1 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
4 hard_regno_nregs instead of HARD_REGNO_NREGS.
5 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
6 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
7 (c6x_expand_epilogue): Likewise.
8 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9 (frv_read_iacc_argument): Likewise.
10 * config/sh/sh.c: Include regs.h.
11 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12 (regs_used): Likewise.
13 (output_stack_adjust): Likewise.
14 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15 * expmed.c: Include regs.h.
16 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
17 * ree.c: Include regs.h.
18 (combine_reaching_defs): Use hard_regno_nregs instead of
20 (add_removable_extension): Likewise.
22 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
24 * regs.h (hard_regno_nregs): Turn into a function.
25 (end_hard_regno): Update accordingly.
26 * caller-save.c (setup_save_areas): Likewise.
27 (save_call_clobbered_regs): Likewise.
28 (replace_reg_with_saved_mem): Likewise.
29 (insert_restore): Likewise.
30 (insert_save): Likewise.
31 * combine.c (can_change_dest_mode): Likewise.
32 (move_deaths): Likewise.
33 (distribute_notes): Likewise.
34 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
35 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
36 (rs6000_split_multireg_move): Likewise.
37 (rs6000_register_move_cost): Likewise.
38 (rs6000_memory_move_cost): Likewise.
39 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
40 (rs6000_split_multireg_move): Likewise.
41 (rs6000_register_move_cost): Likewise.
42 (rs6000_memory_move_cost): Likewise.
43 * cselib.c (cselib_reset_table): Likewise.
44 (cselib_lookup_1): Likewise.
45 * emit-rtl.c (set_mode_and_regno): Likewise.
46 * function.c (aggregate_value_p): Likewise.
47 * ira-color.c (setup_profitable_hard_regs): Likewise.
48 (check_hard_reg_p): Likewise.
49 (calculate_saved_nregs): Likewise.
50 (assign_hard_reg): Likewise.
51 (improve_allocation): Likewise.
52 (calculate_spill_cost): Likewise.
53 * ira-emit.c (modify_move_list): Likewise.
54 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
55 (ira_hard_reg_in_set_p): Likewise.
56 * ira.c (setup_reg_mode_hard_regset): Likewise.
57 (clarify_prohibited_class_mode_regs): Likewise.
58 (check_allocation): Likewise.
59 * lra-assigns.c (find_hard_regno_for_1): Likewise.
60 (lra_setup_reg_renumber): Likewise.
61 (setup_try_hard_regno_pseudos): Likewise.
62 (spill_for): Likewise.
63 (assign_hard_regno): Likewise.
64 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
65 * lra-constraints.c (in_class_p): Likewise.
66 (lra_constraint_offset): Likewise.
67 (simplify_operand_subreg): Likewise.
68 (lra_constraints): Likewise.
69 (split_reg): Likewise.
70 (split_if_necessary): Likewise.
71 (invariant_p): Likewise.
72 (inherit_in_ebb): Likewise.
73 * lra-lives.c (process_bb_lives): Likewise.
74 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
75 (get_hard_regs): Likewise.
77 * lra-spills.c (assign_spill_hard_regs): Likewise.
78 * mode-switching.c (create_pre_exit): Likewise.
79 * postreload.c (reload_combine_recognize_pattern): Likewise.
80 * recog.c (peep2_find_free_register): Likewise.
81 * regcprop.c (kill_value_regno): Likewise.
82 (set_value_regno): Likewise.
83 (copy_value): Likewise.
84 (maybe_mode_change): Likewise.
85 (find_oldest_value_reg): Likewise.
86 (copyprop_hardreg_forward_1): Likewise.
87 * regrename.c (check_new_reg_p): Likewise.
88 (regrename_do_replace): Likewise.
89 * reload.c (push_reload): Likewise.
90 (combine_reloads): Likewise.
91 (find_dummy_reload): Likewise.
92 (operands_match_p): Likewise.
93 (find_reloads): Likewise.
94 (find_equiv_reg): Likewise.
95 (reload_adjust_reg_for_mode): Likewise.
96 * reload1.c (count_pseudo): Likewise.
97 (count_spilled_pseudo): Likewise.
99 (clear_reload_reg_in_use): Likewise.
100 (free_for_value_p): Likewise.
101 (allocate_reload_reg): Likewise.
102 (choose_reload_regs): Likewise.
103 (reload_adjust_reg_for_temp): Likewise.
104 (emit_reload_insns): Likewise.
105 (delete_output_reload): Likewise.
106 * rtlanal.c (subreg_get_info): Likewise.
107 * sched-deps.c (sched_analyze_reg): Likewise.
108 * sel-sched.c (init_regs_for_mode): Likewise.
109 (mark_unavailable_hard_regs): Likewise.
110 (choose_best_reg_1): Likewise.
111 (verify_target_availability): Likewise.
112 * valtrack.c (dead_debug_insert_temp): Likewise.
113 * var-tracking.c (track_loc_p): Likewise.
114 (emit_note_insn_var_location): Likewise.
115 * varasm.c (make_decl_rtl): Likewise.
116 * reginfo.c (choose_hard_reg_mode): Likewise.
117 (init_reg_modes_target): Refer directly to
118 this_target_regs->x_hard_regno_nregs.
120 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
122 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
123 instead of hard_regno_nregs.
125 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
127 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
128 end_hard_regno instead of hard_regno_nregs.
129 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
130 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
131 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
132 * ira-color.c (improve_allocation): Likewise.
133 * lra-assigns.c (find_hard_regno_for_1): Likewise.
134 * lra-lives.c (mark_regno_live): Likewise.
135 (mark_regno_dead): Likewise.
136 * lra-remat.c (operand_to_remat): Likewise.
137 * lra.c (collect_non_operand_hard_regs): Likewise.
138 * postreload.c (reload_combine_note_store): Likewise.
139 (move2add_valid_value_p): Likewise.
140 * reload.c (regno_clobbered_p): Likewise.
142 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
144 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
146 * config/v850/v850.c (v850_reorg): Likewise.
147 * reload.c (refers_to_regno_for_reload_p): Likewise.
148 (find_equiv_reg): Likewise.
149 * reload1.c (reload_reg_reaches_end_p): Likewise.
151 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
153 * caller-save.c (add_used_regs): Use REG_NREGS instead of
155 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
156 * config/arm/arm.c (output_move_neon): Likewise.
157 (arm_attr_length_move_neon): Likewise.
158 (neon_split_vcombine): Likewise.
159 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
160 (c6x_mark_reg_written): Likewise.
161 (c6x_dwarf_register_span): Likewise.
162 * config/i386/i386.c (ix86_save_reg): Likewise.
163 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
164 (rws_access_reg): Likewise.
165 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
166 * mode-switching.c (create_pre_exit): Likewise.
167 * ree.c (combine_reaching_defs): Likewise.
168 (add_removable_extension): Likewise.
169 * regcprop.c (find_oldest_value_reg): Likewise.
170 (copyprop_hardreg_forward_1): Likewise.
171 * reload.c (reload_inner_reg_of_subreg): Likewise.
172 (push_reload): Likewise.
173 (combine_reloads): Likewise.
174 (find_dummy_reload): Likewise.
175 (reload_adjust_reg_for_mode): Likewise.
176 * reload1.c (find_reload_regs): Likewise.
177 (forget_old_reloads_1): Likewise.
178 (reload_reg_free_for_value_p): Likewise.
179 (reload_adjust_reg_for_temp): Likewise.
180 (emit_reload_insns): Likewise.
181 (delete_output_reload): Likewise.
182 * sel-sched.c (choose_best_reg_1): Likewise.
183 (choose_best_pseudo_reg): Likewise.
185 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
186 Alan Hayward <alan.hayward@arm.com>
187 David Sherwood <david.sherwood@arm.com>
189 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
190 * target.def (slow_unaligned_access): New hook.
191 * targhooks.h (default_slow_unaligned_access): Declare.
192 * targhooks.c (default_slow_unaligned_access): New function.
193 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
194 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
195 * doc/tm.texi: Regenerate.
196 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
197 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
198 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
200 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
201 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
203 (rs6000_slow_unaligned_access): New function.
204 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
205 (expand_block_compare): Likewise.
206 (expand_strn_compare): Likewise.
207 (rs6000_rtx_costs): Likewise.
208 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
209 (riscv_slow_unaligned_access): Likewise.
210 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
211 (riscv_slow_unaligned_access_p): ...this and make static.
212 (riscv_option_override): Update accordingly.
213 (riscv_slow_unaligned_access): New function.
214 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
215 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
216 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
217 (rs6000_slow_unaligned_access): New function.
218 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
219 (rs6000_rtx_costs): Likewise.
220 * config/rs6000/rs6000-string.c (expand_block_compare)
221 (expand_strn_compare): Use targetm.slow_unaligned_access instead
222 of SLOW_UNALIGNED_ACCESS.
223 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
224 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
225 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
226 of SLOW_UNALIGNED_ACCESS.
227 * expmed.c (simple_mem_bitfield_p): Likewise.
228 * expr.c (alignment_for_piecewise_move): Likewise.
229 (emit_group_load_1): Likewise.
230 (emit_group_store): Likewise.
231 (copy_blkmode_from_reg): Likewise.
232 (emit_push_insn): Likewise.
233 (expand_assignment): Likewise.
234 (store_field): Likewise.
235 (expand_expr_real_1): Likewise.
236 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
237 * lra-constraints.c (simplify_operand_subreg): Likewise.
238 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
239 * gimple-ssa-store-merging.c: Likewise in block comment at start
241 * tree-ssa-strlen.c: Include target.h.
242 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
243 of SLOW_UNALIGNED_ACCESS.
244 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
246 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
248 PR rtl-optimization/82185
249 * expmed.c (emit_store_flag_int): Only test tem if it has been
252 2017-09-12 Richard Biener <rguenther@suse.de>
255 * match.pd ((FTYPE) N CMP CST): Fix typo.
257 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
259 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
261 (mips_near_type_p): Add 'short_call' attribute as a synonym
263 * doc/extend.texi (short_call): Document new function attribute.
265 2017-09-12 Jakub Jelinek <jakub@redhat.com>
268 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
269 assertion check that in the condition.
270 (get_atomic_generic_size): Likewise. Before testing if parameter
271 has pointer type, if it has array type, call for C++
272 default_conversion to perform array-to-pointer conversion.
274 2017-09-12 Richard Biener <rguenther@suse.de>
276 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
277 for operations we cannot scalarize.
279 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
281 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
282 vectors heap vectors. Clean up comments.
283 Make visited_bbs a reference.
284 (profitable_jump_thread_path): Make GC
285 vectors heap vectors. Clean up comments.
287 (convert_and_register_jump_thread_path): Make GC vectors heap
289 (check_subpath_and_update_thread_path): Same. Clean up comments.
290 Make visited_bbs a reference.
291 (handle_phi): Abstract common code to to
292 register_jump_thread_path_if_profitable.
293 Rename VAR_BB to DEF_BB.
295 Make GC vectors heap vectors.
296 Make visited_bbs a reference.
297 (handle_assignment): Same.
298 (register_jump_thread_path_if_profitable): New.
299 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
301 Make GC vectors heap vectors. Clean up comments.
302 Make visited_bbs a reference.
303 (find_jump_threads_backwards): Make visited_bbs live in the stack.
304 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
307 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
310 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
311 words of E_DImode object are reachable by xtensa_uimm8x4 access.
313 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
315 Revert r251800 and r251799.
317 2017-09-11 Martin Jambor <mjambor@suse.cz>
320 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
321 arguments in advance.
322 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
323 use it to find predecessor edges.
324 (naive_outof_ssa): Collect vector of predecessors.
326 2017-09-08 Jason Merrill <jason@redhat.com>
328 PR c++/70029 - ICE with ref-qualifier and -flto
329 * langhooks.h (struct lang_hooks_for_types): Add
330 copy_lang_qualifiers.
331 * attribs.c (build_type_attribute_qual_variant): Use it.
332 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
334 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
335 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
337 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
340 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
341 (*mulsi3_sp64): New instruction.
342 (mulsi3): New expander.
344 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
346 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
348 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
350 * sancov.c: Include memmodel.h.
352 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
355 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
358 2017-09-07 Carl Love <cel@us.ibm.com>
360 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
361 the sldi instruction.
363 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
365 * sancov.c: Include tm_p.h.
367 2017-09-07 Jakub Jelinek <jakub@redhat.com>
370 * output.h (switch_to_other_text_partition): New declaration.
371 * varasm.c (switch_to_other_text_partition): New function.
372 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
373 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
374 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
375 to the other text partition before emitting LCL label and switch back
376 after emitting the word after it.
378 2017-09-07 Richard Biener <rguenther@suse.de>
380 * passes.def (pass_split_crit_edges): Remove instance before PRE.
381 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
382 critical edges here, after loop init.
383 (pass_data_pre): Remove PROP_no_crit_edges flags.
384 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
385 for valueization of call args to avoid leaking VN_TOP.
386 (visit_use): Assert we do not visit default defs.
387 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
388 Use error_mark_node to more easily detect leaking VN_TOP.
389 All default-defs are varying, not VN_TOP. Mark them visited.
390 (run_scc_vn): Make code match comment.
392 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
394 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
395 OPTION_MASK_FLOAT128_KEYWORD.
396 (POWERPC_MASKS): Likewise.
397 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
398 support for the -mfloat128-type option, and make -mfloat128
399 default on PowerPC Linux systems. Define or undefine
400 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
401 Define __float128 to be __ieee128 if IEEE 128-bit support is
402 enabled, or undefine it.
403 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
404 Delete defining __FLOAT128_TYPE__.
405 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
406 -mfloat128-type option and make -mfloat128 default on PowerPC
408 (TARGET_FLOAT128_TYPE): Likewise.
409 (-mfloat128-type): Likewise.
410 * config/rs6000/rs6000.c (rs6000_option_override_internal):
411 Delete the -mfloat128-type option and make -mfloat128 default on
412 PowerPC Linux systems. Always use __ieee128 to be the keyword for
413 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
414 128-bit floating point is enabled. Change tests from using
415 -mfloat128-type to -mfloat128.
416 (rs6000_mangle_type): Use the correct mangling for the __float128
417 type even if normal long double is restricted to 64-bits.
418 (floatn_mode): Enable the _Float128 type by default on VSX Linux
420 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
421 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
422 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
424 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
425 documentation for -mfloat128.
427 2017-09-06 Olivier Hainque <hainque@adacore.com>
429 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
431 2017-09-06 Wish Wu <wishwu007@gmail.com>
432 Jakub Jelinek <jakub@redhat.com>
434 * asan.c (initialize_sanitizer_builtins): Add
435 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
436 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
437 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
438 BT_FN_VOID_UINT64_PTR variables.
439 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
440 (BT_FN_VOID_UINT16_UINT16): Likewise.
441 (BT_FN_VOID_UINT32_UINT32): Likewise.
442 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
443 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
444 (BT_FN_VOID_UINT64_PTR): Likewise.
445 * common.opt (flag_sanitize_coverage): New variable.
446 (fsanitize-coverage=trace-pc): Remove.
447 (fsanitize-coverage=): Add.
448 * flag-types.h (enum sanitize_coverage_code): New enum.
449 * fold-const.c (fold_range_test): Disable non-short-circuit
450 optimization if flag_sanitize_coverage.
451 (fold_truth_andor): Likewise.
452 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
453 * opts.c (COVERAGE_SANITIZER_OPT): Define.
454 (coverage_sanitizer_opts): New array.
455 (get_closest_sanitizer_option): Add OPTS argument, handle also
456 OPT_fsanitize_coverage_.
457 (parse_sanitizer_options): Adjusted to also handle
458 OPT_fsanitize_coverage_.
459 (common_handle_option): Add OPT_fsanitize_coverage_.
460 * sancov.c (instrument_comparison, instrument_switch): New function.
461 (sancov_pass): Add trace-cmp support.
462 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
463 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
464 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
465 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
466 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
467 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
468 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
469 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
470 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
471 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
473 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
475 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
476 error. Only quit immediately if parsing is complete.
477 (BEGIN): Initialize fatal_err and parse_done.
478 (begin fpu, end fpu): Check number of arguments.
479 (begin arch, end arch): Likewise.
480 (begin cpu, end cpu): Likewise.
481 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
482 (optalias): Likewise.
484 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
486 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
487 * config/arm/arm-isa.h: Delete. Move definitions to ...
488 * arm-cpus.in: ... here. Use new feature and fgroup values.
489 * config/arm/arm.c (arm_option_override): Use lower case for feature
491 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
492 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
493 * config/arm/parsecpu.awk (END): Add new command 'isa'.
495 (print_isa_bits_for): New function.
496 (gen_isa): New function.
497 (gen_comm_data): Use print_isa_bits_for.
498 (define feature): New keyword.
499 (define fgroup): New keyword.
500 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
501 (arm-isa.h): Add rule to generate file.
502 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
503 case for feature bit names.
505 2017-09-06 Richard Biener <rguenther@suse.de>
507 * tree-ssa-pre.c (NECESSARY): Remove.
508 (create_expression_by_pieces): Do not touch pass-local flags.
509 (insert_into_preds_of_block): Likewise.
510 (do_pre_regular_insertion): Likewise.
511 (eliminate_insert): Likewise.
512 (eliminate_dom_walker::before_dom_children): Likewise.
513 (fini_eliminate): Do not look at inserted_exprs.
514 (mark_operand_necessary): Remove.
515 (remove_dead_inserted_code): Replace with simple work-list
516 algorithm based on inserted_exprs and SSA uses.
517 (pass_pre::execute): Re-order fini_eliminate and
518 remove_dead_inserted_code.
520 2017-09-06 Olivier Hainque <hainque@adacore.com>
522 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
523 for VxWorks 7. Adjust surrounding comments.
525 2017-09-06 Richard Biener <rguenther@suse.de>
527 * gimple-ssa-strength-reduction.c
528 (find_candidates_dom_walker::before_dom_children): Also allow
531 2017-09-06 Richard Biener <rguenther@suse.de>
533 PR tree-optimization/82108
534 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
535 for gap in the non-permutation SLP case.
