p,)
12554 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
12555 (mark_nontemporal_store, determine_loop_nest_reuse):
12556 Improve debug output.
12558 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12560 * dbgcnt.def (prefetch): New debug counter.
12561 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
12562 (schedule_prefetches): Stop issueing prefetches if debug counter
12565 2017-06-06 Tom de Vries <tom@codesourcery.com>
12567 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
12568 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
12570 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12572 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
12573 Use aarch64_reg_or_zero predicate for operand 4.
12574 (aarch64_compare_and_swap<mode> define_insn_and_split):
12575 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
12576 (aarch64_store_exclusive<mode>): Likewise for operand 2.
12578 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12580 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
12581 (arm_compute_save_core_reg_mask): This.
12582 (thumb1_compute_save_reg_mask): Rename into ...
12583 (thumb1_compute_save_core_reg_mask): This.
12584 (arm_compute_save_reg0_reg12_mask): Adapt comment.
12585 (arm_compute_frame_layout): Likewise.
12587 2017-06-06 Richard Biener <rguenther@suse.de>
12589 PR tree-optimization/80974
12590 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
12591 keep or clear leaders SSA info.
12593 2017-06-06 Tom de Vries <tom@codesourcery.com>
12595 * config/nvptx/nvptx.c (split_mode_p): New function.
12596 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
12598 2017-06-06 Tom de Vries <tom@codesourcery.com>
12600 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
12602 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12605 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
12608 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12610 * shrink-wrap.c (handle_simple_exit): Update profile.
12611 (try_shrink_wrapping): Upate profile.
12613 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12615 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
12616 (tree_guess_outgoing_edge_probabilities): New.
12617 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
12618 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
12620 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12622 * ipa-split.c (split_function): Initialize return bb profile.
12624 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12626 * profile.c (compute_branch_probabilities): Also initialize
12627 EXIT_BLOCK profile.
12629 2017-06-06 Richard Biener <rguenther@suse.de>
12631 PR tree-optimization/80928
12632 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
12633 (vect_analyze_loop_operations): Properly guard analysis for
12635 (vect_transform_loop): Likewise.
12636 (vect_analyze_loop_2): Also reset SLP type on PHIs.
12637 (vect_model_induction_cost): Do not cost for pure SLP.
12638 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
12639 of induction in inner loop vectorization.
12640 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
12641 (vect_get_and_check_slp_defs): Handle vect_induction_def.
12642 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
12644 (vect_analyze_slp_cost_1): Cost induction.
12645 (vect_detect_hybrid_slp_stmts): Handle PHIs.
12646 (vect_get_slp_vect_defs): Likewise.
12647 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
12648 (vect_transform_stmt): Handle SLP reductions.
12649 * tree-vectorizer.h (vectorizable_induction): Adjust.
12651 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12653 * config/rs6000/rs6000.c (make_resolver_func): Update
12654 init_lowered_empty_function call.
12656 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
12658 * doc/invoke.texi: Document the -fprofile-abs-path option.
12659 * common.opt (fprofile-abs-path): New option.
12660 * gcov-io.h (gcov_write_filename): Declare.
12661 * gcov-io.c (gcov_write_filename): New function.
12662 * coverage.c (coverage_begin_function): Use gcov_write_filename.
12663 * profile.c (output_location): Likewise.
12665 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12667 * shring-wrap.c: Revert accidental commit.
12669 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
12671 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
12673 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12675 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
12677 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
12678 profile in callgraph edge.
12679 * profile-count.h (apply_probability): If THIS is 0, then result is 0
12680 (apply_scale): Likewise.
12681 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
12682 Also scale profile when inlining function with zero profile.
12683 (initialize_cfun): Update exit block profile even when it is zero.
12684 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
12685 when profile is read.
12687 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12689 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
12690 (CLONE_*): New constants to define the processors we can generate
12691 code for with the target_clone attribute.
12692 (rs6000_clone_map): New array to identify which clone processors
12693 the current program is running on.
12694 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
12695 target_clone attribute.
12696 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
12697 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
12698 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
12699 (cpu_expand_builtin): Add support for target_clone attribute.
12700 (rs6000_valid_attribute_p): Allow "default" attribute.
12701 (get_decl_name): New debug function to simplify printing the
12702 current function name in debugging statements.
12703 (rs6000_clone_priority): New functions to support the target_clone
12704 attribute, and be able to generate code to switch between ISA 2.05
12705 through ISA 3.0 (power6 through power9).
12706 (rs6000_compare_version_priority): Likewise.
12707 (rs6000_get_function_versions_dispatcher): Likewise.
12708 (make_resolver_func): Likewise.
12709 (add_condition_to_bb): Likewise.
12710 (dispatch_function_versions): Likewise.
12711 (rs6000_generate_version_dispatcher_body): Likewise.
12712 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
12713 (fusion_gpr_load_p): Fix a spacing issue.
12714 * doc/extend.texi (Common Function Attributes): Document that the
12715 PowerPC supports the target_clone attribute.
12717 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
12719 * config/arm/arm.h: explain F symbol found in description of ARM
12720 register allocation in its legend.
12722 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12724 * config/mips/frame-header-opt.c: Include profile-count.h.
12725 * config/riscv/riscv.c: Include profile-count.h
12727 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12729 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
12731 (sm_set_flag_if_changed): Add bbs field.
12732 (execute_sm_if_changed_flag_set): Pass BBS.
12733 (execute_sm): Update.
12735 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12737 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
12740 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12742 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
12743 (peephole2): New peephole2 to emit the above.
12744 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
12746 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12748 * config/aarch64/aarch64.c (define_peephole2 above
12749 *sub_<shift>_<mode>): New peephole.
12751 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12753 * config/i386/i386.c (make_resolver_func): Update.
12754 * Makefile.in: Add profile-count.h and profile-count.o
12755 * auto-profile.c (afdo_indirect_call): Update to new API.
12756 (afdo_set_bb_count): Update.
12757 (afdo_propagate_edge): Update.
12758 (afdo_propagate_circuit): Update.
12759 (afdo_calculate_branch_prob): Update.
12760 (afdo_annotate_cfg): Update.
12761 * basic-block.h: Include profile-count.h
12762 (struct edge_def): Turn count to profile_count.
12763 (struct basic_block_def): Likewie.
12764 (REG_BR_PROB_BASE): Move to profile-count.h
12765 (RDIV): Move to profile-count.h
12766 * bb-reorder.c (max_entry_count): Turn to profile_count.
12767 (find_traces): Update.
12768 (rotate_loop):Update.
12769 (connect_traces):Update.
12770 (sanitize_hot_paths):Update.
12771 * bt-load.c (migrate_btr_defs): Update.
12772 * cfg.c (RDIV): Remove.
12773 (init_flow): Use alloc_block.
12774 (alloc_block): Uninitialize count.
12775 (unchecked_make_edge): Uninitialize count.
12776 (check_bb_profile): Update.
12777 (dump_edge_info): Update.
12778 (dump_bb_info): Update.
12779 (update_bb_profile_for_threading): Update.
12780 (scale_bbs_frequencies_int): Update.
12781 (scale_bbs_frequencies_gcov_type): Update.
12782 (scale_bbs_frequencies_profile_count): New.
12783 * cfg.h (update_bb_profile_for_threading): Update.
12784 (scale_bbs_frequencies_profile_count): Declare.
12785 * cfgbuild.c (compute_outgoing_frequencies): Update.
12786 (find_many_sub_basic_blocks): Update.
12787 * cfgcleanup.c (try_forward_edges): Update.
12788 (try_crossjump_to_edge): Update.
12789 * cfgexpand.c (expand_gimple_tailcall): Update.
12790 (construct_exit_block): Update.
12791 * cfghooks.c (verify_flow_info): Update.
12792 (dump_bb_for_graph): Update.
12793 (split_edge): Update.
12794 (make_forwarder_block): Update.
12795 (duplicate_block): Update.
12796 (account_profile_record): Update.
12797 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
12798 (get_estimated_loop_iterations): Update.
12799 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
12800 (single_likely_exit): Update.
12801 * cfgloopmanip.c (scale_loop_profile): Update.
12803 (set_zero_probability): Update.
12804 (lv_adjust_loop_entry_edge): Update.
12805 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12806 (purge_dead_edges): Update.
12807 (rtl_account_profile_record): Update.
12808 * cgraph.c (cgraph_node::create): Uninitialize count.
12809 (symbol_table::create_edge): Uninitialize count.
12810 (cgraph_update_edges_for_call_stmt_node): Update.
12811 (cgraph_edge::dump_edge_flags): Update.
12812 (cgraph_node::dump): Update.
12813 (cgraph_edge::maybe_hot_p): Update.
12814 * cgraph.h: Include profile-count.h
12815 (create_clone), create_edge, create_indirect_edge): Update.
12816 (cgraph_node): Turn count to profile_count.
12817 (cgraph_edge0: Likewise.
12818 (make_speculative, clone): Update.
12819 (create_edge): Update.
12820 (init_lowered_empty_function): Update.
12821 * cgraphclones.c (cgraph_edge::clone): Update.
12822 (duplicate_thunk_for_node): Update.
12823 (cgraph_node::create_clone): Update.
12824 * cgraphunit.c (cgraph_node::analyze): Update.
12825 (cgraph_node::expand_thunk): Update.
12826 * final.c (dump_basic_block_info): Update.
12827 * gimple-streamer-in.c (input_bb): Update.
12828 * gimple-streamer-out.c (output_bb): Update.
12829 * graphite.c (print_global_statistics): Update.
12830 (print_graphite_scop_statistics): Update.
12831 * hsa-brig.c: Include basic-block.h.
12832 * hsa-dump.c: Include basic-block.h.
12833 * hsa-gen.c (T sum_slice): Update.
12834 (convert_switch_statements):Update.
12835 * hsa-regalloc.c: Include basic-block.h.
12836 * ipa-chkp.c (chkp_produce_thunks): Update.
12837 * ipa-cp.c (struct caller_statistics): Update.
12838 (init_caller_stats): Update.
12839 (gather_caller_stats): Update.
12840 (ipcp_cloning_candidate_p): Update.
12841 (good_cloning_opportunity_p): Update.
12842 (get_info_about_necessary_edges): Update.
12843 (dump_profile_updates): Update.
12844 (update_profiling_info): Update.
12845 (update_specialized_profile): Update.
12846 (perhaps_add_new_callers): Update.
12847 (decide_about_value): Update.
12848 (ipa_cp_c_finalize): Update.
12849 * ipa-devirt.c (struct odr_type_warn_count): Update.
12850 (struct decl_warn_count): Update.
12851 (struct final_warning_record): Update.
12852 (possible_polymorphic_call_targets): Update.
12853 (ipa_devirt): Update.
12854 * ipa-fnsummary.c (redirect_to_unreachable): Update.
12855 * ipa-icf.c (sem_function::merge): Update.
12856 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
12857 * ipa-inline.c (compute_uninlined_call_time): Update.
12858 (compute_inlined_call_time): Update.
12859 (want_inline_small_function_p): Update.
12860 (want_inline_self_recursive_call_p): Update.
12861 (edge_badness): Update.
12862 (lookup_recursive_calls): Update.
12863 (recursive_inlining): Update.
12864 (inline_small_functions): Update.
12865 (dump_overall_stats): Update.
12866 (dump_inline_stats): Update.
12867 * ipa-profile.c (ipa_profile_generate_summary): Update.
12868 (ipa_propagate_frequency): Update.
12869 (ipa_profile): Update.
12870 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12871 * ipa-utils.c (ipa_merge_profiles): Update.
12872 * loop-doloop.c (doloop_modify): Update.
12873 * loop-unroll.c (report_unroll): Update.
12874 (unroll_loop_runtime_iterations): Update.
12875 * lto-cgraph.c (lto_output_edge): Update.
12876 (lto_output_node): Update.
12877 (input_node): Update.
12878 (input_edge): Update.
12879 (merge_profile_summaries): Update.
12880 * lto-streamer-in.c (input_cfg): Update.
12881 * lto-streamer-out.c (output_cfg): Update.
12882 * mcf.c (create_fixup_graph): Update.
12883 (adjust_cfg_counts): Update.
12884 (sum_edge_counts): Update.
12885 * modulo-sched.c (sms_schedule): Update.
12886 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
12887 * predict.c (maybe_hot_count_p): Update.
12888 (probably_never_executed): Update.
12889 (dump_prediction): Update.
12890 (combine_predictions_for_bb): Update.
12891 (propagate_freq): Update.
12892 (handle_missing_profiles): Update.
12893 (counts_to_freqs): Update.
12894 (rebuild_frequencies): Update.
12895 (force_edge_cold): Update.
12896 * predict.h: Include profile-count.h
12897 (maybe_hot_count_p, counts_to_freqs): UPdate.
12898 * print-rtl-function.c: Do not include cfg.h
12899 * print-rtl.c: Include basic-block.h
12900 * profile-count.c: New file.
12901 * profile-count.h: New file.
12902 * profile.c (is_edge_inconsistent): Update.
12903 (correct_negative_edge_counts): Update.
12904 (is_inconsistent): Update.
12905 (set_bb_counts): Update.
12906 (read_profile_edge_counts): Update.
12907 (compute_frequency_overlap): Update.
12908 (compute_branch_probabilities): Update; Initialize and deinitialize
12910 (branch_prob): Update.
12911 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
12912 (edge_gcov_count): New.
12913 (bb_gcov_count): New.
12914 * shrink-wrap.c (try_shrink_wrapping): Update.
12915 * tracer.c (better_p): Update.
12916 * trans-mem.c (expand_transaction): Update.
12917 (ipa_tm_insert_irr_call): Update.
12918 (ipa_tm_insert_gettmclone_call): Update.
12919 * tree-call-cdce.c: Update.
12920 * tree-cfg.c (gimple_duplicate_sese_region): Update.
12921 (gimple_duplicate_sese_tail): Update.
12922 (gimple_account_profile_record): Update.
12923 (execute_fixup_cfg): Update.
12924 * tree-inline.c (copy_bb): Update.
12925 (copy_edges_for_bb): Update.
12926 (initialize_cfun): Update.
12927 (freqs_to_counts): Update.
12928 (copy_cfg_body): Update.
12929 (expand_call_inline): Update.
12930 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
12931 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
12932 (try_unroll_loop_completely): Update.
12933 (try_peel_loop): Update.
12934 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12935 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
12936 * tree-ssa-loop-split.c (connect_loops): Update.
12937 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
12938 * tree-ssa-reassoc.c (branch_fixup): Update.
12939 * tree-ssa-tail-merge.c (replace_block_by): Update.
12940 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
12941 (compute_path_counts): Update.
12942 (update_profile): Update.
12943 (recompute_probabilities): Update.
12944 (update_joiner_offpath_counts): Update.
12945 (estimated_freqs_path): Update.
12946 (freqs_to_counts_path): Update.
12947 (clear_counts_path): Update.
12948 (ssa_fix_duplicate_block_edges): Update.
12949 (duplicate_thread_path): Update.
12950 * tree-switch-conversion.c (case_bit_test_cmp): Update.
12951 (struct switch_conv_info): Update.
12952 * tree-tailcall.c (decrease_profile): Update.
12953 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
12954 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12955 * value-prof.c (check_counter): Update.
12956 (gimple_divmod_fixed_value): Update.
12957 (gimple_mod_pow2): Update.
12958 (gimple_mod_subtract): Update.
12959 (gimple_ic_transform): Update.
12960 (gimple_stringop_fixed_value): Update.
12961 * value-prof.h (gimple_ic): Update.
12963 2017-06-02 Carl Love <cel@us.ibm.com>
12965 * config/rs6000/rs6000-c: Add support for built-in functions
12966 vector double vec_doublee (vector signed int);
12967 vector double vec_doublee (vector unsigned int);
12968 vector double vec_doublee (vector float);
12969 vector double vec_doubleh (vector signed int);
12970 vector double vec_doubleh (vector unsigned int);
12971 vector double vec_doubleh (vector float);
12972 vector double vec_doublel (vector signed int);
12973 vector double vec_doublel (vector unsigned int);
12974 vector double vec_doublel (vector float);
12975 vector double vec_doubleo (vector signed int);
12976 vector double vec_doubleo (vector unsigned int);
12977 vector double vec_doubleo (vector float);.
12978 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
12979 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
12981 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
12982 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
12983 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
12985 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
12986 vec_doublel, vec_doubleh.
12987 * doc/extend.texi: Update the built-in documentation file for the
12988 new built-in functions.
12990 2017-06-02 David Malcolm <dmalcolm@redhat.com>
12993 * ipa-inline-analysis.c (free_growth_caches): Set
12994 edge_removal_hook_holder to NULL after removing it.
12996 2017-06-02 Sudakshina Das <sudi.das@arm.com>
12998 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
12999 comparision with zero.
13001 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13002 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13003 for early expansion of vec_min and vec_max builtins.
13004 (builtin_function_type): Add min/max unsigned variants to those
13005 identified as having unsigned arguments.
13007 2017-06-02 Olivier Hainque <hainque@adacore.com>
13009 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
13011 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13013 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
13014 Use VALL_F16 iterator rather than VALL.
13016 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13018 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
13019 Emit CBNZ inside loop when doing a strong exchange and comparing
13020 against zero. Generate the CC flags after the loop.
13022 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
13024 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
13025 (dl_section_ref): New.
13026 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
13027 On AIX, append an expression to subtract the size of the
13028 section length to dl_section_ref.
13030 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13032 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13033 for early expansion of vector absolute builtins.
13035 2017-06-02 Richard Biener <rguenther@suse.de>
13037 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
13038 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
13040 2017-06-02 Richard Biener <rguenther@suse.de>
13042 PR tree-optimization/80948
13043 * tree-tailcall.c (find_tail_calls): Track stmts to move in
13044 stmt order as well.
13046 2017-06-02 Richard Biener <rguenther@suse.de>
13048 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
13050 * tree-vect-stmts.c (process_use): Do not mark backedge defs
13051 for inductions as relevant.
13053 2017-06-02 Richard Biener <rguenther@suse.de>
13055 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
13056 (vectorizable_induction): ... this. Remove dead code.
13058 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
13060 * builtins. (expand_builtin_alloca): Remove second parameter and
13061 infer its value from the first parameter instead.
13062 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
13064 2017-06-02 Jakub Jelinek <jakub@redhat.com>
13066 PR rtl-optimization/80903
13067 * loop-doloop.c (add_test): Unshare sequence.
13069 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
13071 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
13073 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
13075 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
13077 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
13078 xlogue_layout::get_instance, logue_layout::xlogue_layout,
13079 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
13080 (xlogue_layout::get_stub_rtx): Make static.
13081 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
13082 (xlogue_layout::compute_stub_managed_regs): Rename to...
13083 (xlogue_layout::count_stub_managed_regs): ...this.
13084 (xlogue_layout::is_stub_managed_reg): New function.
13085 (xlogue_layout::m_stub_names): Rename to...
13086 (xlogue_layout::s_stub_names): ...this, make static.
13087 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
13088 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
13089 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
13090 xlogue_layout::s_stub_names): Instantiate statics.
13091 (stub_managed_regs): Remove.
13092 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
13093 (disable_call_ms2sysv_xlogues): Rename to...
13094 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
13095 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
13097 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
13098 change after reload_completed.
13099 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
13101 (ix86_expand_prologue): Likewise.
13102 (ix86_expand_epilogue): Likewise.
13103 (ix86_expand_split_stack_prologue): Likewise.
13104 (ix86_compute_frame_layout): Remove frame parameter ...
13105 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
13106 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
13108 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
13109 (ix86_frame): Move from here ...
13110 * config/i386/i386.h (ix86_frame): ... to here.
13111 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
13112 complete ix86_frame data structure instead. Remove some_ld_name.
13114 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13116 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
13117 symbols that hold a DECL_VALUE_EXPR.
13119 2017-06-01 Martin Jambor <mjambor@suse.cz>
13121 PR tree-optimization/80898
13122 * tree-sra.c (process_subtree_disqualification): Removed.
13123 (disqualify_candidate): Do not acll
13124 process_subtree_disqualification.
13125 (subtree_mark_written_and_enqueue): New function.
13126 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
13127 RHS has been disqualified and re-queue LHS if necessary. Apart
13128 from that, ignore disqualified RHS.
13130 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13132 * config/s390/s390.c (s390_emit_epilogue): Disable early return
13133 address fetch for z10 or later.
13135 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13137 * config/arc/arc.md (tst_movb): Add guard when splitting.
13139 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13141 * config/arc/arc.c (arc_can_eliminate): Test against
13142 arc_frame_pointer_needed.
13144 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13146 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
13147 to prevent store reordering.
13148 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
13149 (type): Add block type.
13150 (stack_tie): Define special instruction to be used in
13153 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13155 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
13156 constraint. It is not valid for the pattern.
13157 (noncommutative_binary_comparison): Likewise.
13159 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13161 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
13164 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13166 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
13167 be used by the reg-alloc.
13169 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13171 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
13172 reg-alloc when having mul64 or mul32x16 instructions.
13173 (mulsidi3): Likewise.
13174 (umulsidi3): Likewise.
13175 (mulsi32x16): New pattern.
13176 (mulsi64): Likewise.
13177 (mulsidi64): Likewise.
13178 (umulsidi64): Likewise.
13179 (MUL32x16_REG): Define.
13180 (mul64_600): Use MUL32x16_REG.
13181 (mac64_600): Likewise.
13182 (umul64_600): Likewise.
13183 (umac64_600): Likewise.
13185 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13187 * config/arc/arc.md (mulsi3_700): Make it commutative.
13189 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
13191 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
13193 (*sign_extendsidi2_insn): Likewise for movstosw.
13195 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13197 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
13198 the type of the input discriminant value. Convert the
13199 discriminant value of signedness vary.
13201 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
13203 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
13204 Add to -Wall section.
13206 2017-06-01 Richard Biener <rguenther@suse.de>
13208 PR middle-end/66313
13209 * fold-const.c (fold_plusminus_mult_expr): If the factored
13210 factor may be zero use a wrapping type for the inner operation.
13211 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
13212 and handle moved defs.
13213 (process_assignment): Properly guard the unary op case. Return a
13214 tri-state indicating that moving the stmt before the call may allow
13215 to continue. Pass through to_move.
13216 (find_tail_calls): Handle moving unrelated defs before
13219 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
13222 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
13223 splitter result in the canonical way.
13225 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
13227 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
13228 also for 32bit target. Update insn attributes.
13229 (zero-extendsidi2 splitter): Allow all registers for operand 1.
13231 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
13233 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
13234 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
13235 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
13236 (_mm_maskz_min_ss): New intrinsics.
13238 2017-05-31 Martin Liska <mliska@suse.cz>
13240 * tree-vect-loop.c (vect_create_epilog_for_reduction):
13241 Change comment style to one we normally use.
13242 (vectorizable_reduction): Likewise.
13243 (vectorizable_induction): Likewise.
13244 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13245 (vectorizable_call): Likewise.
13246 (vectorizable_simd_clone_call): Likewise.
13247 (vectorizable_conversion): Likewise.
13248 (vectorizable_assignment): Likewise.
13249 (vectorizable_shift): Likewise.
13250 (vectorizable_operation): Likewise.
13251 (vectorizable_store): Likewise.
13252 (vectorizable_load): Likewise.
13253 * tree-vectorizer.h: Likewise.
13255 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
13257 * passes.c (emergency_dump_function): New.
13258 * tree-pass.h (emergency_dump_function): Declare.
13259 * plugin.c (plugins_internal_error_function): Remove.
13260 * plugin.h (plugins_internal_error_function): Remove declaration.
13261 * toplev.c (internal_error_function): New static function. Use it...
13262 (general_init): ...here.
13264 2017-05-31 Graham Markall <graham.markall@embecosm.com>
13266 * config/arc/arc.c (arc_print_operand): Handle constant operands.
13267 (arc_rtx_costs): Add costs for new patterns.
13268 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
13269 * config/arc/predicates.md: Add _1_2_3_operand predicate.
13271 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13273 * tree-ssa-strlen.c (get_next_strinfo): New function.
13274 (get_stridx_plus_constant): Use it.
13275 (zero_length_string): Likewise.
13276 (adjust_related_strinfos): Likewise.
13277 (adjust_last_stmt): Likewise.
13279 2017-05-31 Richard Biener <rguenther@suse.de>
13282 * config/i386/i386.c (ix86_expand_builtin): Remove assert
13283 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
13285 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13287 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
13288 loop_vinfo argument and use of dependence distance vectors.
13289 Check instead whether the two references differ only in their
13290 initial value and assume that they have the same alignment if the
13291 difference is a multiple of the vector alignment.
13292 (vect_analyze_data_refs_alignment): Update call accordingly.
13294 2017-05-31 Martin Liska <mliska@suse.cz>
13297 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
13299 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13301 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
13302 (create_intersect_range_checks): Move from ...
13303 * tree-data-ref.c (create_intersect_range_checks_index)
13304 (create_intersect_range_checks): ... to here.
13305 (create_runtime_alias_checks): New function factored from ...
13306 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
13307 here. Call above function.
13308 * tree-data-ref.h (create_runtime_alias_checks): New function.
13310 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13312 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
13313 segment length for dr_b and compute it in wide_int.
13315 2017-05-31 Richard Biener <rguenther@suse.de>
13317 PR tree-optimization/80906
13318 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
13319 and pass through iv_map.
13320 (copy_bb_and_scalar_dependences): Adjust.
13321 (translate_pending_phi_nodes): Likewise.
13322 (copy_loop_close_phi_args): Handle code-generating IVs instead
13325 2017-05-30 David Malcolm <dmalcolm@redhat.com>
13327 * diagnostic-color.c (color_dict): Add "type-diff".
13328 (parse_gcc_colors): Update comment.
13329 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
13330 -fdiagnostics-show-template-tree and -fno-elide-type.
13331 (GCC_COLORS): Add type-diff to example.
13333 (-fdiagnostics-show-template-tree): New.
13334 (-fno-elide-type): New.
13335 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
13336 the pp_format_decoder callback. Call any m_format_postprocessor's
13338 (pretty_printer::pretty_printer): Initialize
13339 m_format_postprocessor.
13340 (pretty_printer::~pretty_printer): Delete any
13341 m_format_postprocessor.
13342 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
13343 (class format_postprocessor): New class.
13344 (struct pretty_printer::format_decoder): Document the new parameters.
13345 (struct pretty_printer::m_format_postprocessor): New field.
13346 * tree-diagnostic.c (default_tree_printer): Update for new
13347 bool and const char ** params.
13348 * tree-diagnostic.h (default_tree_printer): Likewise.
13350 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
13352 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
13353 (lwa_operand): Delete rs6000_gen_cell_microcode test.
13354 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13355 rs6000_gen_cell_microcode code.
13356 (rs6000_final_prescan_insn): Delete.
13357 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
13358 "warn-cell-microcode" entries.
13359 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
13360 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
13361 throughout. Change cc_reg_not_micro_cr0_operand to
13362 cc_reg_not_cr0_operand throughout.
13363 (*extendhi<mode>2_noload): Delete.
13364 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
13365 (mwarn-cell-microcode): Delete.
13366 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
13367 -mgen-cell-microcode and -mwarn-cell-microcode.
13369 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
13372 * config/i386/constraints.md (Yd): New constraint.
13374 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
13375 and (?Yd, r) alternatives. Update insn attributes.
13376 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
13377 and (?*Yd, r) alternatives. Update insn attributes.
13378 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
13380 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
13382 * gimplify.c (gimplify_modify_expr): Don't create a
13383 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
13386 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
13388 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
13390 2017-05-30 Richard Biener <rguenther@suse.de>
13392 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
13393 and reduc_def fields.
13394 (STMT_VINFO_REDUC_TYPE): New define.
13395 (STMT_VINFO_REDUC_DEF): Likewise.
13396 (vect_force_simple_reduction): Adjust prototype.
13397 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
13398 (vect_is_simple_reduction): Remove check_reduction argument.
13399 (vect_force_simple_reduction): Adjust and set
13400 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13401 (vectorizable_reduction): Do not re-do reduction analysis
13402 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13403 * tree-parloops.c (gather_scalar_reductions): Adjust.
13405 2017-05-30 Richard Biener <rguenther@suse.de>
13407 PR middle-end/80901
13408 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
13411 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
13413 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
13414 Introduce unknown_misalignment parameter and remove vf.
13415 (vect_peeling_hash_get_lowest_cost):
13416 Pass unknown_misalignment parameter.
13417 (vect_enhance_data_refs_alignment):
13418 Fix unsupportable data ref treatment.
13420 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13422 * tree-vect-data-refs.c (vect_get_data_access_cost):
13423 Workaround for SLP handling.
13424 (vect_enhance_data_refs_alignment):
13425 Compute costs for doing no peeling at all, compare to the best
13426 peeling costs so far and avoid peeling if cheaper.
13428 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13430 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
13431 Return peeling info and set costs to zero for unlimited cost
13433 (vect_enhance_data_refs_alignment): Also inspect all datarefs
13434 with unknown misalignment. Compute and costs for unknown
13435 misalignment, compare them to the costs for known misalignment
13436 and choose the cheapest for peeling.
13438 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13440 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
13441 (vect_get_peeling_costs_all_drs): Create function.
13442 (vect_peeling_hash_get_lowest_cost):
13443 Use vect_get_peeling_costs_all_drs.
13444 (vect_peeling_supportable): Create function.
13445 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
13447 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13449 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
13450 DR_HAS_NEGATIVE_STEP.
13451 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
13452 (vect_enhance_data_refs_alignment): Use.
13453 (vect_duplicate_ssa_name_ptr_info): Use.
13454 * tree-vectorizer.h (dr_misalignment): Use.
13455 (known_alignment_for_access_p): Use.
13457 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13460 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
13462 (has_section_name): New function.
13464 2017-05-30 Martin Liska <mliska@suse.cz>
13467 * auto-profile.c (get_function_decl_from_block): Fix
13470 2017-05-30 Richard Biener <rguenther@suse.de>
13472 PR middle-end/80876
13473 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
13475 2017-05-30 Martin Liska <mliska@suse.cz>
13477 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
13478 * dumpfile.h (struct dump_file_info): Remove ctors.
13480 2017-05-30 Martin Liska <mliska@suse.cz>
13482 * predict.def: Fix GNU coding style.
13484 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13486 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13487 Mark 'to' argument with ATTRIBUTE_UNUSED.
13489 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13491 * config/xtensa/xtensa.c (xtensa_emit_call): Use
13492 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
13493 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
13496 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
13498 * doc/install.texi (Options specification): Restore entry of
13499 --enable-sjlj-exceptions.
13501 2017-05-27 Michael Eager <eager@eagercon.com>
13504 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
13506 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
13508 * config/microblaze/microblaze.h
13509 (FIXED_REGISTERS): Update in macro.
13510 (CALL_USED_REGISTERS): Update in macro.
13512 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
13514 * doc/install.texi: Add links to macOS binary distributions.
13516 2017-05-27 Jakub Jelinek <jakub@redhat.com>
13520 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13522 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13524 2017-05-26 Martin Liska <mliska@suse.cz>
13526 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
13528 2017-05-26 Martin Liska <mliska@suse.cz>
13530 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
13531 always leading ';; '.
13532 (dump_bb_info): Likewise.
13533 (brief_dump_cfg): Likewise.
13534 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
13535 * dumpfile.c: Remove usage of TDF_VERBOSE.
13536 * dumpfile.h (enum dump_kind): Likewise.
13537 (dump_gimple_bb_header): Do not use TDF_COMMENT.
13538 * print-tree.c (debug_verbose): Remove.
13539 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
13540 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
13541 * tree-diagnostic.c (default_tree_printer): Replace
13542 TDF_DIAGNOSTIC with TDF_SLIM.
13544 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13546 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
13547 in parameter loop, rather than loop_vinfo.
13548 (create_intersect_range_checks): Ditto.
13549 (vect_create_cond_for_alias_checks): Update call to above functions.
13551 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13553 PR tree-optimization/80815
13554 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
13555 for merging runtime alias checks. Handle negative DR_STEPs.
13557 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13559 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
13561 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
13562 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
13563 out code pruning runtime alias checks.
13564 * tree-data-ref.c (prune_runtime_alias_test_list): New function
13565 factored out from above.
13566 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13568 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13570 (prune_runtime_alias_test_list): New decalaration.
13572 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13574 * tree-vect-data-refs.c (compare_tree): Rename and move ...
13575 * tree-data-ref.c (data_ref_compare_tree): ... to here.
13576 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
13577 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
13578 (operator==, comp_dr_with_seg_len_pair): Ditto.
13579 (vect_prune_runtime_alias_test_list): Ditto.
13581 2017-05-26 Martin Liska <mliska@suse.cz>
13584 * params.def: Bound partial-inlining-entry-probability param.
13586 2017-05-26 Marek Polacek <polacek@redhat.com>
13589 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
13592 2017-05-26 Richard Biener <rguenther@suse.de>
13594 PR tree-optimization/80842
13595 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
13598 2017-05-26 Richard Biener <rguenther@suse.de>
13600 PR tree-optimization/80844
13601 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
13603 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
13605 * doc/md.texi (Machine Constraints): Update x86 family
13606 machine constraints section to match 'config/i386/constraints.md'.
13608 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
13610 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
13612 2017-05-25 Nathan Sidwell <nathan@acm.org>
13614 * doc/invoke.texi (--enable-languages): Update documentation.
13616 2017-05-25 Martin Liska <mliska@suse.cz>
13618 * dumpfile.c: Add TDF_FOLDING.
13619 * dumpfile.h (enum dump_kind): Likewise.
13620 * genmatch.c (dt_simplify::gen_1): Use it.
13622 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13624 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
13626 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13628 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13629 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
13631 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13633 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
13634 * match.pd (X == C): Rewrite it here.
13635 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
13636 with_certain_nonzero_bits2): New predicates.
13637 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
13639 2017-05-24 Nathan Sidwell <nathan@acm.org>
13641 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
13644 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
13647 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13649 * config/powerpcspe: New port. Files are copied from the rs6000
13650 port, with "rs6000" in filenames replaced by "powerpcspe".
13652 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
13654 PR rtl-optimization/80754
13655 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
13657 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
13659 * config/sparc/sparc.md (length): Return the correct value for -mflat
13660 sibcalls to match output_sibcall.
13662 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13666 * config/rs6000/rs6000.c (struct machine_function): Add new field
13668 (rs6000_get_separate_components): Init that field, use it.
13669 (rs6000_components_for_bb): Use the field.
13671 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
13673 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
13675 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
13677 PR middle-end/80823
13678 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
13680 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13683 * config/s390/s390.c (s390_check_qrst_address): Check incoming
13684 address against address_operand predicate.
13685 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
13687 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
13689 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
13690 parameters passed indirectly.
13692 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
13694 * config/i386/i386.md (*movdi_internal): Remove SSE4
13695 alternative 18 (?r, *v). Update insn attributes.
13696 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
13697 Update insn attributes.
13698 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
13699 Update insn attributes.
13700 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
13701 alternative 1 (r, v). Remove isa attribute.
13702 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
13703 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
13704 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
13706 2017-05-23 Tom de Vries <tom@codesourcery.com>
13708 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
13711 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13713 * cgraphunit.c (symbol_table::process_new_functions): Update.
13714 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
13715 (inline_generate_summary): Rename to ...
13716 (ipa_fn_summary_generate): ... this one.
13717 (inline_read_summary): Rename to ...
13718 (ipa_fn_summary_read): ... this one.
13719 (inline_write_summary): Rename to ...
13720 (ipa_fn_summary_write): ... this one.
13721 (inline_free_summary): Rename to ...
13722 (ipa_free_fn_summary): ... this one.
13723 (pass_data_local_fn_summary, pass_local_fn_summary,
13724 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
13725 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
13726 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
13727 make_pass_ipa_fn_summary): New.
13728 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
13729 inline_write_summary, inline_free_summary): Remove.
13730 (ipa_free_fn_summary) : New.
13731 * ipa-inline.c (ipa_inline): Update.
13732 (pass_ipa_inline): Do not generate summaries.
13733 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
13735 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
13736 and add pass_ipa_fn_summary.
13737 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
13739 (make_pass_inline_parameters): Remove.
13741 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
13743 * omp-low.c (struct omp_context): Remove "default_kind" member.
13746 * omp-offload.c (execute_oacc_device_lower): Remove the
13747 parallelism dimensions function attributes for unparallelized
13748 OpenACC kernels constructs.
13750 2017-05-23 Martin Liska <mliska@suse.cz>
13752 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
13754 (cgraph_edge::make_speculative): Likewise.
13755 (cgraph_edge::resolve_speculation): Likewise.
13756 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
13757 (cgraph_node::dump): Likewise.
13758 * cgraph.h: Likewise.
13759 * cgraphunit.c (analyze_functions): Likewise.
13760 (symbol_table::compile): Likewise.
13761 * ipa-cp.c (print_all_lattices): Likewise.
13762 (determine_versionability): Likewise.
13763 (initialize_node_lattices): Likewise.
13764 (ipcp_verify_propagated_values): Likewise.
13765 (estimate_local_effects): Likewise.
13766 (update_profiling_info): Likewise.
13767 (create_specialized_node): Likewise.
13768 (perhaps_add_new_callers): Likewise.
13769 (decide_about_value): Likewise.
13770 (decide_whether_version_node): Likewise.
13771 (identify_dead_nodes): Likewise.
13772 (ipcp_store_bits_results): Likewise.
13773 * ipa-devirt.c (dump_targets): Likewise.
13774 (ipa_devirt): Likewise.
13775 * ipa-icf.c (sem_item::dump): Likewise.
13776 (sem_function::equals): Likewise.
13777 (sem_variable::equals): Likewise.
13778 (sem_item_optimizer::read_section): Likewise.
13779 (sem_item_optimizer::execute): Likewise.
13780 (congruence_class::dump): Likewise.
13781 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
13782 (dump_inline_summary): Likewise.
13783 (estimate_node_size_and_time): Likewise.
13784 (inline_analyze_function): Likewise.
13785 * ipa-inline-transform.c (inline_call): Likewise.
13786 * ipa-inline.c (report_inline_failed_reason): Likewise.
13787 (want_early_inline_function_p): Likewise.
13788 (edge_badness): Likewise.
13789 (update_edge_key): Likewise.
13790 (inline_small_functions): Likewise.
13791 * ipa-profile.c (ipa_profile): Likewise.
13792 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
13793 (ipa_make_edge_direct_to_target): Likewise.
13794 (remove_described_reference): Likewise.
13795 (ipa_impossible_devirt_target): Likewise.
13796 (propagate_controlled_uses): Likewise.
13797 (ipa_print_node_params): Likewise.
13798 (ipcp_transform_function): Likewise.
13799 * ipa-pure-const.c (pure_const_read_summary): Likewise.
13800 (propagate_pure_const): Likewise.
13801 * ipa-reference.c (generate_summary): Likewise.
13802 (read_write_all_from_decl): Likewise.
13803 (propagate): Likewise.
13804 (ipa_reference_read_optimization_summary): Likewise.
13805 * ipa-utils.c (ipa_merge_profiles): Likewise.
13806 * ipa.c (walk_polymorphic_call_targets): Likewise.
13807 (symbol_table::remove_unreachable_nodes): Likewise.
13808 (ipa_single_use): Likewise.
13809 * passes.c (execute_todo): Likewise.
13810 * predict.c (drop_profile): Likewise.
13811 * symtab.c (symtab_node::get_dump_name): New function.
13812 (symtab_node::dump_name): Likewise.
13813 (symtab_node::dump_asm_name): Likewise.
13814 (symtab_node::dump_references): Likewise.
13815 (symtab_node::dump_referring): Likewise.
13816 (symtab_node::dump_base): Likewise.
13817 (symtab_node::debug_symtab): Likewise.
13818 * tree-sra.c (convert_callers_for_node): Likewise.
13819 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
13820 * value-prof.c (init_node_map): Likewise.
13822 2017-05-23 Martin Liska <mliska@suse.cz>
13824 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
13825 and symtab_node::debug_symtab to symbol_table::debug.
13826 * cgraphunit.c (analyze_functions): Use the renamed function.
13827 (symbol_table::compile): Likewise.
13828 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
13829 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
13830 * passes.c (execute_todo): Likewise.
13831 * symtab.c (symbol_table::dump): New function.
13832 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
13834 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13836 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
13837 that nonconst implies exec.
13839 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13841 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
13842 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
13843 (inline_edge_summary_vec): Turn into ...
13844 (ipa_call_summaries): ... this one.
13845 (redirect_to_unreachable, edge_set_predicate,
13846 evaluate_properties_for_edge, inline_summary_alloc,
13847 reset_ipa_call_summary, reset_inline_summary,
13848 inline_summary_t::duplicate): Update.
13849 (inline_edge_duplication_hook): Turn to ...
13850 (ipa_call_summary_t::duplicate): ... this one.
13851 (inline_edge_removal_hook): Turn to ...
13852 (ipa_call_summary_t::remove): ... this one.
13853 (dump_inline_edge_summary): Turn to ...
13854 (dump_ipa_call_summary): ... this one.
13855 (estimate_function_body_sizes): Update.
13856 (inline_update_callee_summaries): Update.
13857 (remap_edge_change_prob): Update.
13858 (remap_edge_summaries): Update.
13859 (inline_merge_summary): Update.
13860 (do_estimate_edge_time): Update.
13861 (inline_generate_summary): Update.
13862 (inline_read_section): Update.
13863 (inline_read_summary): Update.
13864 (inline_free_summary): Update.
13865 * ipa-inline.c (can_inline_edge_p): Update.
13866 (compute_inlined_call_time): Update.
13867 (want_inline_small_function_p): Update.
13868 (edge_badness): Update.
13869 (early_inliner): Update.
13870 * ipa-inline.h (inline_edge_summary): Turn to ...
13871 (ipa_call_summary): ... this one.
13872 (ipa_call_summary_t): New class.
13873 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
13874 (ipa_call_summaries): New.
13875 (inline_edge_summary): Remove.
13876 (estimate_edge_growth): Update.
13877 * ipa-profile.c (ipa_propagate_frequency_1): Update.
13878 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13879 * ipa-split.c (execute_split_functions): Update.
13880 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
13882 2017-05-23 Tom de Vries <tom@codesourcery.com>
13884 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13885 attributes): Document rdrand effective target.
13887 2017-05-23 Tom de Vries <tom@codesourcery.com>
13889 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13890 attributes): Sort alphabetically.
13892 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
13894 * config/avr/genmultilib.awk: Use gsub instead of gensub.
13896 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
13899 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
13900 V2DF/V2DI splat into two separate patterns, one that handles
13901 registers, and the other that only handles memory. Drop support
13902 for splatting from a GPR on ISA 2.07 and then splitting the
13903 splat into direct move and splat.
13904 (vsx_splat_<mode>_reg): Likewise.
13905 (vsx_splat_<mode>_mem): Likewise.
13907 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
13909 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
13911 2017-05-22 Jakub Jelinek <jakub@redhat.com>
13913 PR middle-end/80809
13914 * omp-low.c (finish_taskreg_remap): New function.
13915 (finish_taskreg_scan): If unit size of ctx->record_type
13916 is non-constant, unshare the size expression and replace
13917 decls in it with possible outer var refs.
13919 PR middle-end/80809
13920 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
13921 GOVD_SHARED rather than GOVD_PRIVATE with it.
13922 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
13923 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
13925 PR middle-end/80853
13926 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
13927 as last argument to build_outer_var_ref for pointer bases of array
13928 section reductions.
13930 2017-05-19 Martin Sebor <msebor@redhat.com>
13932 * print-tree.c (print_node): Print DECL_READ_P flag.
13934 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13936 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
13937 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
13938 * cgraph.c: Likewise.
13939 * cgraphunit.c: Likewise.
13940 * gengtype.c: Likewise.
13941 * ipa-cp.c: Likewise.
13942 * ipa-devirt.c: Likewise.
13943 * ipa-icf.c: Likewise.
13944 * ipa-predicate.c: Likewise.
13945 * ipa-profile.c: Likewise.
13946 * ipa-prop.c: Likewise.
13947 * ipa-split.c: Likewise.
13949 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
13950 edge_predicate_pool, dump_inline_hints,
13951 inline_summary::account_size_time, redirect_to_unreachable,
13952 edge_set_predicate, set_hint_predicate,
13953 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
13954 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
13955 inline_summary_t::remove, remap_hint_predicate_after_duplication,
13956 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
13957 ipa_call_summary_t::remove, initialize_growth_caches,
13958 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
13959 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
13960 mark_modified, unmodified_parm_1, unmodified_parm,
13961 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
13962 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
13963 compute_bb_predicates, will_be_nonconstant_expr_predicate,
13964 will_be_nonconstant_predicate, record_modified_bb_info,
13965 get_minimal_bb, record_modified, param_change_prob,
13966 phi_result_unknown_predicate, predicate_for_phi_result,
13967 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
13968 estimate_function_body_sizes, compute_inline_parameters,
13969 compute_inline_parameters_for_curren, pass_data_inline_parameters,
13970 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
13971 inline_update_callee_summaries, remap_edge_change_prob,
13972 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
13973 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
13974 inline_analyze_function, inline_summary_t::insert,
13975 inline_generate_summary, read_ipa_call_summary, inline_read_section,
13976 inline_read_summary, write_ipa_call_summary, inline_write_summary,
13977 inline_free_summary): Move to ipa-fnsummary.h
13978 (predicate_t): Remove.
13979 * ipa-fnsummary.c: New file.
13980 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
13981 (enum inline_hints_vals, inline_hints, agg_position_info,
13982 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
13983 inline_summaries, ipa_call_summary, ipa_call_summary_t,
13984 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
13985 dump_inline_summary, dump_inline_hints, inline_generate_summary,
13986 inline_read_summary, inline_write_summary, inline_free_summary,
13987 inline_analyze_function, initialize_inline_failed,
13988 inline_merge_summary, inline_update_overall_summary,
13989 compute_inline_parameters): Move to ipa-fnsummary.h
13990 * ipa-fnsummary.h: New file.
13991 * ipa-inline-transform.h: Include ipa-inline.h.
13992 * ipa-inline.c: LIkewise.
13994 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13996 * ipa-inline.c (edge_badness): Use inlined_time instead of
13997 inline_summaries->get.
13999 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14001 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
14003 2017-05-22 Nathan Sidwell <nathan@acm.org>
14005 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
14006 (fdump-lang): Document 'raw' option.
14007 * dumpfile.h (TDI_tu): Delete.
14008 * dumpfile.c (dump_files): Remove translation-unit.
14009 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
14011 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
14013 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
14014 command option from $(AWK) call.
14015 * config/avr/genmultilib.awk: Simplify and rewrite so that it
14016 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
14017 [FORMAT]: Remove handling of variable.
14018 * config/avr/t-multilib: Regenerate.
14020 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14022 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
14024 (dump_inline_summary): Do not print self_time.
14025 (estimate_function_body_sizes): Do not set self_time.
14026 (compute_inline_parameters): Likewise.
14027 (inline_read_section, inline_write_summary): Do not stream self_time.
14028 * ipa-inline.h (inline_summary): Drop self_time.
14030 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14032 * ipa-inline-analysis.c (account_size_time): Rename to ...
14033 (inline_summary::account_size_time): ... this one.
14034 (reset_ipa_call_summary): Turn to ...
14035 (ipa_call_summary::reset): ... this one.
14036 (reset_inline_summary): Turn to ...
14037 (inline_summary::reset): ... this one.
14038 (inline_summary_t::remove): Update.
14039 (inline_summary_t::duplicate): Update.
14040 (ipa_call_summary_t::remove): Update.