537 2017-09-06 Martin Jambor <mjambor@suse.cz>
539 PR tree-optimization/82078
540 * tree-sra.c (sort_and_splice_var_accesses): Move call to
541 add_access_to_work_queue...
542 (build_accesses_from_assign): ...here.
543 (propagate_all_subaccesses): Make sure racc is the group
544 representative, if there is one.
546 2017-09-06 Jakub Jelinek <jakub@redhat.com>
549 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
552 2017-09-06 Richard Biener <rguenther@suse.de>
554 * gimple-ssa-strength-reduction.c
555 (find_candidates_dom_walker::before_doom_children): Use a
556 type and not a mode check.
558 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
561 * config/arm/predicates.md (arm_general_adddi_operand): Create new
563 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
565 2017-09-05 Jeff Law <law@redhat.com>
567 PR tree-optimization/64910
568 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
569 cases where we have 3 or more operands.
571 2017-09-05 Jakub Jelinek <jakub@redhat.com>
574 * omp-low.c (lower_omp_for): Recompute tree invariant if
575 gimple_omp_for_initial/final is ADDR_EXPR.
578 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
579 into gimple val before gimplification fo the COND_EXPR.
581 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
583 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
584 REGION_COPY argument.
585 (thread_through_all_blocks): Remove unused argument to
586 duplicate_thread_path.
588 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
589 Alan Hayward <alan.hayward@arm.com>
590 David Sherwood <david.sherwood@arm.com>
592 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
593 Take a scalar_mode rather than a machine_mode.
594 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
595 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
596 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
597 (aarch64_gen_adjusted_ldpstp): Likewise.
598 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
600 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
601 Alan Hayward <alan.hayward@arm.com>
602 David Sherwood <david.sherwood@arm.com>
604 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
605 Take a scalar_int_mode instead of a machine_mode.
606 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
607 (aarch64_output_scalar_simd_mov_immediate): Likewise.
608 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
609 (aarch64_simd_attr_length_rglist): Delete.
610 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
611 a scalar_int_mode instead of a machine_mode.
612 (aarch64_add_offset): Likewise.
613 (aarch64_internal_mov_immediate): Likewise
614 (aarch64_add_constant_internal): Likewise.
615 (aarch64_add_constant): Likewise.
616 (aarch64_movw_imm): Likewise.
617 (aarch64_rtx_arith_op_extract_p): Likewise.
618 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
619 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
620 Remove assert that the mode isn't a vector.
621 (aarch64_output_scalar_simd_mov_immediate): Likewise.
622 (aarch64_expand_mov_immediate): Update calls after above changes.
623 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
624 (aarch64_and_bitmask_imm): Check for scalar integer modes.
625 (aarch64_move_imm): Likewise.
626 (aarch64_can_const_movi_rtx_p): Likewise.
627 (aarch64_strip_extend): Likewise.
628 (aarch64_extr_rtx_p): Likewise.
629 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
630 a CONST_INT when the mode parameter is VOIDmode.
631 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
633 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
635 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
636 * stor-layout.c (bitwise_mode_for_mode): Likewise.
637 (bitwise_type_for_mode): Update accordingly.
639 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
641 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
642 * stor-layout.c (mode_for_size_tree): Likewise.
643 (mode_for_array): Update accordingly.
644 (layout_decl): Likewise.
645 (compute_record_mode): Likewise. Only set the mode once.
647 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
649 * target.def (get_mask_mode): Change return type to opt_mode.
651 * doc/tm.texi: Regenerate.
652 * targhooks.h (default_get_mask_mode): Return an opt_mode.
653 * targhooks.c (default_get_mask_mode): Likewise.
654 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
655 * optabs-query.c (can_vec_mask_load_store_p): Update use of
656 targetm.get_mask_mode.
657 * tree.c (build_truth_vector_type): Likewise.
659 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
661 * machmode.h (mode_for_vector): Return an opt_mode.
662 * stor-layout.c (mode_for_vector): Likewise.
663 (mode_for_int_vector): Update accordingly.
664 (layout_type): Likewise.
665 * config/i386/i386.c (emit_memmov): Likewise.
666 (ix86_expand_set_or_movmem): Likewise.
667 (ix86_expand_vector_init): Likewise.
668 (ix86_get_mask_mode): Likewise.
669 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
671 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
672 * expmed.c (extract_bit_field_1): Likewise.
673 * expr.c (expand_expr_real_2): Likewise.
674 * optabs-query.c (can_vec_perm_p): Likewise.
675 (can_vec_mask_load_store_p): Likewise.
676 * optabs.c (expand_vec_perm): Likewise.
677 * targhooks.c (default_get_mask_mode): Likewise.
678 * tree-vect-stmts.c (vectorizable_store): Likewise.
679 (vectorizable_load): Likewise.
680 (get_vectype_for_scalar_type_and_size): Likewise.
682 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
684 * machmode.h (mode_for_int_vector): New function.
685 * stor-layout.c (mode_for_int_vector): Likewise.
686 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
687 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
688 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
689 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
690 (s390_expand_vcond): Likewise.
692 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
694 * machmode.h (opt_machine_mode): New type.
695 (opt_mode<T>): Allow construction from anything that can be
697 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
698 (mode_for_size): Return an opt_machine_mode.
699 * stor-layout.c (mode_for_size): Likewise.
700 (mode_for_size_tree): Update call accordingly.
701 (bitwise_mode_for_mode): Likewise.
702 (make_fract_type): Likewise.
703 (make_accum_type): Likewise.
704 * caller-save.c (replace_reg_with_saved_mem): Update call
706 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
707 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
708 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
709 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
710 * expmed.c (extract_bit_field_1): Likewise.
711 * reload.c (get_secondary_mem): Likewise.
712 * varasm.c (assemble_integer): Likewise.
713 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
716 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
718 * machmode.h (decimal_float_mode_for_size): New function.
719 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
720 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
721 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
722 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
723 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
724 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
726 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
728 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
729 (get_builtin_sync_mode): Likewise.
730 (expand_ifn_atomic_compare_exchange): Likewise.
731 (expand_builtin_atomic_clear): Likewise.
732 (expand_builtin_atomic_test_and_set): Likewise.
733 (fold_builtin_atomic_always_lock_free): Likewise.
734 * calls.c (compute_argument_addresses): Likewise.
735 (emit_library_call_value_1): Likewise.
736 (store_one_arg): Likewise.
737 * combine.c (combine_instructions): Likewise.
738 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
739 * config/arm/arm.c (arm_function_value): Likewise.
740 (aapcs_allocate_return_reg): Likewise.
741 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
742 * config/i386/i386.c (construct_container): Likewise.
743 (ix86_gimplify_va_arg): Likewise.
744 (ix86_expand_sse_cmp): Likewise.
745 (emit_memmov): Likewise.
746 (emit_memset): Likewise.
747 (expand_small_movmem_or_setmem): Likewise.
748 (ix86_expand_pextr): Likewise.
749 (ix86_expand_pinsr): Likewise.
750 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
751 * config/microblaze/microblaze.c (microblaze_block_move_straight):
753 * config/mips/mips.c (mips_function_value_1) Likewise.
754 (mips_block_move_straight): Likewise.
755 (mips_expand_ins_as_unaligned_store): Likewise.
756 * config/powerpcspe/powerpcspe.c
757 (rs6000_darwin64_record_arg_advance_flush): Likewise.
758 (rs6000_darwin64_record_arg_flush): Likewise.
759 * config/rs6000/rs6000.c
760 (rs6000_darwin64_record_arg_advance_flush): Likewise.
761 (rs6000_darwin64_record_arg_flush): Likewise.
762 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
763 (sparc_function_value_1): Likewise.
764 * config/spu/spu.c (adjust_operand): Likewise.
765 (spu_emit_branch_or_set): Likewise.
766 (arith_immediate_p): Likewise.
767 * emit-rtl.c (gen_lowpart_common): Likewise.
768 * expr.c (expand_expr_real_1): Likewise.
769 * function.c (assign_parm_setup_block): Likewise.
770 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
771 * reload1.c (alter_reg): Likewise.
772 * stor-layout.c (mode_for_vector): Likewise.
773 (layout_type): Likewise.
775 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
777 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
778 (spu_convert_move): Likewise.
779 * lower-subreg.c (resolve_simple_move): Likewise.
781 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
784 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
785 define_insn to a define_expand.
786 (altivec_vsum2sws_direct): New define_insn.
787 (altivec_vsumsws): Convert from a define_insn to a define_expand.
789 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
791 * config/arm/arm.c (arm_option_params_internal): Improve setting of
794 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
799 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
800 realign stack if stack alignment needed is less than incoming
803 2017-09-05 Marek Polacek <polacek@redhat.com>
806 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
809 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
811 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
813 2017-09-05 Richard Biener <rguenther@suse.de>
815 PR tree-optimization/82084
816 * fold-const.c (can_native_encode_string_p): Handle wide characters.
818 2017-09-05 Richard Biener <rguenther@suse.de>
820 PR tree-optimization/82102
821 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
823 2017-09-05 Martin Liska <mliska@suse.cz>
825 PR tree-optimization/82032
826 * tree-cfg.c (generate_range_test): New function.
827 * tree-cfg.h (generate_range_test): Declared here.
828 * tree-cfgcleanup.c (convert_single_case_switch): New function.
829 (cleanup_control_expr_graph): Use it.
830 * tree-switch-conversion.c (try_switch_expansion): Remove
832 (emit_case_nodes): Use generate_range_test.
834 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
837 * config/i386/i386.md (*<btsc><mode>_mask): Add
838 TARGET_USE_BT to insn constraint.
839 (*btr<mode>_mask): Ditto.
841 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
843 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
844 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
846 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
849 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
850 TARGET_NEON and TARGET_IWMMXT.
851 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
852 TARGET_NEON and TARGET_IWMMXT.
853 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
855 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
857 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
858 (ix86_rewrite_tls_address): Ditto.
859 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
860 (ix86_rewrite_tls_address_1): Ditto.
861 (ix86_rewrite_tls_address): Ditto.
862 * config/i386/predicates.md (tls_address_pattern): New predicate.
863 * config/i386/i386.md (TLS address splitter): New splitter.
865 2017-09-04 Richard Biener <rguenther@suse.de>
867 PR tree-optimization/82084
868 * fold-const.h (can_native_encode_string_p): Declare.
869 * fold-const.c (can_native_encode_string_p): Factor out from ...
870 (native_encode_string): ... here.
871 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
872 vectorizing stores from constants we later cannot handle.
874 2017-09-04 Marek Polacek <polacek@redhat.com>
877 * doc/invoke.texi: Update -Wtautological-compare documentation.
879 2017-09-04 Jeff Law <law@redhat.com>
881 PR tree-optimization/64910
882 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
883 swap the first and last operand if the last is a constant.
885 2017-09-04 Marek Polacek <polacek@redhat.com>
888 * convert.c (do_narrow): When sanitizing signed integer overflows,
889 bail out for signed types.
890 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
892 2017-09-04 Richard Biener <rguenther@suse.de>
894 PR tree-optimization/82060
895 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
896 Move devirtualization after stmt folding and before EH/AB/noreturn
897 cleanup to get the stmt refs canonicalized. Use a bool instead
898 of gimple_modified_p since that doesn't work for NOPs. Schedule
899 NOPs generated by folding for removal.
901 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
902 Alan Hayward <alan.hayward@arm.com>
903 David Sherwood <david.sherwood@arm.com>
905 * coretypes.h (pad_direction): New enum.
906 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
907 (FUNCTION_ARG_PADDING): Likewise.
908 * target.def (function_arg_padding): New hook.
909 * targhooks.h (default_function_arg_padding): Declare.
910 * targhooks.c (default_function_arg_padding): New function.
911 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
912 (TARGET_FUNCTION_ARG_PADDING): ...this.
913 * doc/tm.texi: Regenerate.
914 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
915 instead of direction.
916 (compute_argument_addresses): Likewise.
917 (load_register_parameters): Likewise.
918 (emit_library_call_value_1): Likewise.
919 (store_one_arg): Use targetm.calls.function_arg_padding instead
920 of FUNCTION_ARG_PADDING.
921 (must_pass_in_stack_var_size_or_pad): Likewise.
922 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
923 (emit_group_store): Likewise.
924 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
925 instead of FUNCTION_ARG_PADDING.
926 (emit_push_insn): Likewise, and propagate enum change throughout
928 * function.h (direction): Delete.
929 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
931 * function.c (assign_parm_find_stack_rtl): Likewise.
932 (assign_parm_setup_block_p): Likewise.
933 (assign_parm_setup_block): Likewise.
934 (gimplify_parameters): Likewise.
935 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
936 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
938 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
939 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
940 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
941 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
942 (aarch64_function_arg_padding): ...this new function.
943 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
944 (TARGET_FUNCTION_ARG_PADDING): Redefine.
945 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
946 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
947 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
948 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
949 (arm_pad_arg_upward): Replace with...
950 (arm_function_arg_padding): ...this new function.
951 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
953 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
954 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
955 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
956 (ia64_hpux_function_arg_padding): Replace with...
957 (ia64_function_arg_padding): ...this new function. Use pad_direction
958 instead of direction. Check for TARGET_HPUX.
959 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
960 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
961 (iq2000_function_arg_padding): New function.
962 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
963 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
964 (mips_function_arg_padding): ...this new function.
965 (mips_pad_reg_upward): Update accordingly.
966 (TARGET_FUNCTION_ARG_PADDING): Redefine.
967 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
968 targetm.calls.function_arg_padding.
969 (FUNCTION_ARG_PADDING): Delete.
970 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
971 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
972 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
973 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
974 (nios2_block_reg_padding): Return pad_direction instead of direction.
975 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
976 instead of direction.
977 (nios2_function_arg_padding): Likewise. Make static.
978 (TARGET_FUNCTION_ARG_PADDING): Redefine.
979 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
980 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
981 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
982 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
983 (pa_function_arg_padding): Make static. Return pad_direction instead
985 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
986 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
987 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
988 instead of direction. Use targetm.calls.function_arg_padding.
989 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
990 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
991 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
992 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
993 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
995 (function_arg_padding): Rename to...
996 (rs6000_function_arg_padding): ...this. Make static. Return
997 pad_direction instead of direction.
998 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
999 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
1000 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
1001 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
1002 instead of direction. Use targetm.calls.function_arg_padding.
1003 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
1004 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
1005 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
1006 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
1007 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1008 (function_arg_padding): Rename to...
1009 (rs6000_function_arg_padding): ...this. Make static. Return
1010 pad_direction instead of direction.
1011 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
1012 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
1013 * config/s390/s390.c (s390_function_arg_padding): New function.
1014 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1015 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
1016 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
1017 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
1018 (function_arg_padding): Rename to...
1019 (sparc_function_arg_padding): ...this. Make static. Return
1020 pad_direction instead of direction.
1021 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
1022 * config/spu/spu.c (spu_function_arg_padding): New function.
1023 (TARGET_FUNCTION_ARG_PADDING): Redefine.
1024 * system.h (FUNCTION_ARG_PADDING): Poison.
1026 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1027 Alan Hayward <alan.hayward@arm.com>
1028 David Sherwood <david.sherwood@arm.com>
1030 * target.def (modes_tieable_p): New hook.
1031 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
1032 (TARGET_MODES_TIEABLE_P): ...this.
1033 * doc/tm.texi.in: Regenerate.
1034 * hooks.h (hook_bool_mode_mode_true): Declare.
1035 * hooks.c (hook_bool_mode_mode_true): New function.
1036 * combine.c (subst): Use targetm.modes_tieable_p instead of
1038 * dse.c (find_shift_sequence): Likewise.
1039 * expmed.c (extract_low_bits): Likewise.
1040 * lower-subreg.c: Include target.h.
1041 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
1043 * rtlanal.c (rtx_cost): Likewise.
1044 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
1045 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
1046 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
1047 (TARGET_MODES_TIEABLE_P): Redefine.
1048 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
1049 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
1050 (TARGET_MODES_TIEABLE_P): Redefine.
1051 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
1052 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
1053 (arc_modes_tieable_p): New function.
1054 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
1055 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
1056 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
1057 (arm_modes_tieable_p): Make static.
1058 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
1059 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
1060 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
1061 (TARGET_MODES_TIEABLE_P): Redefine.
1062 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
1063 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
1064 (TARGET_MODES_TIEABLE_P): Redefine.
1065 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
1066 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
1067 (cr16_modes_tieable_p): New function.
1068 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
1069 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
1070 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
1071 (TRULY_NOOP_TRUNCATION): Update comment.
1072 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
1073 (TRULY_NOOP_TRUNCATION): Update comment.
1074 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
1075 (frv_modes_tieable_p): New function.
1076 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
1077 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
1078 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
1079 (TARGET_MODES_TIEABLE_P): Redefine.
1080 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
1081 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
1082 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
1083 (TARGET_MODES_TIEABLE_P): Redefine.
1084 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
1085 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
1086 (ia64_modes_tieable_p): New function.
1087 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
1088 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
1089 (iq2000_modes_tieable_p): New function.
1090 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
1091 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
1092 (lm32_modes_tieable_p): New function.
1093 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
1094 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
1095 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
1096 (TARGET_MODES_TIEABLE_P): Redefine.
1097 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
1098 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
1099 (m32r_modes_tieable_p): New function.
1100 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
1101 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
1102 (m68k_modes_tieable_p): New function.
1103 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
1104 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
1105 (mcore_modes_tieable_p): New function.
1106 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
1107 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
1109 (TARGET_MODES_TIEABLE_P): Redefine.
1110 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
1111 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
1112 * config/mips/mips.c (mips_modes_tieable_p): Make static.
1113 (TARGET_MODES_TIEABLE_P): Redefine.
1114 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
1115 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
1116 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
1117 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
1118 (mn10300_modes_tieable_p): ...this and make static.
1119 (TARGET_MODES_TIEABLE_P): Redefine.
1120 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
1121 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
1122 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
1123 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
1124 (msp430_modes_tieable_p): Make static.
1125 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
1126 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
1127 (TARGET_MODES_TIEABLE_P): Redefine.
1128 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
1129 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
1130 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
1131 (TARGET_MODES_TIEABLE_P): Redefine.
1132 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
1133 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
1134 * config/pa/pa.c (pa_modes_tieable_p): Make static.
1135 (TARGET_MODES_TIEABLE_P): Redefine.
1136 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
1137 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
1138 (pdp11_modes_tieable_p): New function.
1139 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
1140 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
1141 (rs6000_modes_tieable_p): New function.
1142 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1143 * config/powerpcspe/powerpcspe.md: Update comment.
1144 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
1145 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
1146 (TARGET_MODES_TIEABLE_P): Redefine.
1147 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
1148 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
1149 (rl78_modes_tieable_p): New function.
1150 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
1151 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
1152 (rs6000_modes_tieable_p): New function.
1153 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
1154 * config/rs6000/rs6000.md: Update comment.
1155 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
1156 * config/rx/rx.c (rx_modes_tieable_p): New function.
1157 (TARGET_MODES_TIEABLE_P): Redefine.
1158 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
1159 * config/s390/s390.c (s390_modes_tieable_p): New function.
1160 (TARGET_MODES_TIEABLE_P): Redefine.
1161 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
1162 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
1163 (sh_modes_tieable_p): New function.
1164 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
1165 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
1166 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
1167 (sparc_modes_tieable_p): Make static.
1168 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
1169 * config/spu/spu.c (spu_modes_tieable_p): New function.
1170 (TARGET_MODES_TIEABLE_P): Redefine.
1171 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
1172 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
1173 (TARGET_MODES_TIEABLE_P): Redefine.
1174 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
1175 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
1176 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
1177 * config/v850/v850.c (v850_modes_tieable_p): New function.
1178 (TARGET_MODES_TIEABLE_P): Redefine.
1179 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
1180 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
1181 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
1182 (visium_modes_tieable_p): New function.
1183 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
1184 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
1185 (xtensa_modes_tieable_p): New function.
1186 * system.h (MODES_TIEABLE_P): Poison.
1188 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1189 Alan Hayward <alan.hayward@arm.com>
1190 David Sherwood <david.sherwood@arm.com>
1192 * target.def (hard_regno_mode_ok): New hook.
1193 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
1194 (TARGET_HARD_REGNO_MODE_OK): ...this.
1195 * doc/tm.texi.in: Regenerate.
1196 * hooks.h (hook_bool_uint_mode_true): Declare.
1197 * hooks.c (hook_bool_uint_mode_true): New function.
1198 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
1200 * genpreds.c (write_insn_preds_c): Add an include of target.h.
1201 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
1202 instead of HARD_REGNO_MODE_OK.
1203 * caller-save.c: Include target.h.
1204 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
1206 * combine.c (can_combine_p): Likewise.
1207 (combinable_i3pat): Likewise.
1208 (can_change_dest_mode): Likewise.
1209 * expr.c (init_expr_target): Likewise.
1210 (convert_move): Likewise.
1211 (convert_modes): Likewise.
1212 * ira.c (setup_prohibited_class_mode_regs): Likewise.
1213 (setup_prohibited_mode_move_regs): Likewise.
1214 * ira.h (target_ira): Likewise.
1215 * lra-assigns.c (find_hard_regno_for_1): Likewise.
1216 * lra-constraints.c (process_alt_operands): Likewise.
1217 (split_reg): Likewise.
1218 * recog.c (peep2_find_free_register): Likewise.
1219 * ree.c (combine_reaching_defs): Likewise.
1220 * regcprop.c (maybe_mode_change): Likewise.
1221 * reginfo.c (init_reg_sets_1): Likewise.
1222 (choose_hard_reg_mode): Likewise.
1223 (simplifiable_subregs): Likewise.
1224 * regrename.c (check_new_reg_p): Likewise.
1225 * reload.c (find_valid_class): Likewise.
1226 (find_valid_class_1): Likewise.
1227 (reload_inner_reg_of_subreg): Likewise.
1228 (push_reload): Likewise.
1229 (combine_reloads): Likewise.
1230 (find_dummy_reload): Likewise.
1231 (find_reloads): Likewise.
1232 * reload1.c (find_reg): Likewise.
1233 (set_reload_reg): Likewise.
1234 (allocate_reload_reg): Likewise.
1235 (choose_reload_regs): Likewise.
1236 (reload_adjust_reg_for_temp): Likewise.
1237 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
1238 (simplify_subreg_regno): Likewise.
1239 * sel-sched.c (init_regs_for_mode): Likewise.
1240 * varasm.c (make_decl_rtl): Likewise.
1241 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
1242 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
1244 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
1245 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
1246 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1247 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
1248 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
1249 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1250 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
1251 (arc_mode_class): Delete.
1252 (HARD_REGNO_MODE_OK): Delete.
1253 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1254 (arc_hard_regno_mode_ok): Rename old array to...
1255 (arc_hard_regno_mode_ok_modes): ...this.
1256 (arc_conditional_register_usage): Update accordingly.
1257 (arc_mode_class): Make static.
1258 (arc_hard_regno_mode_ok): New function.
1259 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
1260 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
1261 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1262 (arm_hard_regno_mode_ok): Make static.
1263 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
1265 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
1266 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
1267 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
1269 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1270 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
1271 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
1272 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
1273 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
1274 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1275 * config/bfin/predicates.md (valid_reg_operand): Use
1276 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
1277 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
1278 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
1279 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1280 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
1281 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
1282 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1283 (cr16_hard_regno_mode_ok): Make static and return a bool.
1284 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
1285 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1286 (cris_hard_regno_mode_ok): New function.
1287 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
1288 (epiphany_mode_class): Delete.
1289 (HARD_REGNO_MODE_OK): Delete.
1290 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
1291 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1292 (hard_regno_mode_ok): Rename to...
1293 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
1294 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
1295 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
1297 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
1298 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
1299 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1300 (frv_hard_regno_mode_ok): Make static and return a bool.
1301 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
1303 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
1304 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
1305 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
1306 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
1308 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1309 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
1310 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
1311 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
1313 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1314 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
1315 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1316 (ia64_hard_regno_mode_ok): New function.
1317 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
1318 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1319 (iq2000_hard_regno_mode_ok): New function.
1320 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
1321 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1322 (lm32_hard_regno_mode_ok): New function.
1323 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
1324 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
1325 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
1326 instead of HARD_REGNO_MODE_OK.
1327 (m32c_hard_regno_ok): Rename to...
1328 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
1329 (m32c_cannot_change_mode_class): Update accordingly.
1330 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1331 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
1332 (m32r_mode_class): Delete.
1333 (HARD_REGNO_MODE_OK): Delete.
1334 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1335 (m32r_hard_regno_mode_ok): Rename to...
1336 (m32r_hard_regno_modes): ...this.
1337 (m32r_mode_class): Make static.
1338 (m32r_hard_regno_mode_ok): New function.
1339 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
1340 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
1341 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1342 (m68k_hard_regno_mode_ok): Make static.
1343 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
1344 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1345 (mcore_hard_regno_mode_ok): New function.
1346 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
1347 (HARD_REGNO_MODE_OK): Delete.
1348 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
1350 (microblaze_hard_regno_mode_ok_p): ...this and make static.
1351 (microblaze_hard_regno_mode_ok): New function.
1352 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1353 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
1354 (mips_hard_regno_mode_ok): Delete.
1355 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
1356 (mips_hard_regno_mode_ok_p): ...this and make static.
1357 (mips_hard_regno_mode_ok_p): Rename to...
1358 (mips_hard_regno_mode_ok_uncached): ...this.
1359 (mips_hard_regno_mode_ok): New function.
1360 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
1361 of HARD_REGNO_MODE_OK.
1362 (mips_option_override): Update after above name changes.
1363 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1364 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
1365 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
1366 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
1367 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
1368 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1369 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
1370 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
1371 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
1372 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1373 (msp430_hard_regno_mode_ok): Make static and return a bool.
1374 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
1375 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
1376 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
1378 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1379 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
1380 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
1381 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
1382 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
1383 (PA_HARD_REGNO_MODE_OK): ...this
1384 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
1385 (PA_HARD_REGNO_MODE_OK): ...this.
1386 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1387 (pa_hard_regno_mode_ok): New function.
1388 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
1389 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1390 (pdp11_hard_regno_mode_ok): New function.
1391 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
1392 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
1394 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
1396 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1397 (rs6000_hard_regno_mode_ok): Rename to...
1398 (rs6000_hard_regno_mode_ok_uncached): ...this.
1399 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1400 (rs6000_hard_regno_mode_ok): New function.
1401 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
1402 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
1403 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
1404 (riscv_hard_regno_mode_ok): ...this and make static.
1405 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1406 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
1407 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
1408 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1409 (rl78_hard_regno_mode_ok): Make static and return bool.
1410 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
1411 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
1413 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
1414 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1415 (rs6000_hard_regno_mode_ok): Rename to...
1416 (rs6000_hard_regno_mode_ok_uncached): ...this.
1417 (rs6000_init_hard_regno_mode_ok): Update accordingly.
1418 (rs6000_hard_regno_mode_ok): New function.
1419 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
1420 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
1421 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1422 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
1423 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
1424 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
1425 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1426 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
1427 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
1428 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1429 (sh_hard_regno_mode_ok): Make static.
1430 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
1431 instead of HARD_REGNO_MODE_OK.
1432 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
1433 (sparc_mode_class): Delete.
1434 (HARD_REGNO_MODE_OK): Delete.
1435 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1436 (hard_regno_mode_classes): Make static.
1437 (sparc_mode_class): Likewise.
1438 (sparc_hard_regno_mode_ok): New function.
1439 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
1440 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
1441 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
1443 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1444 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
1445 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
1446 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
1447 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
1448 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1449 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
1450 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
1451 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
1452 (visium_hard_regno_mode_ok): New function.
1453 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
1454 instead of HARD_REGNO_MODE_OK.
1455 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
1456 (HARD_REGNO_MODE_OK): Delete.
1457 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
1458 (xtensa_hard_regno_mode_ok_p): ...this and make static.
1459 (xtensa_option_override): Update accordingly.
1460 (TARGET_HARD_REGNO_MODE_OK): Redefine.
1461 (xtensa_hard_regno_mode_ok): New function.
1462 * system.h (HARD_REGNO_MODE_OK): Poison.
1464 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1465 Alan Hayward <alan.hayward@arm.com>
1466 David Sherwood <david.sherwood@arm.com>
1468 * target.def (hard_regno_call_part_clobbered): New hook.
1469 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
1470 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
1471 * doc/tm.texi: Regenerate.
1472 * hooks.h (hook_bool_uint_mode_false): Declare.
1473 * hooks.c (hook_bool_uint_mode_false): New function.
1474 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1475 * cselib.c (cselib_process_insn): Use
1476 targetm.hard_regno_call_part_clobbered instead of
1477 HARD_REGNO_CALL_PART_CLOBBERED.
1478 * ira-conflicts.c (ira_build_conflicts): Likewise.
1479 * ira-costs.c (ira_tune_allocno_costs): Likewise.
1480 * lra-constraints.c (need_for_call_save_p): Likewise.
1481 * lra-lives.c: Include target.h.
1482 (check_pseudos_live_through_calls): Use
1483 targetm.hard_regno_call_part_clobbered instead of
1484 HARD_REGNO_CALL_PART_CLOBBERED.
1485 * regcprop.c: Include target.h.
1486 (copyprop_hardreg_forward_1): Use
1487 targetm.hard_regno_call_part_clobbered instead of
1488 HARD_REGNO_CALL_PART_CLOBBERED.
1489 * reginfo.c (choose_hard_reg_mode): Likewise.
1490 * regrename.c (check_new_reg_p): Likewise.
1491 * reload.c (find_equiv_reg): Likewise.
1492 * reload1.c (emit_reload_insns): Likewise.
1493 * sched-deps.c (deps_analyze_insn): Likewise.
1494 * sel-sched.c (init_regs_for_mode): Likewise.
1495 (mark_unavailable_hard_regs): Likewise.
1496 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
1497 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1498 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
1500 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1501 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1502 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
1504 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
1506 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1507 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1508 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
1510 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1511 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1512 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
1514 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1515 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
1517 * config/powerpcspe/powerpcspe.c
1518 (rs6000_hard_regno_call_part_clobbered): New function.
1519 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1520 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1521 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
1523 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1524 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1525 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
1527 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
1528 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
1529 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
1531 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1532 Alan Hayward <alan.hayward@arm.com>
1533 David Sherwood <david.sherwood@arm.com>
1535 * rtl.h (subreg_memory_offset): Declare.
1536 * emit-rtl.c (subreg_memory_offset): New function.
1537 * expmed.c (store_bit_field_1): Use it.
1538 * expr.c (undefined_operand_subword_p): Likewise.
1539 * simplify-rtx.c (simplify_subreg): Likewise.
1541 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
1543 PR rtl-optimization/57448
1546 * optabs.c (expand_atomic_load): Place compiler memory barriers if
1547 using atomic_load pattern.
1548 (expand_atomic_store): Likewise.
1550 2017-09-04 Jakub Jelinek <jakub@redhat.com>
1553 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
1554 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
1555 handling. Use replace_call_with_value with NULL instead of
1556 gsi_replace, unlink_stmt_vdef and release_defs.
1558 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
1561 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
1563 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
1566 * rtl.h (emit_library_call_value_1): Declare.
1567 (emit_library_call): Replace declaration with a series of overloads.
1568 Remove the parameter count argument.
1569 (emit_library_call_value): Likewise.
1570 * calls.c (emit_library_call_value_1): Make global. Replace varargs
1571 with an "rtx_mode_t *".
1572 (emit_library_call_value): Delete.
1573 (emit_library_call): Likewise.
1574 * asan.c (asan_emit_stack_protection): Update calls accordingly.
1575 (asan_emit_allocas_unpoison): Likewise.
1576 * builtins.c (expand_builtin_powi): Likewise.
1577 (expand_asan_emit_allocas_unpoison): Likewise.
1578 * cfgexpand.c (expand_main_function): Likewise.
1579 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
1580 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
1581 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
1582 * config/arm/arm.c (arm_trampoline_init): Likewise.
1583 (arm_call_tls_get_addr): Likewise.
1584 (arm_expand_divmod_libfunc): Likewise.
1585 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
1586 (smulsi3_highpart): Likewise.
1587 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
1588 (c6x_expand_compare): Likewise.
1589 (c6x_expand_movmem): Likewise.
1590 * config/frv/frv.c (frv_trampoline_init): Likewise.
1591 * config/i386/i386.c (ix86_trampoline_init): Likewise.
1592 (ix86_expand_divmod_libfunc): Likewise.
1593 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
1594 (ia64_expand_compare): Likewise.
1595 (ia64_profile_hook): Likewise.
1596 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
1597 (nonlocal_goto): Likewise.
1598 (restore_stack_nonlocal): Likewise.
1599 * config/m32r/m32r.c (block_move_call): Likewise.
1600 (m32r_trampoline_init): Likewise.
1601 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
1602 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
1603 (m68k_call_m68k_read_tp): Likewise.
1604 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
1605 (microblaze_expand_divide): Likewise.
1606 * config/mips/mips.h (mips_args): Likewise.
1607 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
1608 (MIPS_ICACHE_SYNC): Likewise.
1609 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
1610 (nios2_trampoline_init): Likewise.
1611 * config/pa/pa.c (hppa_tls_call): Likewise.
1612 (pa_trampoline_init): Likewise.
1613 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
1614 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
1615 (expand_strn_compare): Likewise.
1616 (rs6000_generate_compare): Likewise.
1617 (rs6000_expand_float128_convert): Likewise.
1618 (output_profile_hook): Likewise.
1619 (rs6000_trampoline_init): Likewise.
1620 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
1621 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
1622 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
1623 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
1624 (rs6000_generate_compare): Likewise.
1625 (rs6000_expand_float128_convert): Likewise.
1626 (output_profile_hook): Likewise.
1627 (rs6000_trampoline_init): Likewise.
1628 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
1629 * config/sh/sh.c (sh_trampoline_init): Likewise.
1630 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
1631 (sparc_emit_float_lib_cmp): Likewise.
1632 (sparc32_initialize_trampoline): Likewise.
1633 (sparc64_initialize_trampoline): Likewise.
1634 (sparc_profile_hook): Likewise.
1635 * config/spu/spu.c (ea_load_store): Likewise.
1636 * config/spu/spu.md (floatunssidf2): Likewise.
1637 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
1638 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
1639 * config/visium/visium.c (expand_block_move_4): Likewise.
1640 (expand_block_move_2): Likewise.
1641 (expand_block_move_1): Likewise.
1642 (expand_block_set_4): Likewise.
1643 (expand_block_set_2): Likewise.
1644 (expand_block_set_1): Likewise.
1645 (visium_trampoline_init): Likewise.
1646 (visium_profile_hook): Likewise.
1647 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
1648 (xtensa_setup_frame_addresses): Likewise.
1649 (xtensa_trampoline_init): Likewise.
1650 * except.c (sjlj_emit_function_enter): Likewise.
1651 (sjlj_emit_function_exit): Likewise.
1652 * explow.c (allocate_dynamic_stack_space): Likewise.
1653 (probe_stack_range): Likewise.
1654 * expr.c (convert_mode_scalar): Likewise.
1655 * optabs.c (expand_binop): Likewise.
1656 (expand_twoval_binop_libfunc): Likewise.
1657 (expand_unop): Likewise.
1658 (prepare_cmp_insn): Likewise.
1659 (prepare_float_lib_cmp): Likewise.
1660 (expand_float): Likewise.
1661 (expand_fix): Likewise.
1662 (expand_fixed_convert): Likewise.
1663 (maybe_emit_sync_lock_test_and_set): Likewise.
1664 (expand_atomic_compare_and_swap): Likewise.