14041 (dump_inline_summary): Update.
14042 (estimate_function_body_sizes): Update.
14043 (compute_inline_parameters): Update.
14044 (estimate_node_size_and_time): Update.
14045 (inline_merge_summary): Update.
14046 (inline_update_overall_summary): Update.
14047 (inline_read_section): Update.
14048 (inline_write_summary): Update.
14049 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
14050 add account_size_time and reset member functions.
14051 (ipa_call_summary): Add reset function.
14052 * ipa-predicate.h (predicate::operator &): Constify.
14054 2017-05-22 Richard Biener <rguenther@suse.de>
14056 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
14058 2017-05-19 Jason Merrill <jason@redhat.com>
14060 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
14062 2017-05-19 Marek Polacek <polacek@redhat.com>
14065 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
14066 TYPE_OVERFLOW_WRAPS checks.
14068 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
14070 * tree-core.h (enum omp_clause_default_kind): Add
14071 "OMP_CLAUSE_DEFAULT_PRESENT".
14072 * tree-pretty-print.c (dump_omp_clause): Handle it.
14073 * gimplify.c (enum gimplify_omp_var_data): Add
14074 "GOVD_MAP_FORCE_PRESENT".
14075 (gimplify_adjust_omp_clauses_1): Map it to
14076 "GOMP_MAP_FORCE_PRESENT".
14077 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
14079 * gimplify.c (oacc_default_clause): Clarify.
14081 2017-05-19 Nathan Sidwell <nathan@acm.org>
14083 LANG_HOOK_REGISTER_DUMPS
14084 * toplev.c (general_init): Call register dump lang hook.
14085 * doc/invoke.texi: Document -fdump-lang option family.
14086 * dumpfile.c (dump_files): Remove class dump here.
14087 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
14088 * dumpfile.h (tree_dump_index): Remove TDI_class.
14089 * langhooks-def.h (lhd_register_dumps): Declare.
14090 (LANG_HOOKS_REGISTER_DUMPS): Define.
14091 (LANG_HOOKS_INITIALIZER): Add it.
14092 * langhooks.c (lhd_register_dumps): Define.
14093 * langhooks.h (struct lang_hooks): Add register_dumps.
14095 2017-05-19 Nathan Sidwell <nathan@acm.org>
14097 * context.h (context::set_passes): New.
14098 * context.c (context::context): Do not create pass manager.
14099 * toplev.c (general_init): Create pass manager here.
14101 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
14103 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
14104 use this splitter if two add or or instructions would also work for
14105 the constant we want to generate.
14107 2017-05-19 Richard Biener <rguenther@suse.de>
14110 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
14111 predicate evaluation.
14113 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14115 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
14117 * ipa-inline.c (want_inline_small_function_p): Do not cast to
14120 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14122 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14123 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14124 (inline_edge_summary_vec): Turn into ...
14125 (ipa_call_summaries): ... this one.
14126 (redirect_to_unreachable, edge_set_predicate,
14127 evaluate_properties_for_edge, inline_summary_alloc,
14128 reset_ipa_call_summary, reset_inline_summary,
14129 inline_summary_t::duplicate): Update.
14130 (inline_edge_duplication_hook): Turn to ...
14131 (ipa_call_summary_t::duplicate): ... this one.
14132 (inline_edge_removal_hook): Turn to ...
14133 (ipa_call_summary_t::remove): ... this one.
14134 (dump_inline_edge_summary): Turn to ...
14135 (dump_ipa_call_summary): ... this one.
14136 (estimate_function_body_sizes): Update.
14137 (inline_update_callee_summaries): Update.
14138 (remap_edge_change_prob): Update.
14139 (remap_edge_summaries): Update.
14140 (inline_merge_summary): Update.
14141 (do_estimate_edge_time): Update.
14142 (inline_generate_summary): Update.
14143 (inline_read_section): Update.
14144 (inline_read_summary): Update.
14145 (inline_free_summary): Update.
14146 * ipa-inline.c (can_inline_edge_p): Update.
14147 (compute_inlined_call_time): Update.
14148 (want_inline_small_function_p): Update.
14149 (edge_badness): Update.
14150 (early_inliner): Update.
14151 * ipa-inline.h (inline_edge_summary): Turn to ...
14152 (ipa_call_summary): ... this one.
14153 (ipa_call_summary_t): New class.
14154 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14155 (ipa_call_summaries): New.
14156 (inline_edge_summary): Remove.
14157 (estimate_edge_growth): Update.
14158 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14159 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14160 * ipa-split.c (execute_split_functions): Update.
14161 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14163 2017-05-19 Richard Biener <rguenther@suse.de>
14165 PR middle-end/80764
14166 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
14168 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
14170 * config/rs6000/rs6000.c (struct machine_function): Add field
14171 fpr_is_wrapped_separately.
14172 (rs6000_get_separate_components): Use 64 components. Handle the
14173 new FPR components.
14174 (rs6000_components_for_bb): Handle the FPR components.
14175 (rs6000_emit_prologue_components): Handle the FPR components.
14176 (rs6000_emit_epilogue_components): Handle the FPR components.
14177 (rs6000_set_handled_components): Handle the FPR components.
14178 (rs6000_emit_prologue): Don't output prologue code for those FPRs
14179 that are already separately shrink-wrapped.
14180 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
14181 that are already separately shrink-wrapped.
14183 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
14186 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
14189 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
14190 (define_peephole2 for Altivec d-form load): Add peepholes to catch
14191 cases where the register allocator uses a move and an offsettable
14192 memory operation to/from a FPR register on ISA 2.06/2.07.
14193 (define_peephole2 for Altivec d-form store): Likewise.
14195 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
14198 * config/i386/mmx.md (*mov<mode>_internal): Enable
14199 alternatives 11, 12, 13 and 14 also for 32bit targets.
14200 Remove alternatives 15, 16, 17 and 18.
14201 * config/i386/sse.md (vec_concatv2di): Change
14202 alternative (!x, *y) to (x, ?!*Yn).
14204 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
14206 * dumpfile.h (enum dump_kind): Remove stray comma.
14208 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14210 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
14211 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
14212 predicate::num_conditions
14213 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
14214 (CHANGED): turn into predicate::changed.
14215 (agg_position_info): Move to ipa-predicate.h
14216 (add_condition, predicate::add_clause, predicate::operator &=,
14217 predicate::or_with, predicate::evaluate, predicate::probability,
14218 dump_condition, dump_clause, predicate::dump,
14219 predicate::remap_after_duplication, predicate::remap_after_inlining,
14220 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
14221 (evaluate_conditions_for_known_args): Update.
14222 (set_cond_stmt_execution_predicate): Update.
14223 * ipa-inline.h: Include ipa-predicate.h
14224 (condition, inline_param_summary, conditions, agg_position_info,
14225 predicate): Move to ipa-predicate.h
14226 * ipa-predicate.c: New file.
14227 * ipa-predicate.h: New file.
14229 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
14231 * final.c (leaf_function_p): Check we are not in a sequence.
14233 2017-05-18 Martin Liska <mliska@suse.cz>
14235 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
14236 * dumpfile.c (dump_register): Use new enum dump_kind.
14237 (get_dump_file_name): Likewise.
14238 (dump_enable_all): Likewise.
14239 (dump_switch_p_1): Likewise.
14240 (enable_rtl_dump_file): Remove usage of TDF_RTL.
14241 * dumpfile.h (enum dump_kind): New enum type.
14242 (struct dump_file_info): Create constructor and
14243 format fields and comments.
14244 * passes.c (pass_manager::register_one_dump_file):
14246 * statistics.c (statistics_early_init): Likewise.
14247 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
14248 TDF_TREE with TDF_SLIM.
14249 (gather_memory_references_ref): Likewise.
14251 2017-05-18 Martin Liska <mliska@suse.cz>
14253 * vec.h (struct vnull): Use it.
14255 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14257 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
14258 (true_predicate, false_predicate, true_predicate_p,
14259 false_predicate_p): Remove.
14260 (single_cond_predicate, not_inlined_predicate): Turn to member function
14262 (add_condition): Update.
14263 (add_clause): Turn to...
14264 (predicate::add_clause): ... this one; update; allow passing NULL
14266 (and_predicates): Turn to ...
14267 (predicate::operator &=): ... this one.
14268 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
14269 (or_predicates): Turn to ...
14270 (predicate::or_with): ... this one.
14271 (evaluate_predicate): Turn to ...
14272 (predicate::evaluate): ... this one.
14273 (predicate_probability): Turn to ...
14274 (predicate::probability): ... this one.
14275 (dump_condition): Update.
14276 (dump_predicate): Turn to ...
14277 (predicate::dump): ... this one.
14278 (account_size_time): Update.
14279 (edge_set_predicate): Update.
14280 (set_hint_predicate): UPdate.
14281 (evaluate_conditions_for_known_args): Update.
14282 (evaluate_properties_for_edge): Update.
14283 (remap_predicate_after_duplication): Turn to...
14284 (predicate::remap_after_duplication): ... this one.
14285 (remap_hint_predicate_after_duplication): Update.
14286 (inline_summary_t::duplicate): UPdate.
14287 (dump_inline_edge_summary): Update.
14288 (dump_inline_summary): Update.
14289 (set_cond_stmt_execution_predicate): Update.
14290 (set_switch_stmt_execution_predicate): Update.
14291 (compute_bb_predicates): Update.
14292 (will_be_nonconstant_expr_predicate): Update.
14293 (will_be_nonconstant_predicate): Update.
14294 (phi_result_unknown_predicate): Update.
14295 (predicate_for_phi_result): Update.
14296 (array_index_predicate): Update.
14297 (estimate_function_body_sizes): Update.
14298 (estimate_node_size_and_time): Update.
14299 (estimate_ipcp_clone_size_and_time): Update.
14300 (remap_predicate): Rename to ...
14301 (predicate::remap_after_inlining): ... this one.
14302 (remap_hint_predicate): Update.
14303 (inline_merge_summary): Update.
14304 (inline_update_overall_summary): Update.
14305 (estimate_size_after_inlining): Update.
14306 (read_predicate): Rename to ...
14307 (predicate::stream_in): ... this one.
14308 (read_inline_edge_summary): Update.
14309 (write_predicate): Rename to ...
14310 (predicate::stream_out): ... this one.
14311 (write_inline_edge_summary): Update.
14312 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
14313 (clause_t): Turn to uint32_t
14314 (predicate): Turn to class; implement constructor and operators
14316 (size_time_entry): Update.
14317 (inline_summary): Update.
14318 (inline_edge_summary): Update.
14320 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
14322 * fold-const.c (fold_binary_loc): Move transformation...
14323 * match.pd (C - X CMP X): ... here.
14325 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
14327 * config/sparc/sparc.c (sparc_option_override): Set function
14328 alignment for -mcpu=niagara7 to 64 to match the I$ line.
14329 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
14331 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
14333 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
14335 2017-05-18 Marek Polacek <polacek@redhat.com>
14338 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
14339 (pass_ubsan::execute): Call gimple_assign_single_p instead of
14340 gimple_assign_load_p.
14342 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
14344 PR middle-end/80692
14345 * real.c (do_compare): Give decimal_do_compare preference over
14346 comparing just the signs.
14348 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
14350 * doc/md.texi (Canonicalization of Instructions): Describe the
14351 canonical form of instructions that inherently set a condition
14354 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
14356 PR middle-end/80775
14357 * tree-cfg.c: Move deletion of unreachable case statements to after
14358 the merging of consecutive case labels.
14360 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
14362 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
14363 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
14364 restoring of callee-saved registers.
14366 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
14368 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
14369 * config/visium/visium.c (single_set_and_flags): Likewise.
14370 * config/visium/visium.md (Substitutions): Likewise.
14372 2017-05-17 Martin Liska <mliska@suse.cz>
14374 * cfg.c: Introduce dump_flags_t type and
14375 use it instead of int type.
14377 * cfghooks.c: Likewise.
14378 * cfghooks.h (struct cfg_hooks): Likewise.
14379 * cfgrtl.c: Likewise.
14380 * cfgrtl.h: Likewise.
14381 * cgraph.c (cgraph_node::get_body): Likewise.
14382 * coretypes.h: Likewise.
14383 * domwalk.c: Likewise.
14384 * domwalk.h: Likewise.
14385 * dumpfile.c (struct dump_option_value_info): Likewise.
14386 (dump_enable_all): Likewise.
14387 (dump_switch_p_1): Likewise.
14388 (opt_info_switch_p): Likewise.
14389 * dumpfile.h (enum tree_dump_index): Likewise.
14390 (struct dump_file_info): Likewise.
14391 * genemit.c: Likewise.
14392 * generic-match-head.c: Likewise.
14393 * gengtype.c (open_base_files): Likewise.
14394 * gimple-pretty-print.c: Likewise.
14395 * gimple-pretty-print.h: Likewise.
14396 * graph.c (print_graph_cfg): Likewise.
14397 * graphite-scop-detection.c (dot_all_sese): Likewise.
14398 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
14399 * loop-unroll.c (report_unroll): Likewise.
14400 * passes.c (pass_manager::register_one_dump_file): Likewise.
14401 * print-tree.c: Likewise.
14402 * statistics.c: Likewise.
14403 * tree-cfg.c: Likewise.
14404 * tree-cfg.h: Likewise.
14405 * tree-dfa.c: Likewise.
14406 * tree-dfa.h: Likewise.
14407 * tree-dump.c (dump_function): Likewise.
14408 * tree-dump.h (struct dump_info): Likewise.
14409 * tree-pretty-print.c: Likewise.
14410 * tree-pretty-print.h: Likewise.
14411 * tree-ssa-live.c: Likewise.
14412 * tree-ssa-live.h: Likewise.
14413 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
14414 * tree-vect-loop.c: Likewise.
14415 * tree-vect-slp.c: Likewise.
14417 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
14418 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14420 PR tree-optimization/80457
14421 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
14422 of all arguments to a statement as scalar_to_vec operations.
14423 (vectorizable_call): Adjust call to vect_model_simple_cost for
14425 (vectorizable_conversion): Likewise.
14426 (vectorizable_assignment): Likewise.
14427 (vectorizable_shift): Likewise.
14428 (vectorizable_operation): Likewise.
14429 (vectorizable_comparison): Likewise.
14430 (vect_is_simple_cond): Record the def types for operands.
14431 (vectorizable_condition): Likewise, call vect_model_simple_cost.
14432 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
14433 for statement argument count.
14435 2017-05-16 Carl Love <cel@us.ibm.com>
14437 * config/rs6000/rs6000-c: Add support for built-in functions
14438 vector unsigned long long vec_bperm (vector unsigned long long,
14439 vector unsigned char)
14440 vector signed long long vec_mule (vector signed int,
14442 vector unsigned long long vec_mule (vector unsigned int,
14443 vector unsigned int)
14444 vector signed long long vec_mulo (vector signed int,
14446 vector unsigned long long vec_mulo (vector unsigned int,
14447 vector unsigned int)
14448 vector signed char vec_sldw (vector signed char,
14449 vector signed char,
14451 vector unsigned char vec_sldw (vector unsigned char,
14452 vector unsigned char,
14454 vector signed short vec_sldw (vector signed short,
14455 vector signed short,
14457 vector unsigned short vec_sldw (vector unsigned short,
14458 vector unsigned short,
14460 vector signed int vec_sldw (vector signed int,
14463 vector unsigned int vec_sldw (vector unsigned int,
14464 vector unsigned int,
14466 vector signed long long vec_sldw (vector signed long long,
14467 vector signed long long,
14469 vector unsigned long long vec_sldw (vector unsigned long long,
14470 vector unsigned long long,
14472 * config/rs6000/rs6000-c: Add support for built-in functions
14473 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
14474 * config/rs6000/altivec.h: Add defintion for vec_sldw.
14475 * doc/extend.texi: Update the built-in documentation for the
14476 new built-in functions.
14478 2017-05-16 Marek Polacek <polacek@redhat.com>
14482 * tree.c (save_expr): Don't fold the expression.
14484 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
14486 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
14487 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
14488 and (?*y,m). Update insn attributes.
14490 2017-05-16 Martin Liska <mliska@suse.cz>
14492 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
14493 flags argument of print_gimple_stmt, print_gimple_expr,
14494 print_generic_stmt and print_generic_expr.
14495 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
14496 * coretypes.h: Likewise.
14497 * except.c (dump_eh_tree): Likewise.
14498 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14499 * gimple-pretty-print.h: Likewise.
14500 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
14501 (backprop::push_to_worklist): Likewise.
14502 (backprop::pop_from_worklist): Likewise.
14503 (backprop::process_use): Likewise.
14504 (backprop::intersect_uses): Likewise.
14505 (note_replacement): Likewise.
14506 * gimple-ssa-store-merging.c
14507 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
14508 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
14509 (pass_store_merging::execute): Likewise.
14510 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
14511 (ssa_base_cand_dump_callback): Likewise.
14512 (dump_incr_vec): Likewise.
14513 (replace_refs): Likewise.
14514 (replace_mult_candidate): Likewise.
14515 (create_add_on_incoming_edge): Likewise.
14516 (create_phi_basis): Likewise.
14517 (insert_initializers): Likewise.
14518 (all_phi_incrs_profitable): Likewise.
14519 (introduce_cast_before_cand): Likewise.
14520 (replace_one_candidate): Likewise.
14521 * gimplify.c (gimplify_expr): Likewise.
14522 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
14523 (set_rename): Likewise.
14524 (rename_uses): Likewise.
14525 (copy_loop_phi_nodes): Likewise.
14526 (add_close_phis_to_merge_points): Likewise.
14527 (copy_loop_close_phi_args): Likewise.
14528 (copy_cond_phi_args): Likewise.
14529 (graphite_copy_stmts_from_block): Likewise.
14530 (translate_pending_phi_nodes): Likewise.
14531 * graphite-poly.c (print_pdr): Likewise.
14532 (dump_gbb_cases): Likewise.
14533 (dump_gbb_conditions): Likewise.
14534 (print_scop_params): Likewise.
14535 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
14536 (build_cross_bb_scalars_use): Likewise.
14537 (gather_bbs::before_dom_children): Likewise.
14538 * hsa-dump.c (dump_hsa_immed): Likewise.
14539 * ipa-cp.c (print_ipcp_constant_value): Likewise.
14540 (get_replacement_map): Likewise.
14541 * ipa-inline-analysis.c (dump_condition): Likewise.
14542 (estimate_function_body_sizes): Likewise.
14543 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
14544 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
14545 * ipa-prop.c (ipa_dump_param): Likewise.
14546 (ipa_print_node_jump_functions_for_edge): Likewise.
14547 (ipa_modify_call_arguments): Likewise.
14548 (ipa_modify_expr): Likewise.
14549 (ipa_dump_param_adjustments): Likewise.
14550 (ipa_dump_agg_replacement_values): Likewise.
14551 (ipcp_modif_dom_walker::before_dom_children): Likewise.
14552 * ipa-pure-const.c (check_stmt): Likewise.
14553 (pass_nothrow::execute): Likewise.
14554 * ipa-split.c (execute_split_functions): Likewise.
14555 * omp-offload.c (dump_oacc_loop_part): Likewise.
14556 (dump_oacc_loop): Likewise.
14557 * trans-mem.c (tm_log_emit): Likewise.
14558 (tm_memopt_accumulate_memops): Likewise.
14559 (dump_tm_memopt_set): Likewise.
14560 (dump_tm_memopt_transform): Likewise.
14561 * tree-cfg.c (gimple_verify_flow_info): Likewise.
14562 (print_loop): Likewise.
14563 * tree-chkp-opt.c (chkp_print_addr): Likewise.
14564 (chkp_gather_checks_info): Likewise.
14565 (chkp_get_check_result): Likewise.
14566 (chkp_remove_check_if_pass): Likewise.
14567 (chkp_use_outer_bounds_if_possible): Likewise.
14568 (chkp_reduce_bounds_lifetime): Likewise.
14569 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
14570 (chkp_mark_completed_bounds): Likewise.
14571 (chkp_register_incomplete_bounds): Likewise.
14572 (chkp_mark_invalid_bounds): Likewise.
14573 (chkp_maybe_copy_and_register_bounds): Likewise.
14574 (chkp_build_returned_bound): Likewise.
14575 (chkp_get_bound_for_parm): Likewise.
14576 (chkp_build_bndldx): Likewise.
14577 (chkp_get_bounds_by_definition): Likewise.
14578 (chkp_generate_extern_var_bounds): Likewise.
14579 (chkp_get_bounds_for_decl_addr): Likewise.
14580 * tree-chrec.c (chrec_apply): Likewise.
14581 * tree-data-ref.c (dump_data_reference): Likewise.
14582 (dump_subscript): Likewise.
14583 (dump_data_dependence_relation): Likewise.
14584 (analyze_overlapping_iterations): Likewise.
14585 * tree-inline.c (expand_call_inline): Likewise.
14586 (tree_function_versioning): Likewise.
14587 * tree-into-ssa.c (dump_defs_stack): Likewise.
14588 (dump_currdefs): Likewise.
14589 (dump_names_replaced_by): Likewise.
14590 (dump_update_ssa): Likewise.
14591 (update_ssa): Likewise.
14592 * tree-object-size.c (pass_object_sizes::execute): Likewise.
14593 * tree-parloops.c (build_new_reduction): Likewise.
14594 (try_create_reduction_list): Likewise.
14595 (ref_conflicts_with_region): Likewise.
14596 (oacc_entry_exit_ok_1): Likewise.
14597 (oacc_entry_exit_single_gang): Likewise.
14598 * tree-pretty-print.h: Likewise.
14599 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
14600 (get_scalar_evolution): Likewise.
14601 (add_to_evolution): Likewise.
14602 (get_loop_exit_condition): Likewise.
14603 (analyze_evolution_in_loop): Likewise.
14604 (analyze_initial_condition): Likewise.
14605 (analyze_scalar_evolution): Likewise.
14606 (instantiate_scev): Likewise.
14607 (number_of_latch_executions): Likewise.
14608 (gather_chrec_stats): Likewise.
14609 (final_value_replacement_loop): Likewise.
14610 (scev_const_prop): Likewise.
14611 * tree-sra.c (dump_access): Likewise.
14612 (disqualify_candidate): Likewise.
14613 (create_access): Likewise.
14614 (reject): Likewise.
14615 (maybe_add_sra_candidate): Likewise.
14616 (create_access_replacement): Likewise.
14617 (analyze_access_subtree): Likewise.
14618 (analyze_all_variable_accesses): Likewise.
14619 (sra_modify_assign): Likewise.
14620 (initialize_constant_pool_replacements): Likewise.
14621 (find_param_candidates): Likewise.
14622 (decide_one_param_reduction): Likewise.
14623 (replace_removed_params_ssa_names): Likewise.
14624 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
14625 * tree-ssa-copy.c (dump_copy_of): Likewise.
14626 (copy_prop_visit_cond_stmt): Likewise.
14627 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
14628 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
14629 (record_equivalences_from_stmt): Likewise.
14630 * tree-ssa-dse.c (compute_trims): Likewise.
14631 (delete_dead_call): Likewise.
14632 (delete_dead_assignment): Likewise.
14633 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
14634 (forward_propagate_into_cond): Likewise.
14635 (pass_forwprop::execute): Likewise.
14636 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14637 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
14639 (move_computations_worker): Likewise.
14640 (execute_sm): Likewise.
14641 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
14642 (remove_exits_and_undefined_stmts): Likewise.
14643 (remove_redundant_iv_tests): Likewise.
14644 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
14645 (adjust_iv_update_pos): Likewise.
14646 * tree-ssa-math-opts.c (bswap_replace): Likewise.
14647 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
14648 (value_replacement): Likewise.
14649 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
14650 * tree-ssa-pre.c (print_pre_expr): Likewise.
14651 (get_representative_for): Likewise.
14652 (create_expression_by_pieces): Likewise.
14653 (insert_into_preds_of_block): Likewise.
14654 (eliminate_insert): Likewise.
14655 (eliminate_dom_walker::before_dom_children): Likewise.
14656 (eliminate): Likewise.
14657 (remove_dead_inserted_code): Likewise.
14658 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
14659 * tree-ssa-reassoc.c (get_rank): Likewise.
14660 (eliminate_duplicate_pair): Likewise.
14661 (eliminate_plus_minus_pair): Likewise.
14662 (eliminate_not_pairs): Likewise.
14663 (undistribute_ops_list): Likewise.
14664 (eliminate_redundant_comparison): Likewise.
14665 (update_range_test): Likewise.
14666 (optimize_range_tests_var_bound): Likewise.
14667 (optimize_vec_cond_expr): Likewise.
14668 (rewrite_expr_tree): Likewise.
14669 (rewrite_expr_tree_parallel): Likewise.
14670 (linearize_expr): Likewise.
14671 (break_up_subtract): Likewise.
14672 (linearize_expr_tree): Likewise.
14673 (attempt_builtin_powi): Likewise.
14674 (attempt_builtin_copysign): Likewise.
14675 (transform_stmt_to_copy): Likewise.
14676 (transform_stmt_to_multiply): Likewise.
14677 (dump_ops_vector): Likewise.
14678 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
14679 (print_scc): Likewise.
14680 (set_ssa_val_to): Likewise.
14681 (visit_reference_op_store): Likewise.
14682 (visit_use): Likewise.
14683 (sccvn_dom_walker::before_dom_children): Likewise.
14684 (run_scc_vn): Likewise.
14685 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14687 (expr_hash_elt::print): Likewise.
14688 (const_and_copies::pop_to_marker): Likewise.
14689 (const_and_copies::record_const_or_copy_raw): Likewise.
14690 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
14691 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
14692 (dump_predicates): Likewise.
14693 (find_uninit_use): Likewise.
14694 (warn_uninitialized_phi): Likewise.
14695 (pass_late_warn_uninitialized::execute): Likewise.
14696 * tree-ssa.c (verify_vssa): Likewise.
14697 (verify_ssa): Likewise.
14698 (maybe_optimize_var): Likewise.
14699 * tree-vrp.c (dump_value_range): Likewise.
14700 (dump_all_value_ranges): Likewise.
14701 (dump_asserts_for): Likewise.
14702 (register_edge_assert_for_2): Likewise.
14703 (vrp_visit_cond_stmt): Likewise.
14704 (vrp_visit_switch_stmt): Likewise.
14705 (vrp_visit_stmt): Likewise.
14706 (vrp_visit_phi_node): Likewise.
14707 (simplify_cond_using_ranges_1): Likewise.
14708 (fold_predicate_in): Likewise.
14709 (evrp_dom_walker::before_dom_children): Likewise.
14710 (evrp_dom_walker::push_value_range): Likewise.
14711 (evrp_dom_walker::pop_value_range): Likewise.
14712 (execute_early_vrp): Likewise.
14714 2017-05-16 Richard Biener <rguenther@suse.de>
14716 * dwarf2out.c (loc_list_from_tree_1): Do not create
14717 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
14719 2017-05-16 Richard Biener <rguenther@suse.de>
14721 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
14723 (note_variable_value_in_expr): If we resolved the decl ref
14724 do not push to the stack.
14726 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
14728 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
14729 operations in fast-math mode.
14730 (vaddq_f16): Likewise.
14731 (vmul_f16): Likewise.
14732 (vmulq_f16): Likewise.
14733 (vsub_f16): Likewise.
14734 (vsubq_f16): Likewise.
14735 * config/arm/neon.md (add<mode>3): New.
14737 (fma:<VH:mode>3): New. Also remove outdated comment.
14740 2017-05-16 Martin Liska <mliska@suse.cz>
14744 * ipa-devirt.c (warn_types_mismatch): Fix typo.
14745 (odr_types_equivalent_p): Likewise.
14747 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
14749 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
14751 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
14754 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
14755 non-interunit SSE move alternatives with '?'.
14756 (zero-extendsidi peephole2): New peephole to skip intermediate
14757 general register in SSE zero-extend sequence.
14759 2017-05-15 Jeff Law <law@redhat.com>
14761 * reorg.c (relax_delay_slots): Create a new variable to hold
14762 the temporary target rather than clobbering TARGET_LABEL.
14764 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
14765 missing argument to extract_bit_field call.
14766 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
14768 2017-05-15 Martin Liska <mliska@suse.cz>
14771 * gcc.c (process_command): Do not allow empty argument of -o option.
14773 2017-05-15 Renlin Li <renlin.li@arm.com>
14775 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
14776 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
14777 * config/aarch64/constraints.md (Usf): Add long call check.
14778 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
14779 (call_value): Likewise.
14780 (sibcall): Likewise.
14781 (sibcall_value): Likewise.
14783 (call_value_insn): New.
14784 (sibcall_insn): Update rtx pattern.
14785 (sibcall_value_insn): Likewise.
14786 (call_internal): Remove.
14787 (call_value_internal): Likewise.
14788 (sibcall_internal): Likewise.
14789 (sibcall_value_internal): Likewise.
14790 (call_reg): Likewise.
14791 (call_symbol): Likewise.
14792 (call_value_reg): Likewise.
14793 (call_value_symbol): Likewise.
14795 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
14798 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
14800 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
14802 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
14803 compatible with CCGOCmode and with CCZmode.
14805 2017-05-14 Martin Sebor <msebor@redhat.com>
14807 PR middle-end/77671
14808 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
14809 (gimple_fold_builtin_snprintf): Same.
14810 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
14811 (gimple_fold_builtin_snprintf): Same.
14812 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
14813 of character types.
14814 (is_call_safe): New function.
14815 (try_substitute_return_value): Call it.
14816 (try_simplify_call): New function.
14817 (pass_sprintf_length::handle_gimple_call): Call it.
14819 2017-05-14 Martin Sebor <msebor@redhat.com>
14821 PR middle-end/80669
14822 * builtins.c (expand_builtin_stpncpy): Simplify.
14824 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
14826 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
14827 * config/i386/i386.h
14828 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
14829 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
14830 (struct machine_function): Add new members call_ms2sysv,
14831 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
14832 (struct machine_frame_state): New fields sp_realigned and
14833 sp_realigned_offset.
14834 * config/i386/i386.c
14835 (enum xlogue_stub): New enum.
14836 (enum xlogue_stub_sets): New enum.
14837 (class xlogue_layout): New class.
14838 (struct ix86_frame): New fields stack_realign_allocate_offset,
14839 stack_realign_offset and outlined_save_offset. Modify comments to
14840 detail stack layout when using out-of-line stubs.
14841 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
14842 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
14843 -mcall-ms2sysv-xlogues.
14844 (stub_managed_regs): New static variable.
14845 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
14846 registers managed by out-of-line stub.
14847 (disable_call_ms2sysv_xlogues): New function.
14848 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
14849 m->call_ms2sysv when appropriate and compute frame layout for
14851 (sp_valid_at, fp_valid_at): New inline functions.
14852 (choose_basereg): New function.
14853 (choose_baseaddr): Add align parameter, use choose_basereg and modify
14855 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
14856 Use align parameter of choose_baseaddr to generated aligned SSE movs
14858 (pro_epilogue_adjust_stack): Modify to track
14859 machine_frame_state::sp_realigned.
14860 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
14861 (ix86_nsaved_sseregs): Likewise.
14862 (ix86_emit_save_regs): Likewise.
14863 (ix86_emit_save_regs_using_mov): Likewise.
14864 (ix86_emit_save_sse_regs_using_mov): Likewise.
14865 (get_scratch_register_on_entry): Likewise.
14866 (gen_frame_set): New function.
14867 (gen_frame_load): Likewise.
14868 (gen_frame_store): Likewise.
14869 (emit_outlined_ms2sysv_save): Likewise.
14870 (emit_outlined_ms2sysv_restore): Likewise.
14871 (ix86_expand_prologue): Modify stack re-alignment code and call
14872 emit_outlined_ms2sysv_save when appropriate.
14873 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
14874 parameter rtx_insn *insn, which allows the function to be used to only
14875 generate the notes.
14876 (ix86_expand_epilogue): Modify validity checks of frame and stack
14877 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
14878 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
14879 * config/i386/predicates.md
14880 (save_multiple): New predicate.
14881 (restore_multiple): Likewise.
14882 * config/i386/sse.md
14883 (save_multiple<mode>): New pattern.
14884 (save_multiple_realign<mode>): Likewise.
14885 (restore_multiple<mode>): Likewise.
14886 (restore_multiple_and_return<mode>): Likewise.
14887 (restore_multiple_leave_return<mode>): Likewise.
14888 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
14890 2017-05-14 Julia Koval <julia.koval@intel.com>
14892 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
14893 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
14894 (__builtin_ia32_xsetbv): New builtins.
14895 * config/i386/i386.c (ix86_expand_special_args_builtin):
14897 (ix86_expand_builtin): Special expand for new intrinsics.
14898 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
14899 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
14900 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
14902 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14904 * cfganal.c (inverted_post_order_compute): Change argument type
14906 * cfganal.h (inverted_post_order_compute): Adjust prototype.
14907 * df-core.c (rest_of_handle_df_initialize): Adjust.
14908 (rest_of_handle_df_finish): Likewise.
14909 (df_analyze_1): Likewise.
14910 (df_analyze): Likewise.
14911 (loop_inverted_post_order_compute): Change argument to be a vec *.
14912 (df_analyze_loop): Adjust.
14913 (df_get_n_blocks): Likewise.
14914 (df_get_postorder): Likewise.
14915 * df.h (struct df_d): Change field to be a vec.
14916 * lcm.c (compute_laterin): Adjust.
14917 (compute_available): Likewise.
14918 * lra-lives.c (lra_create_live_ranges_1): Likewise.
14919 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
14920 * tree-ssa-pre.c (compute_antic): Likewise.
14922 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14924 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
14925 (depth_first_search::depth_first_search): Change structure init
14926 function to this constructor.
14927 (depth_first_search::add_bb): Rename function to this member.
14928 (depth_first_search::execute): Likewise.
14929 (flow_dfs_compute_reverse_finish): Adjust.
14931 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14933 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
14934 (longest_simple_path): Likewise.
14935 * shrink-wrap.c (spread_components): Likewise.
14936 (disqualify_problematic_components): Likewise.
14937 (emit_common_heads_for_components): Likewise.
14938 (emit_common_tails_for_components): Likewise.
14939 (insert_prologue_epilogue_for_components): Likewise.
14941 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14943 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
14946 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14948 * df-core.c (df_set_blocks): Start using auto_bitmap.
14949 (df_compact_blocks): Likewise.
14950 * df-problems.c (df_rd_confluence_n): Likewise.
14951 * df-scan.c (df_insn_rescan_all): Likewise.
14952 (df_process_deferred_rescans): Likewise.
14953 (df_update_entry_block_defs): Likewise.
14954 (df_update_exit_block_uses): Likewise.
14955 (df_entry_block_bitmap_verify): Likewise.
14956 (df_exit_block_bitmap_verify): Likewise.
14957 (df_scan_verify): Likewise.
14958 * lra-constraints.c (lra_constraints): Likewise.
14959 (undo_optional_reloads): Likewise.
14960 (lra_undo_inheritance): Likewise.
14961 * lra-remat.c (calculate_gen_cands): Likewise.
14962 (do_remat): Likewise.
14963 * lra-spills.c (assign_spill_hard_regs): Likewise.
14964 (spill_pseudos): Likewise.
14965 * tree-ssa-pre.c (bitmap_set_and): Likewise.
14966 (bitmap_set_subtract_values): Likewise.
14968 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14970 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
14971 management with auto_bitmap.
14972 (fix_inter_tick): Likewise.
14973 (fix_recovery_deps): Likewise.
14974 * ira.c (add_store_equivs): Likewise.
14975 (find_moveable_pseudos): Likewise.
14976 (split_live_ranges_for_shrink_wrap): Likewise.
14977 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
14978 (rtx_reuse_manager::seen_def_p): Likewise.
14979 (rtx_reuse_manager::set_seen_def): Likewise.
14980 * print-rtl.h (class rtx_reuse_manager): Likewise.
14982 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14984 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
14986 (migrate_btr_def): Likewise.
14987 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
14988 * df-core.c (loop_post_order_compute): Likewise.
14989 (loop_inverted_post_order_compute): Likewise.
14990 * hsa-common.h: Likewise.
14991 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
14992 * init-regs.c (initialize_uninitialized_regs): Likewise.
14993 * ipa-inline.c (resolve_noninline_speculation): Likewise.
14994 (inline_small_functions): Likewise.
14995 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
14996 * ira.c (combine_and_move_insns): Likewise.
14997 (build_insn_chain): Likewise.
14998 * loop-invariant.c (find_invariants): Likewise.
14999 * lower-subreg.c (propagate_pseudo_copies): Likewise.
15000 * predict.c (tree_predict_by_opcode): Likewise.
15001 (predict_paths_leading_to): Likewise.
15002 (predict_paths_leading_to_edge): Likewise.
15003 (estimate_loops_at_level): Likewise.
15004 (estimate_loops): Likewise.
15005 * shrink-wrap.c (try_shrink_wrapping): Likewise.
15006 (spread_components): Likewise.
15007 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
15008 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
15009 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
15010 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
15011 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
15012 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
15013 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
15014 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
15015 (mark_threaded_blocks): Likewise.
15016 (thread_through_all_blocks): Likewise.
15017 * tree-ssa.c (verify_ssa): Likewise.
15018 (execute_update_addresses_taken): Likewise.
15019 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
15021 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15023 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
15025 (post_order_compute): Likewise.
15026 (inverted_post_order_compute): Likewise.
15027 (pre_and_rev_post_order_compute_fn): Likewise.
15029 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15031 * genrecog.c (int_set::int_set): Explicitly construct our
15032 auto_vec base class.
15033 * vec.h (auto_vec::auto_vec): New constructor.
15035 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15037 * bitmap.h (class auto_bitmap): New constructor taking
15038 bitmap_obstack * argument.
15040 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15042 * bitmap.h (class auto_bitmap): Change type of m_bits to
15043 bitmap_head, and adjust ctor / dtor and member operators.
15045 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
15047 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
15048 when returned register mode doesn't match original mode.
15050 2017-05-12 Jeff Law <law@redhat.com>
15051 Jakub Jelinek <jakub@redhat.com>
15053 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
15054 we look for cc setter after the compare-elim changes.
15055 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
15056 within the vector to match what compare-elim now expects.
15057 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
15058 (xorsi3_flags, one_cmplsi2_flags): Likewise.
15060 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
15061 after the compare-elim changes.
15062 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
15063 the vector to match what compare-elim now expects.
15064 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
15065 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
15066 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
15067 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
15068 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
15070 * config/visium/visium.c (single_set_and_flags): Fix where
15071 we look for cc setter after the compare-elim changes.
15072 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
15073 with the vector to match what compare-elim now expects.
15074 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
15075 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
15076 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
15077 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
15078 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
15079 (neg<mode>2_insn_set_overflow): Likewise.
15081 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
15083 PR middle-end/79794
15084 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
15085 maybe_expand_insn call, set ops[0].target. If still set after call,
15086 set alt_rtl. Add extra arg to recursive calls.
15087 (extract_bit_field): Add alt_rtl argument. Pass to
15089 * expmed.h (extract_bit_field): Fix prototype.
15090 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
15091 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
15092 to extract_bit_field_calls.
15093 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
15094 Pass alt_rtl to extract_bit_field calls.
15095 * calls.c (store_unaligned_arguments_into_psuedos)
15096 load_register_parameters): Pass extra NULL to extract_bit_field calls.
15097 * optabs.c (maybe_legitimize_operand): Clear op->target when call
15099 * optabs.h (struct expand_operand): Add target bitfield.
15101 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15103 * compare-elim.c (try_eliminate_compare): Canonicalize
15104 operation with embedded compare to
15105 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
15106 (set (reg) (operation)].
15108 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
15110 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15113 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
15114 cost of adding a carry flag for ADC instruction.
15115 [case MINUS]: Ignore the cost of subtracting a carry flag
15116 for SBB instruction.
15118 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
15120 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
15122 * config/rs6000/bmiintrin.h: New file.
15123 * config/rs6000/bmi2intrin.h: New file.
15124 * config/rs6000/x86intrin.h: New file.
15126 2017-05-12 Jeff Law <law@redhat.com>
15128 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
15131 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
15133 PR middle-end/80707
15134 * tree-cfg.c: Remove cfg edges of unreachable case statements.
15136 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15138 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15139 early expansion of vector divide builtins.
15140 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
15141 builtins identified as having unsigned arguments.
15143 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15145 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
15146 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
15147 expansion of vector logical operations (and, andc, or, xor,
15150 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15152 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
15153 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
15155 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15157 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15158 early GIMPLE expansion of vector multiplies.
15160 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15162 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
15163 TARGET_HAVE_MOVT conditional.
15164 (movt splitter): Likewise.
15166 2017-05-12 Richard Biener <rguenther@suse.de>
15168 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
15169 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15170 Fold all stmts not inplace.
15172 2017-05-12 Richard Biener <rguenther@suse.de>
15174 PR tree-optimization/80713
15175 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
15176 inserted_exprs bit for not removed stmts.
15178 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
15180 PR middle-end/69921
15181 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
15182 parallelized" attribute for parallelized OpenACC kernels.
15183 * omp-offload.c (execute_oacc_device_lower): Use it.
15185 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
15186 Set "oacc kernels" attribute.
15187 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
15188 parameter. Adjust all users.
15189 (oacc_fn_attrib_kernels_p): Remove function.
15190 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
15191 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
15192 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
15193 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
15194 assert "oacc kernels" attribute is set.
15196 2017-05-11 Carl Love <cel@us.ibm.com>
15198 * config/rs6000/rs6000-c: Add support for built-in functions
15199 vector unsigned char vec_popcnt (vector signed char)
15200 vector unsigned char vec_popcnt (vector unsigned char)
15201 vector unsigned short vec_popcnt (vector signed short)
15202 vector unsigned short vec_popcnt (vector unsigned short)
15203 vector unsigned int vec_popcnt (vector signed int)
15204 vector unsigned int vec_popcnt (vector unsigned int)
15205 vector unsigned long long vec_popcnt (vector signed long long)
15206 vector unsigned long long vec_popcnt (vector unsigned long long)
15207 vector signed long long vec_slo (vector signed long long,
15208 vector signed char)
15209 vector signed long long vec_slo (vector signed long long,
15210 vector unsigned char)
15211 vector unsigned long long vec_slo (vector unsigned long long,
15212 vector signed char)
15213 vector unsigned long long vec_slo (vector unsigned long long,
15214 vector unsigned char)
15215 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
15216 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
15217 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
15218 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
15219 * doc/extend.texi: Update the built-in documentation file for the
15220 new built-in functions.
15222 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
15224 * attribs.h (sorted_attr_string): Move machine independent
15225 functions for target clone support from the i386 port to common
15226 code. Rename ix86_function_versions to common_function_versions.
15227 Rename make_name to make_unique_name.
15228 (common_function_versions): Likewise.
15229 (make_unique_name): Likewise.
15230 (make_dispatcher_decl): Likewise.
15231 (is_function_default_version): Likewise.
15232 * attribs.c (attr_strcmp): Likewise.
15233 (sorted_attr_string): Likewise.
15234 (common_function_versions): Likewise.
15235 (make_unique_name): Likewise.
15236 (make_dispatcher_decl): Likewise.
15237 (is_function_default_version): Likewise.
15238 * config/i386/i386.c (attr_strcmp): Likewise.
15239 (sorted_attr_string): Likewise.
15240 (ix86_function_versions): Likewise.
15241 (make_name): Likewise.
15242 (make_dispatcher_decl): Likewise.
15243 (is_function_default_version): Likewise.
15244 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
15246 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15249 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15250 Account for direct move costs for vec_construct of integer
15253 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
15256 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
15257 (UNSPEC_STX_ATOMIC): Ditto.
15258 (loaddi_via_sse): New insn.
15259 (storedi_via_sse): Ditto.
15260 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
15261 Update corresponding peephole2 patterns.
15262 (atomic_storedi_fpu): Ditto.
15264 2017-05-11 Julia Koval <julia.koval@intel.com>
15266 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
15267 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
15269 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
15270 (__builtin_ia32_rsqrt14ss_mask): New builtins.
15271 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
15273 2017-05-11 Nathan Sidwell <nathan@acm.org>
15275 * graphite-poly.c: Include dumpfile.h.
15277 * dumpfle.h (dump_function): Declare here ...
15278 * tree-dump.h (dump_function): ... not here.
15279 * dumpfile.c: #include tree-cfg.h.
15280 (dump_function): Move here from ...
15281 * tree-dump.c (dump_function): ... here.
15282 * gimplify.c: #include splay-tree.h, not tree-dump.h.
15283 * graphite-poly.c: Don't include tree-dump.h.
15284 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
15285 * print-tree.c: Likewise.
15286 * stor-layout.c: Likewise.
15287 * tree-nested.c: Likewise.
15289 * dumpfile.c (dump_start): Use TDF_FLAGS.
15290 (dump_enable_all): Fix TDF_KIND check thinko.
15292 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
15294 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15295 array entries to represent two legal parameterizations of the
15296 overloaded __builtin_cmpb function, as represented by the
15297 P6_OV_BUILTIN_CMPB constant.
15298 (altivec_resolve_overloaded_builtin): Add special case handling
15299 for the __builtin_cmpb function, as represented by the
15300 P6_OV_BUILTIN_CMPB constant.
15301 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
15302 (BU_P6_64BIT_2): New macro.
15303 (BU_P6_OVERLOAD_2): New macro
15304 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
15305 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
15306 (CMPB): Add overload support to represent both 32-bit and 64-bit
15307 compare-bytes function.
15308 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15309 support for TARGET_CMPB.
15310 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
15311 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15312 documentation of the __builtin_cmpb overloaded built-in function.
15314 2017-05-11 Richard Biener <rguenther@suse.de>
15316 PR tree-optimization/80705
15317 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
15318 bases are not vectorizable.
15320 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15322 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
15323 when counting register pressure.
15325 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15327 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
15328 (iv_ca_more_deps): Renamed to ...
15329 (iv_ca_compare_deps): ... this.
15330 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
15332 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15334 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
15336 (determine_group_iv_costs): ... here.
15337 (find_inv_vars_cb): Record inv var if it's not recorded before.
15339 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15341 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
15342 (get_shiftadd_cost): Ditto.
15344 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15346 * tree-ssa-address.c: Include header file.
15347 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
15349 (add_to_parts): Refactor.
15350 (addr_to_parts): New parameter. Update use of move_hint_to_base.
15351 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
15354 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15356 PR tree-optimization/53090
15357 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
15359 (extract_cond_operands): Detect condition with IV on both sides
15360 and return COMP_IV_EXPR_2.
15361 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
15362 (rewrite_use_compare): Simplify by removing call to function
15363 extract_cond_operands.
15365 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15367 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
15368 (extract_cond_operands): Detect condition comparing against non-
15369 invariant bound and return appropriate enum value.
15370 (find_interesting_uses_cond): Update use of extract_cond_operands.
15371 Handle its return value accordingly.
15372 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
15374 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15376 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
15377 nonlinear iv_use computation in loop invariant sensitive way.
15379 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15381 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
15382 (find_iv_candidates): Call relate_compare_use_with_all_cands.
15384 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15386 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
15387 (dump_cand): Support iv_cand.inv_exprs.
15388 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
15390 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
15393 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15395 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
15397 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
15398 as local function. Include necessary header files.
15399 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
15401 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15403 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
15405 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15407 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
15408 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
15409 RSHIFT_EXPR and BIT_NOT_EXPR.
15411 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15413 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
15414 (adjust_setup_cost): New parameter supporting round up adjustment.
15415 (struct address_cost_data): Delete.
15416 (force_expr_to_var_cost): Don't bound cost with spill_cost.