1665 (expand_mem_thread_fence): Likewise.
1666 (expand_atomic_fetch_op): Likewise.
1668 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
1670 * doc/generic.texi (OpenACC): Adjust URL.
1671 * doc/invoke.texi (C Dialect Options): Ditto.
1673 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
1675 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
1676 predicate for operand 1. Add (m,<S>) constraint.
1677 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
1678 Prevent memory operand 1 with register operand 2.
1680 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
1682 PR rtl-optimization/82024
1683 * combine.c (try_combine): If the combination result is a PARALLEL,
1684 and we only need to retain the SET in there that would be placed
1685 at I2, check that we can place that at I3 instead, before doing so.
1687 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1690 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
1692 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
1693 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
1696 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
1697 Jeff Law <law@redhat.com>
1699 * varasm.c (bss_initializer_p): Do not put constants into .bss
1700 (categorize_decl_for_section): Handle bss_initializer_p returning
1701 false when DECL_INITIAL is NULL.
1703 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1706 * config/s390/s390.c (s390_can_inline_p): New function.
1708 2017-09-01 Jeff Law <law@redhat.com>
1710 PR tree-optimization/82052
1711 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
1712 Always initialize the returned slot after a hash table miss
1713 when INSERT is true.
1715 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
1717 * config/s390/s390.md (mem_signal_fence): Remove.
1718 * doc/md.texi (mem_signal_fence): Remove.
1719 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
1721 * target-insns.def (mem_signal_fence): Remove.
1723 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1726 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
1729 * asan.c (create_odr_indicator): Strip name encoding from assembler
1730 name before appending it after __odr_asan_.
1732 2017-09-01 Martin Liska <mliska@suse.cz>
1734 PR tree-optimization/82059
1735 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
1736 frequency only when an edge is redirected.
1738 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1740 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
1741 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
1742 (arc_conditional_register_usage): Remove ARC600 lp_count
1744 (arc_file_start): Emit Tag_ARC_CPU_variation.
1745 (arc_can_use_doloop_p): New conditions to use ZOLs.
1746 (hwloop_fail): New function.
1747 (hwloop_optimize): Likewise.
1748 (hwloop_pattern_reg): Likewise.
1749 (arc_doloop_hooks): New struct, to be used with reorg_loops.
1750 (arc_reorg_loops): New function, calls reorg_loops.
1751 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
1752 (arc600_corereg_hazard): Remove ZOL checking, case handled by
1754 (arc_loop_hazard): Remove function, functionality moved into
1756 (arc_hazard): Remove arc_loop_hazard call.
1757 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
1758 into hwloop_optimize.
1759 (arc_label_align): Remove ZOL handling.
1760 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
1761 * config/arc/arc.md (doloop_begin): Remove pattern.
1762 (doloop_begin_i): Likewise.
1763 (doloop_end_i): Likewise.
1764 (doloop_fallback): Likewise.
1765 (doloop_fallback_m): Likewise.
1766 (doloop_end): Reimplement expand.
1767 (arc_lp): New pattern for LP instruction.
1768 (loop_end): New pattern.
1769 (loop_fail): Likewise.
1770 (decrement_and_branch_until_zero): Likewise.
1771 * config/arc/arc.opt (mlpc-width): New option.
1772 * doc/invoke.texi (mlpc-width): Document option.
1774 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1776 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
1777 (arc_ccfsm_advance): Fix checking for delay slots.
1778 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
1780 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1782 * config/arc/arc.md (movqi_insn): Add stores to save constant long
1784 (movhi_insn): Update store instruction constraint which are saving
1785 6-bit short immediates.
1786 (movsi_insn): Consider also short scaled load operations.
1787 (zero_extendhisi2_i): Use Usd constraint instead of T.
1788 (extendhisi2_i): Add q constraint.
1789 (arc_clzsi2): Add type and length attributes.
1790 (arc_ctzsi2): Likewise.
1791 * config/arc/constraints.md (Usc): Update constraint, the
1792 assembler can parse two relocations for a single instruction.
1794 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1796 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
1797 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
1799 2017-08-31 Olivier Hainque <hainque@adacore.com>
1801 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
1802 match as powerpc-wrs-vxworks*.
1804 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
1806 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
1807 register constraint for by-element operand.
1808 (aarch64_mls_elt_merge<mode>): Likewise.
1810 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1812 * config/arc/arc.c (arc_can_follow_jump): Check for short
1815 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1817 * config.gcc: Use g.opt for arc.
1818 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
1819 functionality moved to ...
1820 (legitimate_scaled_address_p): New function, ...here.
1821 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
1822 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
1823 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
1825 (arc_override_options): Handle G option.
1826 (arc_output_pic_addr_const): Correct function definition.
1827 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
1828 (arc_decl_anon_ns_mem_p): Delete.
1829 (arc_in_small_data_p): Overhaul this function to take into
1830 consideration the value given via G option.
1831 (arc_rewrite_small_data_1): Renamed and corrected old
1832 arc_rewrite_small_data function.
1833 (arc_rewrite_small_data): New function.
1834 (small_data_pattern): Don't use pic_offset_table_rtx.
1835 * config/arc/arc.h (CC1_SPEC): Recognize G option.
1836 * config/arc/simdext.md (movmisalignv2hi): Use
1837 prepare_move_operands function.
1839 (movmisalign*): Likewise.
1840 * doc/invoke.texi (ARC options): Document -G option.
1842 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1844 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
1846 * config/arc/arc.c (arc_print_operand): Output scalled address for
1847 sdata whenever is possible.
1848 (arc_in_small_data_p): Allow sdata for 64bit datum when double
1849 load/stores are available.
1850 (compact_sda_memory_operand): Check for the alignment required by
1851 code density instructions.
1852 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
1854 * config/arc/constraints.md (Usd): Update constraint.
1855 (Us0): New constraint.
1856 (Usc): Update constraint.
1858 2017-08-31 Richard Biener <rguenther@suse.de>
1861 * dwarf2out.c (dwarf2out_early_global_decl): Process each
1864 2017-08-31 Tamar Christina <tamar.christina@arm.com>
1866 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
1867 Resize type_signature.
1869 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1870 Alan Hayward <alan.hayward@arm.com>
1871 David Sherwood <david.sherwood@arm.com>
1873 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
1874 subregs whose inner modes can be stored in GPRs.
1875 (aarch64_classify_index): Likewise.
1877 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1878 Alan Hayward <alan.hayward@arm.com>
1879 David Sherwood <david.sherwood@arm.com>
1881 * config/aarch64/iterators.md (V_cmp_result): Rename to...
1882 (V_INT_EQUIV): ...this.
1883 (v_cmp_result): Rename to...
1884 (v_int_equiv): ...this.
1885 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
1886 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
1887 (copysign<mode>3): Likewise.
1888 (aarch64_simd_bsl<mode>_internal): Likewise.
1889 (aarch64_simd_bsl<mode>): Likewise.
1890 (vec_cmp<mode><mode>): Likewise.
1891 (vcond<mode><mode>): Likewise.
1892 (vcond<v_cmp_mixed><mode>): Likewise.
1893 (vcondu<mode><v_cmp_mixed>): Likewise.
1894 (aarch64_cm<optab><mode>): Likewise.
1895 (aarch64_cmtst<mode>): Likewise.
1896 (aarch64_fac<optab><mode>): Likewise.
1897 (vec_perm_const<mode>): Likewise.
1898 (vcond_mask_<mode><v_cmp_result>): Rename to...
1899 (vcond_mask_<mode><v_int_equiv>): ...this.
1900 (vec_cmp<mode><v_cmp_result>): Rename to...
1901 (vec_cmp<mode><v_int_equiv>): ...this.
1903 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1904 Alan Hayward <alan.hayward@arm.com>
1905 David Sherwood <david.sherwood@arm.com>
1907 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
1909 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
1910 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
1911 (UNSPEC_LD4_DREG): New unspecs.
1912 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
1913 (aarch64_ld2<mode>_dreg_be): Replace with...
1914 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
1916 (aarch64_ld3<mode>_dreg_le)
1917 (aarch64_ld3<mode>_dreg_be): Replace with...
1918 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
1920 (aarch64_ld4<mode>_dreg_le)
1921 (aarch64_ld4<mode>_dreg_be): Replace with...
1922 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
1925 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1927 PR tree-optimization/81987
1928 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
1929 insert an initializer in a location not dominated by the stride
1932 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1934 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
1935 on the entire header of the finally block in the fallthru case.
1937 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1939 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
1941 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
1943 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
1944 rs6000_emit_move_from_cr and call renamed function.
1945 (rs6000_emit_prologue): Call renamed functions.
1946 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
1947 movesi_from_cr, remove volatile CRs.
1949 2017-08-30 Jon Beniston <jon@beniston.com>
1950 Richard Biener <rguenther@suse.de>
1952 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
1953 of VECTOR_MODE_P check.
1954 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
1955 element vector types.
1957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1959 * df.h (df_read_modify_subreg_p): Remove in favor of...
1960 * rtl.h (read_modify_subreg_p): ...this new function. Take a
1961 const_rtx instead of an rtx.
1962 * cprop.c (local_cprop_find_used_regs): Update accordingly.
1963 * df-problems.c (df_word_lr_mark_ref): Likewise.
1964 * ira-lives.c (mark_pseudo_reg_live): Likewise.
1965 (mark_pseudo_reg_dead): Likewise.
1966 (mark_ref_dead): Likewise.
1967 * reginfo.c (init_subregs_of_mode): Likewise.
1968 * sched-deps.c (sched_analyze_1): Likewise.
1969 * df-scan.c (df_def_record_1): Likewise.
1970 (df_uses_record): Likewise.
1971 (df_read_modify_subreg_p): Remove in favor of...
1972 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
1973 const_rtx instead of an rtx.
1975 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1976 Alan Hayward <alan.hayward@arm.com>
1977 David Sherwood <david.sherwood@arm.com>
1979 * rtl.h (partial_subreg_p): New function.
1980 * caller-save.c (save_call_clobbered_regs): Use it.
1981 * calls.c (expand_call): Likewise.
1982 * combine.c (combinable_i3pat): Likewise.
1983 (simplify_set): Likewise.
1984 (make_extraction): Likewise.
1985 (make_compound_operation_int): Likewise.
1986 (gen_lowpart_or_truncate): Likewise.
1987 (force_to_mode): Likewise.
1988 (make_field_assignment): Likewise.
1989 (reg_truncated_to_mode): Likewise.
1990 (record_truncated_value): Likewise.
1991 (move_deaths): Likewise.
1992 * cse.c (record_jump_cond): Likewise.
1993 (cse_insn): Likewise.
1994 * cselib.c (cselib_lookup_1): Likewise.
1995 * expmed.c (extract_bit_field_using_extv): Likewise.
1996 * function.c (assign_parm_setup_reg): Likewise.
1997 * ifcvt.c (noce_convert_multiple_sets): Likewise.
1998 * ira-build.c (create_insn_allocnos): Likewise.
1999 * lra-coalesce.c (merge_pseudos): Likewise.
2000 * lra-constraints.c (match_reload): Likewise.
2001 (simplify_operand_subreg): Likewise.
2002 (curr_insn_transform): Likewise.
2003 * lra-lives.c (process_bb_lives): Likewise.
2004 * lra.c (new_insn_reg): Likewise.
2005 (lra_substitute_pseudo): Likewise.
2006 * regcprop.c (mode_change_ok): Likewise.
2007 (maybe_mode_change): Likewise.
2008 (copyprop_hardreg_forward_1): Likewise.
2009 * reload.c (push_reload): Likewise.
2010 (find_reloads): Likewise.
2011 (find_reloads_subreg_address): Likewise.
2012 * reload1.c (alter_reg): Likewise.
2013 (eliminate_regs_1): Likewise.
2014 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2016 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
2018 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
2019 back to if statements, including unpack.
2021 2017-08-30 Martin Liska <mliska@suse.cz>
2024 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
2026 (func_checker::compare_asm_inputs_outputs): ... this function.
2027 (func_checker::compare_gimple_asm): Use the function to compare
2028 also ASM constrains.
2029 * ipa-icf-gimple.h: Rename the function.
2031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2032 Alan Hayward <alan.hayward@arm.com>
2033 David Sherwood <david.sherwood@arm.com>
2035 * coretypes.h (complex_mode): New type.
2036 * gdbhooks.py (build_pretty_printer): Handle it.
2037 * machmode.h (complex_mode): New class.
2038 (complex_mode::includes_p): New function.
2039 (is_complex_int_mode): Likewise.
2040 (is_complex_float_mode): Likewise.
2041 * genmodes.c (get_mode_class): Handle complex mode classes.
2042 * function.c (expand_function_end): Use is_complex_int_mode.
2044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2045 Alan Hayward <alan.hayward@arm.com>
2046 David Sherwood <david.sherwood@arm.com>
2048 * coretypes.h (scalar_mode_pod): New typedef.
2049 * gdbhooks.py (build_pretty_printer): Handle it.
2050 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
2051 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
2052 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
2053 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
2056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2057 Alan Hayward <alan.hayward@arm.com>
2058 David Sherwood <david.sherwood@arm.com>
2060 * machmode.h (mode_for_vector): Take a scalar_mode instead
2062 * stor-layout.c (mode_for_vector): Likewise.
2063 * explow.c (promote_mode): Use as_a <scalar_mode>.
2064 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
2066 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2067 Alan Hayward <alan.hayward@arm.com>
2068 David Sherwood <david.sherwood@arm.com>
2070 * target.def (preferred_simd_mode): Take a scalar_mode
2071 instead of a machine_mode.
2072 * targhooks.h (default_preferred_simd_mode): Likewise.
2073 * targhooks.c (default_preferred_simd_mode): Likewise.
2074 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
2075 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
2076 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
2077 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
2078 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
2079 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
2080 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
2081 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
2083 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
2084 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
2085 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
2086 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
2087 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
2088 * doc/tm.texi: Regenerate.
2089 * optabs-query.c (can_vec_mask_load_store_p): Return false for
2092 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2093 Alan Hayward <alan.hayward@arm.com>
2094 David Sherwood <david.sherwood@arm.com>
2096 * target.def (scalar_mode_supported_p): Take a scalar_mode
2097 instead of a machine_mode.
2098 * targhooks.h (default_scalar_mode_supported_p): Likewise.
2099 * targhooks.c (default_scalar_mode_supported_p): Likewise.
2100 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
2101 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
2102 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
2103 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
2104 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
2105 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
2106 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
2107 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
2108 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
2109 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
2110 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
2111 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
2113 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
2114 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
2115 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
2116 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
2117 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
2119 * doc/tm.texi: Regenerate.
2121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2122 Alan Hayward <alan.hayward@arm.com>
2123 David Sherwood <david.sherwood@arm.com>
2125 * coretypes.h (opt_scalar_mode): New typedef.
2126 * gdbhooks.py (build_pretty_printers): Handle it.
2127 * machmode.h (mode_iterator::get_2xwider): Add overload for
2129 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
2131 * expr.c (convert_mode_scalar): Likewise.
2132 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2133 * optabs.c (expand_float): Likewise.
2134 (expand_fix): Likewise.
2135 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2137 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2138 Alan Hayward <alan.hayward@arm.com>
2139 David Sherwood <david.sherwood@arm.com>
2141 * optabs.c (expand_float): Explicitly check for scalars before
2142 using a branching expansion.
2143 (expand_fix): Likewise.
2145 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2146 Alan Hayward <alan.hayward@arm.com>
2147 David Sherwood <david.sherwood@arm.com>
2149 * expr.c (convert_mode): Split scalar handling out into...
2150 (convert_mode_scalar): ...this new function. Treat the modes
2153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2154 Alan Hayward <alan.hayward@arm.com>
2155 David Sherwood <david.sherwood@arm.com>
2157 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
2159 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
2161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2162 Alan Hayward <alan.hayward@arm.com>
2163 David Sherwood <david.sherwood@arm.com>
2165 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
2166 rather than a machine_mode.
2167 (fixed_from_string): Likewise.
2168 (fixed_convert): Likewise.
2169 (fixed_convert_from_int): Likewise.
2170 (fixed_convert_from_real): Likewise.
2171 (real_convert_from_fixed): Likewise.
2172 * fixed-value.c (fixed_from_double_int): Likewise.
2173 (fixed_from_string): Likewise.
2174 (fixed_convert): Likewise.
2175 (fixed_convert_from_int): Likewise.
2176 (fixed_convert_from_real): Likewise.
2177 (real_convert_from_fixed): Likewise.
2178 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
2180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2181 Alan Hayward <alan.hayward@arm.com>
2182 David Sherwood <david.sherwood@arm.com>
2184 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
2185 of separate mode class checks. Do not allow vector modes here.
2186 (immed_wide_int_const): Use as_a <scalar_mode>.
2187 * explow.c (trunc_int_for_mode): Likewise.
2188 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
2189 (wi::shwi): Likewise.
2190 (wi::min_value): Likewise.
2191 (wi::max_value): Likewise.
2192 * dwarf2out.c (loc_descriptor): Likewise.
2193 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
2196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2197 Alan Hayward <alan.hayward@arm.com>
2198 David Sherwood <david.sherwood@arm.com>
2200 * tree.h (SCALAR_TYPE_MODE): New macro.
2201 * expr.c (expand_expr_addr_expr_1): Use it.
2202 (expand_expr_real_2): Likewise.
2203 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
2204 (fold_convert_const_fixed_from_int): Likewise.
2205 (fold_convert_const_fixed_from_real): Likewise.
2206 (native_encode_fixed): Likewise
2207 (native_encode_complex): Likewise
2208 (native_encode_vector): Likewise.
2209 (native_interpret_fixed): Likewise.
2210 (native_interpret_real): Likewise.
2211 (native_interpret_complex): Likewise.
2212 (native_interpret_vector): Likewise.
2213 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
2214 (simd_clone_adjust_argument_types): Likewise.
2215 (simd_clone_init_simd_arrays): Likewise.
2216 (simd_clone_adjust): Likewise.
2217 * stor-layout.c (layout_type): Likewise.
2218 * tree.c (build_minus_one_cst): Likewise.
2219 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2220 * tree-inline.c (estimate_move_cost): Likewise.
2221 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
2222 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
2223 (vectorizable_reduction): Likewise.
2224 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
2225 (vect_recog_mixed_size_cond_pattern): Likewise.
2226 (check_bool_pattern): Likewise.
2227 (adjust_bool_pattern): Likewise.
2228 (search_type_for_mask_1): Likewise.
2229 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
2230 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2231 (vectorizable_load): Likewise.
2232 (vectorizable_store): Likewise.
2233 * ubsan.c (ubsan_encode_value): Likewise.
2234 * varasm.c (output_constant): Likewise.
2236 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2237 Alan Hayward <alan.hayward@arm.com>
2238 David Sherwood <david.sherwood@arm.com>
2240 * coretypes.h (scalar_mode): New class.
2241 * machmode.h (scalar_mode): Likewise.
2242 (scalar_mode::includes_p): New function.
2243 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
2244 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
2245 * genmodes.c (get_mode_class): Handle remaining scalar modes.
2246 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
2247 * expmed.c (store_bit_field_1): Likewise.
2248 (extract_bit_field_1): Likewise.
2249 * expr.c (write_complex_part): Likewise.
2250 (read_complex_part): Likewise.
2251 (emit_move_complex_push): Likewise.
2252 (expand_expr_real_2): Likewise.
2253 * function.c (assign_parm_setup_reg): Likewise.
2254 (assign_parms_unsplit_complex): Likewise.
2255 * optabs.c (expand_binop): Likewise.
2256 * rtlanal.c (subreg_get_info): Likewise.
2257 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2258 * varasm.c (output_constant_pool_2): Likewise.
2260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2261 Alan Hayward <alan.hayward@arm.com>
2262 David Sherwood <david.sherwood@arm.com>
2264 * expmed.c (extract_high_half): Use scalar_int_mode and remove
2266 (expmed_mult_highpart_optab): Likewise.
2267 (expmed_mult_highpart): Likewise.
2269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2270 Alan Hayward <alan.hayward@arm.com>
2271 David Sherwood <david.sherwood@arm.com>
2273 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
2274 instead of a machine_mode.
2275 (builtin_memset_read_str): Likewise.
2276 * builtins.c (c_readstr): Likewise.
2277 (builtin_memcpy_read_str): Likewise.
2278 (builtin_strncpy_read_str): Likewise.
2279 (builtin_memset_read_str): Likewise.
2280 (builtin_memset_gen_str): Likewise.
2281 (expand_builtin_signbit): Use scalar_int_mode for local variables.
2282 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
2283 instead of a machine_mode.
2284 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
2286 (make_extraction): Likewise.
2287 (try_widen_shift_mode): Take and return scalar_int_modes instead
2289 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
2290 a scalar_int_mode instead of a machine_mode.
2291 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
2292 (avr_addr_space_pointer_mode): Likewise.
2293 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
2294 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
2295 (msp430_unwind_word_mode): Likewise.
2296 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
2297 (spu_addr_space_pointer_mode): Likewise.
2298 (spu_addr_space_address_mode): Likewise.
2299 (spu_libgcc_cmp_return_mode): Likewise.
2300 (spu_libgcc_shift_count_mode): Likewise.
2301 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
2302 (rl78_addr_space_pointer_mode): Likewise.
2303 (fl78_unwind_word_mode): Likewise.
2304 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
2306 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
2307 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
2308 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
2309 (mips_valid_pointer_mode): Likewise.
2310 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
2311 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
2312 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
2314 (ft32_addr_space_address_mode): Likewise.
2315 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
2316 scalar_int_mode instead of a machine_mode.
2317 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
2319 (m32c_addr_space_address_mode): Likewise.
2320 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
2321 (rs6000_eh_return_filter_mode): Likewise.
2322 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
2323 (rs6000_eh_return_filter_mode): Likewise.
2324 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
2325 (s390_libgcc_shift_count_mode): Likewise.
2326 (s390_unwind_word_mode): Likewise.
2327 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
2329 * target.def (mode_rep_extended): Likewise.
2330 (valid_pointer_mode): Likewise.
2331 (addr_space.valid_pointer_mode): Likewise.
2332 (eh_return_filter_mode): Return a scalar_int_mode rather than
2334 (libgcc_cmp_return_mode): Likewise.
2335 (libgcc_shift_count_mode): Likewise.
2336 (unwind_word_mode): Likewise.
2337 (addr_space.pointer_mode): Likewise.
2338 (addr_space.address_mode): Likewise.
2339 * doc/tm.texi: Regenerate.
2340 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
2342 (do_jump): Use scalar_int_mode for local variables.
2343 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
2344 rather than a machine_mode.
2345 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
2346 (scompare_loc_descriptor_wide): Likewise.
2347 (scompare_loc_descriptor_narrow): Likewise.
2348 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
2350 * except.c (sjlj_emit_dispatch_table): Likewise.
2351 (expand_builtin_eh_copy_values): Likewise.
2352 * explow.c (convert_memory_address_addr_space_1): Likewise.
2353 Take a scalar_int_mode rather than a machine_mode.
2354 (convert_memory_address_addr_space): Take a scalar_int_mode rather
2355 than a machine_mode.
2356 (memory_address_addr_space): Use scalar_int_mode for local variables.
2357 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
2358 rather than a machine_mode.
2359 * expmed.c (mask_rtx): Likewise.
2360 (init_expmed_one_conv): Likewise.
2361 (expand_mult_highpart_adjust): Likewise.
2362 (extract_high_half): Likewise.
2363 (expmed_mult_highpart_optab): Likewise.
2364 (expmed_mult_highpart): Likewise.
2365 (expand_smod_pow2): Likewise.
2366 (expand_sdiv_pow2): Likewise.
2367 (emit_store_flag_int): Likewise.
2368 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
2370 (extract_low_bits): Likewise.
2371 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
2373 * expr.c (pieces_addr::adjust): Likewise.
2374 (can_store_by_pieces): Likewise.
2375 (store_by_pieces): Likewise.
2376 (clear_by_pieces_1): Likewise.
2377 (expand_expr_addr_expr_1): Likewise.
2378 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
2379 (expand_expr_real_1): Likewise.
2380 (try_casesi): Likewise.
2381 * final.c (shorten_branches): Likewise.
2382 * fold-const.c (fold_convert_const_int_from_fixed): Change the
2383 type of "mode" to machine_mode.
2384 * internal-fn.c (expand_arith_overflow_result_store): Take a
2385 scalar_int_mode rather than a machine_mode.
2386 (expand_mul_overflow): Use scalar_int_mode for local variables.
2387 * loop-doloop.c (doloop_modify): Likewise.
2388 (doloop_optimize): Likewise.
2389 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
2390 than a machine_mode.
2391 (expand_doubleword_shift_condmove): Likewise.
2392 (expand_doubleword_shift): Likewise.
2393 (expand_doubleword_clz): Likewise.
2394 (expand_doubleword_popcount): Likewise.
2395 (expand_doubleword_parity): Likewise.
2396 (expand_absneg_bit): Use scalar_int_mode for local variables.
2397 (prepare_float_lib_cmp): Likewise.
2398 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
2399 rather than a machine_mode.
2400 (convert_memory_address_addr_space): Likewise.
2401 (get_mode_bounds): Likewise.
2402 (get_address_mode): Return a scalar_int_mode rather than a
2404 * rtlanal.c (get_address_mode): Likewise.
2405 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
2406 than a machine_mode.
2407 * targhooks.c (default_mode_rep_extended): Likewise.
2408 (default_valid_pointer_mode): Likewise.
2409 (default_addr_space_valid_pointer_mode): Likewise.
2410 (default_eh_return_filter_mode): Return a scalar_int_mode rather
2411 than a machine_mode.
2412 (default_libgcc_cmp_return_mode): Likewise.
2413 (default_libgcc_shift_count_mode): Likewise.
2414 (default_unwind_word_mode): Likewise.
2415 (default_addr_space_pointer_mode): Likewise.
2416 (default_addr_space_address_mode): Likewise.
2417 * targhooks.h (default_eh_return_filter_mode): Likewise.
2418 (default_libgcc_cmp_return_mode): Likewise.
2419 (default_libgcc_shift_count_mode): Likewise.
2420 (default_unwind_word_mode): Likewise.
2421 (default_addr_space_pointer_mode): Likewise.
2422 (default_addr_space_address_mode): Likewise.
2423 (default_mode_rep_extended): Take a scalar_int_mode rather than
2425 (default_valid_pointer_mode): Likewise.
2426 (default_addr_space_valid_pointer_mode): Likewise.
2427 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
2429 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
2430 rather than a machine_mode.
2431 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
2432 for local variables.
2433 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2434 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
2435 than a machine_mode.
2437 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2438 Alan Hayward <alan.hayward@arm.com>
2439 David Sherwood <david.sherwood@arm.com>
2441 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
2442 the mode argument to scalar_int_mode.
2443 (do_jump_by_parts_zero_rtx): Likewise.
2444 (do_jump_by_parts_equality_rtx): Likewise.
2445 (do_jump_by_parts_greater): Take a mode argument.
2446 (do_jump_by_parts_equality): Likewise.
2447 (do_jump_1): Update calls accordingly.
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 * is-a.h (safe_dyn_cast): New function.
2454 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
2455 (jump_table_for_label): Likewise.
2456 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
2457 instead of an rtx_insn *.
2458 (shorten_branches): Use dyn_cast instead of LABEL_P and
2459 JUMP_TABLE_DATA_P. Use jump_table_for_label and
2460 rtx_jump_table_data::get_data_mode.
2461 (final_scan_insn): Likewise.
2463 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2464 Alan Hayward <alan.hayward@arm.com>
2465 David Sherwood <david.sherwood@arm.com>
2467 * combine.c (try_combine): Use is_a <scalar_int_mode> when
2468 trying to combine a full-register integer set with a subreg
2471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2472 Alan Hayward <alan.hayward@arm.com>
2473 David Sherwood <david.sherwood@arm.com>
2475 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
2476 that is always either address_mode or pointer_mode.
2478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2479 Alan Hayward <alan.hayward@arm.com>
2480 David Sherwood <david.sherwood@arm.com>
2482 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
2483 when the two are known to be equal.
2485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2486 Alan Hayward <alan.hayward@arm.com>
2487 David Sherwood <david.sherwood@arm.com>
2489 * simplify-rtx.c (simplify_const_unary_operation): Use
2490 is_a <scalar_int_mode> instead of checking for a nonzero
2491 precision. Forcibly convert op_mode to a scalar_int_mode
2492 in that case. More clearly differentiate the operand and
2493 result modes and use the former when deciding what the value
2494 of a count-bits operation should be. Use is_int_mode instead
2495 of checking for a MODE_INT. Remove redundant check for whether
2496 this mode has a zero precision.
2498 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2499 Alan Hayward <alan.hayward@arm.com>
2500 David Sherwood <david.sherwood@arm.com>
2502 * optabs.c (widen_leading): Change the type of the mode argument
2503 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
2504 (widen_bswap): Likewise.
2505 (expand_parity): Likewise.
2506 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
2507 (expand_ffs): Likewise.
2508 (epand_unop): Check for scalar integer modes before calling the
2511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2512 Alan Hayward <alan.hayward@arm.com>
2513 David Sherwood <david.sherwood@arm.com>
2515 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
2517 (expand_expr_real_1): Update call accordingly.
2519 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2520 Alan Hayward <alan.hayward@arm.com>
2521 David Sherwood <david.sherwood@arm.com>
2523 * expmed.c (store_bit_field_using_insv): Add op0_mode and
2524 value_mode arguments. Use scalar_int_mode internally.
2525 (store_bit_field_1): Rename the new integer mode from imode
2526 to op0_mode and use it instead of GET_MODE (op0). Update calls
2527 to store_split_bit_field, store_bit_field_using_insv and
2528 store_fixed_bit_field.
2529 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
2530 Use scalar_int_mode internally. Use a bit count rather than a mode
2531 when calculating the largest bit size for get_best_mode.
2532 Update calls to store_split_bit_field and store_fixed_bit_field_1.
2533 (store_fixed_bit_field_1): Add mode and value_mode arguments.
2534 Remove assertion that OP0 has a scalar integer mode.
2535 (store_split_bit_field): Add op0_mode and value_mode arguments.
2536 Update calls to extract_fixed_bit_field.
2537 (extract_bit_field_using_extv): Add an op0_mode argument.
2538 Use scalar_int_mode internally.
2539 (extract_bit_field_1): Rename the new integer mode from imode to
2540 op0_mode and use it instead of GET_MODE (op0). Update calls to
2541 extract_split_bit_field, extract_bit_field_using_extv and
2542 extract_fixed_bit_field.
2543 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
2544 to extract_split_bit_field and extract_fixed_bit_field_1.
2545 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
2546 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
2548 (extract_split_bit_field): Add an op0_mode argument. Update call
2549 to extract_fixed_bit_field.
2551 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2552 Alan Hayward <alan.hayward@arm.com>
2553 David Sherwood <david.sherwood@arm.com>
2555 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
2556 * explow.c (hard_function_value): Likewise.
2557 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
2558 convert_to_mode call outside the loop.
2559 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
2560 for the mode iterator. Require the mode specified by max_pieces
2562 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
2564 (copy_blkmode_to_reg): Likewise.
2565 (set_storage_via_setmem): Likewise.
2566 * optabs.c (prepare_cmp_insn): Likewise.
2567 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2568 * stor-layout.c (finish_bitfield_representative): Likewise.
2570 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2571 Alan Hayward <alan.hayward@arm.com>
2572 David Sherwood <david.sherwood@arm.com>
2574 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
2575 * expr.c (convert_move): Use them.
2576 (convert_modes): Likewise.
2577 (store_expr_with_bounds): Likewise.
2579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2580 Alan Hayward <alan.hayward@arm.com>
2581 David Sherwood <david.sherwood@arm.com>
2583 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
2584 parameter for the mode of "x". Remove the "known_x", "known_mode"
2585 and "known_ret" arguments. Change the type of the mode argument
2587 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
2588 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
2589 (reg_num_sign_bit_copies_for_combine): Likewise.
2590 * rtlanal.c (nonzero_bits1): Likewise.
2591 (num_sign_bit_copies1): Likewise.
2592 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
2593 (reg_num_sign_bit_copies_general): Likewise.
2594 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
2595 (reg_nonzero_bits_general): Likewise.
2597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2598 Alan Hayward <alan.hayward@arm.com>
2599 David Sherwood <david.sherwood@arm.com>
2601 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
2602 than in subroutines. Return 1 for non-integer modes.
2603 (cached_num_sign_bit_copies): Change the type of the mode parameter
2605 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
2606 classes. Handle CONST_INT_P first and then check whether X also
2607 has a scalar integer mode. Check the same thing for inner registers
2608 of a SUBREG and for values that are being extended or truncated.
2610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2611 Alan Hayward <alan.hayward@arm.com>
2612 David Sherwood <david.sherwood@arm.com>
2614 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
2615 in subroutines. Return the mode mask for non-integer modes.
2616 (cached_nonzero_bits): Change the type of the mode parameter
2618 (nonzero_bits1): Likewise. Remove early exit for other mode
2619 classes. Handle CONST_INT_P first and then check whether X
2620 also has a scalar integer mode.
2622 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2623 Alan Hayward <alan.hayward@arm.com>
2624 David Sherwood <david.sherwood@arm.com>
2626 * expr.c (widest_int_mode_for_size): Make the comment match the code.
2627 Return a scalar_int_mode and assert that the size is greater than
2629 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
2630 (op_by_pieces_d::op_by_pieces_d): Likewise.
2631 (op_by_pieces_d::run): Likewise.
2632 (can_store_by_pieces): Likewise.
2634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2635 Alan Hayward <alan.hayward@arm.com>
2636 David Sherwood <david.sherwood@arm.com>
2638 * combine.c (extract_left_shift): Add a mode argument and update
2640 (make_compound_operation_int): Change the type of the mode parameter
2641 to scalar_int_mode and update the call to extract_left_shift.
2643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2644 Alan Hayward <alan.hayward@arm.com>
2645 David Sherwood <david.sherwood@arm.com>
2647 * combine.c (simplify_and_const_int): Change the type of the mode
2648 parameter to scalar_int_mode.
2649 (simplify_and_const_int_1): Likewise. Update recursive call.
2651 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2652 Alan Hayward <alan.hayward@arm.com>
2653 David Sherwood <david.sherwood@arm.com>
2655 * combine.c (simplify_compare_const): Check that the mode is a
2656 scalar_int_mode (rather than VOIDmode) before testing its
2658 (simplify_comparison): Move COMPARISON_P handling out of the
2659 loop and restrict the latter part of the loop to scalar_int_modes.
2660 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
2661 and when considering SUBREG_REGs. Use is_int_mode instead of
2662 checking GET_MODE_CLASS against MODE_INT.
2664 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2665 Alan Hayward <alan.hayward@arm.com>
2666 David Sherwood <david.sherwood@arm.com>
2668 * combine.c (try_widen_shift_mode): Move check for equal modes to...
2669 (simplify_shift_const_1): ...here. Use scalar_int_mode for
2670 shift_unit_mode and for modes involved in scalar shifts.
2672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2673 Alan Hayward <alan.hayward@arm.com>
2674 David Sherwood <david.sherwood@arm.com>
2676 * combine.c (force_int_to_mode): New function, split out from...
2677 (force_to_mode): ...here. Keep xmode up-to-date and use it
2678 instead of GET_MODE (x).
2680 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2681 Alan Hayward <alan.hayward@arm.com>
2682 David Sherwood <david.sherwood@arm.com>
2684 * optabs-query.h (extraction_insn::struct_mode): Change type to
2685 opt_scalar_int_mode and update comment.
2686 (extraction_insn::field_mode): Change type to scalar_int_mode.
2687 (extraction_insn::pos_mode): Likewise.