15417 (split_address_cost, ptr_difference_cost): Delete.
15418 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
15419 (struct ainc_cost_data): New struct.
15420 (get_address_cost_ainc): New function.
15421 (get_address_cost, get_computation_cost): Reimplement.
15422 (determine_group_iv_cost_address): Record inv_expr for all uses of
15424 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
15425 (iv_ca_has_deps): Reimplemented to ...
15426 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
15428 (iv_ca_extend): Call iv_ca_more_deps.
15430 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15432 * tree-ssa-address.c (struct mem_address): Move to header file.
15433 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
15434 * tree-ssa-address.h (struct mem_address): Move from C file.
15435 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
15437 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15439 * tree-affine.h (aff_combination_type): New interface.
15440 (aff_combination_zero_p): Remove static.
15441 (aff_combination_const_p): New interface.
15442 (aff_combination_singleton_var_p): New interfaces.
15444 2017-05-11 Richard Biener <rguenther@suse.de>
15446 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15447 Skip unreachable blocks and destinations.
15448 (eliminate): Move stmt removal and fixup ...
15449 (fini_eliminate): ... here. Skip inserted exprs.
15450 (pass_pre::execute): Move fini_pre after fini_eliminate.
15451 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
15452 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
15453 PRE to get rid of dead code that has invalid SSA form and
15454 split critical edges again.
15456 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15458 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15460 2017-05-11 Richard Biener <rguenther@suse.de>
15462 * passes.c (execute_function_todo): Verify loops if they are
15463 said to be up-to-date.
15464 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
15465 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
15467 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
15470 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
15471 handle calling assemble_external ourself.
15474 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
15475 modes with zero size. Enhance comment.
15477 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15479 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
15480 built-ins for vec_xl and vec_xst with short and char pointer
15483 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15485 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
15486 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
15487 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
15488 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
15489 (_mm_maskz_min_round_ss): New intrinsics.
15490 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
15491 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
15492 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
15493 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
15494 (__builtin_ia32_minss_mask_round): New builtins.
15495 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15496 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15497 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
15499 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
15500 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
15502 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
15505 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15507 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
15508 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
15509 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
15510 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
15511 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
15512 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
15513 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
15514 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15515 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
15516 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
15517 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
15518 (__builtin_ia32_mulss_mask_round): New builtins.
15519 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15520 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15521 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
15523 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
15524 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
15526 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
15529 2017-05-10 Julia Koval <julia.koval@intel.com>
15531 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
15532 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
15533 (_mm256_setr_m128i): New intrinsics.
15535 2017-05-10 Julia Koval <julia.koval@intel.com>
15537 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
15538 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
15539 (_mm_maskz_rcp14_ss): New intrinsics.
15540 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
15541 (__builtin_ia32_rcp14ss_mask): New builtins.
15542 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
15544 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
15546 PR tree-optimization/51513
15547 * tree-cfg.c (gimple_seq_unreachable_p): New function.
15548 (assert_unreachable_fallthru_edge_p): Use it.
15549 (group_case_labels_stmt): Likewise.
15550 * tree-cfg.h: Prototype it.
15551 * stmt.c: Include cfghooks.h and tree-cfg.h.
15552 (emit_case_dispatch_table) <gap_label>: New local variable.
15553 Use it to fill dispatch table gaps.
15554 Test for default_label before updating probabilities.
15555 (expand_case) <default_label>: Remove unneeded initialization.
15556 Test for unreachable default case statement and remove its edge.
15557 Set default_label accordingly.
15558 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
15560 2017-05-10 Carl Love <cel@us.ibm.com>
15562 * config/rs6000/rs6000-c: Add support for built-in functions
15563 vector signed char vec_neg (vector signed char)
15564 vector signed short int vec_neg (vector short int)
15565 vector signed int vec_neg (vector signed int)
15566 vector signed long long vec_neg (vector signed long long)
15567 vector float vec_neg (vector float)
15568 vector double vec_neg (vector double)
15569 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
15571 * config/rs6000/altivec.h: Add define for vec_neg
15572 * doc/extend.texi: Update the built-in documentation for the
15573 new built-in functions.
15575 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15577 PR tree-optimization/77644
15578 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
15580 2017-05-10 Nathan Sidwell <nathan@acm.org>
15582 * dumpfile.h (TDI_lang_all): New.
15583 (TDF_KIND): New. Renumber others
15584 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
15586 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
15588 (get_dump_file_name): Adjust suffix generation.
15589 (dump_enable_all): Use TDF_KIND.
15590 * doc/invoke.texi (-fdump-lang-all): Document.
15592 * dumpfile.h: Tabify.
15594 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
15597 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
15598 Move member access before delete.
15600 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
15602 * tree-inline.c (expand_call_inline): Split block at stmt
15605 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
15608 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
15609 are now unused after splitting mov{sf,sd}_hardfloat.
15610 (f32_lr2): Likewise.
15611 (f32_lm): Likewise.
15612 (f32_lm2): Likewise.
15613 (f32_li): Likewise.
15614 (f32_li2): Likewise.
15615 (f32_lv): Likewise.
15616 (f32_sr): Likewise.
15617 (f32_sr2): Likewise.
15618 (f32_sm): Likewise.
15619 (f32_sm2): Likewise.
15620 (f32_si): Likewise.
15621 (f32_si2): Likewise.
15622 (f32_sv): Likewise.
15623 (f32_dm): Likewise.
15624 (f32_vsx): Likewise.
15625 (f32_av): Likewise.
15626 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
15627 For movsf, order stores so the VSX stores occur before the GPR
15628 store which encourages the register allocator to use a traditional
15629 FPR instead of a GPR. For movsd, order the stores so that the GPR
15630 store comes before the VSX stores to allow the power6 to work.
15631 This is due to the power6 not having a 32-bit integer store
15632 instruction from a FPR.
15633 (movsf_hardfloat): Likewise.
15634 (movsd_hardfloat): Likewise.
15636 2017-05-09 Martin Sebor <msebor@redhat.com>
15638 PR translation/80280
15639 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
15642 PR translation/80280
15643 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
15644 data member added in r247778.
15645 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
15647 2017-05-09 Nathan Sidwell <nathan@acm.org>
15649 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
15651 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
15654 2017-05-09 Marek Polacek <polacek@redhat.com>
15656 * doc/invoke.texi: Fix typo.
15658 2017-05-09 Richard Biener <rguenther@suse.de>
15660 * tree-vrp.c (vrp_val_is_max): Adjust comment.
15661 (vrp_val_is_min): Likewise.
15662 (set_value_range_to_value): Likewise.
15663 (set_value_range_to_nonnegative): Likewise.
15664 (gimple_assign_nonzero_p): Likewise.
15665 (gimple_stmt_nonzero_p): Likewise.
15666 (vrp_int_const_binop): Likewise. Remove unreachable case.
15667 (adjust_range_with_scev): Adjust comments.
15668 (compare_range_with_value): Likewise.
15669 (extract_range_from_phi_node): Likewise.
15670 (test_for_singularity): Likewise.
15672 2017-05-09 Richard Biener <rguenther@suse.de>
15674 * tree-vrp.c (get_single_symbol): Add assert that we don't
15675 get overflowed constants as invariant part.
15676 (compare_values_warnv): Add comment before the TREE_NO_WARNING
15677 checks. Use wi::cmp instead of recursing for integer constants.
15678 (compare_values): Just ignore whether we assumed undefined
15679 overflow instead of failing the compare.
15680 (extract_range_for_var_from_comparison_expr): Add comment before the
15681 TREE_NO_WARNING sets.
15682 (test_for_singularity): Likewise.
15683 (extract_range_from_comparison): Do not disable optimization
15684 when we assumed undefined overflow.
15685 (extract_range_basic): Remove init of unused var.
15687 2017-05-09 Richard Biener <rguenther@suse.de>
15689 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
15690 (extract_range_from_multiplicative_op_1): Adjust.
15691 (extract_range_from_binary_expr_1): Use int_const_binop.
15693 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
15696 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
15697 rs6000_store_data_bypass_p in seven define_bypass directives and
15698 in several comments.
15699 * config/rs6000/rs6000-protos.h: Add prototype for
15700 rs6000_store_data_bypass_p function.
15701 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
15702 function implements slightly different (rs6000-specific) semantics
15703 than store_data_bypass_p, returning false rather than aborting
15704 with assertion error when arguments do not satisfy the
15705 requirements of store data bypass.
15706 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
15707 rs6000_store_data_bypass_p.
15709 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
15711 * config/xtensa/xtensa-protos.h
15712 (xtensa_initial_elimination_offset): New declaration.
15713 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15714 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
15715 macro definition, add case for FRAME_POINTER_REGNUM when
15716 FRAME_GROWS_DOWNWARD.
15717 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
15718 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
15719 xtensa_initial_elimination_offset.
15721 2017-05-08 Nathan Sidwell <nathan@acm.org>
15723 * doc/invoke.texi: Alphabetize -fdump options.
15725 2017-05-08 Martin Sebor <msebor@redhat.com>
15727 PR translation/80280
15728 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
15730 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
15732 * target.def (compute_frame_layout): New optional target hook.
15733 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
15734 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
15735 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
15737 * reload1.c (verify_initial_elim_offsets): Likewise.
15738 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
15739 (use_simple_return_p): Call arm_compute_frame_layout if needed.
15740 (arm_get_frame_offsets): Split up into this ...
15741 (arm_compute_frame_layout): ... and this function.
15743 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
15745 * config/aarch64/constraints.md (Usa): New constraint.
15746 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
15748 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
15750 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
15751 with_multilib_list after it has been checked.
15753 2017-05-08 Richard Biener <rguenther@suse.de>
15755 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
15756 (bitmap_set_subtract_values): Likewise.
15758 2017-05-08 Richard Biener <rguenther@suse.de>
15760 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
15761 (gimple_assign_nonzero): ... this and remove strict_overflow_p
15763 (gimple_stmt_nonzero_warnv_p): Rename to ...
15764 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
15766 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
15767 (extract_range_basic): Adjust, do not disable propagation on
15768 strict overflow sensitive simplification.
15769 (vrp_visit_cond_stmt): Likewise.
15771 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
15773 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
15774 body size unconditionally.
15776 2017-05-07 Jeff Law <law@redhat.com>
15779 2017-05-06 Jeff Law <law@redhat.com>
15780 PR tree-optimization/78496
15781 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15784 PR tree-optimization/78496
15785 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15786 (simplify_stmt_using_ranges): Call it.
15787 (vrp_dom_walker::before_dom_children): Extract equivalences
15788 from an ASSERT_EXPR with an equality comparison against a
15791 2017-05-06 Jeff Law <law@redhat.com>
15793 PR tree-optimization/78496
15794 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15797 PR tree-optimization/78496
15798 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15799 (simplify_stmt_using_ranges): Call it.
15800 (vrp_dom_walker::before_dom_children): Extract equivalences
15801 from an ASSERT_EXPR with an equality comparison against a
15804 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15806 * lra-constraints.c (lra_copy_reg_equiv): New function.
15807 (split_reg): Use it to copy equivalence information from the
15808 original register to the spill register.
15810 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15812 PR rtl-optimization/75964
15813 * simplify-rtx.c (simplify_const_relational_operation): Remove
15814 invalid handling of comparisons of integer ABS.
15816 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
15818 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
15819 initialize to zero.
15820 (init_regs): Remove declaration.
15821 (function_arg_advance_32): Initialize error_p as boolean variable.
15823 2017-05-05 Nathan Sidwell <nathan@acm.org>
15825 * store-motion.c (remove_reachable_equiv_notes): Reformat long
15826 lines. Use for (;;).
15828 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15830 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
15831 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
15832 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
15833 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
15834 VF=2 that require versioning.
15836 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15838 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
15841 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15843 * diagnostic.h (diagnostic_override_option_index): Convert from
15844 macro to inline function.
15846 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15848 * diagnostic.c (last_module_changed_p): New function.
15849 (set_last_module): New function.
15850 (diagnostic_report_current_module): Convert macro usage to
15851 the above functions.
15852 * diagnostic.h (diagnostic_context::last_module): Strengthen
15853 from const line_map * to const line_map_ordinary *.
15854 (diagnostic_last_module_changed): Delete macro.
15855 (diagnostic_set_last_module): Delete macro.
15857 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15859 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
15860 with diagnostic_report_diagnostic.
15861 (diagnostic_n_impl_richloc): Likewise.
15862 * diagnostic.h (report_diagnostic): Delete macro.
15863 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
15864 with diagnostic_report_diagnostic.
15865 * substring-locations.c (format_warning_va): Likewise.
15867 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15869 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
15870 save/restor of format_spec. Move option-printing code to...
15871 (print_option_information): ...this new function, and
15872 reimplement by simply printing to the pretty_printer,
15873 rather than appending to the format string.
15875 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15877 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
15878 handling logic into...
15879 (update_effective_level_from_pragmas): ...this new function.
15881 2017-05-04 Andrew Waterman <andrew@sifive.com>
15883 * config/riscv/riscv.opt (mstrict-align): New option.
15884 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
15885 (SLOW_UNALIGNED_ACCESS): Define.
15886 (riscv_slow_unaligned_access): Declare.
15887 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
15889 (riscv_slow_unaligned_access): New variable.
15890 (rocket_tune_info): Set slow_unaligned_access to true.
15891 (optimize_size_tune_info): Set slow_unaligned_access to false.
15892 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
15893 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
15894 (riscv_option_override): Set riscv_slow_unaligned_access.
15895 * doc/invoke.texi: Add -mstrict-align to RISC-V.
15897 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
15899 * config/riscv/riscv.md: Unify indentation.
15901 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15906 * config/rs6000/rs6000.md (u code attribute): Add FIX and
15908 (extendsi<mode>2): Add support for doing sign extension via
15909 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
15910 don't have ISA 3.0 instructions.
15911 (extendsi<mode>2 splitter): Likewise.
15912 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
15913 generate the normal insns since SImode can now go in vector
15914 registers. Disallow the special UNSPECs needed for previous
15915 machines to hide SImode being used. Add new insns
15916 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
15917 (fix_trunc<mode>si2_stfiwx): Likewise.
15918 (fix_trunc<mode>si2_internal): Likewise.
15919 (fixuns_trunc<mode>si2): Likewise.
15920 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15921 (fctiw<u>z_<mode>_smallint): Likewise.
15922 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
15923 of floating point to 32-bit integer from doing a direct move to
15924 the GPR registers to do a store.
15925 (fctiwz_<mode>): Break long line.
15927 2017-05-05 Bin Cheng <bin.cheng@arm.com>
15929 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
15930 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
15931 (addr_list, addr_offset_valid_p): New.
15932 (split_address_groups): Check offset validity with above function.
15933 (gt-tree-ssa-loop-ivopts.h): Include header file.
15935 2017-05-05 Nathan Sidwell <nathan@acm.org>
15937 * config.gcc (arm*-*-*): Add missing 'fi'.
15939 2017-05-05 Steve Ellcey <sellcey@cavium.com>
15941 * doc/invoke.texi (-fopt-info): Explicitly say order of options
15942 included in -fopt-info does not matter.
15943 * doc/optinfo.texi (-fopt-info): Fix description of default
15944 behavour. Explicitly say order of options included in -fopt-info
15947 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
15949 * config.gcc: Allow combinations of aprofile and rmprofile values for
15950 --with-multilib-list.
15951 * config/arm/t-multilib: New file.
15952 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
15953 variables. Remove setting of ISA and floating-point ABI in
15954 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
15955 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
15956 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
15957 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
15959 * config/arm/t-rmprofile: Likewise except for the matches changes.
15960 * doc/install.texi (--with-multilib-list): Document the combination of
15961 aprofile and rmprofile values and warn about pitfalls in doing that.
15963 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
15965 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
15966 (movdi_aarch64): Likewise.
15968 2017-05-05 Jakub Jelinek <jakub@redhat.com>
15970 PR tree-optimization/80632
15971 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
15973 (build_arrays): Initialize it for virtual phis.
15974 (fix_phi_nodes): Use it for virtual phis.
15976 PR tree-optimization/80558
15977 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
15978 [x, y] op z into [x op, y op z] for op & or | if conditions
15981 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
15982 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15985 * config/arm/arm.md (use_literal_pool): Remove.
15986 (64-bit immediate split): No longer takes cost into consideration
15987 if arm_disable_literal_pool is enabled.
15988 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
15989 used when arm_disable_literal_pool is enabled.
15990 (arm_max_const_double_inline_cost): Remove use of
15991 arm_disable_literal_pool.
15992 (push_minipool_fix): Add assert.
15993 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
15994 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
15995 (no_literal_pool_sf_immediate): New.
15997 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15999 PR tree-optimization/80613
16000 * tree-ssa-dce.c (propagate_necessity): Remove cases for
16001 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
16003 2017-05-05 Richard Biener <rguenther@suse.de>
16005 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
16007 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
16009 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
16010 of this flag from insn conditions due to removal from r247495.
16012 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16014 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
16016 (arm_early_store_addr_dep_ptr): Likewise.
16017 * config/arm/aarch-common-protos.h
16018 (arm_early_load_addr_dep_ptr): Add prototype.
16019 (arm_early_store_addr_dep_ptr): Likewise.
16020 * config/arm/cortex-a53.md: Add new bypasses.
16022 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16024 * tree.c (next_type_uid): Change type to unsigned.
16025 (type_hash_canon): Decrement back next_type_uid if
16026 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
16027 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
16030 2017-05-04 Martin Sebor <msebor@redhat.com>
16032 * builtins.c: Fix a trivial typo in a comment.
16034 PR middle-end/79234
16035 * builtins.c (check_sizes): Adjust to handle reading past the end.
16036 Avoid printing excessive upper bound of ranges. Use %E to print
16037 tree nodes instead of converting them to %wu.
16038 (expand_builtin_memchr): New function.
16039 (compute_dest_size): Rename...
16040 (compute_objsize): ...to this.
16041 (expand_builtin_memcpy): Adjust.
16042 (expand_builtin_mempcpy): Adjust.
16043 (expand_builtin_strcat): Adjust.
16044 (expand_builtin_strcpy): Adjust.
16045 (check_strncat_sizes): Adjust.
16046 (expand_builtin_strncat): Adjust.
16047 (expand_builtin_strncpy): Adjust and simplify.
16048 (expand_builtin_memset): Adjust.
16049 (expand_builtin_bzero): Adjust.
16050 (expand_builtin_memcmp): Adjust.
16051 (expand_builtin): Handle memcmp.
16052 (maybe_emit_chk_warning): Check strncat just once.
16054 2017-05-04 Martin Sebor <msebor@redhat.com>
16056 PR preprocessor/79214
16057 PR middle-end/79222
16058 PR middle-end/79223
16059 * builtins.c (check_sizes): Add inlining context and issue
16060 warnings even when -Wno-system-headers is set.
16061 (check_strncat_sizes): Same.
16062 (expand_builtin_strncat): Same.
16063 (expand_builtin_memmove): New function.
16064 (expand_builtin_stpncpy): Same.
16065 (expand_builtin): Handle memmove and stpncpy.
16067 2017-05-04 Bin Cheng <bin.cheng@arm.com>
16069 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
16070 which is not used any more.
16072 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16074 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
16076 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16078 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
16079 (cortexa53_tunings): Likewise.
16080 (cortexa57_tunings): Likewise.
16081 (cortexa72_tunings): Likewise.
16082 (cortexa73_tunings): Likewise.
16084 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16086 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
16087 Set loop alignment to 8.
16089 2017-05-04 Martin Sebor <msebor@redhat.com>
16091 PR translation/80280
16092 * builtins.c (expand_builtin_object_size): Add missing quoting to
16093 %D and like directives.
16094 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
16095 (hsa_type_for_tree_type): Same.
16096 (verify_function_arguments): Same.
16097 * symtab.c (symbol_table::change_decl_assembler_name): Same.
16098 * varasm.c (get_section): Same.
16101 2017-05-04 Martin Sebor <msebor@redhat.com>
16103 PR translation/80280
16104 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
16106 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16108 * config/aarch64/aarch64.c (generic_addrcost_table):
16109 Change HI/TI mode setting.
16111 2017-05-04 Martin Jambor <mjambor@suse.cz>
16113 PR tree-optimization/80622
16114 * tree-sra.c (comes_initialized_p): New function.
16115 (build_accesses_from_assign): Only set write lazily when
16116 comes_initialized_p is false.
16117 (analyze_access_subtree): Use comes_initialized_p.
16118 (propagate_subaccesses_across_link): Assert !comes_initialized_p
16119 instead of testing for PARM_DECL.
16121 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16123 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
16124 constraint on operand 0 to allow more general addressing modes.
16125 Adjust output template.
16126 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
16128 * config/aarch64/aarch64-protos.h
16129 (aarch64_address_valid_for_prefetch_p): Declare prototype.
16130 * config/aarch64/constraints.md (Dp): New address constraint.
16131 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
16134 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
16136 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
16137 update use of estimate_ipcp_clone_size_and_time.
16138 (estimate_local_effects): Update use of
16139 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
16140 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
16141 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
16142 Return nonspecialized time.
16144 2017-05-04 Richard Biener <rguenther@suse.de>
16146 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
16147 for the last VUSE which def dominates the PHI. Directly call
16149 (get_continuation_for_phi_1): Remove.
16151 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
16153 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
16154 to explain the use of truncating division. Cap the number of
16155 iterations to the maximum given by nb_iterations_upper_bound,
16158 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16160 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
16161 * configure: Regenerate.
16162 * config.in: Regenerate.
16163 * config/i386/driver-mingw32.c: new file.
16164 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
16165 * config.host: Link driver-mingw32.o on MinGW host.
16166 * doc/install.texi: Document new --enable-mingw-wildcard configure
16169 2017-05-04 Marek Polacek <polacek@redhat.com>
16171 PR tree-optimization/80612
16172 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
16174 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16175 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
16177 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
16178 (movt splitter): Likewise.
16179 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
16180 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
16181 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
16182 block for Thumb-1 with MOVT.
16183 (thumb2_legitimate_address_p): Move code block ...
16184 (can_avoid_literal_pool_for_label_p): ... into this new function.
16185 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
16187 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
16188 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
16189 "M-profile targets with the MOVT instruction".
16191 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16193 * config/arm/arm-builtins.c (arm_init_builtins): Rename
16194 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
16195 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
16197 2017-05-04 Martin Liska <mliska@suse.cz>
16199 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
16200 variable cond_code.
16202 2017-05-04 Richard Biener <rguenther@suse.de>
16204 * tree.c (array_at_struct_end_p): Handle arrays at struct
16205 end with flexarrays more conservatively. Refactor and treat
16206 arrays of arrays or aggregates more strict. Fix
16207 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
16208 * tree.c (array_at_struct_end_p): Adjust prototype.
16209 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
16210 * gimple-fold.c (get_range_strlen): Likewise.
16211 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
16213 2017-05-04 Richard Biener <rguenther@suse.de>
16215 PR tree-optimization/31130
16216 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
16218 (supports_overflow_infinity): Likewise.
16219 (is_negative_overflow_infinity): Likewise.
16220 (is_positive_overflow_infinity): Likewise.
16221 (is_overflow_infinity): Likewise.
16222 (stmt_overflow_infinity): Likewise.
16223 (overflow_infinity_range_p): Likewise.
16224 (usable_range_p): Remove as always returning true.
16225 (make_overflow_infinity): Remove.
16226 (negative_overflow_infinity): Likewise.
16227 (positive_overflow_infinity): Likewise.
16228 (avoid_overflow_infinity): Likewise.
16229 (set_value_range): Adjust accordingly.
16230 (set_value_range_to_nonnegative): Likewise, remove now unused
16231 overflow_infinity arg.
16232 (vrp_operand_equal_p): Adjust.
16233 (update_value_range): Likewise.
16234 (range_int_cst_singleton_p): Likewise.
16235 (operand_less_p): Likewise.
16236 (compare_values_warnv): Likewise.
16237 (extract_range_for_var_from_comparison_expr): Likewise.
16238 (vrp_int_const_binop): Likewise.
16239 (zero_nonzero_bits_from_vr): Likewise.
16240 (extract_range_from_multiplicative_op_1): Likewise.
16241 (extract_range_from_binary_expr_1): Likewise.
16242 (extract_range_from_unary_expr): Likewise.
16243 (extract_range_from_comparison): Likewise.
16244 (extract_range_basic): Likewise.
16245 (adjust_range_with_scev): Likewise.
16246 (compare_ranges): Likewise.
16247 (compare_range_with_value): Likewise.
16248 (dump_value_range): Likewise.
16249 (test_for_singularity): Likewise, remove strict_overflow_p parameter
16251 (simplify_cond_using_ranges): Adjust.
16253 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16255 * brig-builtins.def: Added a builtin for class_f64.
16256 * builtin-types.def: Added a builtin type needed by class_f64.
16258 2017-05-03 Jason Merrill <jason@redhat.com>
16260 * timevar.def: Add TV_CONSTEXPR.
16262 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16264 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
16266 2017-05-03 Martin Jambor <mjambor@suse.cz>
16268 * ipa-prop.c (ipa_update_after_lto_read): Removed.
16269 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
16270 * ipa-cp.c (ipcp_propagate_stage): Do not call
16271 ipa_update_after_lto_read.
16272 * ipa-inline.c (ipa_inline): Likewise.
16274 2017-05-03 Martin Jambor <mjambor@suse.cz>
16276 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
16277 tag. Added a default constructor and a destructor.
16278 (ipa_edge_args_sum_t): New class;
16279 (ipa_edge_args_sum): Declare.
16280 (ipa_edge_args_vector): Remove declaration.
16281 (IPA_EDGE_REF): Use ipa_edge_args_sum.
16282 (ipa_free_edge_args_substructures): Remove declaration.
16283 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
16284 (ipa_edge_args_info_available_for_edge_p): Likewise.
16285 * ipa-prop.c (ipa_edge_args_vector): Removed.
16286 (edge_removal_hook_holder): Likewise.
16287 (edge_duplication_hook_holder): Likewise.
16288 (ipa_edge_args_sum): New variable.
16289 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
16290 ipa_edge_args_vector.
16291 (ipa_free_edge_args_substructures): Likewise.
16292 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
16293 ipa_edge_args_vector.
16294 (ipa_edge_removal_hook): Turned into method
16295 ipa_edge_args_sum_t::remove.
16296 (ipa_edge_duplication_hook): Turned into method
16297 ipa_edge_args_sum_t::duplicate.
16298 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
16299 registering edge hooks.
16300 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
16301 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
16302 ipa_edge_args_sum instead of ipa_edge_args_vector.
16303 * ipa-profile.c (ipa_profile): Likewise.
16305 2017-05-03 Martin Jambor <mjambor@suse.cz>
16307 * symbol-summary.h (function_summary): New method exists.
16308 (function_summary::symtab_removal): Deallocate through release.
16309 (call_summary): New class.
16310 (gt_ggc_mx): New overload.
16311 (gt_pch_nx): Likewise.
16312 (gt_pch_nx): Likewise.
16314 2017-05-03 Jeff Law <law@redhat.com>
16316 PR tree-optimization/78496
16317 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
16318 from simplify_cond_using_ranges. Split off code to walk
16319 backwards through casts into ...
16320 (simplify_cond_using_ranges_2): New function.
16321 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
16322 (execute_vrp): After identifying jump threads, call
16323 simplify_cond_using_ranges_2.
16325 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
16328 * ipa-inline.h (inline_summary): Add ctor.
16329 (create_ggc): Do not use ggc_cleared_alloc.
16331 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
16332 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16334 * gcc.c (handle_braces): Support escaping in switch matching
16336 * doc/invoke.texi (Spec Files): Document it.
16337 Remove superfluous @code markup in items.
16339 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16341 * diagnostic-show-locus.c (struct column_range): New struct.
16342 (get_affected_columns): New function.
16343 (get_printed_columns): New function.
16344 (struct correction): New struct.
16345 (correction::ensure_capacity): New function.
16346 (correction::ensure_terminated): New function.
16347 (struct line_corrections): New struct.
16348 (line_corrections::~line_corrections): New dtor.
16349 (line_corrections::add_hint): New function.
16350 (layout::print_trailing_fixits): Reimplement in terms of the new
16352 (selftest::test_overlapped_fixit_printing): New function.
16353 (selftest::diagnostic_show_locus_c_tests): Call it.
16355 2017-05-03 Nathan Sidwell <nathan@acm.org>
16357 Canonicalize canonical type hashing
16358 * tree.h (type_hash_canon_hash): Declare.
16359 * tree.c (type_hash_list, attribute_hash_list): Move into
16360 type_hash_canon_hash.
16361 (build_type_attribute_qual_variant): Break out hash code calc into
16362 type_hash_canon_hash.
16363 (type_hash_canon_hash): New. Generic type hash computation.
16364 (build_range_type_1, build_array_type_1, build_function_type,
16365 build_method_type_directly, build_offset_type, build_complex_type,
16366 make_vector_type): Call it.
16368 2017-05-03 Richard Biener <rguenther@suse.de>
16370 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
16371 When all DRs have unknown misaligned do not always peel
16372 when there is a store but apply the same costing model as if
16373 there were only loads.
16375 2017-05-03 Richard Biener <rguenther@suse.de>
16378 PR tree-optimization/80492
16379 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16380 compare_base_decls returning dont-know properly.
16382 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
16384 * config/arm/iterators.md (CCSI): New mode iterator.
16385 (arch): New mode attribute.
16386 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
16387 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
16388 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
16389 code iterator for success result mode.
16390 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
16391 the corresponding new insn generators.
16393 2017-05-03 Bin Cheng <bin.cheng@arm.com>
16396 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16397 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16399 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
16401 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
16402 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
16403 (DDR_A): Wrap DDR argument in brackets.
16404 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
16405 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
16406 (DDR_REVERSED_P): Likewise.
16408 2017-05-03 Jakub Jelinek <jakub@redhat.com>
16410 PR tree-optimization/79472
16411 * tree-switch-conversion.c (struct switch_conv_info): Add
16412 contiguous_range and default_case_nonstandard fields.
16413 (collect_switch_conv_info): Compute contiguous_range and
16414 default_case_nonstandard fields, don't clear final_bb if
16415 contiguous_range and only the default case doesn't have the required
16417 (check_all_empty_except_final): Set default_case_nonstandard instead
16418 of failing if contiguous_range and the default case doesn't have empty
16420 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
16421 and only the default case doesn't have the required constants. Skip
16423 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
16424 if default_case_nonstandard.
16425 (build_constructors): Build constant 1 just once. Assert that default
16426 values aren't inserted in between cases if contiguous_range. Skip
16428 (build_arrays): Skip virtual phis.
16429 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
16430 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
16431 Handle virtual phis.
16432 (gen_inbound_check): Handle default_case_nonstandard case.
16433 (process_switch): Adjust check_final_bb caller. Call
16434 gather_default_values with the first non-default case instead of
16435 default case if default_case_nonstandard.
16437 2017-05-02 Nathan Sidwell <nathan@acm.org>
16439 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
16440 check. Fix formatting.
16442 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
16444 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
16445 errors when comparing specialized and unspecialized times.
16447 2017-05-02 David Malcolm <dmalcolm@redhat.com>
16449 * diagnostic-show-locus.c
16450 (layout::should_print_annotation_line_p): Make private.
16451 (layout::print_annotation_line): Make private.
16452 (layout::annotation_line_showed_range_p): Make private.
16453 (layout::show_ruler): Make private.
16454 (layout::print_source_line): Make private. Pass in line and
16455 line_width, rather than calling location_get_source_line. Drop
16457 (layout::print_leading_fixits): New method.
16458 (layout::print_any_fixits): Rename to...
16459 (layout::print_trailing_fixits): ...this, and make private.
16460 Don't print newline fixits.
16461 (diagnostic_show_locus): Move logic for printing one row into...
16462 (layout::print_line): ...this new function. Move the
16463 location_get_source_line call and error-handling from
16464 print_source_line to here. Call print_leading_fixits, and rename
16465 print_any_fixits to print_trailing_fixits.
16466 (selftest::test_fixit_insert_containing_newline): Update now that
16467 newlines are partially supported.
16468 (selftest::test_fixit_insert_containing_newline_2): New test.
16469 (selftest::test_fixit_replace_containing_newline): Update comments.
16470 (selftest::diagnostic_show_locus_c_tests): Call the new test.
16471 * edit-context.c (class added_line): New class.
16472 (class edited_line): Describe newline handling in comment.
16473 (edited_line::actually_edited_p): New method.
16474 (edited_line::print_content): Delete redundant decl.
16475 (edited_line::m_predecessors): New field.
16476 (edited_file::print_content): Call edited_line::print_content.
16477 (edited_file::print_diff): Update to support newlines.
16478 (edited_file::print_diff_hunk): Likewise.
16479 (edited_file::print_run_of_changed_lines): New function.
16480 (edited_file::print_diff_line): Convert to...
16481 (print_diff_line): ...this.
16482 (edited_file::get_effective_line_count): New function.
16483 (edited_line::edited_line): Initialize new field m_predecessors.
16484 (edited_line::~edited_line): Clean up m_predecessors.
16485 (edited_line::apply_fixit): Handle newlines.
16486 (edited_line::get_effective_line_count): New function.
16487 (edited_line::print_content): New function.
16488 (edited_line::print_diff_lines): New function.
16489 (selftest::test_applying_fixits_insert_containing_newline): New
16491 (selftest::test_applying_fixits_replace_containing_newline): New
16493 (selftest::insert_line): New function.
16494 (selftest::test_applying_fixits_multiple_lines): Add example of
16496 (selftest::edit_context_c_tests): Call the new tests.
16498 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16500 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
16501 parameter cand. Update dump information.
16502 (get_computation_cost): Update uses.
16504 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16506 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
16507 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
16508 (get_computation_at, rewrite_use_address): Update use of
16509 get_computation_aff.
16511 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16513 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
16514 (get_computation): Delete.
16515 (get_computation_cost): Implement like get_computation_cost_at.
16516 Use get_computation_at.
16517 (get_computation_cost_at): Delete.
16518 (rewrite_use_nonlinear_expr): Use get_computation_at.
16519 (rewrite_use_compare, remove_unused_ivs): Ditto.
16521 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16523 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
16525 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16527 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
16528 (ivopts_global_cost_for_size): Rename parameter and update uses.
16529 (iv_ca_recount_cost): Update uses.
16530 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
16531 candidates seperately in n_invs and n_cands.
16532 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
16534 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16536 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
16537 (find_inv_vars_cb): New.
16538 (find_depends): Renamed to ...
16539 (find_inv_vars): ... this.
16540 (add_candidate_1, force_var_cost): Call find_inv_vars.
16541 (split_address_cost, determine_group_iv_cost_cond): Ditto.
16543 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16545 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
16546 inv_vars. Add inv_exprs.
16547 (struct iv_cand): Rename depends_on to inv_vars.
16548 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
16549 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
16550 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
16551 (dump_cand): Dump inv_vars.
16552 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
16553 (record_invariant, find_depends, add_candidate_1): Ditto.
16554 (set_group_iv_cost, force_var_cost): Ditto.
16555 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
16556 (get_computation_cost_at, get_computation_cost): Ditto.
16557 (determine_group_iv_cost_generic): Ditto.
16558 (determine_group_iv_cost_address): Ditto.
16559 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
16560 (determine_group_iv_costs): Ditto.
16561 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
16562 (iv_ca_set_remove_invariants): Renamed to ...
16563 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
16564 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
16565 (iv_ca_set_add_invariants): Renamed to ...
16566 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
16567 (iv_ca_set_cp): Use iv_ca_set_add_invs.
16568 (iv_ca_has_deps): Support inv_vars and inv_exprs.
16569 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
16570 (create_new_ivs): Remove useless dump.
16572 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16574 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
16576 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
16577 (iv_ca_set_no_cp, create_new_iv): Ditto.
16579 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16581 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16583 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16585 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
16586 function tree_check2.
16588 2017-05-02 Martin Liska <mliska@suse.cz>
16590 * doc/gcov.texi: Add missing preposition.
16591 * gcov.c (function_info::function_info): Properly fill up
16592 all member variables.
16594 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16596 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
16598 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16600 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
16602 2017-05-02 Martin Liska <mliska@suse.cz>
16605 * lto-streamer-in.c (lto_read_tree_1): Remove
16606 LTO_STREAMER_DEBUG.
16607 * lto-streamer.c (struct tree_hash_entry): Likewise.
16608 (struct tree_entry_hasher): Likewise.
16609 (tree_entry_hasher::hash): Likewise.
16610 (tree_entry_hasher::equal): Likewise.
16611 (lto_streamer_init): Likewise.
16612 (lto_orig_address_map): Likewise.
16613 (lto_orig_address_get): Likewise.
16614 (lto_orig_address_remove): Likewise.
16615 * lto-streamer.h: Likewise.
16616 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
16617 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
16619 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
16621 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
16622 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
16623 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
16624 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
16625 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
16626 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
16627 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
16628 (mm_maskz_sub_ss): New intrinsics.
16629 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16630 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16631 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
16632 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
16633 (__builtin_ia32_subss_mask_round): New builtins.
16634 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16635 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16636 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
16638 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
16639 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16641 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16644 2017-05-02 Martin Jambor <mjambor@suse.cz>
16646 PR tree-optimization/78687
16647 * tree-sra.c (access): New field parent.
16648 (process_subtree_disqualification): New function.
16649 (disqualify_candidate): Call it.
16650 (build_accesses_from_assign): Reset write flag if creating an
16652 (build_access_subtree): Fill in parent field and also prpagate
16653 down grp_write flag.
16654 (create_artificial_child_access): New parameter set_grp_write, set
16655 grp_write to its value.
16656 (propagate_subaccesses_across_link): Also propagate grp_write flag
16658 (propagate_all_subaccesses): Push the closest parent back to work
16659 queue if add_access_to_work_queue returned true.
16661 2017-05-02 Richard Biener <rguenther@suse.de>
16663 * common.opt (fstrict-overflow): Alias negative to fwrapv.
16664 * doc/invoke.texi (fstrict-overflow): Remove all traces of
16665 -fstrict-overflow documentation.
16666 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
16667 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
16668 flag_strict_overflow.
16669 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
16670 * lto-opts.c (lto_write_options): Do not stream it.
16671 * lto-wrapper.c (merge_and_complain): Do not handle it.
16672 * opts.c (default_options_table): Do not set -fstrict-overflow.
16673 (finish_options): Likewise do not clear it when sanitizing.
16674 * simplify-rtx.c (simplify_const_relational_operation): Do not
16675 test flag_strict_overflow.
16677 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16679 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
16680 using enabled attribute.
16681 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
16682 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
16683 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
16684 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
16685 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
16686 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
16687 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
16688 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
16689 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
16690 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
16692 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16694 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
16696 2017-05-02 Richard Biener <rguenther@suse.de>
16698 PR tree-optimization/80591
16700 2017-04-10 Richard Biener <rguenther@suse.de>
16702 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16705 2017-05-02 Richard Biener <rguenther@suse.de>
16707 PR tree-optimization/80549
16708 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
16709 (cleanup_tree_cfg_noloop): Create forwarders to known loop
16710 headers if they do not have a preheader.
16712 2017-05-02 Martin Liska <mliska@suse.cz>
16715 * common.opt: Fix typo.
16716 * doc/invoke.texi: Likewise.
16718 2017-05-01 Jan Beulich <jbeulich@suse.com>
16720 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
16721 swapping, add (x,x,m,x,n) alternative.
16723 2017-05-01 Nathan Sidwell <nathan@acm.org>
16725 * calls.c (combine_pending_stack_adjustment_and_call): Remove
16726 unnecessary unadjusted_alignment check.
16728 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16731 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
16733 * gimplify.c (gimplify_cilk_detach): New function.
16734 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
16735 * tree-core.h: Document EXPR_CILK_SPAWN.
16736 * tree.h (EXPR_CILK_SPAWN): Define.
16738 2017-05-01 David Malcolm <dmalcolm@redhat.com>
16740 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
16741 to use new fixit_hint representation, using the "replace" logic.
16742 (get_line_span_for_fixit_hint): Likewise.
16743 (layout::print_any_fixits): Likewise.
16744 (selftest::test_one_liner_many_fixits): Rename to...
16745 (selftest::test_one_liner_many_fixits_1): ...this, and update
16746 comment and expected output to reflect that the multiple fix-it
16747 hints are now consolidated into one insertion.
16748 (selftest::test_one_liner_many_fixits_2): New test.
16749 (selftest::test_diagnostic_show_locus_one_liner): Update for
16751 (selftest::test_fixit_consolidation): Update for fix-it API
16753 * diagnostic.c (print_parseable_fixits): Likewise.
16754 * edit-context.c (edited_line::m_line_events): Convert from
16755 auto_vec <line_event *> to auto_vec <line_event>.
16756 (class line_event): Convert from abstract base class to a concrete
16757 class, taking over the role of replace_event.
16758 (class insert_event): Delete.
16759 (class replace_event): Rename to class line_event. Convert to
16761 (edit_context::add_fixits): Reimplement.
16762 (edit_context::apply_insert): Delete.
16763 (edit_context::apply_replace): Rename to...
16764 (edit_context::apply_fixit): ...this. Convert to half-open range.
16765 (edited_file::apply_insert): Delete.
16766 (edited_file::apply_replace): Rename to...
16767 (edited_file::apply_fixit): ...this.
16768 (edited_line::~edited_line): Drop deletion of events.
16769 (edited_line::apply_insert): Delete.
16770 (edited_line::apply_replace): Rename to...
16771 (edited_line::apply_fixit): ...this. Convert to half-open range.
16772 Update for change to type of m_line_events.
16773 * edit-context.h (edit_context::apply_insert): Delete.
16774 (edit_context::apply_replace): Rename to...
16775 (edit_context::apply_fixit): ...this.
16777 2017-05-01 Martin Sebor <msebor@redhat.com>
16779 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
16782 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
16785 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
16786 __get_cpuid_max returns 0.
16787 (__get_cpuid_count): Ditto.
16789 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
16791 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
16792 replacement expression is another instance of one of its arguments.
16794 2017-05-01 Jakub Jelinek <jakub@redhat.com>
16797 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
16798 check for stack push/pop autoinc.
16799 * config/i386/i386.c (ix86_agi_dependent): Return false
16800 if the only reason why modified_in_p returned true is that
16801 addr is SP based and set_insn is a push or pop.
16803 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16805 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
16808 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16811 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
16812 (account_size_time): Use two predicates - exec_pred and
16814 (evaluate_conditions_for_known_args): Compute both clause and
16816 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
16817 (inline_summary_t::duplicate): Update.
16818 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
16820 (compute_inline_parameters): Likewise.
16821 (estimate_edge_size_and_time): Update caluclation of time.
16822 (estimate_node_size_and_time): Compute both time and nonspecialized
16824 (estimate_ipcp_clone_size_and_time): Update.
16825 (inline_merge_summary): Update.
16826 (do_estimate_edge_time): Update.
16827 (do_estimate_edge_size): Update.
16828 (do_estimate_edge_hints): Update.
16829 (inline_read_section, inline_write_summary): Stream both new predicates.
16830 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
16832 (compute_inlined_call_time): Cleanup.
16833 (big_speedup_p): Update.
16834 (edge_badness): Update.
16835 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
16836 (size_time_entry): Replace predicate by exec_predicate and
16837 nonconst_predicate.
16838 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
16839 (estimate_edge_time): Return also nonspec_time.
16840 (reset_edge_growth_cache): Update.
16842 2017-04-29 Jakub Jelinek <jakub@redhat.com>
16844 PR rtl-optimization/80491
16845 * ifcvt.c (noce_process_if_block): When looking for x setter
16846 with missing else_bb, don't check only the insn right before
16847 cond_earliest, but look for the last insn that x is modified in
16848 within the same bb.
16850 PR rtl-optimization/80491
16851 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
16853 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
16855 PR tree-optimization/80487
16856 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
16858 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16860 PR tree-optimization/79697
16861 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
16862 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
16863 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
16865 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
16866 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
16868 2017-04-28 Martin Sebor <msebor@redhat.com>
16870 PR tree-optimization/80523
16871 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
16872 (init_target_to_host_charmap, target_to_host, target_strtol10): New
16874 (maybe_warn, format_directive, parse_directive): Use new functions.
16875 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
16877 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
16879 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
16881 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
16883 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
16884 target_header_dir): Set correctly.
16885 * configure: Regenerated.
16886 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
16887 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
16888 instead of SYSTEM_HEADER_DIR.
16890 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
16892 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
16893 (estimate_local_effects): Likewise.
16894 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
16895 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
16896 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
16897 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
16898 do_estimate_edge_time, estimate_edge_time): Likewise.
16899 * ipa-inline-analysis.c (estimate_node_size_and_time,
16900 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
16901 (estimate_time_after_inlining): Remove.
16903 2017-04-28 Martin Liska <mliska@suse.cz>
16905 * doc/gcov.texi: Enhance documentation of gcov.
16907 2017-04-28 Martin Liska <mliska@suse.cz>
16909 * doc/gcov.texi: Sort options in alphabetic order.
16910 * doc/gcov-dump.texi: Likewise.
16911 * doc/gcov-tool.texi: Likewise.
16912 * gcov.c (print_usage): Likewise.
16913 * gcov-dump.c (print_usage): Likewise.
16914 * gcov-tool.c (print_merge_usage_message): Likewise.
16915 (print_rewrite_usage_message): Likewise.
16916 (print_overlap_usage_message): Likewise.
16918 2017-04-28 Martin Liska <mliska@suse.cz>
16920 PR gcov-profile/53915
16921 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
16923 2017-04-28 Martin Liska <mliska@suse.cz>
16925 PR gcov-profile/79891
16926 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
16927 is marked by compiler as living on a line.
16928 (get_cycles_count): Remove usage of the union.
16929 (output_intermediate_file): Likewise.
16930 (find_source): Fix GNU coding style.
16931 (accumulate_line_counts): Remove old non-all block mode.
16932 (output_lines): Remove usage of the union.
16933 * profile.c (output_location): Include all BBs, even if
16934 belonging to a same line (and file) as a previous BB.
16936 2017-04-28 Martin Liska <mliska@suse.cz>
16938 * gcov.c (process_args): Handle new argument 'w'.
16939 (read_graph_file): Assign ID to BBs.
16940 (output_branch_count): Display BB # if verbose flag is set.
16941 (output_lines): Likewise for arcs.
16942 (print_usage): Add '--verbose' option help.
16943 * doc/gcov.texi: Document --verbose (-w) option.
16945 2017-04-28 Martin Liska <mliska@suse.cz>
16947 * gcov.c (struct block_location_info): New struct.
16948 (process_file): Fill up the new structure.
16949 (read_graph_file): Replace usage of encoding by the newly added
16951 (add_line_counts): Likewise.
16952 (accumulate_line_counts): Remove usage of the union.
16953 (function_info::function_info): New function.
16954 (function_info::~function_info): Likewise.
16955 (process_file): Call delete instead of release_function.
16956 (release_function): Release the function.
16957 (release_structures): Call delete instead of release_function.
16958 (solve_flow_graph): Replace usage of num_blocks.
16959 (find_exception_blocks): Likewise.
16960 (output_lines): Fix GNU coding style.
16962 2017-04-28 Martin Liska <mliska@suse.cz>
16965 * coverage.c (coverage_remove_note_file): New function.
16966 * coverage.h: Declare the function.
16967 * toplev.c (finalize): Clean if an error has been seen.
16969 2017-04-28 Martin Liska <mliska@suse.cz>
16971 PR gcov-profile/80031
16972 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
16973 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
16974 * gcov.c (read_graph_file): Read just number of blocks.
16975 * profile.c (branch_prob): Do not stream 0 flags per a basic
16978 2017-04-28 Martin Liska <mliska@suse.cz>
16980 * gcov-dump.c (tag_*): Add new argument to declarations.
16981 (dump_gcov_file): Likewise.
16982 (tag_blocks): Add and use new argument depth.
16983 (tag_arcs): Likewise.
16984 (tag_lines): Likewise.
16985 (tag_counters): Likewise.
16986 (tag_summary): Likewise.
16987 (dump_working_sets): Use depth to do a proper indentation.
16989 2017-04-28 Jakub Jelinek <jakub@redhat.com>
16992 * cgraph.h (symtab_node::debug_symtab): No longer inline.
16993 * symtab.c (symtab_node::debug_symtab): Move definition here.
16995 2017-04-28 Richard Biener <rguenther@suse.de>
16997 * lto-streamer.h (LTO_major_version): Bump to 7.
16999 2017-04-28 Richard Biener <rguenther@suse.de>
17001 * tree-vrp.c (assert_info): New struct.
17002 (add_assert_info): New helper.
17003 (register_edge_assert_for_2): Refactor to add asserts to a vector
17005 (register_edge_assert_for_1): Likewise.
17006 (register_edge_assert_for): Likewise.
17007 (finish_register_edge_assert_for): New helper actually registering
17008 asserts where live on edge.
17009 (find_conditional_asserts): Adjust.
17010 (find_switch_asserts): Likewise.
17011 (evrp_dom_walker::try_find_new_range): Generalize.
17012 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
17014 2017-04-27 Marek Polacek <polacek@redhat.com>
17017 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
17018 arg10 and arg11 to itype.
17020 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
17022 * doc/extend.texi (Object Size Checking): Improve grammar.
17024 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
17027 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
17028 that the logic for permitting reciprocal estimates matches that
17031 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17034 * tree.c (type_cache_hasher::equal): Only compare
17035 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
17036 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
17037 non-aggregate element types.
17038 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
17039 about the flag on ARRAY_TYPEs in the comment, formatting fix.
17041 2017-04-27 Richard Biener <rguenther@suse.de>
17043 PR middle-end/80533
17044 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
17045 stripping ARRAY_REFs from MEM_EXPR make sure we're not
17046 keeping a reference to a trailing array.
17048 2017-04-27 Richard Biener <rguenther@suse.de>
17050 PR middle-end/80539
17051 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
17052 being in loop-closed SSA form conservatively.
17053 (chrec_fold_multiply_poly_poly): Likewise.
17055 2017-04-27 Tamar Christina <tamar.christina@arm.com>
17057 PR middle-end/79665
17058 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
17059 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
17061 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17064 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
17065 (aarch64_function_arg_alignment): Return unsigned int again, but still
17066 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
17067 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17068 Don't emit -Wpsabi note.
17069 (aarch64_function_arg_boundary): Likewise.
17070 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17073 2017-04-26 Nathan Sidwell <nathan@acm.org>
17075 * tree.h (crc32_unsigned_n): Declare.
17076 (crc32_unsigned, crc32_unsigned): Make inline.
17077 * tree.c (crc32_unsigned_bits): Replace with ...
17078 (crc32_unsigned_n): ... this.
17079 (crc32_unsigned, crc32_byte): Remove.
17080 (crc32_string): Remove unnecessary braces.
17082 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17084 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
17085 * ipa-inline-analysis.c (MAX_TIME): Remove.
17086 (account_size_time): Use sreal for time.
17087 (dump_inline_summary): Update.
17088 (estimate_function_body_sizes): Update.
17089 (estimate_edge_size_and_time): Update.
17090 (estimate_calls_size_and_time): Update.
17091 (estimate_node_size_and_time): Update.
17092 (inline_merge_summary): Update.
17093 (inline_update_overall_summary): Update.
17094 (estimate_time_after_inlining): Update.
17095 (inline_read_section): Update.
17096 (inline_write_summary): Update.
17097 * ipa-inline.c (compute_uninlined_call_time): Update.
17098 (compute_inlined_call_time): Update.
17099 (recursive_inlining): Update.
17100 (inline_small_functions): Update.
17101 (dump_overall_stats): Update.
17102 * ipa-inline.h: Include sreal.h.
17103 (size_time_entry): Turn time to sreal.
17104 (inline_summary): Turn self_time nad time to sreal.
17106 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17108 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
17110 (sreal::stream_out, sreal::stream_in): New.
17111 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
17113 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17115 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
17118 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
17121 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
17123 (dimode_scalar_chain::compute_convert_gain): Ditto.
17124 (dimode_scalar_chain::make_vector_copies): Ditto.
17125 (dimode_scalar_chain::convert_reg): Ditto.
17126 (dimode_scalar_chain::convert_insn): Ditto.
17127 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
17128 (VI248_AVX512BW_1): New mode iterator.
17129 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
17130 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
17133 2017-04-25 Martin Sebor <msebor@redhat.com>
17135 PR tree-optimization/80497
17136 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
17137 constants are representable in HOST_WIDE_INT.
17138 (parse_directive): Ditto.
17140 2017-04-25 Martin Sebor <msebor@redhat.com>
17143 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
17144 (new_zero_array): Adjust signature.
17145 (dom_info::dom_init): Used unsigned rather that size_t.
17146 (dom_info::dom_info): Same.
17148 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17149 Jakub Jelinek <jakub@redhat.com>
17152 * config/arm/arm.c: Include gimple.h.
17153 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17154 returns negative, increment ncrn only if it returned positive.
17155 (arm_needs_doubleword_align): Return int instead of bool,
17156 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
17157 members, but if there is any such non-FIELD_DECL
17158 > PARM_BOUNDARY aligned decl, return -1 instead of false.
17159 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17160 returns negative, increment nregs only if it returned positive.
17161 (arm_setup_incoming_varargs): Likewise.
17162 (arm_function_arg_boundary): Emit -Wpsabi note if
17163 arm_needs_doubleword_align returns negative, return
17164 DOUBLEWORD_ALIGNMENT only if it returned positive.
17166 2017-04-25 Marek Polacek <polacek@redhat.com>
17169 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17170 first argument to type.
17172 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
17175 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
17176 type checks to test for compatibility instead of equality.
17178 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17179 Jakub Jelinek <jakub@redhat.com>
17182 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
17184 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
17185 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
17186 the alignment computation, but return their maximum in warn_alignment.
17187 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17188 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
17190 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
17191 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17194 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17196 * config/arc/simdext.md (dmpyh): Fix typo.
17198 2017-04-25 Richard Biener <rguenther@suse.de>
17200 PR tree-optimization/80492
17201 * alias.c (compare_base_decls): Handle registers with asm
17202 specification conservatively.
17203 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17204 compare_base_decls returning dont-know properly.
17206 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17208 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
17209 (legitimate_offset_address_p): New function.
17210 (arc_legitimate_address_p): Use above function.
17212 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17214 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
17216 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17218 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
17219 ACCH registers whenever they are available.
17221 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17223 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
17224 double regs fix when not used.
17226 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17228 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
17230 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
17231 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
17233 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17235 * config/arc/arc.c (arc_output_addsi): Check for h-register class
17236 when emitting short ADD instructions.
17238 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17240 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
17242 (cmpsi_cc_c_insn): Likewise.
17243 (cbranchsi4_scratch): Compute proper instruction length using
17244 compact_hreg_operand.
17245 * config/arc/predicates.md (compact_hreg_operand): New predicate.
17247 2017-04-25 Richard Biener <rguenther@suse.de>
17249 PR middle-end/80509
17250 * passes.c (pass_manager::pass_manager): Initialize
17251 m_name_to_pass_map.
17253 2017-04-25 Richard Biener <rguenther@suse.de>
17255 PR tree-optimization/79201
17256 * tree-ssa-sink.c (statement_sink_location): Handle calls.
17258 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17261 * config/s390/vector.md: Split MEM->GPR vector moves for
17262 non-s_operand addresses.
17264 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17267 * config/s390/predicates.md (reload_const_wide_int_operand): New
17269 * config/s390/s390.md ("movti"): Remove d/P alternative.
17270 ("movti_bigconst"): New pattern definition.
17272 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17275 * s390-protos.h (s390_expand_cs_hqi): Removed.
17276 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
17277 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
17278 modes as well as CCZ1mode and CCZmode.
17279 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
17280 signature of s390_emit_compare_and_swap.
17281 (s390_expand_cs_hqi): Likewise, make static.
17282 (s390_expand_cs_tdsi): Generate an explicit compare before trying
17283 compare-and-swap, in some cases.
17284 (s390_expand_cs): Wrapper function.
17285 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
17287 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
17288 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
17289 patterns for small and large integers. Forbid symref memory operands.
17290 Move expander to s390.c. Require cc register.
17291 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
17292 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
17293 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
17294 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
17295 symref memory operands. Remove CC mode and call s390_match_ccmode
17297 ("atomic_exchange<mode>"): Allow and implement all integer modes.
17299 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17301 * config/s390/s390.md (define_peephole2): New peephole to help
17302 combining the load-and-test pattern with volatile memory.
17304 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17306 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
17307 with CCZmode for TARGET_Z196.
17309 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17311 PR rtl-optimization/80501
17312 * combine.c (make_compound_operation_int): Set subreg_code to SET
17313 even for AND with mask of the sign bit of mode.
17315 PR rtl-optimization/80500
17316 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
17317 sum's initial value.
17319 2017-04-25 Julian Brown <julian@codesourcery.com>
17320 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17322 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
17324 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
17326 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
17328 2017-04-25 Julian Brown <julian@codesourcery.com>
17329 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17331 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
17332 (thunderx2t99_sha): New Reservation.
17334 2017-04-25 Julian Brown <julian@codesourcery.com>
17335 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17337 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
17338 type for 1-element load.
17340 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
17342 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
17344 2017-04-24 Martin Jambor <mjambor@suse.cz>
17346 PR tree-optimization/80293
17347 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
17348 char arrays not totally scalarizable if it is false.
17349 (analyze_all_variable_accesses): Pass correct value in the new
17350 parameter. Add a statistics counter.
17352 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
17354 PR middle-end/79931
17355 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
17357 2017-04-24 Richard Biener <rguenther@suse.de>
17359 PR tree-optimization/80494
17360 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
17361 out for complex types.
17363 2017-04-24 Richard Biener <rguenther@suse.de>
17365 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
17366 * tree-ssa-sccvn.c (print_scc): Print SCC size.
17367 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
17368 (DFS): Adjust and never fail.
17369 (sccvn_dom_walker::fail): Remove.
17370 (sccvn_dom_walker::before_dom_children): Adjust.
17371 (run_scc_vn): Likewise and never fail.
17372 * tree-ssa-pre.c (pass_pre::execute): Adjust.
17373 (pass_fre::execute): Likewise.
17375 2017-04-24 Richard Biener <rguenther@suse.de>
17377 PR tree-optimization/79725
17378 * tree-ssa-sink.c (statement_sink_location): Return whether
17379 failure reason was zero uses. Move that check later.
17380 (sink_code_in_bb): Deal with zero uses by removing the stmt
17383 2017-04-24 Richard Biener <rguenther@suse.de>
17386 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
17387 pointer-based references.
17389 2017-04-24 Richard Biener <rguenther@suse.de>
17392 * pass_manager.h (pass_manager::operator new): Remove.
17393 (pass_manager::operator delete): Likewise.
17394 * passes.c (pass_manager::operator new): Remove.
17395 (pass_manager::operator delete): Likewise.
17396 (pass_manager::pass_manager): Zero individual pass members.
17398 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
17401 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
17402 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
17403 Check "XEXP (src, 1)" operand here.
17404 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
17405 Check "XEXP (src, 1)" operand here.
17406 (dimode_scalar_chain::make_vector_copies): Detect count register
17407 of a shift instruction. Zero extend count register from QImode
17408 to DImode to satisfy vector shift pattern count operand predicate.
17409 Substitute vector shift count operand with a DImode copy.
17410 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
17413 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
17415 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
17416 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17417 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
17418 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17419 (UNSPEC_NOREX_MEM): Remove definition.
17421 2017-04-21 Richard Biener <rguenther@suse.de>
17423 PR tree-optimization/79547
17424 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17425 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
17426 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
17427 without any constraints.
17429 2017-04-21 Richard Biener <rguenther@suse.de>
17431 PR tree-optimization/78847
17432 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
17434 2017-04-21 Richard Biener <rguenther@suse.de>
17436 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
17437 (build_distinct_type_copy): Likewise.
17438 (build_variant_type_copy): Likewise.
17439 * tree.c (build_qualified_type): Pass down mem-stat info.
17440 (build_distinct_type_copy): Likewise.
17441 (build_variant_type_copy): Likewise.
17443 2017-04-21 Richard Biener <rguenther@suse.de>
17445 PR tree-optimization/80237
17446 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
17448 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
17449 for a simplified result.
17451 2016-04-21 Richard Biener <rguenther@suse.de>
17453 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
17454 sth as strict as a simple_iv but a chrec without symbols and an
17455 operand defined in the loop we are peeling (and not some subloop).
17456 (propagate_constants_for_unrolling): Propagate all constants.
17458 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17461 * config/i386/i386.c (print_reg): Remove assert for disalowed
17462 regno values, call output_operand_lossage instead.
17464 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17467 * config/i386/constraints.md (Yc): New register constraint.
17468 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
17469 Use Yc constraint for alternative 2 of operand 0. Remove
17470 preferred_for_speed attribute.
17472 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17474 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
17475 lastprivate clauses in SIMT case.
17477 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
17479 * doc/invoke.texi (-Wextra-semi): Document new warning option.
17481 2017-04-20 Richard Biener <rguenther@suse.de>
17483 PR tree-optimization/57796
17484 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
17485 as N scalar stores.
17486 (vect_model_load_cost): Cost gathers as N scalar loads.
17488 2017-04-20 Richard Biener <rguenther@suse.de>
17490 * ggc-page.c (ggc_allocated_p): Rename to ...
17491 (safe_lookup_page_table_entry): ... this and return the lookup
17493 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
17495 2017-04-20 Richard Biener <rguenther@suse.de>
17497 PR tree-optimization/80453
17498 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
17499 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
17500 from the conditions.
17501 (vn_phi_eq): Pass them down.
17502 (vn_phi_lookup): Record them.
17503 (vn_phi_insert): Likewise.
17505 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
17507 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
17508 uninitialized variable warning to avoid buffer overrun.
17510 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17513 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
17514 is suppressed for '{ 0 }' in C.
17516 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17518 * BASE-VER: Set to 8.0.0.
17520 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
17522 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
17523 priority .init_array and .fini_array section with SECTION_NOTYPE
17526 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17528 PR middle-end/80423
17529 * tree.h (build_array_type): Add typeless_storage default argument.
17530 * tree.c (type_cache_hasher::equal): Also compare
17531 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
17532 (build_array_type): Add typeless_storage argument, set
17533 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
17535 (build_nonshared_array_type): Adjust build_array_type_1 caller.
17536 (build_array_type): Likewise. Add typeless_storage argument.
17538 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17539 Jakub Jelinek <jakub@redhat.com>
17541 PR tree-optimization/80426
17542 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
17543 operation on symbolic operands, also compute the overflow for the
17544 invariant part when the operation degenerates into a negation.
17546 2017-04-19 Jakub Jelinek <jakub@redhat.com>
17549 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
17550 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
17553 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
17555 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
17558 * config/avr/avr.c (tree.h): Include it.
17559 (cgraph.h): Include it.
17560 (avr_encode_section_info): Don't warn for uninitialized progmem
17561 variable if it's just an alias.
17563 2017-04-19 Richard Biener <rguenther@suse.de>
17566 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
17567 when needed by AutoPGO.
17569 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
17572 * doc/lto.texi: Remove an extra 'that'.
17574 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
17576 PR rtl-optimization/80429
17577 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
17578 are only used in debug insns.
17580 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17581 Vladimir Makarov <vmakarov@redhat.com>
17583 * config/sparc/predicates.md (input_operand): Add comment. Return
17584 true for any memory operand when LRA is in progress.
17585 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
17587 2017-04-18 Jeff Law <law@redhat.com>
17590 * mips.md ({return,simple_return}_internal): Do not overwrite
17593 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17595 PR tree-optimization/80443
17596 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
17597 instead of adding 1, subtract -1 and similarly instead of subtracting
17600 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
17602 PR rtl-optimization/80357
17603 * haifa-sched.c (tmp_bitmap): New variable.
17604 (model_recompute): Handle duplicate use records.
17605 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
17606 (free_global_sched_pressure_data): Free it.
17608 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
17611 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
17612 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17613 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17614 instead of SYSTEM_HEADER_DIR.
17616 2017-04-18 Jeff Law <law@redhat.com>
17618 PR middle-end/80422
17619 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
17620 predecessors after walking up the insn chain.
17622 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17625 * dwarf2out.c (modified_type_die): Try harder not to emit internal
17626 sizetype type into debug info.
17628 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17631 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
17632 unneeded test for TARGET_UPPER_REGS_SF.
17633 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
17635 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17638 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
17639 instead of gsi_after_labels.
17641 2017-04-18 Jeff Law <law@redhat.com>
17643 * regcprop.c (maybe_mode_change): Avoid creating copies of the
17647 2017-04-13 Jeff Law <law@redhat.com>
17648 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17649 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17651 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
17654 * config/avr/avr.c (intl.h): Include it.
17655 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
17657 2017-04-18 Martin Liska <mliska@suse.cz>
17659 PR gcov-profile/78783
17660 * gcov-tool.c (gcov_output_files): Validate that destination
17661 file is either removed by the tool or by a user.
17663 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17664 Guy Benyei <guybe@mellanox.com>
17666 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
17667 block, and do not negate it, the stored id is already negative.
17669 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17671 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
17673 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17676 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
17677 masks of options that should be turned off if the VSX vector
17678 options are turned off.
17679 (OTHER_P8_VECTOR_MASKS): Likewise.
17680 (OTHER_VSX_VECTOR_MASKS): Likewise.
17681 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
17682 rs6000_disable_incompatible_switches to validate no type switches
17684 (rs6000_incompatible_switch): New function to disallow turning on
17685 other vector options if -mno-vsx, -mno-power8-vector, or
17686 -mno-power9-vector are specified.
17688 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17690 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
17692 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17694 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
17695 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
17696 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
17697 (ARG_POINTER_CFA_OFFSET): Likewise.
17699 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17701 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
17702 conditions to take advantage of various optimizations.
17704 2017-04-13 Jeff Law <law@redhat.com>
17706 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17707 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17708 (zero_extendsidi2_dext): Likewise.
17710 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17713 * fold-const.c (fold_ternary_loc): Revert
17714 use op0 instead of fold_convert_loc (loc, type, arg0) part of
17717 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
17719 PR rtl-optimization/80343
17720 * lra-remat.c (update_scratch_ops): Assign original hard reg to
17721 new scratch pseudo.
17723 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
17726 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
17727 to ubsan_encode_value.
17729 2017-04-13 Jeff Law <law@redhat.com>
17731 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
17732 appearing in DEBUG_INSNs.
17734 2017-04-13 Martin Liska <mliska@suse.cz>
17736 PR gcov-profile/80413
17737 * gcov-io.c (gcov_write_string): Copy to buffer just when
17738 allocated size is greater than zero.
17740 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17743 * dwarf2out.c (decls_for_scope): Ignore declarations of
17744 current_function_decl in BLOCK_NONLOCALIZED_VARS.
17746 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
17749 * ipa-visibility.c (non_local_p): Fix typos.
17750 (localize_node): When localizing symbol in same comdat group,
17751 dissolve the group only when we know external symbols are going
17753 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
17755 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17757 PR tree-optimization/79390
17758 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
17759 order does not result in usable sequence, retry with reversed operand
17765 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
17766 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
17767 op0 instead of fold_convert_loc (loc, type, arg0).
17769 2017-04-12 Jeff Law <law@redhat.com>
17771 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
17772 has a delay slot in the generated code.
17774 * config/cris/cris.md (cris_preferred_reload_class): Return
17775 GENNONACR_REGS rather than GENERAL_REGS.
17777 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17780 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
17781 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
17782 signedness of the result type.
17784 2017-04-12 Richard Biener <rguenther@suse.de>
17785 Jeff Law <law@redhat.com>
17787 PR tree-optimization/80359
17788 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
17789 trim stores to TARGET_MEM_REFs.
17791 2017-04-12 Richard Biener <rguenther@suse.de>
17793 PR tree-optimization/79390
17794 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
17795 threading case even more.
17797 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
17800 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
17801 for quad_address_p for TImode, instead of just not indexed_address.
17803 2017-04-12 Richard Biener <rguenther@suse.de>
17804 Bernd Edlinger <bernd.edlinger@hotmail.de>
17806 PR middle-end/79671
17807 * alias.c (component_uses_parent_alias_set_from): Handle
17808 TYPE_TYPELESS_STORAGE.
17809 (get_alias_set): Likewise.
17810 * tree-core.h (tree_type_common): Add typeless_storage flag.
17811 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
17812 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
17813 for types containing members with TYPE_TYPELESS_STORAGE.
17814 (place_field): Likewise.
17815 (layout_type): Likewise for ARRAY_TYPE.
17816 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
17817 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
17818 TYPE_TYPELESS_STORAGE.
17819 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
17821 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17824 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17825 first argument to type.
17827 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17831 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
17832 CONST0_RTX (mode) rather than const0_rtx where appropriate.
17833 (rs6000_expand_binop_builtin): Likewise.
17834 (rs6000_expand_ternop_builtin): Likewise; also add missing
17835 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
17836 vshasigma built-ins.
17837 * doc/extend.texi: Document that vec_xxpermdi's third argument
17838 must be a constant.
17840 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
17842 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
17843 Use shift_const cost parameter when calculating gain of STV shifts.
17845 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
17847 PR rtl-optimization/70478
17848 * lra-constraints.c (process_alt_operands): Check memory for
17849 disfavoring memory insn operand.
17851 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17853 PR middle-end/80100
17854 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
17855 left shift in unsigned HOST_WIDE_INT type.
17857 PR rtl-optimization/80385
17858 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
17859 (not (neg X)) into (plus X -1) for complex or non-integral modes.
17862 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
17863 if they have any depend clauses.
17865 2017-04-11 Martin Liska <mliska@suse.cz>
17868 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
17869 * ipa-split.c (split_function): Create a local comdat symbol
17870 if caller is in a comdat group.
17872 2017-04-11 Martin Liska <mliska@suse.cz>
17875 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
17878 2017-04-11 Martin Sebor <msebor@redhat.com>
17880 PR middle-end/80364
17881 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
17882 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
17884 (directive::set_width, directive::set_precision, format_character):
17886 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
17889 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
17892 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
17893 conflict, set target->arch_name instead of target->cpu_name.
17895 2017-04-11 Richard Biener <rguenther@suse.de>
17897 PR tree-optimization/80374
17898 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
17899 build_zero_cst, remove fold_convertible_p check again.
17901 2017-04-11 Martin Liska <mliska@suse.cz>
17904 * ubsan.c (instrument_object_size): Do not instrument register
17907 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17910 * config/i386/i386-builtin-types.def
17911 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
17912 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
17913 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
17914 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
17915 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
17916 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
17917 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
17918 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
17919 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
17920 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
17921 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
17922 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
17923 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
17924 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
17925 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
17926 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
17927 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
17928 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
17929 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
17930 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
17931 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
17932 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
17933 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
17934 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
17935 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
17936 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
17937 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
17938 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
17939 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
17940 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
17941 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
17942 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
17943 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
17944 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
17945 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
17946 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
17947 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
17948 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
17949 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
17950 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
17951 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
17952 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
17953 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
17954 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
17955 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
17956 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
17958 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
17959 flag to second_arg_count, handle 4 argument function type _COUNT
17960 aliases, handle second_arg_count on second argument rather than last.
17962 2017-04-10 Jeff Law <law@redhat.com>
17964 PR tree-optimization/80374
17965 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
17966 record anything if we can not convert integer_zero_node to the
17969 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
17972 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17973 Enhance special handling given to the TARGET_P9_MINMAX option in
17974 relation to certain other options.
17976 2017-04-10 Bin Cheng <bin.cheng@arm.com>
17978 PR tree-optimization/80153
17979 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
17980 remove POINTER_PLUS_EXPR's base part directly, rather than through
17983 2017-04-10 Richard Biener <rguenther@suse.de>
17984 Bin Cheng <bin.cheng@arm.com>
17986 PR tree-optimization/80153
17987 * tree-affine.c (aff_combination_to_tree): Get base pointer from
17988 the first element of pointer type aff_tree. Build result expr in
17990 (add_elt_to_tree): Convert to type unconditionally. Remove other
17991 fold_convert calls.
17992 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
17993 (rewrite_use_nonlinear_expr): Check invariant using iv information.
17995 2017-04-10 Richard Biener <rguenther@suse.de>
17997 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18000 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
18002 PR rtl-optimization/70478
18003 * lra-constraints.c (curr_small_class_check): New.
18004 (update_and_check_small_class_inputs): New.
18005 (process_alt_operands): Update curr_small_class_check. Disfavor
18006 alternative insn memory operands. Check available regs for small
18009 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18012 * config/mips/mips.opt (-mvirt): Update description.
18013 * doc/invoke.texi (-mvirt): Likewise.
18015 2017-04-10 Richard Biener <rguenther@suse.de>
18017 PR middle-end/80362
18018 * fold-const.c (fold_binary_loc): Look at unstripped ops when
18019 looking for NEGATE_EXPR in -A / -B to A / B folding.
18021 2017-04-10 Martin Liska <mliska@suse.cz>
18023 PR gcov-profile/80224
18024 * gcov.c (print_usage): Fix usage string.
18025 (get_gcov_intermediate_filename): Remove.
18026 (output_gcov_file): Use both for normal and intermediate format.
18027 (generate_results): Do not initialize special file for
18028 intermediate format.
18030 2017-04-10 Richard Biener <rguenther@suse.de>
18032 PR tree-optimization/80304
18033 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
18036 2017-04-10 Nathan Sidwell <nathan@acm.org>
18039 * config/rs6000/rs6000.c (rs6000_vector_type): New.
18040 (rs6000_init_builtins): Use it.
18042 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18044 * config/arm/arm.md (<mrc>): Add mode to SET source.
18045 (<mrrc>): Likewise.
18047 2017-04-10 Richard Biener <rguenther@suse.de>
18049 PR middle-end/80344
18050 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
18052 2017-04-10 Jakub Jelinek <jakub@redhat.com>
18055 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
18056 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
18057 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
18058 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
18059 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
18060 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
18061 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
18062 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
18063 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
18064 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
18065 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
18066 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
18067 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
18068 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
18069 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
18070 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
18071 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
18072 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
18073 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
18074 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
18075 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
18076 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
18077 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
18079 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
18081 PR rtl-optimization/70478
18082 * lra-constraints.c: Reverse the last patch.
18084 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
18086 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
18087 Add comment for WCHAR_T.
18089 2017-04-08 Martin Liska <mliska@suse.cz>
18092 2017-04-07 Martin Liska <mliska@suse.cz>
18095 * ipa-split.c (split_function): Add function part to a same comdat
18098 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18101 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
18103 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
18105 * rs6000/rs6000.c (vec_load_pendulum): Rename...
18106 (vec_pairing): ...to this.
18107 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
18108 (rs6000_sched_init): Adjust for name change.
18109 (struct rs6000_sched_context): Likewise.
18110 (rs6000_init_sched_context): Likewise.
18111 (rs6000_set_sched_context): Likewise.
18113 2017-04-07 Jakub Jelinek <jakub@redhat.com>
18119 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
18121 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
18122 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
18123 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
18125 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
18127 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
18129 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18131 PR rtl-optimization/70703
18132 * ira-color.c (update_conflict_hard_regno_costs): Use
18133 int64_t instead of HOST_WIDE_INT.
18135 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18137 PR rtl-optimization/70478
18138 * lra-constraints.c (process_alt_operands): Disfavor alternative
18139 insn memory operands.
18141 2017-04-07 Jeff Law <law@redhat.com>
18143 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
18144 CALL and NOTE_INSN_CALL_ARG_LOCATION.
18146 2017-04-07 Martin Liska <mliska@suse.cz>
18149 * config/aarch64/aarch64.c (aarch64_process_target_attr):
18150 Show error message instead of an ICE.
18152 2017-04-07 Martin Liska <mliska@suse.cz>
18155 * ipa-split.c (split_function): Add function part to a same comdat
18158 2017-04-07 Richard Biener <rguenther@suse.de>
18160 PR middle-end/80341
18161 * tree.c (get_unwidened): Also handle ! for_type case for
18163 * convert.c (do_narrow): Split out from ...
18164 (convert_to_integer_1): ... here. Do not pass final truncation
18165 type to get_unwidened for TRUNC_DIV_EXPR.
18167 2017-04-07 Richard Biener <rguenther@suse.de>
18169 * tree-affine.c (wide_int_ext_for_comb): Take type rather
18171 (aff_combination_const): Adjust.
18172 (aff_combination_scale): Likewise.
18173 (aff_combination_add_elt): Likewise.
18174 (aff_combination_add_cst): Likewise.
18175 (aff_combination_convert): Likewise.
18176 (add_elt_to_tree): Likewise. Remove unused argument.
18177 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
18179 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
18181 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
18183 * config/arm/arm.c (arm_default_short_enums): Use
18184 ARM_DEFAULT_SHORT_ENUMS.
18185 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
18187 2017-04-06 Jakub Jelinek <jakub@redhat.com>
18190 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
18191 members with redundant out-of-class redeclaration.
18193 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18196 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
18197 * config/i386/i386.md (*zero_extendsidi2):
18198 Add (?*x,*x) and (?*v,*v) alternatives.
18200 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18203 * config/i386/i386.c (ix86_expand_builtin)
18204 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
18205 mode from insn data. Convert operands to insn operand mode.
18206 Copy operands that don't satisfy insn predicate to a register.
18208 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
18210 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
18213 2017-04-06 Richard Biener <rguenther@suse.de>
18215 PR tree-optimization/80334
18216 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
18217 preserve alignment of accesses.
18219 2017-04-06 Richard Biener <rguenther@suse.de>
18221 PR tree-optimization/80262
18222 * tree-sra.c (build_ref_for_offset): Preserve address-space
18224 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
18225 Drop useless address-space information on MEM_REF offsets.
18227 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
18229 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
18231 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18233 PR rtl-optimization/70703
18234 * ira-color.c (update_conflict_hard_regno_costs): Use
18235 HOST_WIDE_INT instead of long.
18237 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
18240 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
18241 not defined for x86_64 target. Add -mmmx target option when __SSE2__
18243 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
18244 for x86_64 target. Handle -m3dnowa option.
18246 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18248 PR rtl-optimization/70703
18249 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
18250 (update_conflict_hard_regno_costs): Use long instead of unsigned
18251 arithmetic for cost calculation.
18253 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18254 Bernd Edlinger <bernd.edlinger@hotmail.de>
18257 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
18260 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
18263 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
18264 ptr_mode with Pmode throughout.
18265 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
18266 into probe_stack_range and use DImode.
18268 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
18271 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
18272 call_eh_return is true.
18274 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18276 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
18277 Initialize last_match_fntype_index.
18279 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18282 * tree-nvr.c: Include internal-fn.h.
18283 (pass_return_slot::execute): Ignore internal calls without
18286 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18287 Richard Biener <rguenther@suse.de>
18290 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
18291 captures used multiple times, except for the last use.
18292 * generic-match-head.c: Include gimplify.h.
18294 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18296 PR tree-optimization/79390
18297 * target.h (struct noce_if_info): Declare.
18298 * targhooks.h (default_noce_conversion_profitable_p): Declare.
18299 * target.def (noce_conversion_profitable_p): New target hook.
18300 * ifcvt.h (struct noce_if_info): New type, moved from ...
18301 * ifcvt.c (struct noce_if_info): ... here.
18302 (noce_conversion_profitable_p): Renamed to ...
18303 (default_noce_conversion_profitable_p): ... this. No longer
18305 (noce_try_store_flag_constants, noce_try_addcc,
18306 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
18307 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
18308 instead of noce_conversion_profitable_p.
18309 * config/i386/i386.c: Include ifcvt.h.
18310 (ix86_option_override_internal): Don't override
18311 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
18312 (ix86_noce_conversion_profitable_p): New function.
18313 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
18314 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
18315 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
18316 * doc/tm.texi: Regenerated.
18318 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18320 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
18323 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18326 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
18327 instructions for small multiply cores.
18329 2017-04-04 Jeff Law <law@redhat.com>
18331 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
18333 (mips_expand_vec_perm_const): Initialize elements in orig_perm
18334 that are not set by the loop over the elements.
18336 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18339 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
18340 int mode, convert_modes it to mode as unsigned, otherwise use
18341 lowpart_subreg to mode rather than SImode.
18342 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
18343 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
18344 Use DImode instead of SImode for the shift count operand.
18345 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
18348 2017-04-04 Richard Biener <rguenther@suse.de>
18350 PR middle-end/80281
18351 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
18352 arithmetic done for the negate or the plus. Simplify.
18353 (A - (-B) -> A + B): Likewise.
18354 * fold-const.c (split_tree): Make sure to not negate pointers.
18356 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
18358 PR rtl-optimization/60818
18359 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
18360 a compare of comparisons with the thing compared if this results
18361 in a different machine mode.
18363 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
18365 * alias.c (base_alias_check): Fix typo in comment.
18366 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
18367 * cgraphunit.c (symbol_table::compile): Likewise.
18368 * collect2.c (maybe_run_lto_and_relink): Likewise.
18369 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
18370 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
18371 * config/avr/avr.c (avr_map_op_t): Likewise.
18372 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
18373 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
18374 * config/epiphany/epiphany.md (movcc): Likewise.
18375 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
18376 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
18378 * config/mips/mips.c (mips_save_restore_reg): Likewise.
18379 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
18380 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
18381 * config/sh/sh.c (sh_rtx_costs): Likewise.
18382 * fold-const.c (fold_truth_andor): Likewise.
18383 * genautomata.c (collapse_flag): Likewise.
18384 * gengtype.h (struct type::u::s): Likewise.
18385 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
18386 * input.c (FORMAT_AMOUNT): Likewise.
18387 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
18388 (known_aggs_to_agg_replacement_list): Likewise.
18389 * ipa-inline-analysis.c: Likewise.
18390 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
18391 * ipa-polymorphic-call.c
18392 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
18393 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
18394 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
18396 * modulo-sched.c (apply_reg_moves): Likewise.
18397 * omp-expand.c (build_omp_regions_1): Likewise.
18398 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
18399 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
18400 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
18401 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
18402 * value-prof.c: Likewise.
18403 * var-tracking.c (val_reset): Likewise.
18405 2017-04-03 Richard Biener <rguenther@suse.de>
18407 PR tree-optimization/80275
18408 * fold-const.c (split_address_to_core_and_offset): Handle
18411 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
18413 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
18414 descriptors is at least equal to that of functions.
18416 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18418 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
18420 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18423 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
18424 (mov<IMOD4:mode>): New expander.
18425 (*mov<IMOD4:mode>_internal): New insn and split pattern.
18427 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
18429 PR rtl-optimization/79405
18430 * fwprop.c (propagations_left): New variable.
18431 (forward_propagate_into): Decrement it.
18432 (fwprop_init): Initialize it.
18433 (fw_prop): If the variable has reached zero, stop propagating.
18434 (fwprop_addr): Ditto.
18436 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18439 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
18440 a FUNCTION_DECL, pass it as decl instead of origin to
18443 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
18445 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
18448 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
18451 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
18452 TARGET_VSX_SMALL_INTEGER.
18454 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18456 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18457 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
18459 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18461 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
18462 extraction from odd-numbered MSA register.
18464 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18466 PR middle-end/80173
18467 * expmed.c (store_bit_field_1): Don't attempt to create
18468 a word subreg out of hard registers wider than word if they
18469 have HARD_REGNO_NREGS of 1 for their mode.
18471 PR middle-end/80163
18472 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
18473 conversions to integer types wider than word and pointer.
18476 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
18477 (rtx_equal_for_cselib_p): Pass 0 to it.
18478 * cselib.c (cselib_hasher::equal): Likewise.
18479 (rtx_equal_for_cselib_1): Add depth argument. If depth
18480 is 128, don't look up VALUE locs and punt. Increment
18481 depth in recursive calls when walking VALUE locs.
18483 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
18485 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
18486 (make_gcov_file_name): Use the canonical path name for generating
18488 (read_line): Fix handling of files with ascii null bytes.
18490 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
18492 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
18493 to initialise a vector register instead
18494 of using a const_int.
18496 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18498 PR translation/80189
18499 * gimplify.c (omp_default_clause): Use %qs instead of %s in
18500 diagnostic messages.
18502 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
18505 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
18506 (dfp_diex_<mode>): Update mode of operand 1.
18507 * doc/extend.texi (dxex, dxexq): Document change to return type.
18508 (diex, diexq): Document change to argument type.
18510 2017-03-30 Martin Jambor <mjambor@suse.cz>
18513 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
18514 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
18515 it reflects the signature changes performed at the callee side.
18516 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
18517 to cgraph_build_function_type_skip_args.
18518 (build_function_decl_skip_args): Adjust call to the above function.
18520 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18523 * config/i386/sse.md
18524 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
18525 register as dest whenever it is a MEM not rtx_equal_p to the
18526 corresponding dup operand, and when forcing into reg move the
18527 reg into the memory afterwards.
18528 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
18529 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
18530 for the force_reg mode.
18531 (avx512vl_vextractf128<mode>): Use register as dest either
18532 always when a MEM, or when it is a MEM not rtx_equal_p to the
18533 corresponding dup operand, or even not when it is a CONST_VECTOR
18534 depending on the mode and lo vs. hi.
18535 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
18537 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
18538 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
18539 Likewise. Require that operands[2] is even.
18540 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
18541 Remove extraneous parens. Require that operands[2] is a multiple
18543 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
18544 operands[0] is a MEM if <mask_applied>, the predicates/constraints
18545 disallow memory then.
18547 2017-03-30 Richard Biener <rguenther@suse.de>
18549 PR tree-optimization/77498
18550 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
18551 to non-constants over backedges.
18553 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
18555 PR rtl-optimization/80233
18556 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
18557 as last_combined_insn. Do not test for BARRIER_P separately.
18559 2017-03-29 Andreas Schwab <schwab@suse.de>
18562 * calls.c (prepare_call_address): Convert funexp to Pmode before
18563 copying to temp reg.
18565 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18567 PR tree-optimization/80158
18568 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
18569 Handle possible future case of more than one alternate
18571 (replace_rhs_if_not_dup): Likewise.
18572 (replace_one_candidate): Likewise.
18574 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
18576 PR rtl-optimization/80193
18577 * ira.c (ira): Do not check allocation for LRA.
18579 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
18581 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
18582 (nvptx_output_simt_exit): Declare.
18583 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
18584 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
18585 (init_softstack_frame): Move initialization of crtl->is_leaf to...
18586 (nvptx_declare_function_name): ...here. Emit declaration of local
18587 memory space buffer for omp_simt_enter insn.
18588 (nvptx_output_unisimt_switch): New.
18589 (nvptx_output_softstack_switch): New.
18590 (nvptx_output_simt_enter): New.
18591 (nvptx_output_simt_exit): New.
18592 * config/nvptx/nvptx.h (struct machine_function): New fields
18593 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
18594 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
18595 (UNSPECV_SIMT_EXIT): Ditto.
18596 (omp_simt_enter_insn): New insn.
18597 (omp_simt_enter): New expansion.
18598 (omp_simt_exit): New insn.
18599 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
18601 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
18602 (expand_GOMP_SIMT_ENTER_ALLOC): New.
18603 (expand_GOMP_SIMT_EXIT): New.
18604 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
18605 (GOMP_SIMT_ENTER_ALLOC): Ditto.
18606 (GOMP_SIMT_EXIT): Ditto.
18607 * target-insns.def (omp_simt_enter): New insn.
18608 (omp_simt_exit): Ditto.
18609 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
18611 (lower_rec_simd_input_clauses): Implement SIMT privatization.
18612 (lower_rec_input_clauses): Likewise.
18613 (lower_lastprivate_clauses): Handle SIMT privatization.
18615 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
18616 (ompdevlow_adjust_simt_enter): New.
18617 (find_simtpriv_var_op): New.
18618 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
18619 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
18621 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
18622 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
18623 (copy_decl_for_dup_finish): Ditto.
18625 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
18627 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
18630 * config/i386/i386.c (ix86_option_override_internal): Always
18631 allow -mpreferred-stack-boundary=3 for 64-bit targets.
18633 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18635 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
18637 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18639 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
18640 mark new edge's irreducible flag accordign to it.
18641 (vect_do_peeling): Check loop preheader edge's irreducible flag
18642 and pass it to function slpeel_add_loop_guard.
18644 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
18646 PR tree-optimization/80218
18647 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18648 Update block frequencies and counts.
18650 2017-03-28 Richard Biener <rguenther@suse.de>
18652 PR tree-optimization/78644
18653 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
18654 of a simplification result we may not use it at all.
18656 2017-03-28 Richard Biener <rguenther@suse.de>
18659 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
18660 without arguments, generate default definition of a SSA name.
18662 2017-03-28 Richard Biener <rguenther@suse.de>
18664 PR middle-end/80222
18665 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
18666 TYPE_REF_CAN_ALIAS_ALL references.
18667 * fold-const.c (fold_indirect_ref_1): Likewise.
18669 2017-03-28 Martin Liska <mliska@suse.cz>
18672 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
18673 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
18675 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18676 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
18678 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
18679 (EXTRA_SPECS): Define.
18680 (SUBTARGET_EXTRA_SPECS): Likewise.
18681 (SUBTARGET_CPP_SPEC): Likewise.
18682 * config/arc/elf.h (EXTRA_SPECS): Renamed to
18683 SUBTARGET_EXTRA_SPECS.
18684 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
18686 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18688 * config/arc/simdext.md (vst64_insn): Update pattern.
18689 (vld32wh_insn): Likewise.
18690 (vld32wl_insn): Likewise.
18691 (vld64_insn): Likewise.
18692 (vld32_insn): Likewise.
18694 2017-03-28 Marek Polacek <polacek@redhat.com>
18697 * fold-const.c (fold_comparison): Use protected_set_expr_location
18698 instead of SET_EXPR_LOCATION.
18700 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
18702 * tree.c (add_expr): Avoid name lookup warning.
18704 2017-03-27 Jeff Law <law@redhat.com>
18706 PR tree-optimization/80216
18707 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
18708 function name. Limit recursion depth.
18709 (record_temporary_equivalences): Corresponding changes.
18711 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
18713 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
18716 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18719 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
18721 * cfgcleanup.c (reg_note_cfa_p): New array.
18722 (insns_have_identical_cfa_notes): New function.
18723 (old_insns_match_p): Don't cross-jump in between /f
18724 and non-/f instructions. If both i1 and i2 are frame related,
18725 verify all CFA notes, their order and content.
18727 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
18730 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
18731 HImode and SImode with zero extend to DImode to one insn.
18732 (bswap<mode>2_extenddi): Likewise.
18733 (bswapsi2_extenddi): Likewise.
18734 (bswaphi2_extendsi): Likewise.
18735 (bswaphi2): Combine bswap HImode and SImode into one insn.
18736 Separate memory insns from swapping register.
18737 (bswapsi2): Likewise.
18738 (bswap<mode>2): Likewise.
18739 (bswaphi2_internal): Delete, no longer used.
18740 (bswapsi2_internal): Likewise.
18741 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
18742 store, and gpr<-gpr swap insns.
18743 (bswap<mode>2_store): Likewise.
18744 (bswaphi2_reg): Register only splitter, combine with the splitter.
18745 (bswaphi2 splitter): Likewise.
18746 (bswapsi2_reg): Likewise.
18747 (bswapsi2 splitter): Likewise.
18748 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
18749 the insns into load, store, and register/register insns.
18750 (bswapdi2_ldbrx): Likewise.
18751 (bswapdi2_load): Likewise.
18752 (bswapdi2_store): Likewise.
18753 (bswapdi2_reg): Likewise.
18755 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
18757 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
18758 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
18760 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
18763 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
18765 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18766 special handling for target option conflicts between dform
18767 options (-mpower9-dform, -mpower9-dform-vector,
18768 -mpower9-dform-scalar) and -mno-direct-move.
18770 2017-03-27 Richard Biener <rguenther@suse.de>
18772 PR tree-optimization/80181
18773 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
18775 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18777 * config/arc/predicates.md (move_double_src_operand): Replace the
18778 call to move_double_src_operand with a call to address_operand.
18780 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18782 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
18783 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
18784 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
18786 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18788 * config/arc/predicates.md (long_immediate_loadstore_operand):
18789 Consider scaled addresses cases.
18791 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18793 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
18794 restored when in interrupt.
18795 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
18796 doesn't have delay slot.
18798 2017-03-27 Richard Biener <rguenther@suse.de>
18801 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
18802 inlined thunk clones.
18804 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18807 * asan.c (instrument_derefs): Copy over last operand from
18808 original COMPONENT_REF to the new COMPONENT_REF with
18809 DECL_BIT_FIELD_REPRESENTATIVE.
18810 * ubsan.c (instrument_object_size): Likewise.
18812 2017-03-27 Richard Biener <rguenther@suse.de>
18814 PR tree-optimization/80170
18815 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
18816 sure DR/SCEV didnt fold in constants we do not see when looking
18817 at the reference base alignment.
18819 2017-03-27 Richard Biener <rguenther@suse.de>
18821 PR middle-end/80171
18822 * gimple-fold.c (fold_ctor_reference): Properly guard against
18823 NULL return value from canonicalize_constructor_val.
18825 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
18828 * config/i386/i386.c (ix86_expand_builtin)
18829 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
18830 flags reg setting and flags reg using instructions.
18831 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
18832 clobbering instructions to zero extend op2.
18834 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
18836 * doc/install.texi (Configuration) <--with-aix-soname>:
18837 Update link to AIX ld.
18839 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
18841 PR rtl-optimization/80160
18842 PR rtl-optimization/80159
18843 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
18844 reg_alternate_class into account.
18846 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
18849 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
18850 to consider in curr_insn_transform.
18852 2017-03-24 Jakub Jelinek <jakub@redhat.com>
18854 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
18855 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
18856 and emit_mode_inner.
18858 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18860 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
18861 argument to the overloaded builtin variants. Use the new flag to
18862 deprecate certain builtin variants.
18863 * config/s390/s390-builtin-types.def: Add new builtin types.
18864 * config/s390/s390-builtins.h: Support new flags field for
18865 overloaded builtins.
18866 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
18867 (s390_macro_to_expand): Enable vector float data type.
18868 (s390_cpu_cpp_builtins_internal): Indicate support of the new
18869 builtins by incrementing the __VEC__ version number.
18870 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
18872 (s390_resolve_overloaded_builtin): Emit error messages depending
18873 on the builtin flags.
18874 * config/s390/s390.c (s390_expand_builtin): Support additional
18875 flags argument. Change error message to match the messages
18876 emitted in s390-c.c.
18877 * config/s390/s390.md: New UNSPEC_* constants.
18878 (op_type): Add new instruction types.
18879 * config/s390/vecintrin.h: Add new builtins and test data class
18881 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
18882 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
18883 (VEC_INEXACT, VEC_NOINEXACT): New constants.
18884 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
18885 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
18886 ("vec_mergel<mode>"): V_HW -> VEC_HW.