2688 * combine.c (make_extraction): Update accordingly.
2689 * optabs-query.c (get_traditional_extraction_insn): Likewise.
2690 (get_optab_extraction_insn): Likewise.
2691 * recog.c (simplify_while_replacing): Likewise.
2692 * expmed.c (narrow_bit_field_mem): Change the type of the mode
2693 parameter to opt_scalar_int_mode.
2695 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2696 Alan Hayward <alan.hayward@arm.com>
2697 David Sherwood <david.sherwood@arm.com>
2699 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
2700 to a scalar_int_mode instead of a machine_mode.
2701 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
2702 (get_best_mode): Return a boolean and use a pointer argument to store
2703 the selected mode. Replace the limit mode parameter with a bit limit.
2704 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
2705 for the values returned by bit_field_mode_iterator::next_mode.
2706 (store_bit_field): Update call to get_best_mode.
2707 (store_fixed_bit_field): Likewise.
2708 (extract_fixed_bit_field): Likewise.
2709 * expr.c (optimize_bitfield_assignment_op): Likewise.
2710 * fold-const.c (optimize_bit_field_compare): Likewise.
2711 (fold_truth_andor_1): Likewise.
2712 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
2713 Update for new type of m_mode.
2714 (get_best_mode): As above.
2716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2717 Alan Hayward <alan.hayward@arm.com>
2718 David Sherwood <david.sherwood@arm.com>
2720 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
2721 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
2722 (store_bit_field): Check is_a <scalar_int_mode> before calling
2723 strict_volatile_bitfield_p.
2724 (extract_bit_field): Likewise.
2726 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2727 Alan Hayward <alan.hayward@arm.com>
2728 David Sherwood <david.sherwood@arm.com>
2730 * target.def (cstore_mode): Return a scalar_int_mode.
2731 * doc/tm.texi: Regenerate.
2732 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
2733 * targhooks.h (default_cstore_mode): Likewise.
2734 * targhooks.c (default_cstore_mode): Likewise, using a forced
2736 * expmed.c (emit_cstore): Expect the target of the cstore to be
2739 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2740 Alan Hayward <alan.hayward@arm.com>
2741 David Sherwood <david.sherwood@arm.com>
2743 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
2745 (niter_desc): Likewise mode.
2746 (iv_analyze): Add a mode parameter.
2748 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
2749 and change its type to scalar_int_mode.
2750 * loop-iv.c: Update commentary at head of file.
2751 (iv_constant): Pass the mode paraeter before the rtx it describes
2752 and change its type to scalar_int_mode. Remove VOIDmode handling.
2753 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
2754 (iv_extend): Likewise.
2755 (shorten_into_mode): Likewise.
2756 (iv_add): Use scalar_int_mode.
2757 (iv_mult): Likewise.
2758 (iv_shift): Likewise.
2759 (canonicalize_iv_subregs): Likewise.
2760 (get_biv_step_1): Pass the outer_mode parameter before the rtx
2761 it describes and change its mode to scalar_int_mode. Also change
2762 the type of the returned inner_mode to scalar_int_mode.
2763 (get_biv_step): Likewise, turning outer_mode from a pointer
2764 into a direct parameter. Update call to get_biv_step_1.
2765 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
2766 iv_constant and get_biv_step.
2767 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
2768 and change its type to scalar_int_mode. Don't initialise iv->mode
2769 to VOIDmode and remove later checks for its still being VOIDmode.
2770 Update calls to iv_analyze_op and iv_analyze_expr. Check
2771 is_a <scalar_int_mode> when changing the mode under consideration.
2772 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
2773 Update call to iv_analyze_expr.
2774 (iv_analyze_op): Add a mode parameter. Reject subregs whose
2775 inner register is not also a scalar_int_mode. Update call to
2777 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
2778 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
2779 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
2780 separate mode class checks. Update calls to iv_analyze. Remove
2781 fix-up of VOIDmodes after iv_analyze_biv.
2782 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
2783 don't have a scalar_int_mode. Update call to biv_p.
2785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2786 Alan Hayward <alan.hayward@arm.com>
2787 David Sherwood <david.sherwood@arm.com>
2789 * cfgexpand.c (convert_debug_memory_address): Use
2790 as_a <scalar_int_mode>.
2791 * combine.c (expand_compound_operation): Likewise.
2792 (make_extraction): Likewise.
2793 (change_zero_ext): Likewise.
2794 (simplify_comparison): Likewise.
2795 * cse.c (cse_insn): Likewise.
2796 * dwarf2out.c (minmax_loc_descriptor): Likewise.
2797 (mem_loc_descriptor): Likewise.
2798 (loc_descriptor): Likewise.
2799 * expmed.c (init_expmed_one_mode): Likewise.
2800 (synth_mult): Likewise.
2801 (emit_store_flag_1): Likewise.
2802 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
2803 of a comparison with size.
2804 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
2805 (reduce_to_bit_field_precision): Likewise.
2806 * function.c (expand_function_end): Likewise.
2807 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
2808 * loop-doloop.c (doloop_modify): Likewise.
2809 * optabs.c (expand_binop): Likewise.
2810 (expand_unop): Likewise.
2811 (expand_copysign_absneg): Likewise.
2812 (prepare_cmp_insn): Likewise.
2813 (maybe_legitimize_operand): Likewise.
2814 * recog.c (const_scalar_int_operand): Likewise.
2815 * rtlanal.c (get_address_mode): Likewise.
2816 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2817 (simplify_cond_clz_ctz): Likewise.
2818 * tree-nested.c (get_nl_goto_field): Likewise.
2819 * tree.c (build_vector_type_for_mode): Likewise.
2820 * var-tracking.c (use_narrower_mode): Likewise.
2822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2823 Alan Hayward <alan.hayward@arm.com>
2824 David Sherwood <david.sherwood@arm.com>
2826 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
2827 * builtins.c (expand_builtin_signbit): Use it.
2828 * cfgexpand.c (expand_debug_expr): Likewise.
2829 * dojump.c (do_jump): Likewise.
2830 (do_compare_and_jump): Likewise.
2831 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
2832 * expmed.c (make_tree): Likewise.
2833 * expr.c (expand_expr_real_2): Likewise.
2834 (expand_expr_real_1): Likewise.
2835 (try_casesi): Likewise.
2836 * fold-const-call.c (fold_const_call_ss): Likewise.
2837 * fold-const.c (unextend): Likewise.
2838 (extract_muldiv_1): Likewise.
2839 (fold_single_bit_test): Likewise.
2840 (native_encode_int): Likewise.
2841 (native_encode_string): Likewise.
2842 (native_interpret_int): Likewise.
2843 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
2844 * internal-fn.c (expand_addsub_overflow): Likewise.
2845 (expand_neg_overflow): Likewise.
2846 (expand_mul_overflow): Likewise.
2847 (expand_arith_overflow): Likewise.
2848 * match.pd: Likewise.
2849 * stor-layout.c (layout_type): Likewise.
2850 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2851 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
2852 * tree-ssanames.c (get_range_info): Likewise.
2853 * tree-switch-conversion.c (array_value_type) Likewise.
2854 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
2855 (vect_recog_divmod_pattern): Likewise.
2856 (vect_recog_mixed_size_cond_pattern): Likewise.
2857 * tree-vrp.c (extract_range_basic): Likewise.
2858 (simplify_float_conversion_using_ranges): Likewise.
2859 * tree.c (int_fits_type_p): Likewise.
2860 * ubsan.c (instrument_bool_enum_load): Likewise.
2861 * varasm.c (mergeable_string_section): Likewise.
2862 (narrowing_initializer_constant_valid_p): Likewise.
2863 (output_constant): Likewise.
2865 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2866 Alan Hayward <alan.hayward@arm.com>
2867 David Sherwood <david.sherwood@arm.com>
2869 * machmode.h (NARROWEST_INT_MODE): New macro.
2870 * expr.c (alignment_for_piecewise_move): Use it instead of
2871 GET_CLASS_NARROWEST_MODE (MODE_INT).
2872 (push_block): Likewise.
2873 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2875 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2878 Alan Hayward <alan.hayward@arm.com>
2879 David Sherwood <david.sherwood@arm.com>
2881 * postreload.c (move2add_valid_value_p): Change the type of the
2882 mode parameter to scalar_int_mode.
2883 (move2add_use_add2_insn): Add a mode parameter and use it instead
2885 (move2add_use_add3_insn): Likewise.
2886 (reload_cse_move2add): Update accordingly.
2888 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2889 Alan Hayward <alan.hayward@arm.com>
2890 David Sherwood <david.sherwood@arm.com>
2892 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
2894 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
2895 * optabs.c (expand_unop): Likewise.
2897 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2898 Alan Hayward <alan.hayward@arm.com>
2899 David Sherwood <david.sherwood@arm.com>
2901 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
2902 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
2903 (popcount_loc_descriptor): Likewise.
2904 (bswap_loc_descriptor): Likewise.
2905 (rotate_loc_descriptor): Likewise.
2906 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
2907 calling the functions above.
2909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2910 Alan Hayward <alan.hayward@arm.com>
2911 David Sherwood <david.sherwood@arm.com>
2913 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
2915 (try_combine): Likewise.
2916 (simplify_if_then_else): Likewise.
2917 * cse.c (cse_insn): Likewise.
2918 * dwarf2out.c (mem_loc_descriptor): Likewise.
2919 * emit-rtl.c (gen_lowpart_common): Likewise.
2920 * simplify-rtx.c (simplify_truncation): Likewise.
2921 (simplify_binary_operation_1): Likewise.
2922 (simplify_const_relational_operation): Likewise.
2923 (simplify_ternary_operation): Likewise.
2924 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
2926 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2927 Alan Hayward <alan.hayward@arm.com>
2928 David Sherwood <david.sherwood@arm.com>
2930 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
2931 * reload.c (push_reload): Likewise.
2932 (find_reloads): Likewise.
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 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
2939 (make_compound_operation_int): Likewise.
2940 (change_zero_ext): Likewise.
2941 * expr.c (convert_move): Likewise.
2942 (convert_modes): Likewise.
2943 * fwprop.c (forward_propagate_subreg): Likewise.
2944 * loop-iv.c (get_biv_step_1): Likewise.
2945 * optabs.c (widen_operand): Likewise.
2946 * postreload.c (move2add_valid_value_p): Likewise.
2947 * recog.c (simplify_while_replacing): Likewise.
2948 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2949 (simplify_binary_operation_1): Likewise. Remove redundant
2950 mode equality check.
2952 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2953 Alan Hayward <alan.hayward@arm.com>
2954 David Sherwood <david.sherwood@arm.com>
2956 * combine.c (combine_simplify_rtx): Add checks for
2957 is_a <scalar_int_mode>.
2958 (simplify_if_then_else): Likewise.
2959 (make_field_assignment): Likewise.
2960 (simplify_comparison): Likewise.
2961 * ifcvt.c (noce_try_bitop): Likewise.
2962 * loop-invariant.c (canonicalize_address_mult): Likewise.
2963 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2966 Alan Hayward <alan.hayward@arm.com>
2967 David Sherwood <david.sherwood@arm.com>
2969 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
2970 is_a <scalar_int_mode> instead of != BLKmode.
2972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2973 Alan Hayward <alan.hayward@arm.com>
2974 David Sherwood <david.sherwood@arm.com>
2976 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
2977 instead of != VOIDmode.
2978 * combine.c (if_then_else_cond): Likewise.
2979 (change_zero_ext): Likewise.
2980 * dwarf2out.c (mem_loc_descriptor): Likewise.
2981 (loc_descriptor): Likewise.
2982 * rtlanal.c (canonicalize_condition): Likewise.
2983 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
2985 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2986 Alan Hayward <alan.hayward@arm.com>
2987 David Sherwood <david.sherwood@arm.com>
2989 * simplify-rtx.c (simplify_binary_operation_1): Use
2990 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
2992 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2993 Alan Hayward <alan.hayward@arm.com>
2994 David Sherwood <david.sherwood@arm.com>
2996 * wide-int.h (int_traits<unsigned char>) New class.
2997 (int_traits<unsigned short>) Likewise.
2998 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
2999 Use GET_MODE_UNIT_PRECISION and remove redundant test for
3001 * combine.c (set_nonzero_bits_and_sign_copies): Use
3002 is_a <scalar_int_mode>.
3003 (find_split_point): Likewise.
3004 (combine_simplify_rtx): Likewise.
3005 (simplify_logical): Likewise.
3006 (expand_compound_operation): Likewise.
3007 (expand_field_assignment): Likewise.
3008 (make_compound_operation): Likewise.
3009 (extended_count): Likewise.
3010 (change_zero_ext): Likewise.
3011 (simplify_comparison): Likewise.
3012 * dwarf2out.c (scompare_loc_descriptor): Likewise.
3013 (ucompare_loc_descriptor): Likewise.
3014 (minmax_loc_descriptor): Likewise.
3015 (mem_loc_descriptor): Likewise.
3016 (loc_descriptor): Likewise.
3017 * expmed.c (init_expmed_one_mode): Likewise.
3018 * lra-constraints.c (lra_constraint_offset): Likewise.
3019 * optabs.c (prepare_libcall_arg): Likewise.
3020 * postreload.c (move2add_note_store): Likewise.
3021 * reload.c (operands_match_p): Likewise.
3022 * rtl.h (load_extend_op): Likewise.
3023 * rtlhooks.c (gen_lowpart_general): Likewise.
3024 * simplify-rtx.c (simplify_truncation): Likewise.
3025 (simplify_unary_operation_1): Likewise.
3026 (simplify_binary_operation_1): Likewise.
3027 (simplify_const_binary_operation): Likewise.
3028 (simplify_const_relational_operation): Likewise.
3029 (simplify_subreg): Likewise.
3030 * stor-layout.c (bitwise_mode_for_mode): Likewise.
3031 * var-tracking.c (adjust_mems): Likewise.
3032 (prepare_call_arguments): Likewise.
3034 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3035 Alan Hayward <alan.hayward@arm.com>
3036 David Sherwood <david.sherwood@arm.com>
3038 * machmode.h (is_int_mode): New fuction.
3039 * combine.c (find_split_point): Use it.
3040 (combine_simplify_rtx): Likewise.
3041 (simplify_if_then_else): Likewise.
3042 (simplify_set): Likewise.
3043 (simplify_shift_const_1): Likewise.
3044 (simplify_comparison): Likewise.
3045 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
3046 * cse.c (notreg_cost): Likewise.
3047 (cse_insn): Likewise.
3048 * cselib.c (cselib_lookup_1): Likewise.
3049 * dojump.c (do_jump_1): Likewise.
3050 (do_compare_rtx_and_jump): Likewise.
3051 * dse.c (get_call_args): Likewise.
3052 * dwarf2out.c (rtl_for_decl_init): Likewise.
3053 (native_encode_initializer): Likewise.
3054 * expmed.c (emit_store_flag_1): Likewise.
3055 (emit_store_flag): Likewise.
3056 * expr.c (convert_modes): Likewise.
3057 (store_field): Likewise.
3058 (expand_expr_real_1): Likewise.
3059 * fold-const.c (fold_read_from_constant_string): Likewise.
3060 * gimple-ssa-sprintf.c (get_format_string): Likewise.
3061 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
3062 * optabs.c (expand_binop): Likewise.
3063 (expand_unop): Likewise.
3064 (expand_abs_nojump): Likewise.
3065 (expand_one_cmpl_abs_nojump): Likewise.
3066 * simplify-rtx.c (mode_signbit_p): Likewise.
3067 (val_signbit_p): Likewise.
3068 (val_signbit_known_set_p): Likewise.
3069 (val_signbit_known_clear_p): Likewise.
3070 (simplify_relational_operation_1): Likewise.
3071 * tree.c (vector_type_mode): Likewise.
3073 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3074 Alan Hayward <alan.hayward@arm.com>
3075 David Sherwood <david.sherwood@arm.com>
3077 * machmode.h (smallest_mode_for_size): Fix formatting.
3078 (smallest_int_mode_for_size): New function.
3079 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
3080 instead of smallest_mode_for_size.
3081 * combine.c (make_extraction): Likewise.
3082 * config/arc/arc.c (arc_expand_movmem): Likewise.
3083 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
3084 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
3085 * config/s390/s390.c (s390_expand_insv): Likewise.
3086 * config/sparc/sparc.c (assign_int_registers): Likewise.
3087 * config/spu/spu.c (spu_function_value): Likewise.
3088 (spu_function_arg): Likewise.
3089 * coverage.c (get_gcov_type): Likewise.
3090 (get_gcov_unsigned_t): Likewise.
3091 * dse.c (find_shift_sequence): Likewise.
3092 * expmed.c (store_bit_field_1): Likewise.
3093 * expr.c (convert_move): Likewise.
3094 (store_field): Likewise.
3095 * internal-fn.c (expand_arith_overflow): Likewise.
3096 * optabs-query.c (get_best_extraction_insn): Likewise.
3097 * optabs.c (expand_twoval_binop_libfunc): Likewise.
3098 * stor-layout.c (layout_type): Likewise.
3099 (initialize_sizetypes): Likewise.
3100 * targhooks.c (default_get_mask_mode): Likewise.
3101 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
3103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3104 Alan Hayward <alan.hayward@arm.com>
3105 David Sherwood <david.sherwood@arm.com>
3107 * machmode.h (opt_mode::else_blk): New function.
3108 (int_mode_for_mode): Declare.
3109 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
3110 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
3112 * cfgexpand.c (expand_debug_expr): Likewise.
3113 * combine.c (gen_lowpart_or_truncate): Likewise.
3114 (gen_lowpart_for_combine): Likewise.
3115 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
3116 * config/avr/avr.c (avr_to_int_mode): Likewise.
3117 (avr_out_plus_1): Likewise.
3118 (avr_out_plus): Likewise.
3119 (avr_out_round): Likewise.
3120 * config/i386/i386.c (ix86_split_to_parts): Likewise.
3121 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
3122 (s390_expand_vcond): Likewise.
3123 * config/spu/spu.c (spu_split_immediate): Likewise.