18888 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
18889 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
18890 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
18891 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
18893 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
18894 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
18895 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
18896 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
18898 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
18899 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
18900 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
18901 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
18902 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
18903 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
18904 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
18906 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
18907 ("vec_scatter_element<V_HW_4:mode>_DI")
18908 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
18909 ("vec_fpint<mode>", "vflls")
18910 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
18911 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
18912 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
18913 ("*vec_cmphe<mode>_cc"): ... these.
18915 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
18916 mode constant instead of magic value.
18918 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18920 * config/s390/s390.c (s390_expand_vec_compare): Support other
18921 vector floating point modes than just V2DF.
18922 (s390_expand_vcond): Likewise.
18923 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
18924 (s390_cannot_change_mode_class): Prevent mode changes between TF
18925 and V1TF in vector registers.
18926 * config/s390/s390.md (DF, SF): New mode attributes.
18927 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
18928 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
18929 SFmode support for VRs.
18930 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
18932 (VFT, VF_HW): New mode iterators.
18933 (vw, sdx): New mode attributes.
18934 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
18935 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
18936 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
18937 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
18938 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
18939 also the new vector floating point modes. Renaming to ...
18941 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
18942 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
18943 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
18944 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
18945 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
18946 ("vec_unordered<mode>"): ... these.
18948 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
18949 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
18950 ("*vec_extendv2df"): New insn definitions.
18952 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18954 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
18955 ("mulditi3_2", "*muldi3_sign"): New patterns.
18956 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
18957 rename the pattern definition.
18959 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18961 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
18963 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
18965 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18967 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
18968 instruction if possible.
18969 * config/s390/vector.md (vec_halfnumelts): New mode
18971 ("*vec_vllezlf<mode>"): New pattern.
18973 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18975 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
18976 ("popcountv4si2", "popcountv2di2"): Rename to ...
18977 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
18978 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
18980 ("popcount<mode>2_vxe"): New pattern.
18982 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18984 * common/config/s390/s390-common.c (processor_flags_table): Add
18986 * config.gcc: Add arch12.
18987 * config/s390/driver-native.c (s390_host_detect_local_cpu):
18988 Default to arch12 for unknown CPU model numbers.
18989 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
18990 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
18991 PROCESSOR_max sanity check.
18992 * config/s390/s390-opts.h (enum processor_type): Add
18994 * config/s390/s390.c (processor_table): Add arch12.
18995 (s390_expand_builtin): Add check for B_VXE flag.
18996 (s390_issue_rate): Add PROCESSOR_ARCH12.
18997 (s390_get_sched_attrmask): Likewise.
18998 (s390_get_unit_mask): Likewise.
18999 (s390_sched_score): Enable z13 scheduling for arch12.
19000 (s390_sched_reorder): Likewise.
19001 (s390_sched_variable_issue): Likewise.
19002 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
19004 (s390_tune_attr): Use z13 scheduling also for arch12.
19005 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
19006 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
19007 (TARGET_VXE_P): New macros.
19008 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
19009 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
19010 * config/s390/s390.opt: Add arch12 as processor_type.
19012 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19014 * config/s390/s390.md
19015 ("fixuns_truncdddi2", "fixuns_trunctddi2")
19016 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
19017 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
19019 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
19020 Rename expanders to ...
19022 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
19023 ("fixuns_truncdddi2_emu"): ... these.
19025 ("fixuns_trunc<mode>si2_emu"): New expander.
19027 ("*fixuns_truncdfdi2_z13"): Rename to ...
19028 ("*fixuns_truncdfdi2_vx"): ... this.
19030 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19032 * config/s390/2964.md: Remove the single element vector compare
19033 instructions which are no longer used.
19034 * config/s390/s390.c (s390_select_ccmode): Remove handling of
19036 (s390_canonicalize_comparison): Remove handling of DFmode
19038 (s390_expand_vec_compare_scalar): Remove function.
19039 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
19040 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
19042 ("*cmp<mode>_ccs"): Add wfcdb instruction.
19044 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19046 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
19048 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
19049 will anyway by matched by mov<mode>_64dfp.
19051 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19053 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
19054 vlef/vstef. Add missing operand to vleif.
19056 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19058 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
19059 pair for all vector types with 64 bit elements.
19060 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
19061 * config/s390/vector.md (V_HW_64): ... here.
19062 (V_128_NOSINGLE): New mode iterator.
19063 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
19064 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
19065 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
19066 ("*vec_load_pairv2di"): Change to ...
19067 ("*vec_load_pair<mode>"): ... this one.
19069 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19071 * config/s390/constraints.md: Add comments.
19072 (jKK): Reject element sizes > 8 bytes.
19073 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
19075 * config/s390/s390.md: Add the s_operand checks formerly in
19076 s390_split_ok_p to various splitters where they are still
19078 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
19079 for 128 bit vectors. Plus two splitters.
19081 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19083 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
19086 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19089 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
19090 error if the boundary argument is not constant.
19092 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19094 PR rtl-optimization/80112
19095 * loop-doloop.c (doloop_condition_get): Don't check condition
19096 if cmp isn't SET with IF_THEN_ELSE src.
19098 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19100 PR tree-optimization/80158
19101 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
19102 replacing a candidate statement, also replace it for the
19103 candidate's alternate interpretation.
19104 (replace_rhs_if_not_dup): Likewise.
19105 (replace_one_candidate): Likewise.
19107 2017-03-24 Richard Biener <rguenther@suse.de>
19109 PR tree-optimization/80167
19110 * graphite-isl-ast-to-gimple.c
19111 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
19113 (translate_isl_ast_to_gimple::get_rename): Likewise.
19115 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
19117 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
19118 handling of certain combinations of target options, including the
19119 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
19120 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
19122 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19125 * config/arm/arm.md (*load_multiple): Add reload_completed to
19126 matching condition.
19128 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19129 Richard Biener <rguenth@suse.de>
19131 PR tree-optimization/79908
19132 PR tree-optimization/80136
19133 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19134 been cast away, gimplify_and_add suffices.
19136 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
19138 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
19140 2017-03-23 Richard Biener <rguenther@suse.de>
19142 PR tree-optimization/80032
19143 * gimplify.c (gimple_push_cleanup): Forced unconditional
19144 cleanups still have to go to the conditional_cleanups
19147 2017-03-22 Jakub Jelinek <jakub@redhat.com>
19149 PR tree-optimization/80072
19150 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
19152 (next_operand_entry_id): Change type to unsigned int.
19153 (sort_by_operand_rank): Make sure to return the right return value
19154 even if unsigned fields are bigger than INT_MAX.
19155 (struct oecount): Change cnt and id type to unsigned int.
19156 (oecount_hasher::equal): Formatting fix.
19157 (oecount_cmp): Make sure to return the right return value
19158 even if unsigned fields are bigger than INT_MAX.
19159 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
19162 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
19163 TREE_READONLY on result if writing it more than once.
19166 * doc/invoke.texi (-fsanitize=thread): Document that with
19167 -fnon-call-exceptions atomics are not able to throw
19171 * tsan.c: Include tree-eh.h.
19172 (instrument_builtin_call): Call maybe_clean_eh_stmt or
19173 maybe_clean_or_replace_eh_stmt where needed.
19174 (instrument_memory_accesses): Add cfg_changed argument.
19175 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
19176 if it returned true.
19177 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
19179 PR rtl-optimization/63191
19180 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
19181 wrapper function, moved the whole old content into ...
19182 (ix86_delegitimize_address_1): ... this. New inline function.
19183 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
19184 true as last argument instead of ix86_delegitimize_address.
19186 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19188 * config/aarch64/aarch64.c (generic_branch_cost): Copy
19189 cortexa57_branch_cost.
19191 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19193 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
19195 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19198 * doc/md.texi (Constraints): Document wA constraint.
19199 * config/rs6000/constraints.md (wA): New.
19200 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
19201 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
19202 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
19203 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
19205 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
19208 * gimplify.c (is_oacc_declared): New function.
19209 (oacc_default_clause): Use it to set default flags for acc declared
19210 variables inside parallel regions.
19211 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
19212 declared variables.
19213 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
19214 declare attribute to any decl as necessary.
19216 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
19219 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
19220 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
19221 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
19222 (arm_arch_lpae): This.
19223 * config/arm/arm.c (arm_arch7ve): Rename into ...
19224 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
19225 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
19228 2017-03-22 Martin Liska <mliska@suse.cz>
19231 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
19232 error message instead of an ICE.
19234 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19236 * doc/extend.texi (6.11 Additional Floating Types): Revise.
19238 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
19240 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19242 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19245 2017-03-21 Martin Sebor <msebor@redhat.com>
19247 * doc/extend.texi: Use "cannot" instead of "can't."
19248 * doc/hostconfig.texi: Same.
19249 * doc/install.texi: Same.
19250 * doc/invoke.texi: Same.
19251 * doc/loop.texi: Same.
19252 * doc/md.texi: Same.
19253 * doc/objc.texi: Same.
19254 * doc/rtl.texi: Same.
19255 * doc/tm.texi: Same.
19256 * doc/tm.texi.in: Same.
19257 * doc/trouble.texi: Same.
19259 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
19262 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
19263 (collect_checksum_attributes): Set it.
19264 (die_checksum_ordered): Use it.
19266 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19268 PR tree-optimization/79908
19269 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
19270 change: For a VA_ARG whose LHS has been cast away, use
19271 force_gimple_operand to construct the side effects.
19273 2017-03-21 David Malcolm <dmalcolm@redhat.com>
19275 PR translation/80001
19276 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
19277 more amenable to translation.
19278 (oacc_loop_auto_partitions): Likewise.
19280 2017-03-21 Marek Polacek <polacek@redhat.com>
19281 Martin Sebor <msebor@redhat.com>
19283 PR tree-optimization/80109
19284 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
19285 on INTEGRAL_TYPE_P.
19287 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19288 Segher Boessenkool <segher@kernel.crashing.org>
19291 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
19292 check reg_used_between_p between insn and one of succ or succ2
19293 depending on if succ is artificial insn not inserted into insn
19296 2017-03-21 Martin Liska <mliska@suse.cz>
19298 PR gcov-profile/80081
19299 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
19300 * doc/gcc.texi: Include gcov-dump stuff.
19301 * doc/gcov-dump.texi: New file.
19303 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
19305 PR rtl-optimization/79150
19306 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
19307 conditional jump, if the jump is the last insn of the loop.
19309 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19310 Richard Biener <rguenth@suse.de>
19312 PR tree-optimization/79908
19313 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19314 been cast away, use force_gimple_operand to construct the side
19317 2017-03-21 Martin Liska <mliska@suse.cz>
19319 PR libfortran/79956
19320 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
19323 2017-03-21 Brad Spengler <spender@grsecurity.net>
19326 * plugin.c (htab_hash_plugin): New function.
19327 (add_new_plugin): Use it and adjust.
19328 (parse_plugin_arg_opt): Adjust.
19329 (init_one_plugin): Likewise.
19331 2017-03-21 Richard Biener <rguenther@suse.de>
19333 PR tree-optimization/80032
19334 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
19335 if set force the cleanup to happen unconditionally.
19336 (gimplify_target_expr): Push inserted clobbers with force_uncond
19337 to avoid them being removed by control-dependent DCE.
19339 2017-03-21 Richard Biener <rguenther@suse.de>
19341 PR tree-optimization/80122
19342 * tree-inline.c (copy_bb): Do not expans va-arg packs or
19343 va_arg_pack_len when the inlined call stmt requires pack
19345 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
19347 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19350 * tsan.c (instrument_builtin_call): If the memory model argument
19351 is not a constant, assume it is valid.
19354 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
19357 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19359 PR rtl-optimization/79910
19360 * combine.c (can_combine_p): Do not allow combining an I0 or I1
19361 if its dest is used by an insn before I2 (other than the combined
19362 insns themselves, which are properly handled already).
19364 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19367 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19369 * combine.c (record_used_regs): New static function.
19370 (try_combine): Handle situations where there is an additional
19371 instruction between I2 and I3 which needs to have a LOG_LINK
19375 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19377 * combine.c (try_combine): Delete redundant i1 test. Call
19378 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19380 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19383 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
19384 alternatives 13/14.
19386 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19388 PR tree-optimization/80054
19389 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
19390 the optimization if a PHI or any of its arguments is not dominated
19391 by the candidate's basis. Use gphi* rather than gimple* as
19393 (replace_profitable_candidates): Clean up a gimple* variable that
19394 should be a gphi* variable.
19396 2017-03-20 Martin Sebor <msebor@redhat.com>
19399 * doc/extend.texi (attribute constructor): Document present limitation.
19401 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
19404 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
19405 __POWER9_VECTOR__ #ifdef control, change template definition to
19406 use Power9-specific built-in function.
19407 (vec_any_eq): Likewise.
19408 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
19409 to control outcomes from this test.
19410 (vector_ae_<mode>p): For VEC_F modes, likewise.
19412 2017-03-20 Ian Lance Taylor <iant@google.com>
19414 * config/i386/i386.c (ix86_function_regparm): Save an extra
19415 register for -fsplit-stack with DECL_STATIC_CHAIN.
19417 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19420 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
19421 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
19423 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19425 * config/riscv/riscv.c (riscv_print_operand): Use "fence
19427 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
19430 2017-03-20 Marek Polacek <polacek@redhat.com>
19433 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
19435 2017-03-20 Richard Biener <rguenther@suse.de>
19437 PR tree-optimization/80113
19438 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
19439 allocate extra SSA name for PHI def.
19440 (add_close_phis_to_outer_loops): Likewise.
19441 (add_close_phis_to_merge_points): Likewise.
19442 (copy_loop_close_phi_args): Likewise.
19443 (copy_cond_phi_nodes): Likewise.
19445 2017-03-20 Martin Liska <mliska@suse.cz>
19447 PR middle-end/79753
19448 * tree-chkp.c (chkp_build_returned_bound): Do not build
19449 returned bounds for a LHS that's not a BOUNDED_P type.
19451 2017-03-20 Martin Liska <mliska@suse.cz>
19455 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
19456 COMPLEX_CST and VECTOR_CST.
19458 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19461 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
19462 target operand. A new splitter adds the clobber statement in case
19463 the target operand is dead anyway.
19465 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
19467 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
19468 to age-old versions of binutils and glibc.
19470 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
19472 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
19474 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19476 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
19478 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19480 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
19481 requirement for binutils 2.13.
19483 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19485 * combine.c (try_combine): Delete redundant i1 test. Call
19486 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19488 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
19490 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
19491 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
19493 <riscv64-*-elf>: Re-arrange section
19494 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
19495 <riscv32-*-linux>: Likewise.
19496 <riscv64-*-elf>: Likewise
19497 <riscv64-*-linux>: Likewise.
19499 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
19502 * aarch64.opt(verbose-cost-dump): Fix typo.
19504 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
19507 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
19508 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
19510 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19512 * reload.c (find_reloads): When reloading a nonoffsettable address,
19513 use RELOAD_OTHER for it and its address reloads.
19515 PR rtl-optimization/79910
19516 * combine.c (record_used_regs): New static function.
19517 (try_combine): Handle situations where there is an additional
19518 instruction between I2 and I3 which needs to have a LOG_LINK
19521 2017-03-17 Jeff Law <law@redhat.com>
19523 PR tree-optimization/71437
19524 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
19525 conditional in the hash table first.
19526 (vrp_dom_walker::before_dom_children): Extract condition from
19527 ASSERT_EXPR. Record condition, its inverion and any implied
19528 conditions as well.
19530 2017-03-17 Marek Polacek <polacek@redhat.com>
19531 Markus Trippelsdorf <markus@trippelsdorf.de>
19533 PR tree-optimization/80079
19534 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
19537 2017-03-17 Richard Biener <rguenther@suse.de>
19539 PR middle-end/80075
19540 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
19541 Properly verify the LHS before the RHS possibly claims to be
19543 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
19546 2017-03-17 Martin Jambor <mjambor@suse.cz>
19548 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
19549 (List of -O2 options): Likewise.
19550 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
19553 2017-03-17 Tom de Vries <tom@codesourcery.com>
19555 * gcov-dump.c (print_usage): Print bug_report_url.
19557 2017-03-17 Richard Biener <rguenther@suse.de>
19559 PR middle-end/80050
19560 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
19561 (parser::peek): Likewise.
19563 2017-03-17 Richard Biener <rguenther@suse.de>
19565 PR tree-optimization/80048
19566 * sese.c (free_sese_info): Properly release rename_map and
19567 copied_bb_map elements.
19569 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
19571 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
19572 Add linked-list forward and backlinks. Insert on
19573 construction, remove on destruction.
19574 (class pass_store_merging): Add m_stores_head field.
19575 (pass_store_merging::terminate_and_process_all_chains):
19576 Iterate over m_stores_head list.
19577 (pass_store_merging::terminate_all_aliasing_chains):
19579 (pass_store_merging::execute): Check for debug stmts first.
19580 Push new chains onto the m_stores_head stack.
19582 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
19585 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
19586 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
19587 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
19589 2017-03-16 Jeff Law <law@redhat.com>
19591 PR tree-optimization/71437
19592 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
19593 member function. Implementation moved into after_dom_children
19594 member function and into the threader's thread_outgoing_edges
19596 (dom_opt_dom_walker::after_dom_children): Simplify by moving
19597 some code into new thread_outgoing_edges.
19598 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
19599 definition. Simplify marker handling (do it here). Assume we always
19600 have the available expression and the const/copies tables.
19601 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
19603 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
19604 * tree-vrp.c (equiv_stack): No longer file scoped.
19605 (vrp_dom_walker): New class.
19606 (vrp_dom_walker::before_dom_children): New member function.
19607 (vrp_dom_walker::after_dom_children): Likewise.
19608 (identify_jump_threads): Setup domwalker. Use it rather than
19609 walking edges in a random order by hand. Simplify setup/finalization.
19610 (finalize_jump_threads): Remove.
19611 (vrp_finalize): Do not call identify_jump_threads here.
19612 (execute_vrp): Do it here instead and call thread_through_all_blocks
19615 PR tree-optimization/71437
19616 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
19618 (simplify_stmt_for_jump_threading): Add basic_block argument. All
19620 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
19621 (dom_opt_dom_walker::thread_across_edge): Remove
19622 handle_dominating_asserts argument. All callers changed.
19623 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
19624 changes. Remove calls to lhs_of_dominating_assert. Other
19625 uses of handle_dominating_asserts turn into unconditional code
19626 (simplify_control_stmt_condition_1): Likewise.
19627 (simplify_control_stmt_condition): Likewise.
19628 (thread_through_normal_block, thread_across_edge): Likewise.
19629 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
19630 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
19631 object if it is not an SSA_NAME.
19632 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
19633 before calling into the VRP specific simplifiers.
19634 (identify_jump_threads): Remove handle_dominating_asserts
19637 2017-03-16 Jakub Jelinek <jakub@redhat.com>
19640 * tree-diagnostic.c (default_tree_printer): No longer static.
19641 * tree-diagnostic.h (default_tree_printer): New prototype.
19643 2017-03-16 Tamar Christina <tamar.christina@arm.com>
19645 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
19646 Change ins into fmov.
19648 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19650 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
19651 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
19652 Use h_con constraint for operand 1.
19653 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
19654 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
19656 2017-03-15 Jeff Law <law@redhat.com>
19658 PR tree-optimization/71437
19659 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
19660 (record_temporary_equivalences): Use it.
19662 PR tree-optimization/71437
19663 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
19664 tree-ssa-scopedtables.
19665 (lookup_avail_expr, build_and_record_new_cond): Likewise.
19666 (record_conditions, record_cond, vuse_eq): Likewise.
19667 (record_edge_info): Adjust to API tweak of record_conditions.
19668 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
19669 (record_temporary_equivalences, optimize_stmt): Likewise.
19670 (eliminate_redundant_computations): Likewise.
19671 (record_equivalences_from_stmt): Likewise.
19672 * tree-ssa-scopedtables.c: Include options.h and params.h.
19673 (vuse_eq): New function, moved from tree-ssa-dom.c
19674 (build_and_record_new_cond): Likewise.
19675 (record_conditions): Likewise. Accept vector of conditions rather
19676 than edge_equivalence structure for first argument.
19677 for the first argument.
19678 (avail_exprs_stack::lookup_avail_expr): New member function, moved
19679 from tree-ssa-dom.c.
19680 (avail_exprs_stack::record_cond): Likewise.
19681 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
19682 from tree-ssa-dom.c.
19683 (avail_exprs_stack): Add new member functions lookup_avail_expr
19685 (record_conditions): Declare.
19687 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
19690 * lra-constraints.c (process_alt_operands): Increase reject for
19691 reloading an input/output operand.
19693 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
19696 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
19697 insns to convert from signed/unsigned char/short to IEEE 128-bit
19699 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
19701 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
19704 * config/i386/i386.c (ix86_vector_duplicate_value): Create
19705 subreg of inner mode for values already in registers.
19707 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
19709 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
19710 iteration reg is used after the loop.
19712 2017-03-14 Martin Sebor <msebor@redhat.com>
19714 PR tree-optimization/79800
19715 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
19716 precision in negative-positive range.
19717 (format_floating): Call non-const overload with adjusted precision.
19719 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19722 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
19725 2017-03-14 Martin Sebor <msebor@redhat.com>
19727 PR middle-end/80020
19728 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
19729 * builtins.def (aligned_alloc): Use it.
19732 * Makefile.in (GTFILES): Add calls.c.
19733 * calls.c: Include "gt-calls.h".
19735 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
19737 PR rtl-optimization/79728
19738 * regs.h (struct target_regs): New field
19739 x_contains_allocatable_regs_of_mode.
19740 (contains_allocatable_regs_of_mode): New macro.
19741 * reginfo.c (init_reg_sets_1): Initialize it, and change
19742 contains_reg_of_mode so it includes global regs as well.
19743 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
19744 rather than contains_regs_of_mode.
19746 2017-03-14 Martin Liska <mliska@suse.cz>
19748 * doc/invoke.texi: Document options that can't be combined with
19749 -fcheck-pointer-bounds.
19751 2017-03-14 Martin Liska <mliska@suse.cz>
19753 PR middle-end/79831
19754 * doc/invoke.texi (-Wchkp): Document the option.
19756 2017-03-14 Martin Liska <mliska@suse.cz>
19758 * Makefile.in: Install gcov-dump.
19760 2017-03-14 Martin Liska <mliska@suse.cz>
19762 * multiple_target.c (expand_target_clones): Bail out for
19763 an invalid attribute.
19765 2017-03-14 Richard Biener <rguenther@suse.de>
19767 * alias.c (struct alias_set_entry): Pack properly.
19768 * cfgloop.h (struct loop): Likewise.
19769 * cse.c (struct set): Likewise.
19770 * ipa-utils.c (struct searchc_env): Likewise.
19771 * loop-invariant.c (struct invariant): Likewise.
19772 * lra-remat.c (struct cand): Likewise.
19773 * recog.c (struct change_t): Likewise.
19774 * rtl.h (struct address_info): Likewise.
19775 * symbol-summary.h (function_summary): Likewise.
19776 * tree-loop-distribution.c (struct partition): Likewise.
19777 * tree-object-size.c (struct object_size_info): Likewise.
19778 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
19779 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
19780 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
19781 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
19782 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
19783 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
19784 (struct _stmt_vec_info): Likewise.
19786 2017-03-14 Martin Liska <mliska@suse.cz>
19789 * multiple_target.c (create_dispatcher_calls): Check that
19790 a target can create a function dispatcher.
19792 2017-03-14 Martin Liska <mliska@suse.cz>
19795 * multiple_target.c (expand_target_clones): Drop local.local
19796 flag for default implementation.
19798 2017-03-14 Richard Biener <rguenther@suse.de>
19800 PR tree-optimization/80030
19801 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
19803 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
19805 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
19806 gcc_fallthrough() instead of __attribute__((fallthrough));
19808 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19810 * doc/gcc.texi: Remove "up" link to (DIR).
19811 * doc/gccint.texi: Ditto.
19813 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19815 * doc/install.texi (Specific) <avr>: Remove reference to
19818 2017-03-13 Jeff Law <law@redhat.com>
19820 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
19821 attribute rather than comments.
19823 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
19824 match_scratch operand is highest.
19826 2017-03-13 Martin Liska <mliska@suse.cz>
19828 PR middle-end/78339
19829 * ipa-pure-const.c (warn_function_noreturn): If the declarations
19830 is a CHKP clone, use original declaration.
19832 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19834 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
19835 (arc_conditional_register_usage): Use a different allocation order
19836 when optimizing for size.
19837 * common/config/arc/arc-common.c (arc_option_optimization_table):
19838 Section anchors default on when optimizing for size.
19840 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19842 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
19844 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19846 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
19847 * config/arc/arc.md (cpu_facility): Add cd variant.
19848 (*movqi_insn): Add code density variant.
19849 (*movhi_insn): Likewise.
19850 (*movqi_insn): Likewise.
19851 (*addsi3_mixed): Likewise.
19852 (subsi3_insn): Likewise.
19854 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19856 * config/arc/arc.md (movsi_cond_exec): Update constraint.
19858 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19860 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
19861 expressions with MINUS and UNARY ops.
19863 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19866 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
19868 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
19869 between vec_select and vector argument.
19870 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
19871 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
19872 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
19873 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
19874 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
19875 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
19877 2017-03-13 Richard Biener <rguenther@suse.de>
19880 * params.def (vect-max-peeling-for-alignment): Fix typo.
19882 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19884 * doc/install.texi (Specific) <mips-*-*>: Remove description of
19885 issue that only occurred with binutils below 2.18.
19887 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19889 * doc/install.texi (Specific) <cris-axis-elf>: No longer
19890 refer to binutils 2.11/2.12 minimum.
19892 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19894 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
19895 ftp.kernel.org and simplify binutils requirement.
19897 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
19899 * doc/invoke.texi (Warning Options): Fix spelling of link-time
19901 (Optimize Options): Ditto. Also remove redundancy.
19903 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19905 PR translation/79848
19906 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
19908 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
19909 to G_ to avoid double translation.
19911 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19913 PR translation/79923
19914 * auto-profile.c (get_combined_location): Convert leading
19915 character of diagnostics to lower case and remove trailing period.
19916 (read_profile): Likewise for various diagnostics.
19917 * config/arm/arm.c (arm_option_override): Remove trailing period
19918 from various diagnostics.
19919 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
19920 (msp430_expand_delay_cycles): Likewise.
19922 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19925 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
19926 full command-line argument, rather than just "str".
19927 (aarch64_validate_march): Likewise.
19928 (aarch64_validate_mtune): Likewise.
19930 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
19932 PR rtl-optimization/78911
19933 * lra-assigns.c (must_not_spill_p): New function.
19934 (spill_for): Use it.
19936 2017-03-10 Jakub Jelinek <jakub@redhat.com>
19938 PR tree-optimization/79981
19939 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
19940 ATOMIC_COMPARE_EXCHANGE ifn result.
19941 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
19942 IFN_ATOMIC_COMPARE_EXCHANGE.
19944 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19947 * opts.c (parse_sanitizer_options): Add missing question mark to
19948 "did you mean" message.
19950 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19952 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
19954 (VMULEUH_UNS): Likewise.
19955 (VMULOUB_UNS): Likewise.
19956 (VMULOUH_UNS): Likewise.
19957 * config/rs6000/rs6000.c (builtin_function_type): Remove
19958 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
19960 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19963 * read-rtl-function.c (function_reader::read_rtx_operand): Update
19964 x with result of extra_parsing_for_operand_code_0.
19965 (function_reader::extra_parsing_for_operand_code_0): Convert
19966 return type from void to rtx, returning x. When reading
19967 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
19968 larger size containing struct block_symbol.
19970 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
19972 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
19973 -mfloat128-hardware without -m64.
19975 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
19978 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
19979 entries to the case statement that marks unsigned arguments to
19980 overloaded functions.
19982 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19984 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
19985 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
19987 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
19990 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
19991 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
19993 2017-03-10 Martin Liska <mliska@suse.cz>
19997 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
19998 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
20001 2017-03-10 Olivier Hainque <hainque@adacore.com>
20003 * tree-switch-conversion (array_value_type): Start by resetting
20004 candidate type to it's main variant.
20006 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20008 PR rtl-optimization/79909
20009 * combine.c (try_combine): Use simplify_replace_rtx on individual
20010 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
20011 of the whole CALL_INSN_FUNCTION_USAGE.
20013 PR tree-optimization/79972
20014 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
20015 get_range_info on SSA_NAMEs. Formatting fixes.
20017 2017-03-10 Richard Biener <rguenther@suse.de>
20018 Jakub Jelinek <jakub@redhat.com>
20020 PR tree-optimization/77975
20021 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
20022 edge to be constant.
20023 (get_val_for): For constant x return it. Formatting fix.
20024 (loop_niter_by_eval): Avoid pointless looping if the next iteration
20025 would use the same bases as the current one.
20027 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20029 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
20030 instead of vec_select for V1TImode.
20031 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
20033 (VSX_LE_128): Add V1TI to this mode iterator.
20034 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
20035 (*vsx_le_perm_store_<mode>): Likewise.
20036 (pre-reload splitter for VSX stores): Likewise.
20037 (post-reload splitter for VSX stores): Likewise.
20038 (*vsx_xxpermdi2_le_<mode>): Likewise.
20039 (*vsx_lxvd2x2_le_<mode>): Likewise.
20040 (*vsx_stxvd2x2_le_<mode>): Likewise.
20042 2017-03-09 Michael Eager <eager@eagercon.com>
20044 Correct failures with --enable-checking=yes,rtl.
20046 * config/microblaze/microblaze.c (microblaze_expand_shift):
20047 Replace GET_CODE test with CONST_INT_P and INTVAL test with
20048 test for const0_rtx.
20049 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
20050 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
20052 2017-03-09 Richard Biener <rguenther@suse.de>
20054 PR tree-optimization/79977
20055 * graphite-scop-detection.c (scop_detection::merge_sese):
20056 Handle the case of extra exits to blocks dominating the entry.
20058 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
20060 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20063 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
20065 PR rtl-optimization/79949
20066 * lra-constraints.c (process_alt_operands): Check memory when
20067 trying to predict a cycle. Print about the overall increase.
20069 2017-03-09 Richard Biener <rguenther@suse.de>
20071 PR middle-end/79971
20072 * gimple-expr.c (useless_type_conversion_p): Preserve
20073 TYPE_SATURATING for fixed-point types.
20075 2017-03-09 Richard Biener <rguenther@suse.de>
20078 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
20079 alignment of BLKmode params.
20081 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20084 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
20085 (VALL_NO_V2Q): Likewise.
20087 * config/aarch64/aarch64-simd.md
20088 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
20090 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
20091 VALL_NO_V2Q mode iterator.
20092 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
20094 2017-03-09 Martin Liska <mliska@suse.cz>
20096 PR tree-optimization/79631
20097 * tree-chkp-opt.c (chkp_is_constant_addr): Call
20098 tree_int_cst_sign_bit just for INTEGER constants.
20100 2017-03-09 Martin Liska <mliska@suse.cz>
20104 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
20107 2017-03-09 Marek Polacek <polacek@redhat.com>
20110 * tree.c (inchash::add_expr): Handle TREE_VEC.
20112 2017-03-09 Martin Liska <mliska@suse.cz>
20115 (chkp_narrow_size_and_offset): New function.
20116 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
20117 (void chkp_parse_bit_field_ref): New function.
20118 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
20119 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
20121 2017-03-09 Martin Liska <mliska@suse.cz>
20124 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
20125 (chkp_find_bounds_1): Remove gcc_unreachable.
20127 2017-03-09 Jakub Jelinek <jakub@redhat.com>
20130 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
20131 BUILT_IN_SYNC*, determine the access type from the size suffix and
20132 always build a MEM_REF with that type. Handle forgotten
20133 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
20136 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
20137 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
20138 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
20139 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
20140 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
20141 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
20142 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
20143 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
20144 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
20145 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
20146 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
20147 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
20148 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
20149 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
20150 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
20151 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
20152 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
20153 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
20154 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
20155 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
20156 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
20157 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
20158 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
20159 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
20160 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
20161 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
20162 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
20163 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
20164 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
20165 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
20166 definitions outside of __OPTIMIZE__ guarded section.
20169 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
20170 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
20171 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
20172 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
20175 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20177 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
20178 ("vfenez<mode>"): Add missing constraints.
20180 2017-03-08 Martin Sebor <msebor@redhat.com>
20183 * config/nds32/nds32.c (nds32_option_override):
20184 Fix misspelled diagnostic.
20186 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20189 * gimplify.c (gimplify_omp_for): Replace index var in outer
20190 taskloop statement with an artificial variable and add
20191 OMP_CLAUSE_PRIVATE clause for it.
20193 2017-03-08 Richard Biener <rguenther@suse.de>
20195 PR tree-optimization/79955
20196 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
20197 for accesses that are completely outside of the variable.
20199 2017-03-08 Andrew Haley <aph@redhat.com>
20201 PR tree-optimization/79943
20202 * tree-ssa-loop-split.c (compute_new_first_bound): When
20203 calculating the new upper bound, (END-BEG) should be added, not
20206 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20208 * config/avr/avr.md (setmemhi): Make sure match_dup
20209 operand number comes before match_scratch.
20211 2017-03-08 Richard Biener <rguenther@suse.de>
20213 PR tree-optimization/79920
20214 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
20215 with ncopies == 1 to ...
20216 (vect_transform_slp_perm_load): ... here. Properly compute
20217 all element loads by iterating VF times over the group. Do
20218 not handle ncopies (computed in a broken way) in
20219 vect_create_mask_and_perm.
20221 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20224 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
20225 is a uniform vector, use uniform_vector_p return value instead of
20226 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
20228 2017-03-07 Marek Polacek <polacek@redhat.com>
20230 PR middle-end/79809
20231 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
20232 (alloca_call_type): Likewise.
20234 2017-03-07 Martin Liska <mliska@suse.cz>
20236 * gcov.c (process_args): Put comment to correct location.
20238 2017-03-07 Martin Liska <mliska@suse.cz>
20240 PR middle-end/68270
20241 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
20242 Use array_at_struct_end_p instead of DECL_CHAIN (field).
20243 (chkp_narrow_bounds_for_field): Likewise.
20244 (chkp_parse_array_and_component_ref): Pass one more argument to
20247 2017-03-07 Richard Biener <rguenther@suse.de>
20249 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
20252 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
20254 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
20255 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
20257 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20260 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
20261 to end of description.
20262 (PARAM_MAX_STORES_TO_MERGE): Likewise.
20264 2017-03-07 Jakub Jelinek <jakub@redhat.com>
20266 PR rtl-optimization/79901
20267 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
20269 (*avx512f_<code><mode>3<mask_name>): ... this.
20270 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
20271 iterator instead of VI8_AVX2_AVX512BW.
20273 PR rtl-optimization/79901
20274 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
20275 min/max expander, expand it using expand_vec_cond_expr.
20278 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
20281 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20284 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
20285 to void * for PCH reasons.
20286 * dwarf2out.c (output_loc_operands, output_die): Cast
20287 v.val_vec.array to unsigned char *.
20289 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
20292 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
20295 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
20297 PR rtl-optimization/79571
20298 * lra-constraints.c (process_alt_operands): Calculate static
20299 reject and subtract it from overall when only addresses will be
20302 2017-03-06 Julia Koval <julia.koval@intel.com>
20305 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
20306 incoming stack boundary to 128 for 64-bit targets.
20308 2017-03-06 Richard Biener <rguenther@suse.de>
20310 PR tree-optimization/79894
20311 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
20312 to NULL after folding it.
20314 2017-03-06 Richard Biener <rguenther@suse.de>
20316 PR tree-optimization/79824
20317 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
20318 check disabling peeling for gaps.
20320 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
20322 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20323 attributes): Document gettimeofday.
20325 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20327 * config/s390/s390.c (s390_option_override_internal): Set
20328 PARAM_MIN_VECT_LOOP_BOUND
20330 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20332 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
20333 * config/s390/s390.md: Likewise.
20335 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20338 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
20339 (<avx2_avx512>_perm<mode>): Rename to ...
20340 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
20342 (<avx512>_perm<mode>_mask): Rename to ...
20343 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
20345 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
20346 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
20347 instead of VI8F_256_512.
20348 (avx512f_perm<mode>): New define_expand.
20349 (avx512f_perm<mode>_mask): Likewise.
20350 (avx512f_perm<mode>_1<mask_name>): New define_insn.
20351 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
20353 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20355 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
20356 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
20358 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
20360 2017-03-06 Martin Liska <mliska@suse.cz>
20363 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
20364 when having a SSA NAME w/o VAR_DECL assigned to it.
20366 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20368 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
20369 msa_dpsub_<su>_d): Fix MODE for vec_select.
20371 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20373 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
20375 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
20377 2017-03-06 Richard Biener <rguenther@suse.de>
20379 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
20380 * plugin.c (register_plugin_info): Likewise.
20381 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
20383 2017-03-05 Jakub Jelinek <jakub@redhat.com>
20385 * config/i386/sse.md (sse_storehps, sse_storelps,
20386 avx_<castmode><avxsizesuffix>_<castmode>,
20387 avx512f_<castmode><avxsizesuffix>_<castmode>,
20388 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
20389 in condition that at least one operand is not a MEM.
20391 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20393 PR middle-end/79805
20394 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
20395 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
20397 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
20398 gimple_call_nothrow_p flag based on whether original builtin can throw.
20399 If it can, emit following stmts on the fallthrough edge.
20400 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
20401 don't create new bb if inserting just debug stmts on the edge, try to
20402 insert them on the fallthru bb or just reset debug stmts.
20404 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
20407 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
20408 restore recog_data (including the operand rtxes inside it) around
20409 the call to get_insn_template.
20411 2017-03-03 Martin Sebor <msebor@redhat.com>
20413 PR tree-optimization/79699
20414 * context.c (context::~context): Free MPFR caches to avoid
20415 a memory leak on program exit.
20417 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20419 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
20420 Use wide_int::ulow () instead of .elt (0).
20422 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20424 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
20425 (*pushxf): Limit oF constraint to 32bit targets and add oC
20426 constraint for 64bit targets.
20427 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
20428 (*pushdf): Change rmF constraint to rmC.
20430 2017-03-03 Martin Liska <mliska@suse.cz>
20432 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
20433 Remove unused variable.
20435 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20438 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
20439 is a memory operand, increase num_memory.
20440 (ix86_expand_args_builtin): Likewise.
20442 2017-03-03 Jan Hubicka <jh@suse.cz>
20445 * ipa-devirt.c (maybe_record_node): Properly handle
20446 __cxa_pure_virtual visibility.
20448 2017-03-03 Martin Liska <mliska@suse.cz>
20450 PR tree-optimization/79803
20451 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
20453 (pass_loop_prefetch::execute): Disabled optimization if an
20454 assumption about L1 cache size is not met.
20456 2017-03-03 Martin Liska <mliska@suse.cz>
20458 PR rtl-optimization/79574
20459 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
20460 (hash_scan_set): Likewise.
20461 (dump_hash_table): Likewise.
20462 (hoist_code): Likewise.
20464 2017-03-03 Richard Biener <rguenther@suse.de>
20466 * fixed-value.c (fixed_from_string): Restore use of elt (1)
20467 in place of uhigh ().
20468 (fixed_convert_from_real): Likewise.
20470 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20473 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
20475 2017-03-03 Richard Biener <rguenther@suse.de>
20477 PR middle-end/79818
20478 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
20479 TYPE_OVERFLOW_UNDEFINED check.
20481 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20483 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
20485 (vector_ae_<mode>_p): Likewise.
20486 (vector_nez_<mode>_p): Likewise.
20487 (vector_ne_v2di_p): Likewise.
20488 (vector_ae_v2di_p): Likewise.
20489 (vector_ne_<mode>_p): Likewise.
20490 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
20492 (vsx_tsqrt<mode>2_fe): Likewise.
20494 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
20497 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
20499 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20501 PR rtl-optimization/79780
20502 * cprop.c (one_cprop_pass): When second and further conditional trap
20503 in a single basic block is turned into an unconditional trap, turn it
20504 into a deleted note to avoid RTL verification failures.
20506 2017-03-02 Richard Biener <rguenther@suse.de>
20508 * fold-const.c (const_binop): Use ulow () instead of elt (0).
20510 2017-03-02 Richard Biener <rguenther@suse.de>
20512 PR tree-optimization/79345
20514 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
20515 param and abort the walk, returning -1 if it is hit.
20516 (walk_aliased_vdefs): Take a limit param and pass it on.
20517 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
20518 defaulting to 0 and return a signed int.
20519 * tree-ssa-uninit.c (struct check_defs_data): New struct.
20520 (check_defs): New helper.
20521 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
20522 about uninitialized memory.
20523 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
20524 bogus uninitialized warning.
20525 (fixed_convert_from_real): Likewise.
20527 2017-03-02 Bin Cheng <bin.cheng@arm.com>
20529 PR tree-optimization/66768
20530 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
20531 iv_use if base object can't be determined.
20533 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20535 PR tree-optimization/79345
20536 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
20537 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
20538 (get_pattern_stats): Initialize it.
20539 * genemit.c (gen_expand): Verify match_scratch numbers come after
20540 match_operand/match_dup numbers.
20541 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
20542 match_scratch numbers.
20543 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
20545 * config/s390/s390.md (trunctdsd2): Likewise.
20547 2017-03-02 Richard Biener <rguenther@suse.de>
20549 * wide-int.h (wide_int_storage::operator=): Implement in terms
20552 2017-03-02 Richard Biener <rguenther@suse.de>
20554 PR tree-optimization/79777
20555 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
20556 the to insert expression to sth existing.
20558 2017-03-01 Martin Sebor <msebor@redhat.com>
20560 PR middle-end/79692
20561 * gimple-ssa-sprintf.c
20562 (directive::known_width_and_precision): New function.
20563 (format_integer): Use it.
20564 (get_mpfr_format_length): Consider the full range of precision
20565 when computing %g output with the # flag. Set the likely byte
20566 count to 3 rather than 1 when precision is indeterminate.
20567 (format_floating): Correct the lower bound of precision.
20569 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20571 * doc/invoke.texi: Document default code model for 64-bit Linux.
20573 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20576 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
20577 udiv rather than div since input pattern is unsigned.
20579 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
20581 * config/i386/i386.c (print_reg): Warn for values of
20582 unsupported size in integer register.
20584 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
20587 * config/rs6000/predicates.md (current_file_function_operand): Do
20588 not allow self calls to be local if the function is replaceable.
20590 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
20593 * config/rs6000/altivec.h (vec_ctz and others): Change the
20594 preprocessor macro that controls conditional compilation from
20595 _ARCH_PWR9 to __POWER9_VECTOR__.
20596 (vec_all_ne): Change parameterization of __altivec_scalar_pred
20597 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
20598 control (instead of _ARCH_PWR9 control) so that template
20599 definition uses power9-specific function.
20600 (vec_any_eq): Likewise.
20601 (vec_all_ne): Change macro definition to use a power9-specific
20602 expansion under #ifdef __POWER9_VECTOR__ control (instead of
20603 _ARCH_PWR9 control).
20604 (vec_any_eq) Likewise.
20605 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
20606 expansion for CMPNEF to remove support for xvcmpnesp instruction.
20607 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
20608 support for xvcmpnedp instruction.
20609 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
20610 macro expansion so that Power9 implementation of vec_all_ne does
20611 not use the AltiVec predicate framework.
20612 (VCMPNEH_P): Likewise.
20613 (VCMPNEW_P): Likewise.
20614 (VCMPNED_P): Likewise.
20615 (VCMPNEFP_P): Likewise.
20616 (VCMPNEDP_P): Likewise.
20617 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
20618 implementation of vec_any_eq to not use AltiVec predicate
20620 (VCMPAEH_P): Likewise.
20621 (VCMPAEW_P): Likewise.
20622 (VCMPAED_P): Likewise.
20623 (VCMPAEFP_P): Likewise.
20624 (VCMPAEDP_P): Likewise.
20625 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
20626 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
20627 not use the AltiVec predicate framework.
20628 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
20629 of vec_any_eq to not use AltiVec predicate framework.
20630 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20631 support for predefined __POWER9_VECTOR__ macro to indicate that
20632 Power9 instruction selection is enabled.
20633 (altivec_overloaded_builtins): Remove extraneous
20634 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
20635 function argument types RS6000_BTI_bool_V16QI and
20636 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
20637 entry for overloaded function argument types RS6000_BTI_bool_V4SI
20638 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
20639 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
20640 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
20641 Power9 for implementations of vec_cmpne. Change the signature for
20642 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
20643 (representing vec_all_ne) to remove the previously described first
20644 argument of type RS6000_BTI_INTSI, as this was an artifact of
20645 reliance on the AltiVec predicate framework, which is no longer
20646 used in the implementation of these functions. Add
20647 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
20648 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
20649 since, unlike the AltiVec predicate framework implementation, we
20650 do not share function descriptors between vec_alle and vec_anyeq.
20651 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
20652 set of modes that receive special treatment even when
20653 TARGET_P9_VECTOR is true. The special treatment emits code that
20654 does not depend on Power9 instructions.
20655 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
20656 define_expand to not rely on AltiVec predicate framework.
20657 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20659 (vector_ne_v2di_p): Change this define_expand to not rely on
20660 AltiVec predicate framework.
20661 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
20663 (vector_ne_<mode>_p): Change this define_expand to not rely on
20664 AltiVec predicate framework.
20665 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20667 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
20668 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
20669 define_insn pattern.
20670 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
20671 define_insn pattern because the xvcmpne<VSs>. instruction is not
20673 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
20674 instruction is not supported.
20676 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20678 * config/nvptx/nvptx.c: Include intl.h.
20680 2017-03-01 Martin Jambor <mjambor@suse.cz>
20683 * ipa-prop.h (ipa_bits): Removed field known.
20684 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
20685 to pointers. Adjusted their comments to warn about their sharing.
20686 (ipcp_transformation_summary): Change bits to a vector of pointers.
20687 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
20688 (ipa_get_ipa_bits_for_value): Declare.
20689 * tree-vrp.h (value_range): Mark as GTY((for_user)).
20690 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
20691 (ipa_bits_hash_table): Likewise.
20692 (ipa_vr_ggc_hash_traits): Likewise.
20693 (ipa_vr_hash_table): Likewise.
20694 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
20695 being pointers and vr_known being removed.
20696 (ipa_set_jf_unknown): Likewise.
20697 (ipa_get_ipa_bits_for_value): New function.
20698 (ipa_set_jfunc_bits): Likewise.
20699 (ipa_get_value_range): New overloaded functions.
20700 (ipa_set_jfunc_vr): Likewise.
20701 (ipa_compute_jump_functions_for_edge): Use the above functions to
20702 construct bits and vr parts of jump functions.
20703 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
20704 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20706 (ipcp_grow_transformations_if_necessary): Also allocate
20707 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20709 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
20710 them. Fix too long lines.
20711 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
20712 vr_known being removed.
20713 (ipa_read_jump_function): Use new setter functions to construct bits
20714 and vr parts of jump functions or set them to NULL.
20715 (write_ipcp_transformation_info): Adjust for bits being pointers.
20716 (read_ipcp_transformation_info): Likewise.
20717 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
20719 Include gt-ipa-prop.h.
20720 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
20722 (ipcp_store_bits_results): Likewise.
20723 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
20724 Do not write to existing jump functions but use a temporary instead.
20726 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20729 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
20730 attempt to use its first operand as BIT_FIELD_REF base.
20732 2017-03-01 Richard Biener <rguenther@suse.de>
20734 PR middle-end/79721
20735 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
20736 interpolating formula in wrapping arithmetic.
20737 (chrec_apply): Convert chrec_evaluate return value to wanted type.