3124 (spu_expand_mov): Likewise.
3125 * dse.c (get_stored_val): Likewise.
3126 * expmed.c (store_bit_field_1): Likewise.
3127 (convert_extracted_bit_field): Use int_mode_for_mode instead of
3129 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
3130 (extract_low_bits): Likewise.
3131 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
3132 handling rather than repeating the check.
3133 (emit_group_store): Likewise.
3134 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
3135 * optabs.c (expand_absneg_bit): Likewise.
3136 (expand_copysign_absneg): Likewise.
3137 (expand_copysign_bit): Likewise.
3138 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
3139 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
3140 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
3141 * var-tracking.c (prepare_call_arguments): Likewise.
3142 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
3143 int_mode_for_mode instead of mode_for_size.
3144 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
3146 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3147 Alan Hayward <alan.hayward@arm.com>
3148 David Sherwood <david.sherwood@arm.com>
3150 * machmode.h (int_mode_for_size): New function.
3151 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
3152 instead of mode_for_size.
3153 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
3155 * combine.c (expand_field_assignment): Use int_mode_for_size
3156 instead of mode_for_size.
3157 (make_extraction): Likewise.
3158 (simplify_shift_const_1): Likewise.
3159 (simplify_comparison): Likewise.
3160 * dojump.c (do_jump): Likewise.
3161 * dwarf2out.c (mem_loc_descriptor): Likewise.
3162 * emit-rtl.c (init_derived_machine_modes): Likewise.
3163 * expmed.c (flip_storage_order): Likewise.
3164 (convert_extracted_bit_field): Likewise.
3165 * expr.c (copy_blkmode_from_reg): Likewise.
3166 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
3167 * internal-fn.c (expand_mul_overflow): Likewise.
3168 * lower-subreg.c (simple_move): Likewise.
3169 * optabs-libfuncs.c (init_optabs): Likewise.
3170 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3171 * tree.c (vector_type_mode): Likewise.
3172 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
3173 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
3174 * tree-vect-generic.c (expand_vector_parallel): Likewise.
3175 * tree-vect-stmts.c (vectorizable_load): Likewise.
3176 (vectorizable_store): Likewise.
3178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3179 Alan Hayward <alan.hayward@arm.com>
3180 David Sherwood <david.sherwood@arm.com>
3182 * coretypes.h (pod_mode): New type.
3183 (scalar_int_mode_pod): New typedef.
3184 * machmode.h (pod_mode): New class.
3185 (int_n_data_t::m): Change type to scalar_int_mode_pod.
3186 * genmodes.c (emit_mode_int_n): Update accordingly.
3187 * lower-subreg.h (target_lower_subreg): Change type to
3188 scalar_int_mode_pod.
3189 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
3190 scalar_int_mode_pod.
3192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3193 Alan Hayward <alan.hayward@arm.com>
3194 David Sherwood <david.sherwood@arm.com>
3196 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
3197 machine_mode to scalar_int_mode.
3198 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
3199 (rs6000_option_override_internal): Remove cast to int.
3200 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
3201 machine_mode to scalar_int_mode.
3202 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
3203 (rs6000_option_override_internal): Remove cast to int.
3204 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
3205 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
3207 * config/s390/s390.c (s390_expand_builtin): Likewise.
3208 * coretypes.h (scalar_int_mode): New type.
3209 (opt_scalar_int_mode): New typedef.
3210 * machmode.h (scalar_int_mode): New class.
3211 (scalar_int_mode::includes_p): New function.
3212 (byte_mode): Change type to scalar_int_mode.
3213 (word_mode): Likewise.
3214 (ptr_mode): Likewise.
3215 * emit-rtl.c (byte_mode): Likewise.
3216 (word_mode): Likewise.
3217 (ptr_mode): Likewise.
3218 (init_derived_machine_modes): Update accordingly.
3219 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
3220 and MODE_PARTIAL_INT.
3221 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
3222 opt_scalar_int_mode.
3224 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3225 Alan Hayward <alan.hayward@arm.com>
3226 David Sherwood <david.sherwood@arm.com>
3228 * target.def (libgcc_floating_mode_supported_p): Take a
3230 * doc/tm.texi: Regenerate.
3231 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
3233 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3234 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
3237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3238 Alan Hayward <alan.hayward@arm.com>
3239 David Sherwood <david.sherwood@arm.com>
3241 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
3242 * doc/tm.texi: Regenerate.
3243 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
3244 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
3245 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
3246 * targhooks.h (default_floatn_mode): Likewise.
3247 * targhooks.c (default_floatn_mode): Likewise.
3248 * tree.c (build_common_tree_nodes): Update accordingly.
3250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3251 Alan Hayward <alan.hayward@arm.com>
3252 David Sherwood <david.sherwood@arm.com>
3254 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
3255 (mode_iterator::iterate_p): Likewise.
3256 (mode_iterator::get_wider): Likewise.
3257 * expr.c (init_expr_target): Use opt_scalar_float_mode.
3259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3260 Alan Hayward <alan.hayward@arm.com>
3261 David Sherwood <david.sherwood@arm.com>
3263 * coretypes.h (opt_scalar_float_mode): New typedef.
3264 * machmode.h (float_mode_for_size): New function.
3265 * emit-rtl.c (double_mode): Delete.
3266 (init_emit_once): Use float_mode_for_size.
3267 * stor-layout.c (layout_type): Likewise.
3268 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
3270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3271 Alan Hayward <alan.hayward@arm.com>
3272 David Sherwood <david.sherwood@arm.com>
3274 * output.h (assemble_real): Take a scalar_float_mode.
3275 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
3276 * config/arm/arm.md (consttable_4): Likewise.
3277 (consttable_8): Likewise.
3278 (consttable_16): Likewise.
3279 * config/mips/mips.md (consttable_float): Likewise.
3280 * config/s390/s390.c (s390_output_pool_entry): Likewise.
3281 * varasm.c (assemble_real): Take a scalar_float_mode.
3282 (output_constant_pool_2): Update accordingly.
3283 (output_constant): Likewise.
3285 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3286 Alan Hayward <alan.hayward@arm.com>
3287 David Sherwood <david.sherwood@arm.com>
3289 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
3290 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
3291 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
3292 (native_encode_real): Likewise.
3293 (native_interpret_real): Likewise.
3294 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
3295 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3297 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3298 Alan Hayward <alan.hayward@arm.com>
3299 David Sherwood <david.sherwood@arm.com>
3301 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
3302 <scalar_float_mode>. Simplify.
3303 (gen_extend_conv_libfunc): Likewise.
3305 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3306 Alan Hayward <alan.hayward@arm.com>
3307 David Sherwood <david.sherwood@arm.com>
3309 * coretypes.h (scalar_float_mode): New type.
3310 * machmode.h (mode_traits::from_int): Use machine_mode if
3311 USE_ENUM_MODES is defined.
3312 (is_a): New function.
3314 (dyn_cast): Likewise.
3315 (scalar_float_mode): New class.
3316 (scalar_float_mode::includes_p): New function.
3317 (is_float_mode): Likewise.
3318 * gdbhooks.py (MachineModePrinter): New class.
3319 (build_pretty_printer): Use it for scalar_float_mode.
3320 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
3321 (format_helper::format_helper): Turn into a template.
3322 * genmodes.c (get_mode_class): New function.
3323 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
3324 or machine_mode if none.
3325 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
3326 as_a <scalar_float_mode>.
3327 * dwarf2out.c (mem_loc_descriptor): Likewise.
3328 (insert_float): Likewise.
3329 (add_const_value_attribute): Likewise.
3330 * simplify-rtx.c (simplify_immed_subreg): Likewise.
3331 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
3332 (expand_unop): Update accordingly.
3333 (expand_abs_nojump): Likewise.
3334 (expand_copysign_absneg): Take a scalar_float_mode.
3335 (expand_copysign_bit): Likewise.
3336 (expand_copysign): Update accordingly.
3338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3339 Alan Hayward <alan.hayward@arm.com>
3340 David Sherwood <david.sherwood@arm.com>
3342 * coretypes.h (opt_mode): New class.
3343 * machmode.h (opt_mode): Likewise.
3344 (opt_mode::else_void): New function.
3345 (opt_mode::require): Likewise.
3346 (opt_mode::exists): Likewise.
3347 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
3348 (GET_MODE_2XWIDER_MODE): Likewise.
3349 (mode_iterator::get_wider): Update accordingly.
3350 (mode_iterator::get_2xwider): Likewise.
3351 (mode_iterator::get_known_wider): Likewise, turning into a template.
3352 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
3353 forcing a wider mode to exist.
3354 * config/cr16/cr16.h (LONG_REG_P): Likewise.
3355 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3356 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
3357 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
3358 * lower-subreg.c (init_lower_subreg): Likewise.
3359 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
3360 on the final iteration.
3361 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
3362 a wider mode exists before asking for a move pattern.
3363 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
3364 forcing a wider mode to exist.
3365 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
3366 returning false if no such mode exists.
3367 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
3368 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
3369 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
3370 Avoid checking for a MODE_INT if we already know the mode is not a
3372 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
3373 forcing a wider mode to exist.
3374 (expmed_mult_highpart_optab): Likewise.
3375 (expmed_mult_highpart): Likewise.
3376 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
3378 * lto-streamer-in.c (lto_input_mode_table): Likewise.
3379 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
3380 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3381 * internal-fn.c (expand_mul_overflow): Update use of
3382 GET_MODE_2XWIDER_MODE.
3383 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3384 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
3385 GET_MODE_WIDER_MODE.
3386 (convert_plusminus_to_widen): Likewise.
3387 * tree-switch-conversion.c (array_value_type): Likewise.
3388 * var-tracking.c (emit_note_insn_var_location): Likewise.
3389 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
3390 Return false inside rather than outside the loop if no wider mode
3392 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
3393 and GET_MODE_2XWIDER_MODE
3394 (can_compare_p): Use else_void.
3395 * gdbhooks.py (OptMachineModePrinter): New class.
3396 (build_pretty_printer): Use it for opt_mode.
3398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3399 Alan Hayward <alan.hayward@arm.com>
3400 David Sherwood <david.sherwood@arm.com>
3402 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
3403 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
3405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3406 Alan Hayward <alan.hayward@arm.com>
3407 David Sherwood <david.sherwood@arm.com>
3409 * machmode.h (mode_traits): New structure.
3410 (get_narrowest_mode): New function.
3411 (mode_iterator::start): Likewise.
3412 (mode_iterator::iterate_p): Likewise.
3413 (mode_iterator::get_wider): Likewise.
3414 (mode_iterator::get_known_wider): Likewise.
3415 (mode_iterator::get_2xwider): Likewise.
3416 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
3417 (FOR_EACH_MODE): Likewise.
3418 (FOR_EACH_MODE_FROM): Likewise.
3419 (FOR_EACH_MODE_UNTIL): Likewise.
3420 (FOR_EACH_WIDER_MODE): Likewise.
3421 (FOR_EACH_2XWIDER_MODE): Likewise.
3422 * builtins.c (expand_builtin_strlen): Use new mode iterators.
3423 * combine.c (simplify_comparison): Likewise
3424 * config/i386/i386.c (type_natural_mode): Likewise.
3425 * cse.c (cse_insn): Likewise.
3426 * dse.c (find_shift_sequence): Likewise.
3427 * emit-rtl.c (init_derived_machine_modes): Likewise.
3428 (init_emit_once): Likewise.
3429 * explow.c (hard_function_value): Likewise.
3430 * expmed.c (extract_fixed_bit_field_1): Likewise.
3431 (extract_bit_field_1): Likewise.
3432 (expand_divmod): Likewise.
3433 (emit_store_flag_1): Likewise.
3434 * expr.c (init_expr_target): Likewise.
3435 (convert_move): Likewise.
3436 (alignment_for_piecewise_move): Likewise.
3437 (widest_int_mode_for_size): Likewise.
3438 (emit_block_move_via_movmem): Likewise.
3439 (copy_blkmode_to_reg): Likewise.
3440 (set_storage_via_setmem): Likewise.
3441 (compress_float_constant): Likewise.
3442 * omp-low.c (omp_clause_aligned_alignment): Likewise.
3443 * optabs-query.c (get_best_extraction_insn): Likewise.
3444 * optabs.c (expand_binop): Likewise.
3445 (expand_twoval_unop): Likewise.
3446 (expand_twoval_binop): Likewise.
3447 (widen_leading): Likewise.
3448 (widen_bswap): Likewise.
3449 (expand_parity): Likewise.
3450 (expand_unop): Likewise.
3451 (prepare_cmp_insn): Likewise.
3452 (prepare_float_lib_cmp): Likewise.
3453 (expand_float): Likewise.
3454 (expand_fix): Likewise.
3455 (expand_sfix_optab): Likewise.
3456 * postreload.c (move2add_use_add2_insn): Likewise.
3457 * reg-stack.c (reg_to_stack): Likewise.
3458 * reginfo.c (choose_hard_reg_mode): Likewise.
3459 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
3460 * stor-layout.c (mode_for_size): Likewise.
3461 (smallest_mode_for_size): Likewise.
3462 (mode_for_vector): Likewise.
3463 (finish_bitfield_representative): Likewise.
3464 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
3465 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
3466 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3467 * var-tracking.c (prepare_call_arguments): Likewise.
3469 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3470 Alan Hayward <alan.hayward@arm.com>
3471 David Sherwood <david.sherwood@arm.com>
3473 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
3474 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
3475 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
3476 * machmode.h (mode_size): Move earlier in file.
3477 (mode_precision): Likewise.
3478 (mode_inner): Likewise.
3479 (mode_nunits): Likewise.
3480 (mode_unit_size): Likewise.
3481 (unit_unit_precision): Likewise.
3482 (mode_wider): Likewise.
3483 (mode_2xwider): Likewise.
3484 (machine_mode): New class.
3485 (mode_to_bytes): New function.
3486 (mode_to_bits): Likewise.
3487 (mode_to_precision): Likewise.
3488 (mode_to_inner): Likewise.
3489 (mode_to_unit_size): Likewise.
3490 (mode_to_unit_precision): Likewise.
3491 (mode_to_nunits): Likewise.
3492 (GET_MODE_SIZE): Use mode_to_bytes.
3493 (GET_MODE_BITSIZE): Use mode_to_bits.
3494 (GET_MODE_PRECISION): Use mode_to_precision.
3495 (GET_MODE_INNER): Use mode_to_inner.
3496 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
3497 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
3498 (GET_MODE_NUNITS): Use mode_to_nunits.
3499 * system.h (ALWAYS_INLINE): New macro.
3500 * config/powerpcspe/powerpcspe-c.c
3501 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
3502 int for arg1_mode and arg2_mode.
3504 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3505 Alan Hayward <alan.hayward@arm.com>
3506 David Sherwood <david.sherwood@arm.com>
3508 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
3509 Prefix mode names with E_ in case statements.
3510 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3511 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
3512 (aarch64_split_simd_move): Likewise.
3513 (aarch64_gen_storewb_pair): Likewise.
3514 (aarch64_gen_loadwb_pair): Likewise.
3515 (aarch64_gen_store_pair): Likewise.
3516 (aarch64_gen_load_pair): Likewise.
3517 (aarch64_get_condition_code_1): Likewise.
3518 (aarch64_constant_pool_reload_icode): Likewise.
3519 (get_rsqrte_type): Likewise.
3520 (get_rsqrts_type): Likewise.
3521 (get_recpe_type): Likewise.
3522 (get_recps_type): Likewise.
3523 (aarch64_gimplify_va_arg_expr): Likewise.
3524 (aarch64_simd_container_mode): Likewise.
3525 (aarch64_emit_load_exclusive): Likewise.
3526 (aarch64_emit_store_exclusive): Likewise.
3527 (aarch64_expand_compare_and_swap): Likewise.
3528 (aarch64_gen_atomic_cas): Likewise.
3529 (aarch64_emit_bic): Likewise.
3530 (aarch64_emit_atomic_swap): Likewise.
3531 (aarch64_emit_atomic_load_op): Likewise.
3532 (aarch64_evpc_trn): Likewise.
3533 (aarch64_evpc_uzp): Likewise.
3534 (aarch64_evpc_zip): Likewise.
3535 (aarch64_evpc_ext): Likewise.
3536 (aarch64_evpc_rev): Likewise.
3537 (aarch64_evpc_dup): Likewise.
3538 (aarch64_gen_ccmp_first): Likewise.
3539 (aarch64_gen_ccmp_next): Likewise.
3540 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
3541 (alpha_emit_xfloating_libcall): Likewise.
3542 (emit_insxl): Likewise.
3543 (alpha_arg_type): Likewise.
3544 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
3545 (arc_preferred_simd_mode): Likewise.
3546 (arc_secondary_reload): Likewise.
3547 (get_arc_condition_code): Likewise.
3548 (arc_print_operand): Likewise.
3549 (arc_legitimate_constant_p): Likewise.
3550 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3551 * config/arc/arc.md (casesi_load): Likewise.
3552 (casesi_compact_jump): Likewise.
3553 * config/arc/predicates.md (proper_comparison_operator): Likewise.
3554 (cc_use_register): Likewise.
3555 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3556 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
3557 (arm_init_iwmmxt_builtins): Likewise.
3558 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
3559 (neon_expand_vector_init): Likewise.
3560 (arm_attr_length_move_neon): Likewise.
3561 (maybe_get_arm_condition_code): Likewise.
3562 (arm_emit_vector_const): Likewise.
3563 (arm_preferred_simd_mode): Likewise.
3564 (arm_output_iwmmxt_tinsr): Likewise.
3565 (thumb1_output_casesi): Likewise.
3566 (thumb2_output_casesi): Likewise.
3567 (arm_emit_load_exclusive): Likewise.
3568 (arm_emit_store_exclusive): Likewise.
3569 (arm_expand_compare_and_swap): Likewise.
3570 (arm_evpc_neon_vuzp): Likewise.
3571 (arm_evpc_neon_vzip): Likewise.