20739 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20741 PR tree-optimization/79734
20742 * tree-vect-generic.c (expand_vector_condition): Optimize
20743 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
20744 Handle VEC_COND_EXPR where comparison has different inner width from
20745 type's inner width.
20747 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
20749 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
20750 markup, and similar issues. Remove @opindex entries for things
20751 that aren't options. Add missing -mmpy-option entries.
20753 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20755 PR tree-optimization/79737
20756 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
20757 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
20758 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
20759 instead of byte_size. Formatting fix.
20760 (shift_bytes_in_array_right): Formatting fix.
20762 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
20765 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
20766 condition on optimize for the leaf function test.
20768 2017-02-28 Martin Liska <mliska@suse.cz>
20771 * read-rtl-function.c (function_reader::handle_unknown_directive):
20772 Bail out when one uses -flto.
20774 2017-02-28 Martin Liska <mliska@suse.cz>
20776 * common.opt: Replace space with tabular for options of <number>
20778 * config/i386/i386.opt: Show <number> value for
20779 -mlarge-data-threshold.
20780 * opts.c (print_filtered_help): Do not display number in hexadecimal
20783 2017-02-28 Martin Liska <mliska@suse.cz>
20785 * common.opt: Fix --help=option -Q for options which are of
20788 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
20790 * config/i386/i386.c (print_reg): Error out for values
20791 of 8-bit size in invalid integer register.
20793 2017-02-28 Martin Sebor <msebor@redhat.com>
20795 PR tree-optimization/79691
20796 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
20798 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20801 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
20802 gcc_unreachable with output_operand_lossage.
20804 2017-02-28 Richard Biener <rguenther@suse.de>
20806 PR tree-optimization/79740
20807 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
20809 (visit_nary_op): Insert the nary into the hashtable if we
20810 pattern-matched sth.
20811 * tree-ssa-pre.c (eliminate_insert): Robustify.
20813 2017-02-28 Richard Biener <rguenther@suse.de>
20815 PR middle-end/79731
20816 * fold-const.c (decode_field_reference): Reject out-of-bound
20819 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20821 * config/i386/i386.c: Include intl.h.
20822 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
20823 instead of just cond ? "..." : "...".
20824 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
20825 * coverage.c (read_counts_file): Likewise.
20826 * omp-offload.c: Include intl.h.
20827 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
20828 of just cond ? "..." : "...".
20829 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
20830 of just cond ? "..." : "...".
20832 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
20835 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
20837 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
20838 'tune for' CPU name.
20839 * config/arm/arm-cpu-data.h: Regenerated.
20841 2017-02-28 Richard Biener <rguenther@suse.de>
20843 PR tree-optimization/79732
20844 * tree-inline.c (expand_call_inline): Do not shadow var.
20846 2017-02-28 Richard Biener <rguenther@suse.de>
20848 PR tree-optimization/79723
20849 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
20850 address-space properly.
20852 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
20854 * doc/optinfo.texi (Optimization groups): Fix option used for
20856 * doc/invoke.texi (-fopt-info): Document "omp".
20857 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
20858 (OPTGROUP_ALL): Add OPTGROUP_OMP.
20859 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
20860 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
20861 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
20863 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
20865 * dumpfile.c (optgroup_options): Instead of "openmp", associate
20866 OPTGROUP_OMP with "omp".
20868 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
20871 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
20872 for arithmetic shift of unsigned V2DI.
20874 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
20876 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
20877 arc/linux.h headers.
20878 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
20879 (LINK_SPEC): Likewise.
20880 (ARC_TLS_EXTRA_START_SPEC): Likewise.
20881 (EXTRA_SPECS): Likewise.
20882 (STARTFILE_SPEC): Likewise.
20883 (ENDFILE_SPEC): Likewise.
20884 (LIB_SPEC): Likewise.
20885 (TARGET_SDATA_DEFAULT): Likewise.
20886 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
20887 (MULTILIB_DEFAULTS): Likewise.
20888 (DWARF2_UNWIND_INFO): Likewise.
20889 * config/arc/big.h: New file.
20890 * config/arc/elf.h: Likewise.
20891 * config/arc/linux.h: Likewise.
20892 * config/arc/t-uClibc: Remove.
20894 2017-02-27 Bin Cheng <bin.cheng@arm.com>
20896 PR tree-optimization/77536
20897 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
20898 (tree_transform_and_unroll_loop): Use above function to compute the
20899 estimated niter of unrolled loop and use it when scaling profile.
20900 Also use count info rather than frequency if it's non-zero.
20901 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
20902 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
20903 (vect_transform_loop): Call above function.
20905 2017-02-27 Richard Biener <rguenther@suse.de>
20907 PR tree-optimization/45397
20908 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
20909 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
20910 (visit_nary_op): Add pattern matching for CSEing sign-changed
20911 or truncated operations with wider ones.
20913 2017-02-27 Richard Biener <rguenther@suse.de>
20915 PR tree-optimization/79690
20916 * tree-vect-stmts.c (vectorizable_store): Use vector type
20917 built from the DR with address-space.
20919 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
20921 * doc/invoke.texi (Optimize Options): Refine the description
20922 of asan-use-after-return.
20924 2017-02-25 Alan Modra <amodra@gmail.com>
20926 PR rtl-optimization/79584
20927 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
20928 base, not ad->base_term, the reg within base. Remove assertion
20929 that ad->base == ad->base_term. Replace gen_int_mode using
20930 bogus mode with const0_rtx.
20932 2017-02-25 Jakub Jelinek <jakub@redhat.com>
20934 PR middle-end/79396
20935 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
20936 FMA_EXPR like tcc_binary or tcc_unary.
20938 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
20941 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
20943 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
20944 (output_loc_operands): Handle DW_OP_call_ref and
20945 DW_OP_GNU_variable_value.
20946 (struct variable_value_struct): New type.
20947 (struct variable_value_hasher): Likewise.
20948 (variable_value_hash): New variable.
20949 (string_types): Remove.
20950 (copy_loc_descr): New function.
20951 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
20952 (prepend_loc_descr_to_each): New function.
20953 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
20954 instead of add_loc_descr_to_each if the first argument is single
20955 location list and the second has multiple.
20956 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
20957 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
20958 when looking for variable value which doesn't have other location info.
20959 (loc_list_from_tree): Formatting fix.
20960 (gen_array_type_die): Simplify DW_AT_string_length handling.
20961 (adjust_string_types): Remove.
20962 (gen_subprogram_die): Don't call adjust_string_types nor test/set
20963 string_types. Call resolve_variable_values.
20964 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
20965 (resolve_addr_in_expr): Likewise. Add A argument.
20966 (copy_deref_exprloc): Remove deref argument. Adjust for the
20967 original expression being DW_OP_GNU_variable_value with optionally
20968 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
20969 optionally after it.
20970 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
20971 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
20972 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
20973 (variable_value_hasher::hash, variable_value_hasher::equal): New
20975 (resolve_variable_value_in_expr, resolve_variable_value,
20976 resolve_variable_values, note_variable_value_in_expr,
20977 note_variable_value): New functions.
20978 (dwarf2out_early_finish): Call note_variable_value on all toplevel
20981 2017-02-24 Jakub Jelinek <jakub@redhat.com>
20984 * opts.h (handle_generated_option): Add GENERATED_P argument.
20985 * opts-common.c (handle_option): Adjust function comment.
20986 (handle_generated_option): Add GENERATED_P argument, pass it to
20988 (control_warning_option): Pass false to handle_generated_option
20990 * opts.c (maybe_default_option): Pass true to handle_generated_option
20992 * optc-gen.awk: Likewise.
20994 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20996 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
20997 a REG, look at the REG it is a SUBREG of.
20998 (splitter for cmpeqsi_t): Ditto.
21000 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21002 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
21003 the special USEs with the pattern of the insn, not the insn itself.
21005 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
21008 * doc/invoke.texi: Document -mload-store-pairs.
21010 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21011 Sandra Loosemore <sandra@codesourcery.com>
21013 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
21014 argument isn't a CONST_INT.
21015 (nios2_alternate_compare_const): Assert op is a CONST_INT.
21016 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
21017 (nios2_validate_compare): Bypass alternate compare logic if *op2
21018 is not a CONST_INT.
21019 (ldstwm_operation_p): Return false if first_base is not a REG or
21020 if first_offset is not a CONST_INT.
21022 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21024 * config/cris/cris.md: Use correct operand in a define_peephole2.
21026 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21028 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
21030 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21032 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
21033 this_insn if it is an INSN or JUMP_INSN.
21034 (force_offsettable): Look at base, not at addr.
21035 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
21036 on things that aren't necessarily CONST_INTs.
21038 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
21040 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
21041 -mfpmath=sse is the default also for x86-32 targets with SSE2
21042 instruction set when @option{-ffast-math} is enabled
21044 2017-02-24 Jeff Law <law@redhat.com>
21046 PR rtl-optimizatoin/79286
21047 * ira.c (update_equiv_regs): Drop may_trap_p exception to
21050 2017-02-24 Richard Biener <rguenther@suse.de>
21052 PR tree-optimization/79389
21053 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
21056 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
21058 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
21059 function comment to reflect reality.
21060 (loop_exits_before_overflow): Fix typo in function description.
21062 2017-02-24 Richard Biener <rguenther@suse.de>
21064 PR tree-optimization/79389
21065 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
21066 properly that a threading opportunity exists. Detect conditional
21067 copy/constant propagation opportunities.
21069 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
21071 * config/visium/visium.md (type): Add trap.
21072 (b): New mode attribute.
21073 (*btst): Rename into...
21074 (*btst<mode>): ...this and adjust.
21075 (*cbranchsi4_btst_insn): Rename into...
21076 (*cbranch<mode>4_btst_insn): ...this and adjust.
21077 (trap): New define_insn.
21079 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21081 PR tree-optimization/79389
21082 * ifcvt.c (struct noce_if_info): Add rev_cond field.
21083 (noce_reversed_cond_code): New function.
21084 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
21085 reversed_comparison_code. Formatting fix.
21086 (noce_try_store_flag): Test rev_cond != NULL in addition to
21087 reversed_comparison_code.
21088 (noce_try_store_flag_constants): Likewise.
21089 (noce_try_store_flag_mask): Likewise.
21090 (noce_try_addcc): Use rev_cond if non-NULL instead of
21091 reversed_comparison_code.
21092 (noce_try_cmove_arith): Likewise. Formatting fixes.
21093 (noce_try_minmax, noce_try_abs): Clear rev_cond.
21094 (noce_find_if_block): Initialize rev_cond.
21095 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
21096 instead of false as last argument never attempt to reverse it
21099 2017-02-23 Bin Cheng <bin.cheng@arm.com>
21101 PR tree-optimization/79663
21102 * tree-predcom.c (combine_chains): Process refs in reverse order
21103 only for ZERO length chains, and add explaining comment.
21105 2017-02-23 Jeff Law <law@redhat.com>
21107 PR tree-optimization/79578
21108 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
21109 in call to operand_equal_p.
21111 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
21114 * config/i386/cpuid.h: Fix another undefined behavior.
21116 2017-02-23 Richard Biener <rguenther@suse.de>
21118 PR tree-optimization/79683
21119 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
21120 vector types for data-refs.
21122 2017-02-23 Martin Liska <mliska@suse.cz>
21124 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
21126 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21128 PR middle-end/79665
21129 * internal-fn.c (get_range_pos_neg): Moved to ...
21130 * tree.c (get_range_pos_neg): ... here. No longer static.
21131 * tree.h (get_range_pos_neg): New prototype.
21132 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
21133 are known to be in between 0 and signed maximum inclusive, try to
21134 expand both unsigned and signed divmod and use the cheaper one from
21137 2017-02-22 Jeff Law <law@redhat.com>
21139 PR tree-optimization/79578
21140 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
21141 to compare base operands.
21143 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
21146 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
21147 gpc_reg_operand instead of fpr_reg_operand.
21149 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
21151 * config/mips/mips.c (mips_return_in_memory): Force FP
21152 vector types to be returned in memory for o32 ABI.
21154 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21156 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
21157 instead of DW_TAG_member for static data member declarations and don't
21158 set no_linkage_name for static inline data members.
21159 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
21162 2017-02-22 Martin Liska <mliska@suse.cz>
21164 * doc/invoke.texi: Replace inequality signs with square brackets
21167 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21169 PR tree-optimization/68644
21170 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
21172 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
21175 * lra-constraints.c (simplify_operand_subreg): Handle
21176 WORD_REGISTER_OPERATIONS targets.
21178 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21181 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
21182 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
21183 elimination by swapping fld*.
21185 2017-02-22 Richard Biener <rguenther@suse.de>
21187 PR tree-optimization/79673
21188 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
21189 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
21190 irrelevant address-space qualifiers and avoiding a
21191 ADDR_SPACE_CONVERT_EXPR from fold_convert.
21193 2017-02-22 Richard Biener <rguenther@suse.de>
21195 PR tree-optimization/79666
21196 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
21197 to not symbolically negate if that may introduce undefined
21200 2017-02-22 Martin Liska <mliska@suse.cz>
21203 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
21204 * data-streamer-out.c (streamer_write_gcov_count_stream):
21206 * value-prof.c (stream_out_histogram_value): Make assert more
21207 precise based on type of counter.
21209 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
21212 * config/i386/i386.md (standard_x87sse_constant_load splitter):
21213 Use nonimmediate_operand instead of memory_operand for operand 1.
21214 (float-extend standard_x87sse_constant_load splitter): Ditto.
21216 2017-02-21 Jeff Law <law@redhat.com>
21218 PR tree-optimization/79621
21219 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
21220 blocks with edges to themselves.
21222 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21225 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
21226 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
21227 Use gimple_call_builtin_p.
21230 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
21231 on temporarily removed DEBUG_INSNs.
21233 PR tree-optimization/79649
21234 * tree-loop-distribution.c (classify_partition): Give up on
21235 non-generic address space loads/stores.
21237 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
21239 * doc/loop.texi (Loop manipulation): Remove nonexistent
21240 tree_ssa_loop_version from the documentation.
21241 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
21243 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21246 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
21247 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
21248 * config/rs6000/rs6000.c: Include except.h.
21249 (rs6000_expand_split_stack_prologue): Call
21250 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
21252 2017-02-21 Martin Jambor <mjambor@suse.cz>
21255 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
21256 have been analyzed.
21258 2017-02-21 Martin Jambor <mjambor@suse.cz>
21260 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
21261 for backward compatibility only.
21262 * doc/invoke.texi (Option Summary): Remove all references to
21263 -fipa-cp-alignment.
21265 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
21269 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21271 * lra-constraints.c (curr_insn_transform): Handle
21272 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21274 2017-02-21 Martin Liska <mliska@suse.cz>
21276 * config/i386/i386.opt: Replace -masm-dialect with -masm.
21278 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
21280 PR translation/79638
21281 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
21283 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
21286 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
21287 (arm_function_ok_for_sibcall): Return false for an indirect call by
21288 descriptor if all the argument registers are used.
21289 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
21290 alignment of the function.
21292 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21294 PR tree-optimization/61441
21295 * simplify-rtx.c (simplify_const_unary_operation): For
21296 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
21297 the sNaN unmodified.
21299 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21301 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21302 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21303 instead of SYSTEM_HEADER_DIR.
21305 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
21306 Martin Liška <mliska@suse.cz>
21308 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
21309 Fix typos and grammar, use active voice, and clarify.
21311 2017-02-20 Marek Polacek <polacek@redhat.com>
21313 PR middle-end/79537
21314 * gimplify.c (gimplify_expr): Handle unused *&&L;.
21317 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
21319 2017-02-20 Jakub Jelinek <jakub@redhat.com>
21322 * config/i386/i386.c (ix86_expand_builtin): Handle
21323 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
21324 ix86_builtins_isa[fcode].isa as a requirement of those
21325 flags and any other flag in the bitmask.
21326 (ix86_init_mmx_sse_builtins): Use 0 instead of
21327 ~OPTION_MASK_ISA_64BIT as mask.
21328 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
21329 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
21330 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
21331 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
21333 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21336 * lra-constraints.c (split_reg): Check requested split mode
21337 is supported by the register.
21339 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21341 * lra-constraints.c (simplify_operand_subreg): Remove early
21344 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21347 * lra-constraints.c (curr_insn_transform): Tighten condition
21348 for converting SUBREG reloads from OP_OUT to OP_INOUT.
21350 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21353 * lra-constraints.c (curr_insn_transform): Handle
21354 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21356 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
21359 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
21361 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
21363 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
21366 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
21369 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21371 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
21374 2017-02-18 Jakub Jelinek <jakub@redhat.com>
21376 * final.c (last_columnnum, override_columnnum): New variables.
21377 (final_start_function): Set last_columnnum, pass it to begin_prologue
21378 hook and pass 0 to dwarf2out_begin_prologue.
21379 (final_scan_insn): Update override_columnnum. Pass last_columnnum
21380 to source_line debug hook.
21381 (notice_source_line): Compute last_columnnum and for debug_column_info
21382 return true on column changes.
21383 * debug.h (struct gcc_debug_hooks): Add column argument to
21384 source_line and begin_prologue hooks.
21385 (debug_nothing_int_charstar_int_bool): Remove prototype.
21386 (debug_nothing_int_int_charstar,
21387 debug_nothing_int_int_charstar_int_bool): New prototypes.
21388 (dwarf2out_begin_prologue): Add column argument.
21389 * debug.c (do_nothing_debug_hooks): Adjust source_line and
21390 begin_prologue hooks.
21391 (debug_nothing_int_charstar_int_bool): Remove.
21392 (debug_nothing_int_int_charstar,
21393 debug_nothing_int_int_charstar_int_bool): New functions.
21394 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
21395 through to dwarf2out_source_line.
21396 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
21397 (dwarf2out_source_line): Add column argument, emit it if requested.
21398 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
21400 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21401 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21402 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
21403 through to dwarf2out_begin_prologue.
21404 (vmsdbgout_source_line): Add column argument, pass it through to
21405 dwarf2out_source_line.
21406 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
21407 dbxout_source_line caller.
21408 (dbxout_source_line): Add column argument.
21410 * common.opt (gno-column-info, gcolumn-info): New options.
21411 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
21412 (check_die): Also test for multiple DW_AT_decl_column attributes.
21413 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
21414 DW_AT_decl_column if requested.
21415 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
21417 (gen_variable_die): Likewise.
21418 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
21419 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
21422 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
21423 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
21424 (ix86_handle_option): Handle OPT_m3dnowa.
21425 * doc/invoke.texi (-m3dnowa): Document.
21426 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
21427 -m3dnowa instead of -m3dnow -march=athlon.
21430 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
21431 instead of gcc_assert for K, r and R code checks. Formatting fixes.
21433 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21436 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
21437 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
21438 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
21439 generator for vsx_xxpermdi_<mode>_be.
21440 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
21441 force big-endian semantics.
21442 (vsx_xxpermdi_<mode>_be): New define_expand with same
21443 implementation as previous version of vsx_xxpermdi_<mode>.
21445 2017-02-17 Jakub Jelinek <jakub@redhat.com>
21447 PR tree-optimization/79327
21448 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
21449 variable, its initialization and use.
21451 2017-02-17 Julia Koval <julia.koval@intel.com>
21453 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
21454 (OPTION_MASK_ISA_PKU_UNSET): New.
21455 (ix86_handle_option): Handle -mrdpid.
21456 * config/i386/cpuid.h (bit_RDPID): New.
21457 * config/i386/driver-i386.c (host_detect_local_cpu):
21458 Detect RDPID feature.
21459 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
21460 * config/i386/i386-c.c (ix86_target_macros_internal):
21462 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
21463 (ix86_valid_target_attribute_inner_p): Add "rdpid".
21464 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
21465 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
21466 * config/i386/i386.md (define_insn "rdpid"): New.
21467 * config/i386/i386.opt Add -mrdpid.
21468 * config/i386/immintrin.h (_rdpid_u32): New.
21470 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
21472 PR rtl-optimization/79541
21473 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
21474 instead of transforming it into USE.
21476 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
21478 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
21479 If HONOR_SNANS (SFmode) force the input to a register.
21480 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
21481 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
21482 an frsp or similar insn.
21484 2017-02-17 Martin Liska <mliska@suse.cz>
21486 PR rtl-optimization/79577
21487 * params.def (selsched-max-sched-times): Increase minimum to 1.
21489 2017-02-17 Martin Liska <mliska@suse.cz>
21491 PR rtl-optimization/79574
21492 * gcse.c (want_to_gcse_p): Prevent integer overflow.
21494 2017-02-17 Martin Liska <mliska@suse.cz>
21496 PR tree-optimization/79529
21497 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
21498 ssa_defined_default_def_p to handle cases which are implicitly
21500 * tree-ssa.c (ssa_defined_default_def_p): New function.
21501 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
21502 which are implicitly defined.
21503 * tree-ssa.h (ssa_defined_default_def_p): Declare.
21505 2017-02-17 Richard Biener <rguenther@suse.de>
21507 PR middle-end/79576
21508 * params.def (max-ssa-name-query-depth): Limit to 10.
21510 2017-02-17 Richard Biener <rguenther@suse.de>
21512 PR tree-optimization/79552
21513 * tree-ssa-structalias.c (visit_loadstore): Properly verify
21516 2017-02-17 Richard Biener <rguenther@suse.de>
21519 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
21521 2017-02-17 Marek Polacek <polacek@redhat.com>
21523 PR middle-end/79536
21524 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
21525 (fold_negate_expr): New wrapper.
21527 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
21529 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
21530 Correct terminology and de-emphasize pre-standard behavior.
21532 2017-02-16 Alan Modra <amodra@gmail.com>
21534 PR rtl-optimization/79286
21535 * ira.c (def_dominates_uses): New function.
21536 (update_equiv_regs): Don't create an equivalence for insns that
21537 may trap where the register def does not dominate the use.
21539 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
21541 PR rtl-optimization/78127
21542 * lra.c (lra): Call lra_eliminate before finish the loop after
21545 2017-02-16 Richard Biener <rguenther@suse.de>
21547 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
21549 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
21550 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
21551 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
21552 (isl_val_int_from_wi): New function.
21553 (extract_affine_gmp): Rename to ...
21554 (extract_affine_wi): ... this, take a widest_int.
21555 (extract_affine_int): Just wrap extract_affine_wi.
21556 (add_param_constraints): Use isl_val_int_from_wi.
21557 (add_loop_constraints): Likewise, and extract_affine_wi.
21559 2017-02-15 Jeff Law <law@redhat.com>
21561 PR middle-end/79521
21562 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
21563 ira_init_register_move_cost_if_necessary.
21565 2017-02-15 Martin Sebor <msebor@redhat.com>
21567 PR middle-end/32003
21568 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
21569 removed in a prior commit.
21571 2017-02-15 Bin Cheng <bin.cheng@arm.com>
21573 PR tree-optimization/79347
21574 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
21575 counters during peeling.
21577 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
21579 * Makefile.in (site.exp): Remove "set ISLVER".
21581 2017-02-15 Jakub Jelinek <jakub@redhat.com>
21584 * real.c (real_from_integer): Call real_convert even for decimal.
21586 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
21589 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
21591 2017-02-14 Andrew Pinski <apinski@cavium.com>
21593 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
21594 cores and change the partno/implementer to be correct.
21595 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
21596 the 'B" as the implementer.
21597 * config/aarch64/aarch64-tune.md: Regenerate.
21599 2017-02-14 Carl Love <cel@us.ibm.com>
21601 * config/rs6000/rs6000.c: Add case statement entry to make the
21602 xvcvuxdsp built-in argument unsigned.
21603 * config/rs6000/vsx.md: Fix the source and return operand types so they
21604 match the instruction definitions from the ISA document. Fix typo
21605 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
21608 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
21611 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
21612 member early_clobber_alts.
21613 * lra-lives.c (reg_early_clobber_p): New.
21614 (process_bb_lives): Use it.
21615 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
21616 (debug_operand_data): Initialize early_clobber_alts.
21617 (setup_operand_alternative): Set up early_clobber_alts.
21618 (collect_non_operand_hard_regs): Ditto. Pass early clobber
21619 alternatives to new_insn_reg.
21620 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
21622 (lra_update_insn_regno_info): Pass the new arg.
21624 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21626 PR middle-end/79505
21627 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
21628 (new_oacc_loop_raw): Don't clear already cleared fields.
21631 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
21632 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
21633 _mm512_prefetch_i64gather_ps): New inline functions and macros.
21635 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
21638 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
21640 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
21643 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
21644 the extra instruction to the right place to store 128-bit constant
21647 2017-02-14 Martin Sebor <msebor@redhat.com>
21649 PR middle-end/79448
21650 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
21651 warning for strings of unknown length.
21653 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
21655 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
21657 2017-02-14 Jeff Law <law@redhat.com>
21660 * ira-costs.c (scan_one_insn): Initialize register move costs
21661 for pseudos seen in USE/CLOBBER insns.
21663 PR tree-optimization/79095
21664 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
21665 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
21666 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
21667 if the operands are known to be not equal, then the resulting range
21669 (intersect_ranges): If the new range is ~[0,0] and the old range is
21670 wide, then prefer ~[0,0].
21671 * tree-vrp.c (overflow_comparison_p_1): New function.
21672 (overflow_comparison_p): New function.
21673 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
21674 if NAME is used in an overflow test.
21675 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
21676 overflow check that can be expressed as an equality test, then adjust
21677 ops to be that equality test.
21679 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21681 * config/s390/s390-builtin-types.def: Remove flags argument.
21682 * config/s390/s390.c (s390_init_builtins): Likewise.
21684 2017-02-14 Martin Liska <mliska@suse.cz>
21686 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
21687 vector. Fix trailing white spaces.
21689 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
21691 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
21694 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21696 PR rtl-optimization/68664
21697 * config/arm/arm.c (arm_sched_can_speculate_insn):
21698 New function. Declare prototype.
21699 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21701 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21703 PR rtl-optimization/68664
21704 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
21706 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21708 2017-02-14 Amit Pawar <amit.pawar@amd.com>
21710 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
21711 max skip bytes for function, loop and jump.
21713 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21715 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
21716 ABS_EXPR for gimple dump.
21718 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21721 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
21723 PR tree-optimization/79408
21724 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
21725 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
21726 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
21727 also if rhs1 is INTEGER_CST.
21729 2017-02-14 Richard Biener <rguenther@suse.de>
21731 PR middle-end/79432
21732 * tree-into-ssa.c (insert_phi_nodes): When the function can
21733 have abnormal edges rewrite SSA names with broken use-def
21734 dominance out of SSA and register them for PHI insertion.
21736 2017-02-13 Martin Sebor <msebor@redhat.com>
21738 PR middle-end/79496
21739 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
21740 clearing info.nowrite flag when snprintf size argument is a range.
21742 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21744 * cprop.c (cprop_jump): Add missing space in string literal.
21745 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
21746 (get_constraint_for_component_ref): Likewise.
21747 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
21748 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
21749 * lra-constraints.c (process_alt_operands): Likewise.
21750 * ipa-inline.c (inline_small_functions): Likewise.
21751 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
21752 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21753 * trans-mem.c (diagnose_tm_1_op): Likewise.
21754 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
21755 (grid_parallel_clauses_gridifiable): Likewise.
21757 * config/nvptx/mkoffload.c (process): Add space in between
21760 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
21761 "MOD4_SSE_REGS" and "ALL_REGS".
21763 * spellcheck.c (test_data): Add , in between "foo" and "food".
21765 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21768 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
21769 boundary crossing check and subsequent code generation agree.
21771 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21773 * config/aarch64/aarch64.c (has_memory_op): Delete.
21774 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
21777 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21779 PR rtl-optimization/79388
21780 PR rtl-optimization/79450
21781 * combine.c (distribute_notes): When removing TEM_INSN for which
21782 corresponding dest has last value recorded, invalidate that last
21785 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21787 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
21788 of explicit '@'. Add missing assembly comment marker on branch costs
21791 2017-02-13 Nathan Sidwell <nathan@acm.org>
21793 * gengtype-lex.l (<in_struct>): Add '/'.
21795 2017-02-13 Martin Liska <mliska@suse.cz>
21798 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
21800 2017-02-13 Richard Biener <rguenther@suse.de>
21802 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
21804 * configure: Re-generate.
21805 * config.in: Likewise.
21806 * graphite-dependences.c: Simplify as if
21807 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
21808 * graphite-isl-ast-to-gimple.c: Likewise.
21809 * graphite-optimize-isl.c: Likewise.
21810 * graphite-poly.c: Likewise.
21811 * graphite-sese-to-poly.c: Likewise.
21812 * graphite.h: Likewise.
21813 * toplev.c: Include isl/version.h and use isl_version () for
21814 printing the ISL version.
21815 * doc/install.texi: Update ISL requirement.
21817 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21819 * doc/standards.texi (Standards): Update reference to
21822 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21824 * doc/extend.texi (Named Address Spaces): sourceware.org now
21826 * doc/install.texi (Binaries): Ditto.
21829 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21831 * doc/cpp.texi: Replace "stringify"/"stringification" with C
21832 standard terminology "stringize"/"stringizing" throughout.
21833 * doc/cppinternals.texi: Likewise.
21835 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21837 * doc/extend.texi: Fix some spelling mistakes and typos.
21838 * doc/invoke.texi: Likewise.
21840 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21843 * params.def (inline-min-speedup) Change from 10 to 8.
21845 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21847 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
21850 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21853 * ipa-inline-analysis.c (get_minimal_bb): New function.
21854 (record_modified): Use it.
21855 (remap_edge_change_prob): Handle also ancestor functions.
21857 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
21859 * doc/contrib.texi (Contributors): Remove broken link into
21860 the Mauve CVS repository.
21862 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21864 PR middle-end/79454
21865 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
21866 result computation whenever lhs doesn't have vector mode, not
21867 just when it has BLKmode.
21869 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21871 * doc/makefile.texi (profiledbootstrap): Refer to the
21872 installation instructions only in textual form.
21874 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21877 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
21879 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21881 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
21882 (Specific): Update mingw-w64 reference.
21884 (Specific): Remove broken link to Renesas RX processor.
21886 2017-02-10 Richard Biener <rguenther@suse.de>
21888 * toplev.c (process_options): Do not mention obsolete graphite
21889 options when printing sorry message about missing graphite support.
21890 Mention -floop-nest-optimize.
21892 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
21894 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
21895 (vtst_p16): Likewise.
21896 (vtstq_p8): Likewise.
21897 (vtstq_p16): Likewise.
21899 (vtstq_p64): Likewise.
21900 * config/arm/arm_neon.h (vgetq_lane_p64): New.
21901 (vset_lane_p64): New.
21902 (vsetq_lane_p64): New.
21904 2017-02-10 Jakub Jelinek <jakub@redhat.com>
21906 PR tree-optimization/79411
21907 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
21908 stmt operands are SSA_NAMEs used in abnormal phis.
21909 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
21912 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21915 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
21918 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21920 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
21922 2017-02-09 Jakub Jelinek <jakub@redhat.com>
21924 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
21928 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
21929 not arbitrary TREE_CONSTANT.
21932 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
21933 "omp declare target link" attribute unless is_global_var.
21934 * omp-offload.c (find_link_var_op): Likewise.
21936 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
21937 Chung-Lin Tang <cltang@codesourcery.com>
21939 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
21941 (gimplify_adjust_omp_clauses): Don't delete TILE.
21942 (gimplify_omp_for): Deal with TILE.
21943 * internal-fn.c (expand_GOACC_TILE): New function.
21944 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
21946 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
21947 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
21949 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
21950 avoid DIV for outermost collapse var.
21951 (expand_oacc_for): Insert tile element loop as needed. Adjust.
21952 Remove out of date comments, fix whitespace.
21953 * omp-general.c (omp_extract_for_data): Deal with tiling.
21954 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
21955 adjust OLF_DIM_BASE value.
21956 (struct omp_for_data): Add tiling field.
21957 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
21958 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
21959 for auto loops. Remove default auto determining, moved to
21960 oacc_loop_fixed_partitions.
21961 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
21962 stmts, add e_mask field.
21963 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
21964 (oacc_thread_numbers): Use oacc_dim_call.
21965 (oacc_xform_tile): New.
21966 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
21967 (finish_oacc_loop): Adjust for ifns vector.
21968 (oacc_loop_discover_walk): Append loop abstraction sites to list,
21969 add case for GOACC_TILE fns.
21970 (oacc_loop_xform_loop): Delete.
21971 (oacc_loop_process): Iterate over call list directly, and add
21972 handling for GOACC_TILE fns.
21973 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
21975 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
21976 vector partitioning to outer loops. Assign 2 partitions to loops
21977 when available. Add TILE handling.
21978 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
21979 (execite_oacc_device_lower): Process GOACC_TILE fns,
21980 ignore unknown specs.
21981 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
21982 * tree.c (omp_clause_num_ops): Adjust TILE ops.
21983 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
21985 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
21987 * configure.ac (ACX_BUGURL): Update.
21988 * configure: Regenerate.
21990 2017-02-09 Richard Biener <rguenther@suse.de>
21992 PR tree-optimization/69823
21993 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
21994 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
21996 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
21998 * config/arc/arc-c.def: Add __NPS400__ definition.
21999 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
22000 (TARGET_NPS400): Define.
22002 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22004 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
22006 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
22007 pointer, arch_info.
22008 (arc_cpu_types): Fill the arch_info field with a pointer into the
22009 arc_arch_types table.
22010 (arc_selected_cpu): Declare.
22011 * config/arc/arc.c (arc_selected_cpu): Make global.
22012 (arc_selected_arch): Delete.
22013 (arc_base_cpu): Delete.
22014 (arc_override_options): Remove references to deleted variables,
22015 update access to arch information.
22016 (ARC_OPT): Update access to arch information.
22017 (ARC_OPTX): Likewise.
22018 * config/arc/arc.h (arc_base_cpu): Remove declaration.
22019 (TARGET_ARC600): Update access to arch information.
22020 (TARGET_ARC601): Likewise.
22021 (TARGET_ARC700): Likewise.
22022 (TARGET_EM): Likewise.
22023 (TARGET_HS): Likewise.
22024 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
22027 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
22030 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
22031 condition/operands for integer GE/LE/GEU/LEU operations.
22033 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
22035 PR translation/79397
22036 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
22039 2017-02-08 Martin Jambor <mjambor@suse.cz>
22042 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
22043 whether allocation happened.
22044 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
22045 nothing was allocated.
22047 2017-02-08 Jakub Jelinek <jakub@redhat.com>
22049 PR tree-optimization/79408
22050 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
22051 constant, but SSA_NAME with a known integer range, use the minimum
22052 of that range instead of op1 to determine if modulo can be replaced
22053 with its first operand.
22055 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22057 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
22059 2017-02-08 Richard Biener <rguenther@suse.de>
22061 PR tree-optimization/71824
22062 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22063 Check all loops contained in the merged region.
22065 2017-02-07 Andrew Pinski <apinski@cavium.com>
22067 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
22069 2017-02-07 Andrew Pinski <apinski@cavium.com>
22071 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
22072 (thunderxt88): Likewise.
22073 (thunderxt81): Disable LSE and change v8.1 to v8.
22074 (thunderxt83): Likewise.
22076 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22077 Richard Biener <rguenther@suse.de>
22079 PR middle-end/79399
22080 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
22081 type from int to size_t.
22082 * ira-costs.c (struct_costs_size): Change type from int to size_t.
22084 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22086 PR rtl-optimization/79386
22087 * cprop.c (bypass_conditional_jumps): Initialize
22088 bypass_last_basic_block already before splitting bbs after
22089 unconditional traps...
22090 (bypass_conditional_jumps): ... rather than here.
22093 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
22094 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
22095 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
22096 fix -masm=intel patterns.
22098 2017-02-07 Richard Biener <rguenther@suse.de>
22100 PR tree-optimization/79256
22101 PR middle-end/79278
22102 * builtins.c (get_object_alignment_2): Use min_align_of_type
22103 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
22104 and ADJUST_FIELD_ALIGN.
22106 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
22108 * doc/tm.texi: Regenerate.
22109 * stor-layout.c (layout_decl): Adjust.
22110 (update_alignment_for_field): Likewise.
22111 (place_field): Likewise.
22112 (min_align_of_type): Likewise.
22113 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
22114 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
22115 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
22116 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
22117 * config/frv/frv.c (frv_adjust_field_align): Likewise.
22118 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
22119 * config/i386/i386.c (x86_field_alignment): Likewise.
22120 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
22121 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
22122 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
22123 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
22124 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
22125 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
22129 2017-01-30 Richard Biener <rguenther@suse.de>
22131 PR tree-optimization/79256
22132 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22133 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22136 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
22138 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
22139 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
22140 builtins to SImode and emit a zero-extend, if necessary.
22142 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22144 * docs/invoke.texi (RISC-V Options): Alphabetize.
22146 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22148 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
22151 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22153 * config/riscv/riscv.c: New file.
22154 * common/config/riscv/riscv-common.c: Likewise.
22155 * config.gcc: Likewise.
22156 * config/riscv/constraints.md: Likewise.
22157 * config/riscv/elf.h: Likewise.
22158 * config/riscv/generic.md: Likewise.
22159 * config/riscv/linux.h: Likewise.
22160 * config/riscv/multilib-generator: Likewise.
22161 * config/riscv/peephole.md: Likewise.
22162 * config/riscv/pic.md: Likewise.
22163 * config/riscv/predicates.md: Likewise.
22164 * config/riscv/riscv-builtins.c: Likewise.
22165 * config/riscv/riscv-c.c: Likewise.
22166 * config/riscv/riscv-ftypes.def: Likewise.
22167 * config/riscv/riscv-modes.def: Likewise.
22168 * config/riscv/riscv-opts.h: Likewise.
22169 * config/riscv/riscv-protos.h: Likewise.
22170 * config/riscv/riscv.h: Likewise.
22171 * config/riscv/riscv.md: Likewise.
22172 * config/riscv/riscv.opt: Likewise.
22173 * config/riscv/sync.md: Likewise.
22174 * config/riscv/t-elf-multilib: Likewise.
22175 * config/riscv/t-linux: Likewise.
22176 * config/riscv/t-linux-multilib: Likewise.
22177 * config/riscv/t-riscv: Likewise.
22178 * configure.ac: Likewise.
22179 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
22180 Waterman as RISC-V maintainers.
22181 * doc/install.texi: Add RISC-V entries.
22182 * doc/invoke.texi: Add RISC-V options section.
22183 * doc/md.texi: Add RISC-V constraints section.
22184 * configure: Regenerated.
22186 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
22189 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
22190 false values to be constant vectors with all 0 or all 1 bits set.
22191 (vcondu<mode><mode>): Likewise.
22192 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
22194 (fpmask_comparison_operator): Update comment.
22195 (vecint_comparison_operator): New predicate.
22196 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
22197 vector conditionals when the true and false values are constant
22198 vectors with all 0 bits or all 1 bits set.
22200 2017-02-06 Martin Sebor <msebor@redhat.com>
22202 PR tree-optimization/79376
22203 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
22205 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
22207 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
22208 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
22209 to simplify split condition.
22211 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22213 * omp-expand.c (oxpand_omp_atomic_fetch_op,
22214 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
22217 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
22219 PR rtl-optimization/68664
22220 * target.def (can_speculate_insn): New hook.
22221 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
22222 * doc/tm.texi: Regenerate.
22223 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
22224 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
22225 (rs6000_sched_can_speculate_insn): New function.
22227 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22229 PR tree-optimization/79284
22230 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
22231 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
22232 vectorizable_mask_load_store, vectorizable_operation,
22233 vect_is_simple_cond, get_same_sized_vectype): Use it instead
22234 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
22235 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
22236 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
22237 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
22238 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
22239 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
22240 is_gimple_assign (stmt). Replace another such test with
22241 is_gimple_assign (stmt).
22243 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22246 * config/avr/avr.c (rtl-iter.h): Include it.
22247 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
22248 (avr_legitimate_combined_insn): ...and implementation.
22250 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
22252 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
22253 * config/s390/s390.c (s390_const_operand_ok)
22254 (s390_canonicalize_comparison, s390_extract_part)
22255 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
22256 (s390_contiguous_bitmask_p, s390_rtx_costs)
22257 (legitimize_pic_address): Likewise.
22258 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
22259 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
22260 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
22261 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
22262 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
22264 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22266 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
22267 REGNO($0) == REGNO($1).
22269 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22271 * config/s390/linux.h(SIZE_TYPE): Add comment.
22273 2017-02-06 Julian Brown <julian@codesourcery.com>
22274 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22275 Virendra Pathak <virendra.pathak@broadcom.com>
22277 * config/aarch64/aarch64-cores.def: Change the scheduler
22279 * config/aarch64/aarch64.md: Include thunderx2t99.md.
22280 * config/aarch64/thunderx2t99.md: New file.
22282 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22284 * doc/standards.texi (Go Language): Update link to language
22287 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22289 * tree-eh.c (lower_resx): Sanitize profile.
22290 (cleanup_empty_eh_move_lp): Likewise.
22292 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22295 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
22297 * cfgloopmanip.h (loop_version): Update prototype.
22298 * modulo-sched.c (sms_schedule): Update call of loop_version.
22299 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
22300 * tree-parloops.c (gen_parallel_loop): Likewise.
22301 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
22302 * tree-ssa-loop-split.c (split_loop): Likewise.
22303 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
22304 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22306 2017-02-05 Martin Liska <mliska@suse.cz>
22309 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
22311 (print_operand_address): Initialize a struct to zero.
22313 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22315 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
22316 garbage collector only in textual form.
22318 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22320 * doc/extend.texi (x86 specific memory model extensions for
22321 transactional memory): Simplify a phrase.
22323 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
22326 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
22327 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
22328 (atomic_storedi_1): Likewise.
22330 2017-02-04 Jakub Jelinek <jakub@redhat.com>
22332 PR tree-optimization/79338
22333 * tree-parloops.c (gather_scalar_reductions): Don't call
22334 vect_analyze_loop_form for loop->inner before destroying loop's
22337 2017-02-03 Martin Sebor <msebor@redhat.com>
22339 PR tree-optimization/79327
22340 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
22341 when precision has resulted in leading zeros.
22342 (format_integer): Adjust the likely counter to assume an unknown
22343 argument that may be zero is non-zero.
22345 2017-02-03 Jason Merrill <jason@redhat.com>
22348 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
22349 avoid copying non-taken branch.
22351 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22353 PR tree-optimization/79340
22354 * tree-vect-loop.c (vectorizable_reduction): Release
22355 vec_defs elements after safe_splicing them into other vectors.
22358 PR tree-optimization/79327
22359 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
22360 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
22362 (format_integer): Use wide_int_to_tree instead of build_int_cst
22363 + to_?hwi. If argmin is NULL, just set argmin and argmax to
22364 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
22365 of shortest and longest sequence.
22367 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
22369 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
22370 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
22372 2017-02-03 Walter Lee <walt@tilera.com>
22375 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
22376 after initial stackframe link reg save.
22377 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
22379 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22382 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
22383 wu for stxssp alternative.
22385 2017-02-03 Martin Sebor <msebor@redhat.com>
22387 PR tree-optimization/79352
22388 * gimple-fold.c (get_range_strlen): Add argument.
22389 (get_range_strlen): Change return type to bool.
22390 (get_maxval_strlen): Pass in a dummy argument.
22391 * gimple-fold.h (get_range_strlen): Change return type to bool.
22392 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
22393 * tree.h (array_at_struct_end_p): Add argument.
22394 * tree.c (array_at_struct_end_p): Handle it.
22396 2017-02-03 Martin Liska <mliska@suse.cz>
22399 * multiple_target.c (create_dispatcher_calls): Redirect edge
22400 from a caller of a dispatcher.
22401 (expand_target_clones): Make the clones local.
22402 (ipa_target_clone): Do both target clones and resolvers.
22403 (ipa_dispatcher_calls): Remove the pass.
22404 (pass_dispatcher_calls::gate): Likewise.
22405 (make_pass_dispatcher_calls): Likewise.
22406 * passes.def (pass_target_clone): Put as very first IPA early
22409 2017-02-03 Martin Liska <mliska@suse.cz>
22411 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
22412 in case of a function with ifunc attribute.
22414 2017-02-03 Martin Liska <mliska@suse.cz>
22416 * cgraph.c (cgraph_node::dump): Dump function version info.
22417 * symtab.c (symtab_node::dump_base): Add missing new line.
22419 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22421 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
22422 (ifcombine_ifandif): Use it.
22424 2017-02-03 Martin Liska <mliska@suse.cz>
22426 * doc/invoke.texi: Document default value for
22427 use-after-scope-direct-emission-threshold.
22429 2017-02-03 Martin Liska <mliska@suse.cz>
22431 PR tree-optimization/79339
22432 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
22433 (format_floating): Likewise.
22435 2017-02-03 Martin Liska <mliska@suse.cz>
22438 * ipa-prop.c (ipa_node_params_t::insert): Remove current
22440 (ipa_node_params_t::remove): Likewise.
22441 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
22442 initialization from removed ipa_node_params_t::insert.
22443 (ipa_node_params::~ipa_node_params): Move from removed
22444 ipa_node_params_t::release.
22445 * symbol-summary.h (symbol_summary::m_released): New member.
22446 Do not release a summary twice. Do not allow to call finalizer
22447 for types of a summary that live in GGC memory.
22449 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22451 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
22454 2017-02-02 Martin Sebor <msebor@redhat.com>
22456 PR middle-end/79275
22457 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
22458 (format_string): Tighten up the range of output for non-constant
22459 strings and correct the expected range for wide non-constant strings.
22461 2017-02-02 Martin Sebor <msebor@redhat.com>
22463 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
22465 PR middle-end/32003
22466 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
22468 (-fdump-tree-@var): Add to index and document how to come up
22469 with pass-specific option and dump file names.
22470 (-fdump-passes): Clarify where to look for output.
22472 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22474 PR middle-end/77445
22475 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
22476 statistics of the analyzed path; allow threading for speed when
22477 any of BBs along the path are optimized for speed.
22479 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
22481 PR middle-end/78468
22482 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
22483 settings of the virtual registers.
22486 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
22488 * explow.c (get_dynamic_stack_size): Take known alignment of stack
22489 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
22492 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22494 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
22495 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
22497 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22499 * config/s390/s390.md: Add missing comments with the expanded
22501 * config/s390/vector.md: Likewise.
22502 * config/s390/vx-builtins.md: Likewise.
22504 2017-02-02 Jakub Jelinek <jakub@redhat.com>
22507 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
22508 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
22509 conditions on a single line.
22511 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22513 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22514 __S390_VX__ to __VX__.
22516 2017-02-01 Andrew Pinski <apinski@cavium.com>
22518 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
22519 stmt_info to record_stmt_cost.
22520 (vect_get_known_peeling_cost): Pass stmt_info if known to
22522 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
22523 cpu_vector_cost field into
22524 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
22525 field into vec_int_stmt_cost and vec_fp_stmt_cost.
22526 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
22527 splitting of scalar_stmt_cost and vec_stmt_cost.
22528 (thunderx_vector_cost): Likewise.
22529 (cortexa57_vector_cost): LIkewise.
22530 (exynosm1_vector_cost): Likewise.
22531 (xgene1_vector_cost): Likewise.
22532 (thunderx2t99_vector_cost): Improve after the splitting of the two
22534 (aarch64_builtin_vectorization_cost): Update for the splitting of
22535 scalar_stmt_cost and vec_stmt_cost.
22537 2017-02-01 Torvald Riegel <triegel@redhat.com>
22538 Richard Henderson <rth@redhat.com>
22540 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
22541 conditional on existance of a fast atomic load.
22542 * optabs-query.c (can_atomic_load_p): New function.
22543 * optabs-query.h (can_atomic_load_p): Declare it.
22544 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
22545 no fast atomic load is available for the particular size of access.
22546 (expand_atomic_compare_and_swap): Likewise.
22547 (expand_atomic_load): Likewise.
22548 (expand_atomic_store): Likewise.
22549 (expand_atomic_fetch_op): Likewise.
22550 * testsuite/lib/target-supports.exp
22551 (check_effective_target_sync_int_128): Remove x86 because it provides
22552 no fast atomic load.
22553 (check_effective_target_sync_int_128_runtime): Likewise.
22555 2017-02-01 Richard Biener <rguenther@suse.de>
22557 * graphite.c: Include tree-vectorizer.h for find_loop_location.
22558 (graphite_transform_loops): Provide opt-info for optimized nests.
22559 * tree-parloop.c (parallelize_loops): Provide opt-info for
22560 parallelized loops.
22562 2017-02-01 Richard Biener <rguenther@suse.de>
22564 PR middle-end/79315
22565 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
22566 was not set before.
22568 2017-02-01 Richard Biener <rguenther@suse.de>
22570 PR tree-optimization/71824
22571 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22572 Verify the loops are valid in the merged SESE region.
22573 (scop_detection::can_represent_loop_1): Check analyzing the
22574 evolution of the number of iterations in the region succeeds.
22576 2017-01-31 Ian Lance Taylor <iant@golang.org>
22578 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
22579 REG_ARGS_SIZE note to 32-bit push insns and call insn.
22581 2017-01-31 David Malcolm <dmalcolm@redhat.com>
22583 PR preprocessor/79210
22584 * input.c (get_substring_ranges_for_loc): Replace line_width
22585 assertion with error-handling.
22587 2017-01-31 Richard Biener <rguenther@suse.de>
22589 PR tree-optimization/77318
22590 * graphite-sese-to-poly.c (extract_affine): Fix assert.
22591 (create_pw_aff_from_tree): Take loop parameter.
22592 (add_condition_to_pbb): Pass loop of the condition to
22593 create_pw_aff_from_tree.
22595 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22597 * config/s390/s390.c (s390_asan_shadow_offset): New function.
22598 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
22600 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
22604 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
22606 (convert_int_to_float128): Likewise.
22607 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
22608 (convert_int_to_float128): Likewise.
22609 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
22610 (UNSPEC_IEEE128_CONVERT): Likewise.
22611 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
22612 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
22613 Use local variables for IBM extended format.
22614 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
22615 (fix_trunc<mode>si2_fprs): Likewise.
22616 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
22617 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
22618 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
22619 to know that we can now have integers of all sizes in vector
22621 (fix<uns>_<mode>di2_hw): Likewise.
22622 (float<uns>_<mode>si2_hw): Likewise.
22623 (fix_<mode>si2_hw): Likewise.
22624 (fixuns_<mode>si2_hw): Likewise.
22625 (float<uns>_<mode>di2_hw): Likewise.
22626 (float_<mode>di2_hw): Likewise.
22627 (float_<mode>si2_hw): Likewise.
22628 (floatuns_<mode>di2_hw): Likewise.
22629 (floatuns_<mode>si2_hw): Likewise.
22630 (xscvqp<su>wz_<mode>): Delete, no longer used.
22631 (xscvqp<su>dz_<mode>): Likewise.
22632 (xscv<su>dqp_<mode>): Likewise.
22633 (ieee128_mfvsrd_64bit): Likewise.
22634 (ieee128_mfvsrd_32bit): Likewise.
22635 (ieee128_mfvsrwz): Likewise.
22636 (ieee128_mtvsrw): Likewise.
22637 (ieee128_mtvsrd_64bit): Likewise.
22638 (ieee128_mtvsrd_32bit): Likewise.
22640 2017-01-31 Martin Liska <mliska@suse.cz>
22643 * ipa-prop.c (ipa_free_all_node_params): Call release method
22644 instead of ~sumbol_summary to not to trigger double times
22647 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
22649 PR tree-optimization/71691
22650 * bitmap.h (class auto_bitmap): New.
22651 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
22652 is_maybe_undefined instead of ssa_undefined_value_p.
22654 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22656 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22657 __S390_ARCH_LEVEL__ to __ARCH__.
22659 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22661 PR tree-optimization/79267
22662 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
22663 if should_remove_lhs_p is true.
22665 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
22668 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
22669 (add_alignment_attribute): New.
22670 (base_type_die): Add alignment attribute.
22671 (subrange_type_die): Likewise.
22672 (modified_type_die): Likewise.
22673 (gen_array_type_die): Likewise.
22674 (gen_descr_array_type_die: Likewise.
22675 (gen_enumeration_type_die): Likewise.
22676 (gen_subprogram_die): Likewise.
22677 (gen_variable_die): Likewise.
22678 (gen_field_die): Likewise.
22679 (gen_ptr_to_mbr_type_die): Likewise.
22680 (gen_struct_or_union_type_die): Likewise.
22681 (gen_subroutine_type_die): Likewise.
22682 (gen_typedef_die): Likewise.
22683 (base_type_cmp): Compare alignment attribute.
22685 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22688 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
22689 (setb_unsigned) New pattern for setb with CCUNS.
22690 * config/rs6000/rs6000.c (expand_block_compare): Use a different
22691 subfc./subfe sequence to avoid overflow problems. Generate a
22692 shorter sequence with cmpld/setb for power9.
22693 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
22694 for generating subfc. instruction.
22695 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
22696 now uses this instruction.
22698 2017-01-30 Ian Lance Taylor <iant@google.com>
22701 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
22702 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
22704 2017-01-30 Martin Sebor <msebor@redhat.com>
22706 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
22707 Move constant to the right of a relational operator.
22708 (get_mpfr_format_length, format_character, format_string): Ditto.
22709 (should_warn_p, maybe_warn): Same.
22711 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
22713 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
22716 * asan.c (get_translation_unit_decl): Remove function.
22717 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
22719 2017-01-30 Martin Liska <mliska@suse.cz>
22721 PR gcov-profile/79259
22722 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
22723 -fprofile-generate.
22725 2017-01-30 Martin Liska <mliska@suse.cz>
22728 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
22729 Initialize variables with NULL value.
22731 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
22734 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
22736 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
22738 2017-01-30 Richard Biener <rguenther@suse.de>
22740 PR tree-optimization/79276
22741 * tree-vrp.c (process_assert_insertions): Properly adjust common
22742 when removing a duplicate.
22744 * gcc.dg/torture/pr79276.c: New testcase.
22746 2017-01-30 Richard Biener <rguenther@suse.de>
22748 PR tree-optimization/79256
22749 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22750 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22752 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
22754 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
22757 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
22758 ("*r<noxa>sbg_<mode>_sll_bitmask")
22759 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
22760 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
22761 Use contiguous_bitmask_nowrap_operand.
22763 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22766 * config/rs6000/altivec.h (vec_xl): Revise #define.
22767 (vec_xst): Likewise.
22769 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
22771 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
22773 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
22775 PR rtl-optimization/79194
22776 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
22777 traps before call to bypass_conditional_jumps.
22779 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22781 PR tree-optimization/71374
22782 * lra-constraints.c (check_conflict_input_operands): New.
22783 (match_reload): Use it.
22785 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22788 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
22789 account to calculate conflict_set.
22791 2017-01-27 Bin Cheng <bin.cheng@arm.com>
22793 PR rtl-optimization/78559
22794 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
22795 other_insn in combine.
22797 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22799 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
22800 uint16_type_node for BT_UINT16.
22802 2017-01-27 David Malcolm <dmalcolm@redhat.com>
22804 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
22805 "RTL Tests" to menu.
22806 (GIMPLE Tests): New node.
22807 (RTL Tests): New node.
22809 2017-01-27 Richard Biener <rguenther@suse.de>
22811 PR tree-optimization/79245
22812 * tree-loop-distribution.c (distribute_loop): Apply cost
22813 modeling also to detected patterns.
22815 2017-01-27 Richard Biener <rguenther@suse.de>
22817 PR tree-optimization/71433
22818 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
22819 (compare_assert_loc): New function.
22820 (process_assert_insertions): Sort and optimize assert locations
22821 to remove duplicates and push down identical assertions on
22822 edges to their destination block.
22824 2017-01-27 Richard Biener <rguenther@suse.de>
22826 PR tree-optimization/79244
22827 * tree-vrp.c (remove_range_assertions): Forcefully propagate
22828 out SSA names even if abnormal.
22830 2017-01-27 Jakub Jelinek <jakub@redhat.com>
22832 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
22833 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
22834 instead of MPFR_RNDN.
22836 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
22839 * arm.c (arm_option_override): Don't call build_target_option_node
22840 until after doing all option overrides.
22841 (arm_valid_target_attribute_tree): Likewise.
22843 2017-01-27 Martin Liska <mliska@suse.cz>
22845 * doc/invoke.texi (-fprofile-arcs): Document profiling support
22846 for {cd}tors and C++ {cd}tors.
22848 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
22850 * config/s390/s390.md ("*setmem_long_and")
22851 ("*setmem_long_and_31z"): Use zero_extend instead of and.
22853 2017-01-26 Martin Sebor <msebor@redhat.com>
22855 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
22858 2017-01-26 Martin Sebor <msebor@redhat.com>
22860 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
22861 HAVE_DFmode before using XFmode or DFmode.
22862 (parse_directive): Avoid using the z length modifier to avoid
22863 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
22865 PR middle-end/78703
22866 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
22867 to accept adjustment as an array.
22868 (get_int_range): New function.
22869 (struct directive): Make width and prec arrays.
22870 (directive::set_width, directive::set_precision): Call get_int_range.
22871 (format_integer, format_floating): Handle width and precision ranges.
22872 (format_string, parse_directive): Same.
22874 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22877 * dwarf2out.c (generate_skeleton_bottom_up): For children with
22878 comdat_type_p set, just clone them, but keep the children in the
22882 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
22883 which have direct callers with -fvar-tracking-assignments enabled
22885 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
22886 inside of type units.
22888 2017-01-26 Martin Sebor <msebor@redhat.com>
22890 PR middle-end/78703
22891 * gimple-ssa-sprintf.c (struct result_range): Add likely and
22893 (struct format_result): Replace number_chars, number_chars_min,
22894 and number_chars_max with a single member of struct result_range.
22896 (format_result::operator+=): Adjust.
22897 (struct fmtresult): Remove bounded. Handle likely and unlikely
22899 (fmtresult::adjust_for_width_or_precision): New function.
22900 (fmtresult:type_max_digits): New function.
22901 (bytes_remaining): Handle likely and unlikely counters.
22902 (min_bytes_remaining): Remove.
22903 (format_percent): Simplify.
22904 (format_integer, format_floating): Set likely and unlikely counters.
22905 (get_string_length, format_character, format_string): Same.
22906 (format_plain, should_warn_p): New function.
22907 (maybe_warn): Call should_warn_p. Update diagnostic messages
22908 and handle those for all directives, including plain strings.
22909 (format_directive): Handle likely and unlikely counters.
22910 Remove unnecessary quoting from diagnostics. Add an informational
22912 (add_bytes): Remove.
22913 (pass_sprintf_length::compute_format_length): Simplify.
22914 (try_substitute_return_value): Handle likely and unlikely counters.
22916 2017-01-26 Carl Love <cel@us.ibm.com>
22918 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
22919 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
22921 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
22924 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
22925 endianess for subregs into account.
22926 * lra-constraints.c (lra_constraints): Do risky transformations
22927 always on the first iteration.
22928 * lra-lives.c (check_pseudos_live_through_calls): Add arg
22929 last_call_used_reg_set.
22930 (process_bb_lives): Define and use last_call_used_reg_set.
22931 * lra.c (lra): Always continue after lra_constraints on the first
22934 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
22936 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
22938 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
22940 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22942 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
22943 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
22944 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
22945 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
22946 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
22947 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
22948 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
22949 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
22950 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
22952 2017-01-26 Marek Polacek <polacek@redhat.com>
22955 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
22956 for the third operand.
22958 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22960 PR middle-end/79236
22961 * omp-low.c (struct omp_context): Add simt_stmt field.
22962 (scan_omp_for): Return omp_context *.
22963 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
22964 context to the _simt_ SIMD stmt.
22965 (lower_omp_for): For combined SIMD with sibling _simt_
22966 SIMD, make sure to use the same decls in _looptemp_
22967 clauses as in the sibling.
22969 2017-01-26 David Sherwood <david.sherwood@arm.com>
22971 PR middle-end/79212
22972 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
22975 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22978 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
22979 emit fld b; fld a; if possible.
22981 * brig-builtins.def: Update copyright years.
22982 * config/arm/arm_acle_builtins.def: Update copyright years.
22984 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
22987 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
22988 constraint instead of o for the stxsd instruction.
22990 2017-01-25 Carl Love <cel@us.ibm.com>
22992 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
22993 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
22995 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
22997 * doc/invoke.texi (C++ Dialect Options): Fix typo.
22999 2017-01-25 Richard Biener <rguenther@suse.de>
23001 PR tree-optimization/69264
23002 * target.def (vector_alignment_reachable): Improve documentation.
23003 * doc/tm.texi: Regenerate.
23004 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
23006 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
23007 earlier changes with respect to TYPE_USER_ALIGN.
23008 (vector_alignment_reachable_p): Likewise. Improve dumping.
23010 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23013 * config/arm/arm.md (xordi3): Force constant operand into a register
23016 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23018 * doc/invoke.texi (-fstore-merging): Correct default optimization
23019 levels at which it is enabled.
23020 (-O): Move -fstore-merging from list to...
23023 2017-01-25 Richard Biener <rguenther@suse.de>
23026 * omp-expand.c: Include debug.h.
23027 (expand_omp_taskreg): Make sure to generate early debug before
23028 outlining anything from a function.
23029 (expand_omp_target): Likewise.
23030 (grid_expand_target_grid_body): Likewise.
23032 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
23035 * asan.c (get_translation_unit_decl): New function.
23036 (asan_add_global): Extract modules file name from globals
23037 TRANSLATION_UNIT_DECL name.
23039 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
23042 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
23043 for long calls with APCS frame and VFP.
23045 2017-01-24 David Malcolm <dmalcolm@redhat.com>
23047 * cfg.c (original_copy_tables_initialized_p): New function.
23048 * cfg.h (original_copy_tables_initialized_p): New decl.
23049 * cfgrtl.c (relink_block_chain): Guard the call to
23050 free_original_copy_tables with a call to
23051 original_copy_tables_initialized_p.
23052 * cgraph.h (symtab_node::native_rtl_p): New decl.
23053 * cgraphunit.c (symtab_node::native_rtl_p): New function.
23054 (symtab_node::needed_p): Don't assert for early assembly output
23055 for __RTL functions.
23056 (cgraph_node::finalize_function): Set "force_output" for __RTL
23058 (cgraph_node::analyze): Bail out early for __RTL functions.
23059 (analyze_functions): Update assertion to support __RTL functions.
23060 (cgraph_node::expand): Bail out early for __RTL functions.
23061 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
23063 * function.h (struct function): Update comment for field
23065 * gimple-expr.c: Include "tree-pass.h".
23066 (gimple_has_body_p): Return false for __RTL functions.
23067 * Makefile.in (OBJS): Add run-rtl-passes.o.
23068 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
23070 (gcc::pass_manager::get_clean_slate): New accessor.
23071 * passes.c: Include "insn-addr.h".
23072 (should_skip_pass_p): Add logging. Update logic for running
23073 "expand" to be compatible with both __GIMPLE and __RTL. Guard
23074 property-provider override so it is only done for gimple passes.
23076 (skip_pass): New function.
23077 (execute_one_pass): Call skip_pass when skipping passes.
23078 * read-md.c (md_reader::read_char): Support filtering
23079 the input to a subset of line numbers.
23080 (md_reader::md_reader): Initialize fields
23081 m_first_line and m_last_line.
23082 (md_reader::read_file_fragment): New function.
23083 * read-md.h (md_reader::read_file_fragment): New decl.
23084 (md_reader::m_first_line): New field.
23085 (md_reader::m_last_line): New field.
23086 * read-rtl-function.c (function_reader::create_function): Only
23087 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
23088 curr_properties. Set DECL_INITIAL to a dummy block.
23089 (read_rtl_function_body_from_file_range): New function.
23090 * read-rtl-function.h (read_rtl_function_body_from_file_range):
23092 * run-rtl-passes.c: New file.
23093 * run-rtl-passes.h: New file.
23095 2017-01-24 Jeff Law <law@redhat.com>
23097 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
23100 2017-01-24 Bin Cheng <bin.cheng@arm.com>
23102 PR tree-optimization/79159
23103 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
23104 (record_nonwrapping_iv): Improve boundary using above function if no
23105 value range information.
23107 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
23108 Martin Jambor <mjambor@suse.cz>
23110 * brig-builtins.def: New file.
23111 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
23112 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
23113 (DEF_HSAIL_SAT_BUILTIN): Likewise.
23114 (DEF_HSAIL_INTR_BUILTIN): Likewise.
23115 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
23116 * builtin-types.def (BT_INT8): New.
23117 (BT_INT16): Likewise.
23118 (BT_UINT8): Likewise.
23119 (BT_UINT16): Likewise.
23120 (BT_FN_ULONG): Likewise.
23121 (BT_FN_UINT_INT): Likewise.
23122 (BT_FN_UINT_ULONG): Likewise.
23123 (BT_FN_UINT_LONG): Likewise.
23124 (BT_FN_UINT_PTR): Likewise.
23125 (BT_FN_ULONG_PTR): Likewise.
23126 (BT_FN_INT8_FLOAT): Likewise.
23127 (BT_FN_INT16_FLOAT): Likewise.
23128 (BT_FN_UINT32_FLOAT): Likewise.
23129 (BT_FN_UINT16_FLOAT): Likewise.
23130 (BT_FN_UINT8_FLOAT): Likewise.
23131 (BT_FN_UINT64_FLOAT): Likewise.
23132 (BT_FN_UINT16_UINT32): Likewise.
23133 (BT_FN_UINT32_UINT16): Likewise.
23134 (BT_FN_UINT16_UINT16_UINT16): Likewise.
23135 (BT_FN_INT_PTR_INT): Likewise.
23136 (BT_FN_UINT_PTR_UINT): Likewise.
23137 (BT_FN_LONG_PTR_LONG): Likewise.
23138 (BT_FN_ULONG_PTR_ULONG): Likewise.
23139 (BT_FN_VOID_UINT64_UINT64): Likewise.
23140 (BT_FN_UINT8_UINT8_UINT8): Likewise.
23141 (BT_FN_INT8_INT8_INT8): Likewise.
23142 (BT_FN_INT16_INT16_INT16): Likewise.
23143 (BT_FN_INT_INT_INT): Likewise.
23144 (BT_FN_UINT_FLOAT_UINT): Likewise.
23145 (BT_FN_FLOAT_UINT_UINT): Likewise.
23146 (BT_FN_ULONG_UINT_UINT): Likewise.
23147 (BT_FN_ULONG_UINT_PTR): Likewise.
23148 (BT_FN_ULONG_ULONG_ULONG): Likewise.
23149 (BT_FN_UINT_UINT_UINT): Likewise.
23150 (BT_FN_VOID_UINT_PTR): Likewise.
23151 (BT_FN_UINT_UINT_PTR: Likewise.
23152 (BT_FN_UINT32_UINT64_PTR): Likewise.
23153 (BT_FN_INT_INT_UINT_UINT): Likewise.
23154 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
23155 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
23156 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
23157 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
23158 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
23159 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
23160 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
23161 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
23162 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
23163 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
23164 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
23165 * doc/frontends.texi: List BRIG FE.
23166 * doc/install.texi (Testing): Add BRIG tesring requirements.
23167 * doc/invoke.texi (Overall Options): Mention BRIG.
23168 * doc/standards.texi (Standards): Doucment BRIG HSA version.
23170 2017-01-24 Richard Biener <rguenther@suse.de>
23172 PR translation/79208
23173 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
23175 2017-01-24 Martin Jambor <mjambor@suse.cz>
23178 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
23179 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
23180 and known_contexts.
23182 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
23184 PR middle-end/79123
23185 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
23186 casts from signed to unsigned really don't have a range.
23188 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
23190 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
23191 GMP_RNDx for compatiblity.
23193 2017-01-24 Martin Liska <mliska@suse.cz>
23196 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
23197 that would prevent us to call alloca with -1 as argument.
23199 2017-01-24 Jakub Jelinek <jakub@redhat.com>
23201 * dwarf2out.c (output_compilation_unit_header, output_file_names):
23202 Avoid -Wformat-security warning.
23204 2017-01-23 Andrew Pinski <apinski@cavium.com>
23206 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
23209 2017-01-23 Martin Sebor <msebor@redhat.com>
23211 PR middle-end/78703
23212 * gimple-ssa-sprintf.c (warn_level): New global.
23213 (format_integer): Use it here and throughout the rest of the file.
23214 Use the same switch to compute sign as base.
23215 (maybe_warn): New function.
23216 (format_directive): Factor out warnings into maybe_warn.
23217 Add debugging output. Use warn_level.
23218 (add_bytes): Use warn_level.
23219 (pass_sprintf_length::compute_format_length): Add debugging output.
23220 (try_substitute_return_value): Same.
23221 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
23223 PR middle-end/78703
23224 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
23225 (struct fmtresult, format_integer, format_floating): Adjust.
23226 (fmtresult::fmtresult): Set max correctly in two argument ctor.
23227 (get_string_length, format_string,format_directive): Same.
23228 (pass_sprintf_length::compute_format_length): Same.
23229 (try_substitute_return_value): Simplify slightly.
23231 PR middle-end/78703
23232 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
23233 (fmtresult::operator+=): Outlined.
23234 (struct fmtresult): Add ctors.
23235 (struct conversion_spec): Rename...
23236 (struct directive): ...to this. Add and remove data members.
23237 (directive::set_width, directive::set_precision): New functions.
23238 (format_percent): Use fmtresult ctor.
23239 (get_width_and_precision): Remove.
23240 (format_integer): Make naming changes. Avoid computing width and
23242 (format_floating): Same. Adjust indentation.
23243 (format_character, format_none): New functions.
23244 (format_string): Moved character handling to format_character.
23245 (format_directive): Remove arguments, change return type.
23246 (parse_directive): New function.
23247 (pass_sprintf_length::compute_format_length): Move directive
23248 parsing to parse_directive.
23250 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23252 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
23253 (assign_assembler_name_if_needed): ... this.
23254 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
23255 (assign_assembler_name_if_needed): ... this.
23256 (free_lang_data_in_cgraph): Adjust callers.
23257 * cgraphunit.c (cgraph_node::analyze): Likewise.
23258 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
23261 2017-01-23 Richard Biener <rguenther@suse.de>
23263 PR tree-optimization/79088
23264 PR tree-optimization/79188
23265 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
23266 resetting loop bounds after last path deletion. Reset loop
23267 bounds of the target loop, make code match the comments.
23268 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
23269 Make sure loops need no fixups.
23271 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23273 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
23274 exponent support with double type for first argument.
23275 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
23276 type returned by __builtin_vec_extract_sig,
23277 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
23278 functions from "vector int" to "vector unsigned int" or from
23279 "vector long long int" to "vector unsigned long long int".
23280 Changed type returned by __builtin_vec_extract_exp,
23281 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
23282 functions from "vector int" to "vector unsigned int" or from
23283 "vector long long int" to "vector unsigned long long int".
23284 Changed return type of __builtin_vec_test_data_class,
23285 __builtin_vec_test_data_class_sp, and
23286 __builtin_vec_test_data_class_dp from "vector int" to
23287 "vector bool int" or from "vector long long int" to "vector bool
23288 long long int" and changed second argument type from "unsigned
23289 int" to "int". Added new overloaded function forms "vector float
23290 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
23291 "vector float __builtin_vec_insert_exp_sp (vector float, vector
23292 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
23293 double, vector unsigned long long int)" and "vector double
23294 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
23295 long int)". Changed return type of
23296 __builtin_scalar_test_data_class and
23297 __builtin_scalar_test_data_class_sp and
23298 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
23299 int" and changed second argument from "unsigned int" to "int".
23300 Changed type returned by __builtin_scalar_test_neg,
23301 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
23302 from "int" to "bool int". Added new overloaded function form
23303 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
23304 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
23305 exponent double-precision with floating point first argument.
23306 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
23307 documentation of scalar_test_data_class, scalar_test_neg,
23308 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
23309 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
23310 vec_test_data_class built-in functions to reflect refinements in
23311 their type signatures.
23313 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
23315 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
23317 (aarch64_elf_asm_destructor): Likewise.
23319 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
23321 PR rtl-optimization/78634
23322 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
23323 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
23324 * ifcvt.c (noce_try_cmove): Add missing cost check.
23326 PR rtl-optimization/71724
23327 * combine.c (if_then_else_cond): Look for situations where it is
23328 beneficial to undo the work of one of the recursive calls.
23330 2017-01-23 Bin Cheng <bin.cheng@arm.com>
23332 PR tree-optimization/70754
23333 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
23334 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
23335 combined stmt before it if not NULL.
23336 (combine_chains): Process refs reversely and compute dominance point
23339 2017-01-23 Martin Liska <mliska@suse.cz>
23341 PR tree-optimization/79196
23342 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
23343 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
23345 (strlen_optimize_stmt): Call the renamed function.
23347 2017-01-23 Michael Matz <matz@suse.de>
23349 PR tree-optimization/78384
23350 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
23352 2017-01-23 Richard Biener <rguenther@suse.de>
23354 PR tree-optimization/79186
23355 * tree-vrp.c (register_new_assert_for): Make sure we've seen
23356 both incoming edges before moving an assert.
23358 2017-01-23 Martin Jambor <mjambor@suse.cz>
23360 * ipa-prop.c (load_from_param_1): Removed.
23361 (load_from_unmodified_param): Bits from load_from_param_1 put back
23363 (load_from_param): Removed.
23364 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
23365 with stmt. Reverted back to use of load_from_unmodified_param.
23367 2017-01-23 Martin Jambor <mjambor@suse.cz>
23370 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
23371 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
23372 field a pointer to garbage collected vector, mark lattices and
23373 ipcp_orig_node with GTY((skip)).
23374 (ipa_get_param_count): Adjust to descriptors being a pointer.
23375 (ipa_get_param): Likewise.
23376 (ipa_get_type): Likewise.
23377 (ipa_get_param_move_cost): Likewise.
23378 (ipa_set_param_used): Likewise.
23379 (ipa_get_controlled_uses): Likewise.
23380 (ipa_set_controlled_uses): Likewise.
23381 (ipa_is_param_used): Likewise.
23382 (ipa_node_params_t): Move into garbage collector. New methods insert
23384 (ipa_node_params_sum): Annotate wth GTY(()).
23385 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
23387 (ipa_load_from_parm_agg): Adjust declaration.
23388 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
23389 * ipa-profile.c (ipa_profile): Likewise.
23390 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
23391 (ipa_populate_param_decls): Make descriptors parameter garbage
23393 (ipa_dump_param): Adjust to descriptors being a pointer.
23394 (ipa_alloc_node_params): Likewise.
23395 (ipa_initialize_node_params): Likewise.
23396 (load_from_param_1): Make descriptors parameter garbage collected.
23397 (load_from_unmodified_param): Likewise.
23398 (load_from_param): Likewise.
23399 (ipa_load_from_parm_agg): Likewise.
23400 (ipa_node_params::~ipa_node_params): Removed.
23401 (ipa_free_all_node_params): Remove call to delete operator.
23402 (ipa_node_params_t::insert): New.
23403 (ipa_node_params_t::remove): Likewise.
23404 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
23405 copy known_csts and known_contexts vectors.
23406 (ipa_read_node_info): Adjust to descriptors being a pointer.
23407 (ipcp_modif_dom_walker): Make m_descriptors field garbage
23409 (ipcp_transform_function): Make descriptors variable garbage
23412 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
23414 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
23415 * config/i386/avx512dqintrin.h: Ditto.
23416 * config/i386/avx512fintrin.h: Ditto.
23417 * config/i386/i386.c: Handle new builtins.
23418 * config/i386/i386-builtin.def: Add new builtins.
23419 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
23420 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
23422 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23423 Martin Liska <mliska@suse.cz>
23425 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
23426 * asan.c (asan_expand_poison_ifn): Support stores and use
23427 appropriate ASAN report function.
23428 * internal-fn.c (expand_ASAN_POISON_USE): New function.
23429 * internal-fn.def (ASAN_POISON_USE): Declare.
23430 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
23431 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
23432 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
23433 ASAN_POISON calls w/o LHS.
23434 * tree-ssa.c (execute_update_addresses_taken): Create clobber
23435 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
23436 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
23437 * gimplify.c (asan_poison_variables): Add attribute
23438 use_after_scope_memory to variables that really needs to live
23440 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
23441 having the attribute.
23443 2017-01-23 Martin Liska <mliska@suse.cz>
23445 * asan.c (create_asan_shadow_var): New function.
23446 (asan_expand_poison_ifn): Likewise.
23447 * asan.h (asan_expand_poison_ifn): New declaration.
23448 * internal-fn.c (expand_ASAN_POISON): Likewise.
23449 * internal-fn.def (ASAN_POISON): New builtin.
23450 * sanopt.c (pass_sanopt::execute): Expand
23451 asan_expand_poison_ifn.
23452 * tree-inline.c (copy_decl_for_dup_finish): Make function
23454 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
23455 * tree-ssa.c (is_asan_mark_p): New function.
23456 (execute_update_addresses_taken): Rewrite local variables
23457 (identified just by use-after-scope as addressable) into SSA.
23459 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23461 * doc/install.texi (Specific): opensource.apple.com uses https
23462 now. Remove trailing slash.
23464 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23466 * README.Portability: Remove note on an Irix compatibility issue.
23468 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
23470 * gcov.c (INCLUDE_ALGORITHM): Define.
23471 (INCLUDE_VECTOR): Define.
23472 No longer include <vector> and <algorithm> directly.
23474 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
23476 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
23478 * doc/invoke.texi (Code Gen Options): Ditto.
23480 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
23483 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
23485 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
23487 rtl-optimization/79125
23488 * cprop.c (local_cprop_pass): Handle cases where we make an
23489 unconditional trap.
23491 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
23495 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
23496 read from, for big endian.
23498 2017-01-20 Jiong Wang <jiong.wang@arm.com>
23500 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
23501 register pauth builtins for LP64 only.
23503 2017-01-20 Marek Polacek <polacek@redhat.com>
23506 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
23509 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23511 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
23513 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
23514 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
23515 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
23517 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23520 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
23521 in big-endian mode when they are not a single duplicated value.
23523 2017-01-20 Richard Biener <rguenther@suse.de>
23525 * BASE-VER: Bump to 7.0.1.
23527 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23529 * omp-low.c (omplow_simd_context): New struct. Use it...
23530 (lower_rec_simd_input_clauses): ...here and...
23531 (lower_rec_input_clauses): ...here to hold common data. Adjust all
23532 references to idx, lane, max_vf, is_simt.
23534 2017-01-20 Graham Markall <graham.markall@embecosm.com>
23536 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
23539 2017-01-20 Martin Jambor <mjambor@suse.cz>
23541 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
23542 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
23544 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
23545 (GTFILES): Rename hsa.c to hsa-common.c.
23546 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
23547 * hsa-dump.c: Likewise.
23548 * hsa-gen.c: Likewise.
23549 * hsa-regalloc.c: Likewise.
23550 * ipa-hsa.c: Likewise.
23551 * omp-expand.c: Likewise.
23552 * omp-low.c: Likewise.
23553 * toplev.c: Likewise.
23555 2017-01-20 Marek Polacek <polacek@redhat.com>
23558 * doc/invoke.texi: Document -Wduplicated-branches.
23559 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
23560 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
23561 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
23562 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
23563 return 0 only when not OEP_LEXICOGRAPHIC.
23564 (fold_build_cleanup_point_expr): Use the expression
23565 location when building CLEANUP_POINT_EXPR.
23566 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
23567 * tree.c (add_expr): Handle error_mark_node.
23569 2017-01-20 Martin Liska <mliska@suse.cz>
23572 * tree-profile.c (init_ic_make_global_vars): Do not call
23574 (gimple_init_gcov_profiler): Likewise.
23576 2017-01-20 Martin Liska <mliska@suse.cz>
23579 * cgraph.h (maybe_create_reference): Remove argument and
23581 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
23583 * ipa-cp.c (create_specialized_node): Likewise.
23584 * symtab.c (symtab_node::maybe_create_reference): Handle
23585 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
23587 2017-01-20 Martin Liska <mliska@suse.cz>
23589 * read-rtl-function.c (function_reader::create_function): Use
23590 build_decl instread of build_decl_stat.
23592 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
23594 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
23595 * config/i386/avx512dqintrin.h: Ditto.
23596 * config/i386/avx512fintrin.h: Ditto.
23597 * config/i386/i386-builtin-types.def: Add new types.
23598 * config/i386/i386.c: Handle new types.
23599 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
23600 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
23601 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
23602 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
23603 (__builtin_ia32_kshiftridi): New.
23604 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
23606 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
23610 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
23611 define to rs6000_init_stack_protect_guard.
23612 (rs6000_init_stack_protect_guard): New function.
23614 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23615 Yunqiang Su <yunqiang.su@imgtec.com>
23617 * config.gcc (supported_defaults): Add madd4.
23618 (with_madd4): Add validation.
23619 (all_defaults): Add madd4.
23620 * config/mips/mips.opt (mmadd4): New option.
23621 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23623 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23625 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
23626 (ISA_HAS_FUSED_MADD4): Likewise.
23627 * doc/invoke.texi (-mmadd4): Document the new option.
23628 * doc/install.texi (--with-madd4): Document the new option.
23630 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23632 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
23633 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
23634 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
23635 (aarch64_init_pauth_hint_builtins): New.
23636 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
23637 (aarch64_expand_builtin): Expand new builtins.
23639 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23641 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
23642 * combine-stack-adj.c (no_unhandled_cfa): Handle
23643 REG_CFA_TOGGLE_RA_MANGLE.
23644 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
23645 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
23646 info for return address signing.
23647 (aarch64_expand_epilogue): Likewise.
23649 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23651 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
23652 * config/aarch64/aarch64-protos.h
23653 (aarch64_return_address_signing_enabled): New declaration.
23654 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
23656 (aarch64_expand_prologue): Sign return address before it's pushed onto
23658 (aarch64_expand_epilogue): Authenticate return address fetched from
23660 (aarch64_override_options): Sanity check for ILP32 and ISA level.
23661 (aarch64_attributes): New function attributes for "sign-return-address".
23662 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
23663 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
23664 ("*do_return"): Generate combined instructions according to key index.
23665 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
23666 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
23668 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
23669 * config/aarch64/aarch64.opt (msign-return-address=): New.
23670 * doc/extend.texi (AArch64 Function Attributes): Documents
23671 "sign-return-address=".
23672 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
23674 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23676 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
23677 overall option summary.
23679 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23681 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
23682 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
23683 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
23684 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
23686 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
23688 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
23689 -mpower9-minmax by default for -mcpu=power9.
23690 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
23691 128-bit floating point.
23693 2017-01-20 Alan Modra <amodra@gmail.com>
23695 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
23696 optimizing for size.
23698 2017-01-20 Alan Modra <amodra@gmail.com>
23701 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
23702 for strcmp and strncmp from corresponding builtin decl.
23704 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23706 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
23707 instead of i386/rtems-64.h.
23708 * config/i386/rtems-64.h: Remove.
23710 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23714 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
23716 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
23718 2017-01-19 Tamar Christina <tamar.christina@arm.com>
23720 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
23721 Change int to HOST_WIDE_INT.
23722 * config/aarch64/aarch64-protos.h
23723 (aarch64_simd_gen_const_vector_dup): Likewise.
23724 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
23726 2017-01-19 David Malcolm <dmalcolm@redhat.com>
23728 * langhooks-def.h (lhd_type_for_size): New decl.
23729 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
23730 * langhooks.c (lhd_type_for_size): New function, taken from
23733 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
23735 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
23736 define_bypass for CR latency.
23737 (power9-cracked-alu): Update bypass latency and remove power9-branch.
23738 (power9-alu2): Add define_bypass for CR latency.
23740 (power9-mul): Update insn latency.
23741 (power9-mul-compare): Update insn latency, bypass latency and remove
23744 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23746 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
23748 * config/aarch64/aarch64.md
23749 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
23750 aarch64_nopcrelative_literal_loads.
23751 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
23753 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
23755 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
23756 TARGET_LOONGSON_3A.
23757 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
23759 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
23762 * config.gcc (supported_defaults): Add lxc1-sxc1.
23763 (with_lxc1_sxc1): Add validation.
23764 (all_defaults): Add lxc1-sxc1.
23765 * config/mips/mips.opt (mlxc1-sxc1): New option.
23766 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23768 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23769 __mips_no_lxc1_sxc1.
23770 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
23771 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
23772 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
23774 2017-01-19 Richard Biener <rguenther@suse.de>
23776 PR tree-optimization/72488
23777 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
23778 sure to restore SSA info.
23779 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
23781 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
23783 PR rtl-optimization/79121
23784 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
23785 of the inner type when shifting an extended value.
23787 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
23790 * symtab.c (symtab_node::equal_address_to): Fix comparing of
23791 interposable aliases.
23793 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
23796 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
23797 Use the evmergelohi instruction.
23798 (mov_si<mode>_e500_subreg4_2_le): Likewise.
23799 (mov_sitf_e500_subreg8_2_be): Likewise.
23800 (mov_sitf_e500_subreg12_2_le): Likewise.
23801 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
23802 (mov_si<mode>_e500_subreg4_2_be): Likewise.
23803 (mov_sitf_e500_subreg8_2_le): Likewise.
23804 (mov_sitf_e500_subreg12_2_be): Likewise.
23806 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23808 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
23809 attribute from vecsimple to vecperm.
23810 (altivec_vbpermq2): Likewise.
23812 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23815 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
23817 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23818 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
23819 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
23820 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
23821 case where N arg is SIZE_MAX.
23822 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
23823 (cmpstrsi): Add pattern.
23825 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23827 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23828 __builtin_vec_revb builtins.
23829 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
23830 built-in functions to support generation of the ISA 3.0 XXBR<x>
23831 vector byte reverse instructions.
23832 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
23833 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
23834 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
23835 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
23836 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
23837 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
23838 (P9V_BUILTIN_VEC_REVB): Likewise.
23839 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
23840 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
23841 (p9_xxbrq_v16qi): Likewise.
23842 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
23843 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
23844 (p9_xxbrh_v8hi): Likewise.
23845 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
23846 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
23847 vec_revb built-in functions.
23849 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
23851 PR rtl-optimization/78952
23852 * config/i386/i386.md (any_extract): New code iterator.
23853 (*insvqi_2): Use any_extract for source operand.
23854 (*insvqi_3): Use any_shiftrt for source operand.
23856 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
23858 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
23860 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
23862 2017-01-18 Matthias Klose <doko@ubuntu.com>
23864 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
23866 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23868 * config/rs6000/altivec.h (vec_bperm): Change #define.
23869 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
23870 (altivec_vbpermq2): New define_insn.
23871 (altivec_vbpermd): Likewise.
23872 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
23873 function interface.
23874 (VBPERMD): Likewise.
23875 (VBPERM): New polymorphic function interface.
23876 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
23877 Add entries for P9V_BUILTIN_VEC_VBPERM.
23878 * doc/extend.texi: Add interfaces for vec_bperm.
23880 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23882 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
23883 first letter of error messages.
23884 (s390_resolve_overloaded_builtin): Likewise.
23885 * config/s390/s390.c (s390_expand_builtin): Likewise.
23886 (s390_invalid_arg_for_unprototyped_fn): Likewise.
23887 (s390_valid_target_attribute_inner_p): Likewise.
23888 * config/s390/s390.md ("tabort"): Likewise.
23890 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
23892 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
23893 (ISA_AVOID_DIV_HILO): New macro.
23894 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
23895 (ISA_HAS_DDIV): Likewise.
23897 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23899 * doc/invoke.texi (fabi-version): Correct number of occurrences.
23901 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23903 * doc/invoke.texi (fabi-version): Spelling fix.
23905 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23908 * doc/invoke.texi (fabi-version): Mention mangling fix for
23911 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23914 * doc/invoke.texi (fabi-version): Document discriminator mangling.
23916 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
23919 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
23920 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
23922 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
23924 (stack_protect_test): Ditto.
23925 * config/rs6000/rs6000.opt (mstack-protector-guard=,
23926 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
23928 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
23929 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
23930 -mstack-protector-guard-offset=.
23931 (RS/6000 and PowerPC Options): Ditto.
23933 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
23935 * config/i386/i386.h (MASK_CLASS_P): New define.
23936 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
23937 there are no registers from different register sets also when
23938 mask registers are used. Update function comment.
23939 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
23940 to (*k/*r) and (*k/*km) alternatives.
23942 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
23944 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
23945 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
23946 (EH_RETURN_HANDLER_RTX): New define.
23947 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
23948 Force frame pointer in EH return functions.
23949 (aarch64_expand_epilogue): Add barrier for eh_return.
23950 (aarch64_final_eh_return_addr): Remove.
23951 (aarch64_eh_return_handler_rtx): New function.
23952 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
23954 (aarch64_eh_return_handler_rtx): New prototype.
23956 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23958 * config/rs6000/altivec.h (vec_rlmi): New #define.
23959 (vec_vrlnm): Likewise.
23960 (vec_rlnm): Likewise.
23961 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
23962 (UNSPEC_VRLNM): Likewise.
23963 (VIlong): New mode iterator.
23964 (altivec_vrl<VI_char>mi): New define_insn.
23965 (altivec_vrl<VI_char>nm): Likewise.
23966 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
23968 (VRLDNM): Likewise.
23969 (RLNM): New polymorphic function entry.
23970 (VRLWMI): New monomorphic function entry.
23971 (VRLDMI): Likewise.
23972 (RLMI): New polymorphic function entry.
23973 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
23974 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
23975 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
23978 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23981 * dwarf2out.c (field_byte_offset): Restore the
23982 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
23983 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
23984 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
23987 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
23990 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
23992 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23995 * dwarf2out.c (add_data_member_location_attribute): For constant
23996 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
23997 instead of DW_AT_data_member_location, DW_AT_bit_offset and
23998 DW_AT_byte_size attributes.
24000 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24002 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
24003 after forcing to constant memory when the code model is medium.
24005 2017-01-17 Julia Koval <julia.koval@intel.com>
24008 * config/i386/avx512fintrin.h
24009 (_mm512_i32gather_ps): Change __addr type to void const*.
24010 (_mm512_mask_i32gather_ps): Ditto.
24011 (_mm512_i32gather_pd): Ditto.
24012 (_mm512_mask_i32gather_pd): Ditto.
24013 (_mm512_i64gather_ps): Ditto.
24014 (_mm512_mask_i64gather_ps): Ditto.
24015 (_mm512_i64gather_pd): Ditto.
24016 (_mm512_mask_i64gather_pd): Ditto.
24017 (_mm512_i32gather_epi32): Ditto.
24018 (_mm512_mask_i32gather_epi32): Ditto.
24019 (_mm512_i32gather_epi64): Ditto.
24020 (_mm512_mask_i32gather_epi64): Ditto.
24021 (_mm512_i64gather_epi32): Ditto.
24022 (_mm512_mask_i64gather_epi32): Ditto.
24023 (_mm512_i64gather_epi64): Ditto.
24024 (_mm512_mask_i64gather_epi64): Ditto.
24025 (_mm512_i32scatter_ps): Change __addr type to void*.
24026 (_mm512_mask_i32scatter_ps): Ditto.
24027 (_mm512_i32scatter_pd): Ditto.
24028 (_mm512_mask_i32scatter_pd): Ditto.
24029 (_mm512_i64scatter_ps): Ditto.
24030 (_mm512_mask_i64scatter_ps): Ditto.
24031 (_mm512_i64scatter_pd): Ditto.
24032 (_mm512_mask_i64scatter_pd): Ditto.
24033 (_mm512_i32scatter_epi32): Ditto.
24034 (_mm512_mask_i32scatter_epi32): Ditto.
24035 (_mm512_i32scatter_epi64): Ditto.
24036 (_mm512_mask_i32scatter_epi64): Ditto.
24037 (_mm512_i64scatter_epi32): Ditto.
24038 (_mm512_mask_i64scatter_epi32): Ditto.
24039 (_mm512_i64scatter_epi64): Ditto.
24040 (_mm512_mask_i64scatter_epi64): Ditto.
24041 * config/i386/avx512pfintrin.h
24042 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
24043 (_mm512_mask_prefetch_i32gather_ps): Ditto.
24044 (_mm512_mask_prefetch_i64gather_pd): Ditto.
24045 (_mm512_mask_prefetch_i64gather_ps): Ditto.
24046 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
24047 (_mm512_prefetch_i32scatter_ps): Ditto.
24048 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
24049 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
24050 (_mm512_prefetch_i64scatter_pd): Ditto.
24051 (_mm512_prefetch_i64scatter_ps): Ditto.
24052 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
24053 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
24054 * config/i386/avx512vlintrin.h
24055 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
24056 (_mm_mmask_i32gather_ps): Ditto.
24057 (_mm256_mmask_i32gather_pd): Ditto.
24058 (_mm_mmask_i32gather_pd): Ditto.
24059 (_mm256_mmask_i64gather_ps): Ditto.
24060 (_mm_mmask_i64gather_ps): Ditto.
24061 (_mm256_mmask_i64gather_pd): Ditto.
24062 (_mm_mmask_i64gather_pd): Ditto.
24063 (_mm256_mmask_i32gather_epi32): Ditto.
24064 (_mm_mmask_i32gather_epi32): Ditto.
24065 (_mm256_mmask_i32gather_epi64): Ditto.
24066 (_mm_mmask_i32gather_epi64): Ditto.
24067 (_mm256_mmask_i64gather_epi32): Ditto.
24068 (_mm_mmask_i64gather_epi32): Ditto.
24069 (_mm256_mmask_i64gather_epi64): Ditto.
24070 (_mm_mmask_i64gather_epi64): Ditto.
24071 (_mm256_i32scatter_ps): Change __addr type to void*.
24072 (_mm256_mask_i32scatter_ps): Ditto.
24073 (_mm_i32scatter_ps): Ditto.
24074 (_mm_mask_i32scatter_ps): Ditto.
24075 (_mm256_i32scatter_pd): Ditto.
24076 (_mm256_mask_i32scatter_pd): Ditto.
24077 (_mm_i32scatter_pd): Ditto.
24078 (_mm_mask_i32scatter_pd): Ditto.
24079 (_mm256_i64scatter_ps): Ditto.
24080 (_mm256_mask_i64scatter_ps): Ditto.
24081 (_mm_i64scatter_ps): Ditto.
24082 (_mm_mask_i64scatter_ps): Ditto.
24083 (_mm256_i64scatter_pd): Ditto.
24084 (_mm256_mask_i64scatter_pd): Ditto.
24085 (_mm_i64scatter_pd): Ditto.
24086 (_mm_mask_i64scatter_pd): Ditto.
24087 (_mm256_i32scatter_epi32): Ditto.
24088 (_mm256_mask_i32scatter_epi32): Ditto.
24089 (_mm_i32scatter_epi32): Ditto.
24090 (_mm_mask_i32scatter_epi32): Ditto.
24091 (_mm256_i32scatter_epi64): Ditto.
24092 (_mm256_mask_i32scatter_epi64): Ditto.
24093 (_mm_i32scatter_epi64): Ditto.
24094 (_mm_mask_i32scatter_epi64): Ditto.
24095 (_mm256_i64scatter_epi32): Ditto.
24096 (_mm256_mask_i64scatter_epi32): Ditto.
24097 (_mm_i64scatter_epi32): Ditto.
24098 (_mm_mask_i64scatter_epi32): Ditto.
24099 (_mm256_i64scatter_epi64): Ditto.
24100 (_mm256_mask_i64scatter_epi64): Ditto.
24101 (_mm_i64scatter_epi64): Ditto.
24102 (_mm_mask_i64scatter_epi64): Ditto.
24103 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
24104 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
24105 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
24106 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
24107 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
24108 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
24109 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
24110 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
24111 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
24112 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
24113 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
24114 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
24115 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
24116 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
24117 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
24118 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
24119 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
24120 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
24121 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
24122 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
24123 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
24124 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
24125 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
24126 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
24127 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
24128 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
24129 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
24130 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
24131 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
24132 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
24133 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
24134 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
24135 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
24136 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
24137 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
24138 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
24139 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
24140 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
24141 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
24142 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
24143 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
24144 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
24145 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
24146 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
24147 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
24148 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
24149 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
24150 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
24151 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
24152 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
24153 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
24154 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
24155 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
24156 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
24157 definitions accordingly.
24159 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
24160 Kuan-Lin Chen <kuanlinchentw@gmail.com>
24163 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
24166 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
24169 * ira-conflicts.c (ira_build_conflicts): Update total conflict
24170 hard regs for inner regno.
24172 2017-01-17 Martin Liska <mliska@suse.cz>
24175 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
24176 assumption and add comment.
24178 2017-01-17 Nathan Sidwell <nathan@acm.org>
24180 * ipa-visibility.c (localize_node): New function, broken out of ...
24181 (function_and_variable_visibility): ... here. Call it.
24183 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24185 PR middle-end/77445
24186 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
24187 correctly set frequency of oudgoing edge.
24188 (duplicate_thread_path): Fix profile updating.
24190 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24193 * configure.ac: Add GCC_BASE_VER.
24194 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
24195 version from BASE-VER file.
24196 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
24197 (gcc.o): Depend on $(BASEVER).
24198 * common.opt (dumpfullversion): New option.
24199 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
24200 * doc/invoke.texi: Document -dumpfullversion.
24201 * doc/install.texi: Document --with-gcc-major-version-only.
24202 * configure: Regenerated.
24204 2017-01-17 Richard Biener <rguenther@suse.de>
24206 PR tree-optimization/71433
24207 * tree-vrp.c (register_new_assert_for): Merge same asserts
24208 on all incoming edges.
24209 (process_assert_insertions_for): Handle insertions at the
24212 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
24214 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
24215 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
24217 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
24220 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
24223 2017-01-17 Alan Modra <amodra@gmail.com>
24226 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
24227 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
24228 symbolic stack limit when pic.
24230 2017-01-16 Martin Sebor <msebor@redhat.com>
24232 PR tree-optimization/78608
24233 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
24235 2017-01-16 Jeff Law <law@redhat.com>
24238 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
24239 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
24240 for several include directories that may be relative to sysroot.
24241 * config/i386/x-mingw32 (gplus_includedir): Define.
24242 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
24243 (native_system_includedir): Likewise.
24244 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
24245 override if TARGET_SYSTEM_ROOT is defined.
24246 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
24248 PR tree-optimization/79090
24249 PR tree-optimization/33562
24250 PR tree-optimization/61912
24251 PR tree-optimization/77485
24252 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
24253 and computed trims into the dump file.
24255 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24257 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
24259 2017-01-16 Jakub Jelinek <jakub@redhat.com>
24262 * gimplify.c (gimplify_init_constructor): If want_value and
24263 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
24267 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
24268 sequence. Formatting fixes.
24269 (doloop_optimize): Formatting fixes.
24272 * gcc.c (debug_level_greater_than_spec_func): New function.
24273 (static_spec_functions): Add debug-level-gt spec function.
24274 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
24276 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24277 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24278 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
24279 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
24280 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
24281 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
24283 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
24285 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
24286 QImode fixups to general and mask registers only.
24288 2017-01-16 Carl Love <cel@us.ibm.com>
24290 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
24291 for built-in functions
24292 vector signed char vec_nabs (vector signed char)
24293 vector signed short vec_nabs (vector signed short)
24294 vector signed int vec_nabs (vector signed int)
24295 vector signed long long vec_nabs (vector signed long long)
24296 vector float vec_nabs (vector float)
24297 vector double vec_nabs (vector double)
24298 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
24300 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
24301 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
24302 * doc/extend.texi: Update the documentation file for the new built-in
24305 2017-01-16 Martin Sebor <msebor@redhat.com>
24307 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
24310 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24312 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
24313 UNSPEC_VSX__XXSPLTD to require special splat handling.
24315 2017-01-16 David Malcolm <dmalcolm@redhat.com>
24318 * system.h: Poison strndup.
24320 2017-01-16 Alan Modra <amodra@gmail.com>
24323 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
24326 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
24328 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
24330 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
24332 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
24333 call recog here. Assert that INSN_CODE (insn) is non-negative.
24335 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24338 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
24340 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
24341 in the currently scheduled RTL fragment.
24343 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24345 PR rtl-optimization/78751
24346 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
24349 2017-01-14 Jeff Law <law@redhat.com>
24351 PR tree-optimization/79090
24352 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
24353 variable length stores.
24354 (compute_trims): Delete dead assignment to *trim_tail.
24355 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
24358 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
24360 PR rtl-optimization/78626
24361 PR rtl-optimization/78727
24362 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
24363 of a block, and split such blocks after everything else is finished.
24365 2017-01-14 Alan Modra <amodra@gmail.com>
24368 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
24369 target legitimate_combined_insn.
24370 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
24371 (rs6000_legitimate_combined_insn): New function.
24372 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
24374 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
24375 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
24376 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
24378 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
24380 * doc/frontends.texi (G++ and GCC): Remove references to Java.
24382 2017-01-13 Jeff Law <law@redhat.com>
24384 PR tree-optimization/33562
24385 PR tree-optimization/61912
24386 PR tree-optimization/77485
24387 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
24389 (delete_dead_assignment): Likewise.
24390 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
24391 statement to delete_dead_call and delete_dead_assignment.
24393 2017-01-13 David Malcolm <dmalcolm@redhat.com>
24396 * substring-locations.c (format_warning_va): Strengthen case 1 so
24397 that both endpoints of the substring must be within the format
24398 range for just the substring to be printed.
24400 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
24402 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
24403 * config/i386/i386.c (ix86_target_string): Add missing options
24404 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
24405 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
24406 flags_other and ix86_target_other to flags2_other. Display unknown
24408 (ix86_valid_target_attribute_inner_p): Add missing options and
24409 reorder options by implied ISAs, as in ix86_target_string.
24411 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24413 * hash-table.h (hash_table::too_empty_p): New function.
24414 (hash_table::expand): Use it.
24415 (hash_table::traverse): Likewise.
24416 (hash_table::empty_slot): Use sizeof (value_type) instead of
24417 sizeof (PTR) to convert bytes to elements. Shrink the table
24418 if the current size is excessive for the current number of
24421 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24423 * ira-costs.c (record_reg_classes): Break from the inner loop
24424 early once alt_fail is known to be true. Update outer loop
24425 handling accordingly.
24427 2017-01-13 Jeff Law <law@redhat.com>
24429 * tree-ssa-dse.c (decrement_count): New function.
24430 (increment_start_addr, maybe_trim_memstar_call): Likewise.
24431 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
24432 when we know the partially dead statement is a mem* function.
24434 PR tree-optimization/61912
24435 PR tree-optimization/77485
24436 * tree-ssa-dse.c: Include expr.h.
24437 (maybe_trim_constructor_store): New function.
24438 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
24440 PR tree-optimization/33562
24441 PR tree-optimization/61912
24442 PR tree-optimization/77485
24443 * doc/invoke.texi: Document new dse-max-object-size param.
24444 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
24445 * tree-ssa-dse.c: Include params.h.
24446 (dse_store_status): New enum.
24447 (initialize_ao_ref_for_dse): New, partially extracted from
24449 (valid_ao_ref_for_dse, normalize_ref): New.
24450 (setup_live_bytes_from_ref, compute_trims): Likewise.
24451 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
24452 (maybe_trim_partially_dead_store): Likewise.
24453 (maybe_trim_complex_store): Likewise.
24454 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
24455 Track what bytes live from the original store. Return tri-state
24456 for dead, partially dead or live.
24457 (dse_dom_walker): Add constructor, destructor and new private members.
24458 (delete_dead_call, delete_dead_assignment): New extracted from
24460 (dse_optimize_stmt): Make a member of dse_dom_walker.
24461 Use initialize_ao_ref_for_dse.
24463 PR tree-optimization/33562
24464 PR tree-optimization/61912
24465 PR tree-optimization/77485
24466 * sbitmap.h (bitmap_count_bits): Prototype.
24467 (bitmap_clear_range, bitmap_set_range): Likewise.
24468 * sbitmap.c (bitmap_clear_range): New function.
24469 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
24471 2017-01-13 Martin Liska <mliska@suse.cz>
24474 * function.c (set_cfun): Add new argument force.
24475 * function.h (set_cfun): Likewise.
24476 * ipa-inline-transform.c (inline_call): Use the function when
24477 strict alising from is dropped for function we inline to.
24479 2017-01-13 Richard Biener <rguenther@suse.de>
24481 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
24482 for dumping GIMPLE INTEGER_CSTs.
24484 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24486 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
24487 to 201112L since C++17.
24489 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24492 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
24493 if -fsanitize=kernel-address is present.
24495 2017-01-13 Richard Biener <rguenther@suse.de>
24497 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
24498 as _Literal ( type ) number in case usual suffixes do not
24499 preserve all information.
24501 2017-01-13 Richard Biener <rguenther@suse.de>
24503 PR tree-optimization/77283
24504 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
24505 and ssa-iterators.h.
24506 (is_feasible_trace): Implement a cost model based on joiner
24509 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24512 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
24513 char or short to __float128/_Float128 directly.
24515 2017-01-12 Martin Sebor <msebor@redhat.com>
24517 to -Wformat-overflow.
24518 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
24519 (min_bytes_remaining): Same.
24520 (get_string_length): Same.
24521 (format_string): Same.
24522 (format_directive): Same.
24524 (pass_sprintf_length::handle_gimple_call): Same.
24526 2017-01-12 Jakub Jelinek <jakub@redhat.com>
24528 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
24529 info.nowrite calls with no lhs that can't throw. Return bool
24530 whether gsi_remove has been called or not.
24531 (pass_sprintf_length::handle_gimple_call): Return bool whether
24532 try_substitute_return_value called gsi_remove. Formatting fix.
24533 (pass_sprintf_length::execute): Don't use gsi_remove if
24534 handle_gimple_call returned true.
24537 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
24538 be removed due to side-effects, don't remove following barrier nor
24539 turn the successor edge into fallthru edge.
24541 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24544 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
24545 element-reversing loads and stores as not swappable.
24547 2017-01-12 Nathan Sidwell <nathan@acm.org>
24548 Nicolai Stange <nicstange@gmail.com>
24550 * combine.c (try_combine): Don't ignore result of overlap checking
24551 loop. Combine overlap & asm check into single loop.
24553 2017-01-12 Richard Biener <rguenther@suse.de>
24555 * tree-pretty-print.c (dump_generic_node): Provide -gimple
24556 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
24558 2017-01-12 Richard Biener <rguenther@suse.de>
24560 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
24561 and TS_TARGET_OPTION directly derive from TS_BASE.
24562 * tree-core.h (tree_optimization_option): Derive from tree_base.
24563 (tree_target_option): Likewise.
24565 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
24567 * config/i386/i386.c (memory_address_length): Increase len
24568 only when rip_relative_addr_p returns false.
24570 2017-01-11 Julia Koval <julia.koval@intel.com>
24572 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
24573 (OPTION_MASK_ISA_SGX_SET): New.
24574 (ix86_handle_option): Handle OPT_msgx.
24575 * config.gcc: Added sgxintrin.h.
24576 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
24577 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
24578 * config/i386/i386.c (ix86_target_string): Add -msgx.
24580 (ix86_option_override_internal): Handle new options.
24581 (ix86_valid_target_attribute_inner_p): Add sgx.
24582 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
24583 * config/i386/i386.opt: Add msgx.
24584 * config/i386/sgxintrin.h: New file.
24585 * config/i386/x86intrin.h: Add sgxintrin.h.
24587 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24590 * fold-const.c (maybe_nonzero_address): Return 1 for function
24592 (tree_single_nonzero_warnv_p): Don't handle function local objects
24596 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
24599 2017-01-11 David Malcolm <dmalcolm@redhat.com>
24602 * opts.c: Include "spellcheck.h"
24603 (struct string_fragment): New struct.
24604 (struct edit_distance_traits<const string_fragment &>): New
24606 (get_closest_sanitizer_option): New function.
24607 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
24609 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24611 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
24613 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
24614 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
24615 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
24616 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
24617 for initial die_offset if dwarf_split_debug_info.
24618 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
24619 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
24621 (output_skeleton_debug_sections): Formatting fix. Use
24622 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
24623 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
24625 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
24627 * config/arm/cortex-a53.md: Add bypasses for
24628 cortex_a53_r2f_cvt.
24629 (cortex_a53_r2f): Only use for transfers.
24630 (cortex_a53_f2r): Likewise.
24631 (cortex_a53_r2f_cvt): Add reservation for conversions.
24632 (cortex_a53_f2r_cvt): Likewise.
24634 2017-01-11 Tamar Christina <tamar.christina@arm.com>
24636 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
24637 to all inlined functions, change static to extern.
24639 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
24642 * config/arm/arm.c (legitimize_pic_address): Handle reference to
24644 (arm_assemble_integer): Likewise.
24646 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
24648 * config.gcc: Use new awk script to check CPU, FPU and architecture
24649 parameters for --with-... options.
24650 * config/arm/parsecpu.awk: New file
24651 * config/arm/arm-cpus.in: New file.
24652 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
24654 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
24656 * config/arm/t-arm: Update dependency rules.
24657 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
24658 of processing .def files.
24659 * config/arm/genopt.sh: Deleted.
24660 * config/arm/gentune.sh: Deleted.
24661 * config/arm/arm-cores.def: Deleted.
24662 * config/arm/arm-arches.def: Deleted.
24663 * config/arm/arm-fpus.def: Deleted.
24664 * config/arm/arm-tune.md: Regenerated.
24665 * config/arm/arm-tables.opt: Regenerated.
24666 * config/arm/arm-cpu.h: New generated file.
24667 * config/arm/arm-cpu-data.h: New generated file.
24668 * config/arm/arm-cpu-cdata.h: New generated file.
24670 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
24673 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
24675 (input_varpool_node): Unpack dynamically_initialized bit.
24677 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
24679 PR rtl-optimization/79032
24680 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
24681 the alignment of the adjusted memory reference against that of MODE,
24682 instead of the alignment of the original memory reference.
24684 2017-01-11 Martin Jambor <mjambor@suse.cz>
24686 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
24688 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
24689 decorated functions.
24691 2017-01-11 Richard Biener <rguenther@suse.de>
24693 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
24694 set range/nonnull info for PHI results. Do not set it on
24695 stmts marked for removal.
24697 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
24699 * expr.c (store_field): In the bitfield case, fetch the return value
24700 from the registers before applying a single big-endian adjustment.
24701 Always do a final load for a BLKmode value not larger than a word.
24703 2017-01-10 David Malcolm <dmalcolm@redhat.com>
24706 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
24707 that we correctly handle column numbers greater than
24708 LINE_MAP_MAX_COLUMN_NUMBER.
24710 2017-01-10 Martin Sebor <msebor@redhat.com>
24712 PR middle-end/78245
24713 * gimple-ssa-sprintf.c (get_destination_size): Call
24714 {init,fini}object_sizes.
24715 * tree-object-size.c (addr_object_size): Adjust.
24716 (pass_through_call): Adjust.
24717 (pass_object_sizes::execute): Adjust.
24718 * tree-object-size.h (fini_object_sizes): Declare.
24720 2017-01-10 Martin Sebor <msebor@redhat.com>
24722 PR tree-optimization/78775
24723 * builtins.c (get_size_range): Move...
24724 * calls.c: ...to here.
24725 (alloc_max_size): Accept zero argument.
24726 (operand_signed_p): Remove.
24727 (maybe_warn_alloc_args_overflow): Call get_size_range.
24728 * calls.h (get_size_range): Declare.
24730 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
24732 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
24733 from TI's devices.csv file as of September 2016.
24734 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
24736 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
24738 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
24739 * doc/invoke.texi: Likewise.
24740 * doc/md.texi: Likewise.
24741 * doc/objc.texi: Likewise.
24743 2017-01-10 Joshua Conner <joshconner@google.com>
24745 * config/arm/fuchsia-elf.h: New file.
24746 * config/fuchsia.h: New file.
24747 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
24748 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
24750 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
24752 2016-01-10 Richard Biener <rguenther@suse.de>
24754 PR tree-optimization/79034
24755 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24756 Propagate out degenerate PHIs in the joiner.
24758 2017-01-10 Martin Liska <mliska@suse.cz>
24760 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
24761 (sort_congruence_classes_by_decl_uid): Likewise.
24762 (sort_congruence_class_groups_by_decl_uid): Likewise.
24763 (sem_item_optimizer::merge_classes): Sort class, groups in these
24764 classes and members in the groups by DECL_UID of declarations.
24765 This would make merge operations stable.
24767 2017-01-10 Martin Liska <mliska@suse.cz>
24769 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
24770 usage of m_classes_vec.
24771 (sem_item_optimizer::~sem_item_optimizer): Likewise.
24772 (sem_item_optimizer::get_group_by_hash): Likewise.
24773 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
24774 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
24775 (sem_item_optimizer::verify_classes): Likewise.
24776 (sem_item_optimizer::process_cong_reduction): Likewise.
24777 (sem_item_optimizer::dump_cong_classes): Likewise.
24778 (sem_item_optimizer::merge_classes): Likewise.
24779 * ipa-icf.h (congruence_class_hash): Rename from
24780 congruence_class_group_hash. Remove declaration of m_classes_vec.
24782 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
24784 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
24785 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
24786 * config.gcc: Add avx512vpopcntdqintrin.h.
24787 * config/i386/avx512vpopcntdqintrin.h: New.
24788 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
24789 * config/i386/i386-builtin-types.def: Add new types.
24790 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
24791 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
24792 __builtin_ia32_vpopcountq_v8di_mask): New.
24793 * config/i386/i386-c.c (ix86_target_macros_internal): Define
24794 __AVX512VPOPCNTDQ__.
24795 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
24796 (PTA_AVX512VPOPCNTDQ): Define.
24797 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
24798 TARGET_AVX512VPOPCNTDQ_P): Define.
24799 * config/i386/i386.opt: Add mavx512vpopcntdq.
24800 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
24801 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
24803 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24805 PR middle-end/77484
24806 * predict.def (PRED_CALL): Set to 67.
24808 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
24810 * expr.c (store_field): In the bitfield case, if the value comes from
24811 a function call and is of an aggregate type returned in registers, do
24812 not modify the field mode; extract the value in all cases if the mode
24813 is BLKmode and the size is not larger than a word.
24815 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
24818 * config/i386/cpuid.h: Fix undefined behavior.
24820 2017-01-04 Jeff Law <law@redhat.com>
24822 PR tree-optimization/79007
24823 PR tree-optimization/67955
24824 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
24825 conservative for pt.null when flag_non_call_exceptions is on.
24827 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24829 PR translation/79019
24830 PR translation/79020
24831 * params.def (PARAM_INLINE_MIN_SPEEDUP,
24832 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
24833 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
24835 * config/avr/avr.opt (maccumulate-args): Likewise.
24836 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
24837 * common.opt (freport-bug): Likewise.
24838 * cif-code.def (CIF_FINAL_ERROR): Likewise.
24839 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
24840 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
24841 translatable string.
24842 * config/i386/i386.c (function_value_32): Likewise.
24843 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
24844 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
24846 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
24847 * common/config/msp430/msp430-common.c (msp430_handle_option):
24849 * symtab.c (symtab_node::verify_base): Likewise.
24850 * opts.c (set_debug_level): Likewise.
24851 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
24852 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
24853 missing whitespace to translatable strings.
24854 * config/avr/avr.md (bswapsi2): Fix typo in comment.
24855 * config/sh/superh.h: Likewise.
24856 * config/i386/xopintrin.h: Likewise.
24857 * config/i386/znver1.md: Likewise.
24858 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
24859 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
24860 * double-int.h (struct double_int): Likewise.
24861 * double-int.c (div_and_round_double): Likewise.
24862 * wide-int.cc: Likewise.
24863 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
24864 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
24865 * cfgcleanup.c (crossjumps_occured): Renamed to ...
24866 (crossjumps_occurred): ... this.
24867 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
24870 PR tree-optimization/78899
24871 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
24872 returning bool return struct loop *, NULL for failure and the new
24874 (versionable_outer_loop_p): Don't version outer loop if it has
24875 dont_vectorized bit set.
24876 (tree_if_conversion): When versioning outer loop, ensure
24877 tree_if_conversion is performed also on the inner loop of the
24878 non-vectorizable outer loop copy.
24879 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
24880 LOOP_VECTORIZED in inner loop of the scalar outer loop and
24881 prevent vectorization of it.
24882 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
24883 the outer loop vectorization of the non-scalar version is attempted
24884 before vectorization of the inner loop in scalar version. If
24885 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
24886 vectorization of its inner loop.
24887 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
24888 has 2 inner loops, rename also on edges from bb whose single pred
24889 is outer_loop->header. Fix typo in function comment.
24891 2017-01-09 Martin Sebor <msebor@redhat.com>
24894 * asan.c (asan_emit_stack_protection): Increase local buffer size
24895 to avoid snprintf truncation warning.
24897 2017-01-09 Andrew Pinski <apinski@cavium.com>
24899 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
24900 to reference thunderx2t99 for the tuning structure
24901 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
24903 (thunderx2t99_extra_costs): This.
24904 * config/aarch64/aarch64-tune.md: Regenerate.
24905 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
24906 (vulcan_addrcost_table): This.
24907 (vulcan_regmove_cost): Rename to ...
24908 (thunderx2t99_regmove_cost): This.
24909 (vulcan_vector_cost): Rename to ...
24910 (thunderx2t99_vector_cost): this.
24911 (vulcan_branch_cost): Rename to ...
24912 (thunderx2t99_branch_cost): This.
24913 (vulcan_tunings): Rename to ...
24914 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
24915 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
24917 2017-01-09 Martin Jambor <mjambor@suse.cz>
24921 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
24922 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
24923 (propagate_vr_accross_jump_function): Use the above function for all
24924 value range computations for pass-through jump functions and type
24925 converasion from explicit value range values.
24926 (ipcp_propagate_stage): Do not attempt to deduce types of formal
24927 parameters from TYPE_ARG_TYPES.
24928 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
24929 (ipa_write_node_info): Stream type of the actual argument.
24930 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
24932 2017-01-09 Martin Liska <mliska@suse.cz>
24935 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
24936 (lookup_compiler): Do not show error message with have_E.
24938 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24940 PR tree-optimization/78938
24941 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
24942 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
24943 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
24944 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
24947 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24949 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
24952 2017-01-09 Richard Biener <rguenther@suse.de>
24954 PR tree-optimization/78997
24955 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
24956 name condition properly.
24958 2017-01-09 Richard Biener <rguenther@suse.de>
24961 * dwarf2out.c (is_cxx): New overload with context.
24962 (is_naming_typedef_decl): Use it.
24964 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24966 * invoke.texi (Option Summary): Correct spacing in option lists
24967 and add line breaks to fix over-long lines.
24969 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24971 PR middle-end/17660
24973 * extend.texi (Common Variable Attributes): Add xref to GCC
24974 Internals manual to explain mode attribute keywords.
24976 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24979 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
24980 and Preprocessor Options.
24981 (Options for Linking): Document -pthread here....
24982 (RS/6000 and PowerPC Options): ...not here.
24983 (Solaris 2 Options): ...or here.
24984 * doc/cppopts.texi: Document -pthread.
24986 2017-01-08 Martin Sebor <msebor@redhat.com>
24988 PR middle-end/77708
24989 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
24990 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
24991 New member functions.
24992 (format_directive): Used them.
24994 (pass_sprintf_length::handle_gimple_call): Same.
24995 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
24996 to avoid truncation for any argument.
24997 (extract_affine_mul): Same.
24998 * tree.c (get_file_function_name): Same.
25000 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25002 PR middle-end/77484
25003 * predict.def (PRED_INDIR_CALL): Set to 86.
25005 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25007 PR preprocessor/54124
25008 * doc/cppopts.texi: Reformat -d subtable to list the full name
25009 of the options. Add cross-reference to the docs for the general
25010 compiler -d options.
25011 * doc/invoke.texi (Developer Options): Add cross-reference to the
25012 preprocessor-specific -d option documentation.
25014 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25016 PR preprocessor/13498
25017 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
25018 redudant material, and reflect new command-line options.
25019 (System Headers): Likewise.
25021 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25023 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
25024 -isystem, and -idirafter. Copy-edit.
25025 * doc/cppopts.texi: Copy-edit. Remove contradiction about
25026 default for -ftrack-macro-expansion. Delete obsolete and
25027 badly-formatted implementation details about -fdebug-cpp output.
25028 * doc/cppwarnopts.texi: Copy-edit.
25030 2017-01-07 David Malcolm <dmalcolm@redhat.com>
25033 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25034 that the transition from a max line width >= 1<<10 to narrower
25035 lines works correctly.
25037 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
25039 * doc/options.texi (PerFunction): New.
25040 * opt-functions.awk (switch_flags): Map both Optimization and
25041 PerFunction to CL_OPTIMIZATION.
25042 * opth-gen.awk: Test for PerFunction flag along with
25044 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
25045 it only when the latter is present. Skip those that don't in
25046 the hash function generator.
25047 * common.opt (fvar-tracking): Mark as PerFunction instead of
25049 (fvar-tracking-assignments): Likewise.
25050 (fvar-tracking-assignments-toggle): Likewise.
25051 (fvar-tracking-uninit): Likewise.
25053 2017-01-07 Jakub Jelinek <jakub@redhat.com>
25055 PR translation/79018
25056 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
25059 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
25062 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
25063 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
25064 TARGET_LONG_JUMP_TABLE_OFFSETS.
25065 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
25066 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
25067 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
25068 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
25069 * config/m68k/m68k.md (tablejump expander): Likewise.
25070 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
25071 TARGET_LONG_JUMP_TABLE_OFFSETS.
25072 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
25073 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
25075 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
25076 David Holsgrove <david.holsgrove@xilinx.com>
25078 * common/config/microblaze/microblaze-common.c
25079 (TARGET_EXCEPT_UNWIND_INFO): Remove.
25080 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
25082 * config/microblaze/microblaze.c (microblaze_must_save_register)
25083 (microblaze_expand_epilogue, microblaze_return_addr): Handle
25085 (microblaze_eh_return): New function.
25086 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
25087 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
25088 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
25089 * config/microblaze/microblaze.md (eh_return): New pattern.
25091 2017-01-06 Jakub Jelinek <jakub@redhat.com>
25093 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
25094 GCC_DIAGNOSTIC_STRINGIFY): Define.
25096 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
25098 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25100 * config/arm/arm.md (<mcrr>): New.
25102 * config/arm/arm.c (arm_arch5te): New.
25103 (arm_option_override): Set arm_arch5te.
25104 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
25106 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
25107 (arm_mcrr_qualifiers): ... this. New.
25108 (MRRC_QUALIFIERS): Define to...
25109 (arm_mrrc_qualifiers): ... this. New.
25110 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
25112 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
25113 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
25114 (MRRCI, mrrc, MRRC): New.
25115 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
25116 VUNSPEC_MRRC2): New.
25118 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25120 * config/arm/arm.md (<mcr>): New.
25122 * config/arm/arm.c (arm_coproc_builtin_available): Add
25123 support for mcr, mrc, mcr2 and mrc2.
25124 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
25125 (arm_mcr_qualifiers): ... this. New.
25126 (MRC_QUALIFIERS): Define to ...
25127 (arm_mrc_qualifiers): ... this. New.
25128 (MCR_QUALIFIERS): Define to ...
25129 (arm_mcr_qualifiers): ... this. New.
25130 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
25132 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
25133 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
25134 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
25135 VUNSPEC_MRC2): New.
25137 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25139 * config/arm/arm.md (*ldc): New.
25143 * config/arm/arm.c (arm_coproc_builtin_available): Add
25144 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
25145 (arm_coproc_ldc_stc_legitimate_address): New.
25146 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
25147 'qualifier_const_pointer'.
25148 (LDC_QUALIFIERS): Define to...
25149 (arm_ldc_qualifiers): ... this. New.
25150 (STC_QUALIFIERS): Define to...
25151 (arm_stc_qualifiers): ... this. New.
25152 * config/arm/arm-protos.h
25153 (arm_coproc_ldc_stc_legitimate_address): New.
25154 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
25155 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
25156 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
25157 stc2, stcl, stc2l): New.
25158 * config/arm/constraints.md (Uz): New.
25159 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
25160 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
25161 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
25162 VUNSPEC_STC2L): New.
25164 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25166 * config/arm/arm.md (<cdp>): New.
25167 * config/arm/arm.c (neon_const_bounds): Rename this ...
25168 (arm_const_bounds): ... this.
25169 (arm_coproc_builtin_available): New.
25170 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
25171 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
25172 (CDP_QUALIFIERS): Define to...
25173 (arm_cdp_qualifiers): ... this. New.
25175 (arm_expand_builtin_args): Add case for 6 arguments.
25176 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
25177 (arm_const_bounds): ... this.
25178 (arm_coproc_builtin_available): New.
25179 * config/arm/arm_acle.h (__arm_cdp): New.
25181 * config/arm/arm_acle_builtins.def (cdp): New.
25183 * config/arm/iterators.md (CDPI,CDP,cdp): New.
25184 * config/arm/neon.md: Rename all 'neon_const_bounds' to
25185 'arm_const_bounds'.
25186 * config/arm/types.md (coproc): New.
25187 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
25188 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
25189 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
25190 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
25192 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25194 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
25195 (UBINOP_QUALIFIERS): New.
25197 (acle_builtin_data): New. Change comment.
25198 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
25199 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
25200 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
25201 arm_acle_builtins.def.
25202 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
25203 (arm_init_acle_builtins): New.
25204 (CRC32_BUILTIN): Remove.
25205 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
25206 crc32cb, crc32ch and crc32cw.
25207 (arm_init_crc32_builtins): Remove.
25208 (arm_init_builtins): Use arm_init_acle_builtins rather
25209 than arm_init_crc32_builtins.
25210 (arm_expand_acle_builtin): New.
25211 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
25212 * config/arm/arm_acle_builtins.def: New.
25214 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25216 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
25217 (arm_builtin_datum): ... this.
25218 (arm_init_neon_builtin): Rename to ...
25219 (arm_init_builtin): ... this. Add a new parameters PREFIX
25220 and USE_SIG_IN_NAME.
25221 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
25222 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
25223 'arm_builtin_datum'.
25224 (arm_init_vfp_builtins): Likewise.
25225 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
25226 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
25227 (arm_expand_neon_args): Rename to ...
25228 (arm_expand_builtin_args): ... this. Rename builtin_arg
25229 enum values and differentiate between ARG_BUILTIN_MEMORY
25230 and ARG_BUILTIN_NEON_MEMORY.
25231 (arm_expand_neon_builtin_1): Rename to ...
25232 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
25233 values, arm_expand_builtin_args and add bool parameter NEON.
25234 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
25235 (arm_expand_vfp_builtin): Likewise.
25236 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
25238 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25240 PR middle-end/77484
25241 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
25242 * predict.c (tree_estimate_probability_bb): Reverse direction of
25243 polymorphic call predictor.
25245 2017-01-06 David Malcolm <dmalcolm@redhat.com>
25247 * passes.c (execute_one_pass): Split out pass-skipping logic into...
25248 (determine_pass_name_match): ...this new function and...
25249 (should_skip_pass_p): ...this new function.
25251 2017-01-06 Nathan Sidwell <nathan@acm.org>
25253 * ipa-visibility.c (function_and_variable_visibility): Reformat
25254 comments and long lines. Remove extrneous if.
25255 * symtab.c (symtab_node::make_decl_local): Fix code format.
25256 (symtab_node::set_section_for_node): Fix comment typo.
25258 2017-01-06 Martin Liska <mliska@suse.cz>
25261 * lra-constraints.c: Rename invariant to lra_invariant.
25262 * predict.c (set_even_probabilities): Initialize e to NULL.
25264 2017-01-05 Martin Sebor <msebor@redhat.com>
25266 PR tree-optimization/78910
25267 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
25268 (format_integer): Correct off-by-one error in the handling
25269 of precision with negative numbers in signed conversions..
25271 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
25273 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
25275 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25277 PR tree-optimization/71016
25278 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
25279 factor_out_conditional_conversion. Formatting fix.
25280 (factor_out_conditional_conversion): Add cond_stmt argument.
25281 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
25282 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
25285 2017-01-05 David Malcolm <dmalcolm@redhat.com>
25287 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
25288 read-rtl-function.o, and selftest-rtl.o.
25289 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
25290 (selftest::aarch64_test_loading_full_dump): New function.
25291 (selftest::aarch64_run_selftests): New function.
25292 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
25293 selftest::aarch64_run_selftests.
25294 * config/i386/i386.c
25295 (selftest::ix86_test_loading_dump_fragment_1): New function.
25296 (selftest::ix86_test_loading_call_insn): New function.
25297 (selftest::ix86_test_loading_full_dump): New function.
25298 (selftest::ix86_test_loading_unspec): New function.
25299 (selftest::ix86_run_selftests): Call the new functions.
25300 * emit-rtl.c (maybe_set_max_label_num): New function.
25301 * emit-rtl.h (maybe_set_max_label_num): New decl.
25302 * function.c (instantiate_decls): Guard call to
25303 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
25304 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
25306 * gensupport.c (gen_reader::gen_reader): Pass "false"
25307 for new "compact" param of rtx_reader.
25308 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
25309 rather than an empty string for NULL strings.
25310 * read-md.c: Potentially include config.h rather than bconfig.h.
25311 Wrap include of errors.h with #ifdef GENERATOR_FILE.
25312 (have_error): New global, copied from errors.c.
25313 (md_reader::read_name): Rename to...
25314 (md_reader::read_name_1): ...this, adding "out_loc" param,
25315 and converting "missing name or number" to returning false, rather
25317 (md_reader::read_name): Reimplement in terms of read_name_1.
25318 (md_reader::read_name_or_nil): New function.
25319 (md_reader::read_string): Handle "(nil)" by returning NULL.
25320 (md_reader::md_reader): Add new param "compact".
25321 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
25322 (md_reader::read_file): New method.
25323 * read-md.h (md_reader::md_reader): Add new param "compact".
25324 (md_reader::read_file): New method.
25325 (md_reader::is_compact): New accessor.
25326 (md_reader::read_name): Convert return type from void to file_location.
25327 (md_reader::read_name_or_nil): New decl.
25328 (md_reader::read_name_1): New decl.
25329 (md_reader::m_compact): New field.
25330 (noop_reader::noop_reader): Pass "false" for new "compact" param
25332 (rtx_reader::rtx_reader): Add new "compact" param.
25333 (rtx_reader::read_rtx_operand): Make virtual and convert return
25334 type from void to rtx.
25335 (rtx_reader::read_until): New decl.
25336 (rtx_reader::handle_any_trailing_information): New virtual function.
25337 (rtx_reader::postprocess): New virtual function.
25338 (rtx_reader::finalize_string): New virtual function.
25339 (rtx_reader::m_in_call_function_usage): New field.
25340 (rtx_reader::m_reuse_rtx_by_id): New field.
25341 * read-rtl-function.c: New file.
25342 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
25343 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
25344 (selftest::verify_three_block_rtl_cfg): New decl.
25345 * read-rtl-function.h: New file.
25346 * read-rtl.c: Potentially include config.h rather than bconfig.h.
25347 For host, include function.h, memmodel.h, and emit-rtl.h.
25348 (one_time_initialization): New function.
25349 (struct compact_insn_name): New struct.
25350 (compact_insn_names): New array.
25351 (find_code): Handle insn codes in compact dumps.
25352 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
25353 (bind_subst_iter_and_attr): Likewise.
25354 (add_condition_to_string): Likewise.
25355 (add_condition_to_rtx): Likewise.
25356 (apply_attribute_uses): Likewise.
25357 (add_current_iterators): Likewise.
25358 (apply_iterators): Likewise.
25359 (initialize_iterators): Guard usage of apply_subst_iterator with
25360 #ifdef GENERATOR_FILE.
25361 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
25362 (md_reader::read_mapping): Likewise.
25363 (add_define_attr_for_define_subst): Likewise.
25364 (add_define_subst_attr): Likewise.
25365 (read_subst_mapping): Likewise.
25366 (check_code_iterator): Likewise.
25367 (rtx_reader::read_rtx): Likewise. Move one-time initialization
25369 (one_time_initialization): New function.
25370 (rtx_reader::read_until): New method.
25371 (read_flags): New function.
25372 (parse_reg_note_name): New function.
25373 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
25374 Handle reuse_rtx ids.
25375 Wrap iterator lookup within #ifdef GENERATOR_FILE.
25376 Add parsing support for RTL dumps, mirroring the special-cases in
25377 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
25378 values, and calling handle_any_trailing_information.
25379 (rtx_reader::read_rtx_operand): Convert return type from void
25380 to rtx, returning return_rtx. Handle case 'e'. Call
25381 finalize_string on XSTR and XTMPL fields.
25382 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
25383 "(nil)" values were omitted. Call the postprocess vfunc on the
25385 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
25386 class ctor. Initialize m_in_call_function_usage. Call
25387 one_time_initialization.
25388 * rtl-tests.c (selftest::test_uncond_jump): Call
25389 set_new_first_and_last_insn.
25390 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
25391 * selftest-rtl.c: New file.
25392 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
25393 (selftest::get_insn_by_uid): New decl.
25394 * selftest-run-tests.c (selftest::run_tests): Call
25395 read_rtl_function_c_tests.
25396 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
25397 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
25400 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
25402 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
25403 operands in a special way. Assert that pos+len <= mode precision.
25405 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25407 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
25408 3 argument Alias with unlimited for the negative form.
25409 (fno-vect-cost-model): Removed.
25411 2017-01-05 Martin Liska <mliska@suse.cz>
25413 * hsa-gen.c (gen_hsa_divmod): New function.
25414 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
25416 2017-01-05 Martin Liska <mliska@suse.cz>
25419 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
25422 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25424 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
25425 small constant length operands.
25427 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25429 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
25430 between loop iterations.
25432 2017-01-05 Martin Liska <mliska@suse.cz>
25435 * gimplify.c (gimplify_decl_expr): Compare to
25436 asan_poisoned_variables instread of checking flags.
25437 (gimplify_target_expr): Likewise.
25438 (gimplify_expr): Likewise.
25439 (gimplify_function_tree): Conditionally initialize
25440 asan_poisoned_variables.
25442 2017-01-04 Jeff Law <law@redhat.com>
25444 PR tree-optimizatin/78812
25445 * rtl.h (contains_mem_rtx_p): Prototype.
25446 * ifcvt.c (containts_mem_rtx_p): Move from here to...
25447 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
25448 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
25449 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
25450 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
25452 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25454 * input.c (assert_char_at_range): Default-initialize actual_range.
25456 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25458 * df-scan.c (df_ref_create_structure): Make regno unsigned,
25459 to match the caller.
25461 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25463 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
25464 insns after final jump in test to emit dummy move.
25466 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25468 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
25469 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
25471 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25473 * multiple_target.c (create_dispatcher_calls): Init e_next.
25474 * tree-ssa-loop-split.c (split_loop): Init border.
25475 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
25478 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
25483 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
25484 (altivec_register_operand): Do not return true if the operand
25485 contains a SUBREG mixing SImode and SFmode.
25486 (vsx_register_operand): Likewise.
25487 (vsx_reg_sfsubreg_ok): New predicate.
25488 (vfloat_operand): Do not return true if the operand contains a
25489 SUBREG mixing SImode and SFmode.
25490 (vint_operand): Likewise.
25491 (vlogical_operand): Likewise.
25492 (gpc_reg_operand): Likewise.
25493 (int_reg_operand): Likewise.
25494 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
25495 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
25496 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
25498 (rs6000_emit_move_si_sf_subreg): New helper function.
25499 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
25500 fixup SUBREGs involving SImode and SFmode.
25501 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
25502 numbers for the new peephole2 optimization.
25503 (peephole2 for SFmode unions): New peephole2 to optimize cases in
25504 the GLIBC math library that do AND/IOR/XOR operations on single
25505 precision floating point.
25506 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
25507 target macros to say whether we need to avoid SUBREGs mixing
25509 (TARGET_ALLOW_SF_SUBREG): Likewise.
25510 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
25511 (UNSPEC_SI_FROM_SF): Likewise.
25512 (iorxor): Change spacing.
25513 (and_ior_xor): New iterator for AND, IOR, and XOR.
25514 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
25515 (movdi_from_sf_zero_ext): Likewise.
25516 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
25517 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
25518 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
25519 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
25520 (fms<mode>4): Likewise.
25521 (fnma<mode>4): Likewise.
25522 (fnms<mode>4): Likewise.
25523 (nfma<mode>4): Likewise.
25524 (nfms<mode>4): Likewise.
25526 2017-01-04 Marek Polacek <polacek@redhat.com>
25529 * doc/invoke.texi: Document -Wpointer-compare.
25531 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25533 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
25536 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
25537 descriptions for -gdwarf-5 and emit them as uleb128 instead of
25540 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
25543 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
25544 documentation of the powerpc_popcntb_ok attribute.
25545 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25546 code to issue warning messages if a requested CPU configuration is
25547 not supported by the binary (assembler and loader) toolchain.
25548 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
25549 made to define a built-in function that has been disabled.
25550 (paired_init_builtins): Add assertion to prevent ICE if attempt is
25551 made to define a built-in function that has been disabled.
25552 (altivec_init_builtins): Add comment explaining why definition
25553 of the DST built-in functions is not preceded by an assertion
25554 check. Add assertions to prevent ICE if attempts are made to
25555 define an altivec predicate or an abs* built-in function that has
25557 (htm_init_builtins): Add comment explaining why definition of the
25558 htm built-in functions is not preceded by an assertion check.
25560 2017-01-04 Jeff Law <law@redhat.com>
25562 PR tree-optimizatin/67955
25563 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
25564 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
25565 the points-to solution does not include pt_null. Use DECL_PT_UID
25568 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
25570 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
25571 Use gen_int_mode instead of gen_lopwart for const_int operands.
25573 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25575 PR tree-optimization/71563
25576 * match.pd: Simplify X << Y into X if Y is known to be 0 or
25577 out of range value - has low bits known to be zero.
25579 2017-01-04 Alan Modra <amodra@gmail.com>
25581 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
25582 * configure: Regenerate.
25583 * config.in: Regenerate.
25585 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25588 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
25589 a substring of the message, but strcmp with the whole message. Ifdef
25590 ENABLE_NLS, translate the message first using dgettext.
25592 2017-01-03 Jeff Law <law@redhat.com>
25594 PR tree-optimizatin/78856
25595 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
25596 (mark_threaded_blocks): Remove code to truncate thread paths that
25597 cross multiple loop headers. Instead invalidate the cached loop
25598 iteration information and handle case of a thread path walking
25599 into an irreducible region.
25601 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
25604 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
25605 assertions. Add support for doing the signbit if the IEEE 128-bit
25606 floating point value is in a GPR.
25607 * config/rs6000/rs6000.md (Fsignbit): Delete.
25608 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
25609 Update the length attribute if the value is in a GPR.
25610 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
25611 the sign or zero extension instruction, since the value is always 0/1.
25612 (signbit<mode>2_dm2): Delete using <Fsignbit>.
25615 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
25616 extracting SImode to a GPR register so that we can generate a
25617 store, limit the vector to be in a traditional Altivec register
25618 for the vextuwrx instruction.
25620 2017-01-03 Ian Lance Taylor <iant@google.com>
25622 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
25624 2017-01-03 Martin Sebor <msebor@redhat.com>
25626 PR tree-optimization/78696
25627 * gimple-ssa-sprintf.c (format_floating): Correct handling of
25628 precision. Use MPFR for %f for greater fidelity. Correct handling
25630 (pass_sprintf_length::compute_format_length): Set width and precision
25631 specified by asrerisk to void_node for vararg functions.
25632 (try_substitute_return_value): Adjust dump output.
25634 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
25636 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
25638 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
25640 * doc/invoke.texi (SPARC options): Document -mlra as the default.
25641 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
25642 -mlra/-mno-lra was passed to the compiler.
25644 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
25646 PR rtl-optimization/65618
25647 * emit-rtl.c (try_split): Move initialization of "before" and
25648 "after" to just before the call to emit_insn_after_setloc.
25650 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
25652 * doc/md.texi (Standard Names): Remove reference to Java frontend.
25654 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
25656 * dwarf2out.c (gen_enumeration_type_die): When
25657 -gno-strict-dwarf, add a DW_AT_encoding attribute.
25659 2017-01-03 Jakub Jelinek <jakub@redhat.com>
25661 PR tree-optimization/78965
25662 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
25663 Change first argument from const call_info & to call_info &. For %n
25664 set info.nowrite to false.
25666 PR middle-end/78901
25667 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
25668 possibly throwing calls.
25670 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
25671 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
25672 and fns handling, rather than in a separate case SSA_NAME.
25674 2017-01-02 Jeff Law <law@redhat.com>
25676 * config/darwin-driver.c (darwin_driver_init): Const-correctness
25677 fixes for first_period and second_period variables.
25679 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
25682 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
25683 (*insvqi_1): New insn pattern.
25684 (*insvqi_1_mem_rex64): Ditto.
25685 (*insvqi_2): Ditto.
25686 (*insvqi_3): Rename from *insvqi.
25688 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
25690 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
25692 * doc/cfg.texi (Edges): Remove reference to Java.
25693 (Maintaining the CFG): Ditto.
25695 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25697 PR middle-end/77674
25698 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
25699 transparent aliases.
25701 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25703 PR middle-end/77484
25704 * predict.def (PRED_CALL): Update hitrate.
25705 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
25706 * predict.c (tree_estimate_probability_bb): Split CALL predictor
25707 into direct/indirect/polymorphic variants.
25709 2017-01-01 Jakub Jelinek <jakub@redhat.com>
25711 Update copyright years.
25713 * gcc.c (process_command): Update copyright notice dates.
25714 * gcov-dump.c (print_version): Ditto.
25715 * gcov.c (print_version): Ditto.
25716 * gcov-tool.c (print_version): Ditto.
25717 * gengtype.c (create_file): Ditto.
25718 * doc/cpp.texi: Bump @copying's copyright year.
25719 * doc/cppinternals.texi: Ditto.
25720 * doc/gcc.texi: Ditto.
25721 * doc/gccint.texi: Ditto.
25722 * doc/gcov.texi: Ditto.
25723 * doc/install.texi: Ditto.
25724 * doc/invoke.texi: Ditto.
25726 Copyright (C) 2017 Free Software Foundation, Inc.
25728 Copying and distribution of this file, with or without modification,
25729 are permitted in any medium without royalty provided the copyright
25730 notice and this notice are preserved.