3572 (arm_evpc_neon_vrev): Likewise.
3573 (arm_evpc_neon_vtrn): Likewise.
3574 (arm_evpc_neon_vext): Likewise.
3575 (arm_validize_comparison): Likewise.
3576 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
3577 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
3578 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
3579 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
3580 (c6x_preferred_simd_mode): Likewise.
3581 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
3582 (epiphany_rtx_costs): Likewise.
3583 * config/epiphany/predicates.md (proper_comparison_operator):
3585 * config/frv/frv.c (condexec_memory_operand): Likewise.
3586 (frv_emit_move): Likewise.
3587 (output_move_single): Likewise.
3588 (output_condmove_single): Likewise.
3589 (frv_hard_regno_mode_ok): Likewise.
3590 (frv_matching_accg_mode): Likewise.
3591 * config/h8300/h8300.c (split_adds_subs): Likewise.
3592 (h8300_rtx_costs): Likewise.
3593 (h8300_print_operand): Likewise.
3594 (compute_mov_length): Likewise.
3595 (output_logical_op): Likewise.
3596 (compute_logical_op_length): Likewise.
3597 (compute_logical_op_cc): Likewise.
3598 (h8300_shift_needs_scratch_p): Likewise.
3599 (output_a_shift): Likewise.
3600 (compute_a_shift_length): Likewise.
3601 (compute_a_shift_cc): Likewise.
3602 (expand_a_rotate): Likewise.
3603 (output_a_rotate): Likewise.
3604 * config/i386/i386.c (classify_argument): Likewise.
3605 (function_arg_advance_32): Likewise.
3606 (function_arg_32): Likewise.
3607 (function_arg_64): Likewise.
3608 (function_value_64): Likewise.
3609 (ix86_gimplify_va_arg): Likewise.
3610 (ix86_legitimate_constant_p): Likewise.
3611 (put_condition_code): Likewise.
3612 (split_double_mode): Likewise.
3613 (ix86_avx256_split_vector_move_misalign): Likewise.
3614 (ix86_expand_vector_logical_operator): Likewise.
3615 (ix86_split_idivmod): Likewise.
3616 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
3617 (ix86_build_const_vector): Likewise.
3618 (ix86_build_signbit_mask): Likewise.
3619 (ix86_match_ccmode): Likewise.
3620 (ix86_cc_modes_compatible): Likewise.
3621 (ix86_expand_branch): Likewise.
3622 (ix86_expand_sse_cmp): Likewise.
3623 (ix86_expand_sse_movcc): Likewise.
3624 (ix86_expand_int_sse_cmp): Likewise.
3625 (ix86_expand_vec_perm_vpermi2): Likewise.
3626 (ix86_expand_vec_perm): Likewise.
3627 (ix86_expand_sse_unpack): Likewise.
3628 (ix86_expand_int_addcc): Likewise.
3629 (ix86_split_to_parts): Likewise.
3630 (ix86_vectorize_builtin_gather): Likewise.
3631 (ix86_vectorize_builtin_scatter): Likewise.
3632 (avx_vpermilp_parallel): Likewise.
3633 (inline_memory_move_cost): Likewise.
3634 (ix86_tieable_integer_mode_p): Likewise.
3635 (x86_maybe_negate_const_int): Likewise.
3636 (ix86_expand_vector_init_duplicate): Likewise.
3637 (ix86_expand_vector_init_one_nonzero): Likewise.
3638 (ix86_expand_vector_init_one_var): Likewise.
3639 (ix86_expand_vector_init_concat): Likewise.
3640 (ix86_expand_vector_init_interleave): Likewise.
3641 (ix86_expand_vector_init_general): Likewise.
3642 (ix86_expand_vector_set): Likewise.
3643 (ix86_expand_vector_extract): Likewise.
3644 (emit_reduc_half): Likewise.
3645 (ix86_emit_i387_round): Likewise.
3646 (ix86_mangle_type): Likewise.
3647 (ix86_expand_round_sse4): Likewise.
3648 (expand_vec_perm_blend): Likewise.
3649 (canonicalize_vector_int_perm): Likewise.
3650 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
3651 (expand_vec_perm_1): Likewise.
3652 (expand_vec_perm_interleave3): Likewise.
3653 (expand_vec_perm_even_odd_pack): Likewise.
3654 (expand_vec_perm_even_odd_1): Likewise.
3655 (expand_vec_perm_broadcast_1): Likewise.
3656 (ix86_vectorize_vec_perm_const_ok): Likewise.
3657 (ix86_expand_vecop_qihi): Likewise.
3658 (ix86_expand_mul_widen_hilo): Likewise.
3659 (ix86_expand_sse2_abs): Likewise.
3660 (ix86_expand_pextr): Likewise.
3661 (ix86_expand_pinsr): Likewise.
3662 (ix86_preferred_simd_mode): Likewise.
3663 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
3664 * config/i386/sse.md (*andnot<mode>3): Likewise.
3665 (<mask_codefor><code><mode>3<mask_name>): Likewise.
3666 (*<code><mode>3): Likewise.
3667 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
3668 (ia64_expand_atomic_op): Likewise.
3669 (ia64_arg_type): Likewise.
3670 (ia64_mode_to_int): Likewise.
3671 (ia64_scalar_mode_supported_p): Likewise.
3672 (ia64_vector_mode_supported_p): Likewise.
3673 (expand_vec_perm_broadcast): Likewise.
3674 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
3675 (iq2000_function_arg_advance): Likewise.
3676 (iq2000_function_arg): Likewise.
3677 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
3678 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
3679 (m68k_libcall_value): Likewise.
3680 (m68k_function_value): Likewise.
3681 (sched_attr_op_type): Likewise.
3682 * config/mcore/mcore.c (mcore_output_move): Likewise.
3683 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
3685 (microblaze_function_arg): Likewise.
3686 * config/mips/mips.c (mips16_build_call_stub): Likewise.
3687 (mips_print_operand): Likewise.
3688 (mips_mode_ok_for_mov_fmt_p): Likewise.
3689 (mips_vector_mode_supported_p): Likewise.
3690 (mips_preferred_simd_mode): Likewise.
3691 (mips_expand_vpc_loongson_even_odd): Likewise.
3692 (mips_expand_vec_unpack): Likewise.
3693 (mips_expand_vi_broadcast): Likewise.
3694 (mips_expand_vector_init): Likewise.
3695 (mips_expand_vec_reduc): Likewise.
3696 (mips_expand_msa_cmp): Likewise.
3697 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
3698 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
3699 (cc_flags_for_mode): Likewise.
3700 * config/msp430/msp430.c (msp430_print_operand): Likewise.
3701 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
3702 (nds32_output_casesi_pc_relative): Likewise.
3703 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3704 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
3705 (nvptx_gen_unpack): Likewise.
3706 (nvptx_gen_pack): Likewise.
3707 (nvptx_gen_shuffle): Likewise.
3708 (nvptx_gen_wcast): Likewise.
3709 (nvptx_preferred_simd_mode): Likewise.
3710 * config/pa/pa.c (pa_secondary_reload): Likewise.
3711 * config/pa/predicates.md (base14_operand): Likewise.
3712 * config/powerpcspe/powerpcspe-c.c
3713 (altivec_resolve_overloaded_builtin): Likewise.
3714 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
3716 (rs6000_preferred_simd_mode): Likewise.
3717 (output_vec_const_move): Likewise.
3718 (rs6000_expand_vector_extract): Likewise.
3719 (rs6000_split_vec_extract_var): Likewise.
3720 (reg_offset_addressing_ok_p): Likewise.
3721 (rs6000_legitimate_offset_address_p): Likewise.
3722 (rs6000_legitimize_address): Likewise.
3723 (rs6000_emit_set_const): Likewise.
3724 (rs6000_const_vec): Likewise.
3725 (rs6000_emit_move): Likewise.
3726 (spe_build_register_parallel): Likewise.
3727 (rs6000_darwin64_record_arg_recurse): Likewise.
3728 (swap_selector_for_mode): Likewise.
3729 (spe_init_builtins): Likewise.
3730 (paired_init_builtins): Likewise.
3731 (altivec_init_builtins): Likewise.
3732 (do_load_for_compare): Likewise.
3733 (rs6000_generate_compare): Likewise.
3734 (rs6000_expand_float128_convert): Likewise.
3735 (emit_load_locked): Likewise.
3736 (emit_store_conditional): Likewise.
3737 (rs6000_output_function_epilogue): Likewise.
3738 (rs6000_handle_altivec_attribute): Likewise.
3739 (rs6000_function_value): Likewise.
3740 (emit_fusion_gpr_load): Likewise.
3741 (emit_fusion_p9_load): Likewise.
3742 (emit_fusion_p9_store): Likewise.
3743 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
3744 (fusion_gpr_mem_load): Likewise.
3745 (fusion_addis_mem_combo_load): Likewise.
3746 (fusion_addis_mem_combo_store): Likewise.
3747 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
3748 (fusion_gpr_mem_load): Likewise.
3749 (fusion_addis_mem_combo_load): Likewise.
3750 (fusion_addis_mem_combo_store): Likewise.
3751 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3753 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
3754 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
3755 (rs6000_preferred_simd_mode): Likewise.
3756 (output_vec_const_move): Likewise.
3757 (rs6000_expand_vector_extract): Likewise.
3758 (rs6000_split_vec_extract_var): Likewise.
3759 (reg_offset_addressing_ok_p): Likewise.
3760 (rs6000_legitimate_offset_address_p): Likewise.
3761 (rs6000_legitimize_address): Likewise.
3762 (rs6000_emit_set_const): Likewise.
3763 (rs6000_const_vec): Likewise.
3764 (rs6000_emit_move): Likewise.
3765 (rs6000_darwin64_record_arg_recurse): Likewise.
3766 (swap_selector_for_mode): Likewise.
3767 (paired_init_builtins): Likewise.
3768 (altivec_init_builtins): Likewise.
3769 (rs6000_expand_float128_convert): Likewise.
3770 (emit_load_locked): Likewise.
3771 (emit_store_conditional): Likewise.
3772 (rs6000_output_function_epilogue): Likewise.
3773 (rs6000_handle_altivec_attribute): Likewise.
3774 (rs6000_function_value): Likewise.
3775 (emit_fusion_gpr_load): Likewise.
3776 (emit_fusion_p9_load): Likewise.
3777 (emit_fusion_p9_store): Likewise.
3778 * config/rx/rx.c (rx_gen_move_template): Likewise.
3779 (flags_from_mode): Likewise.
3780 * config/s390/predicates.md (s390_alc_comparison): Likewise.
3781 (s390_slb_comparison): Likewise.
3782 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
3783 (s390_vector_mode_supported_p): Likewise.
3784 (s390_cc_modes_compatible): Likewise.
3785 (s390_match_ccmode_set): Likewise.
3786 (s390_canonicalize_comparison): Likewise.
3787 (s390_emit_compare_and_swap): Likewise.
3788 (s390_branch_condition_mask): Likewise.
3789 (s390_rtx_costs): Likewise.
3790 (s390_secondary_reload): Likewise.
3791 (__SECONDARY_RELOAD_CASE): Likewise.
3792 (s390_expand_cs): Likewise.
3793 (s390_preferred_simd_mode): Likewise.
3794 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
3795 * config/sh/sh.c (sh_print_operand): Likewise.
3796 (dump_table): Likewise.
3797 (sh_secondary_reload): Likewise.
3798 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3799 * config/sh/sh.md (casesi_worker_1): Likewise.
3800 (casesi_worker_2): Likewise.
3801 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
3802 (fcc_comparison_operator): Likewise.
3803 * config/sparc/sparc.c (sparc_expand_move): Likewise.
3804 (emit_soft_tfmode_cvt): Likewise.
3805 (sparc_preferred_simd_mode): Likewise.
3806 (output_cbranch): Likewise.
3807 (sparc_print_operand): Likewise.
3808 (sparc_expand_vec_perm_bmask): Likewise.
3809 (vector_init_bshuffle): Likewise.
3810 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3811 (spu_vector_mode_supported_p): Likewise.
3812 (spu_expand_insv): Likewise.
3813 (spu_emit_branch_or_set): Likewise.
3814 (spu_handle_vector_attribute): Likewise.
3815 (spu_builtin_splats): Likewise.
3816 (spu_builtin_extract): Likewise.
3817 (spu_builtin_promote): Likewise.
3818 (spu_expand_sign_extend): Likewise.
3819 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3820 (tilegx_simd_int): Likewise.
3821 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
3822 (tilepro_simd_int): Likewise.
3823 * config/v850/v850.c (const_double_split): Likewise.
3824 (v850_print_operand): Likewise.
3825 (ep_memory_offset): Likewise.
3826 * config/vax/vax.c (vax_rtx_costs): Likewise.
3827 (vax_output_int_move): Likewise.
3828 (vax_output_int_add): Likewise.
3829 (vax_output_int_subtract): Likewise.
3830 * config/visium/predicates.md (visium_branch_operator): Likewise.
3831 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
3832 (visium_print_operand_address): Likewise.
3833 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3834 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
3835 (xtensa_expand_conditional_branch): Likewise.
3836 (xtensa_copy_incoming_a7): Likewise.
3837 (xtensa_output_literal): Likewise.
3838 * dfp.c (decimal_real_maxval): Likewise.
3839 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3841 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3842 Alan Hayward <alan.hayward@arm.com>
3843 David Sherwood <david.sherwood@arm.com>
3845 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
3846 (mode_nunits_inline): Likewise.
3847 (mode_inner_inline): Likewise.
3848 (mode_unit_size_inline): Likewise.
3849 (mode_unit_precision_inline): Likewise.
3850 (emit_insn_modes_h): Likewise. Also emit a #define of the
3852 (emit_mode_wider): Add an E_ prefix to mode names.
3853 (emit_mode_complex): Likewise.
3854 (emit_mode_inner): Likewise.
3855 (emit_mode_adjustments): Likewise.
3856 (emit_mode_int_n): Likewise.
3857 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
3858 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
3859 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
3860 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
3861 (CRC32_BUILTIN, ENTRY): Likewise.
3862 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
3863 (aarch64_pop_regs): Likewise.
3864 (aarch64_process_components): Likewise.
3865 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
3866 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
3867 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
3868 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
3869 * config/arm/arm.c (arm_init_libfuncs): Likewise.
3870 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
3872 * config/i386/i386-builtin.def (pcmpestr): Likewise.
3873 (pcmpistr): Likewise.
3874 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
3875 * config/mmix/mmix.c (mmix_output_condition): Likewise.
3876 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
3878 * config/rl78/rl78.c (mduc_regs): Likewise.
3879 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3880 (htm_expand_builtin): Likewise.
3881 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
3882 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
3883 * config/xtensa/xtensa.c (print_operand): Likewise.
3884 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
3885 (NUM_MODE_VECTOR_INT): Likewise.
3886 * genoutput.c (null_operand): Likewise.
3887 (output_operand_data): Likewise.
3888 * genrecog.c (print_parameter_value): Likewise.
3889 * lra.c (debug_operand_data): Likewise.
3891 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3893 * dwarf2out.c (scompare_loc_descriptor_wide)
3894 (scompare_loc_descriptor_narrow): New functions, split out from...
3895 (scompare_loc_descriptor): ...here.
3896 * expmed.c (emit_store_flag_int): New function, split out from...
3897 (emit_store_flag): ...here.
3899 2017-08-30 Richard Biener <rguenther@suse.de>
3901 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
3902 (dwarf2out_early_finish): Move setting of AT_pubnames from
3903 early debug output to early finish.
3905 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3907 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
3908 and -mdata-region to the assembler.
3910 2017-08-30 Richard Biener <rguenther@suse.de>
3912 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
3914 (gen_subprogram_die): Add DW_AT_object_pointer only early.
3915 (dwarf2out_early_global_decl): Only generate a DIE for the
3916 abstract origin if it doesn't already exist or is a declaration DIE.
3917 (resolve_addr): Do not add the linkage name twice when
3918 generating a stub DIE for the DW_TAG_GNU_call_site target.
3920 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3922 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3923 Use machine_mode rather than int for arg1_mode.
3925 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
3928 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
3929 that the second argument of the built-in functions to unpack
3930 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
3931 switch statement instead a lot of if statements.
3932 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
3933 Allow 64-bit values to be in Altivec registers as well as
3934 traditional floating point registers.
3935 (pack<mode>, FMOVE128_VSX iterator): Likewise.
3937 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
3939 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
3940 MAX_REGS_PER_ADDRESS == 1.
3942 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
3944 * config/i386/i386.opt (flag_fentry): Do not init to -1.
3945 * config/i386/i386.c (ix86_option_override_internal): Simplify
3946 setting of opts->x_flag_entry.
3948 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3949 Jakub Jelinek <jakub@redhat.com>
3950 Richard Biener <rguenther@suse.de>
3952 PR tree-optimization/81503
3953 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
3954 folded constant fits in the target type; reorder tests for clarity.
3956 2017-08-29 Martin Liska <mliska@suse.cz>
3958 * passes.def: Include pass_lower_switch.
3959 * stmt.c (dump_case_nodes): Remove and move to
3960 tree-switch-conversion.
3961 (case_values_threshold): Likewise.
3962 (expand_switch_as_decision_tree_p): Likewise.
3963 (emit_case_decision_tree): Likewise.
3964 (expand_case): Likewise.
3965 (balance_case_nodes): Likewise.
3966 (node_has_low_bound): Likewise.
3967 (node_has_high_bound): Likewise.
3968 (node_is_bounded): Likewise.
3969 (emit_case_nodes): Likewise.
3970 (struct simple_case_node): New struct.
3971 (add_case_node): Remove.
3972 (emit_case_dispatch_table): Use vector instead of case_list.
3973 (reset_out_edges_aux): Remove.
3974 (compute_cases_per_edge): Likewise.
3975 (expand_case): Build list of simple_case_node.
3976 (expand_sjlj_dispatch_table): Use it.
3977 * tree-switch-conversion.c (struct case_node): Moved from
3978 stmt.c and adjusted.
3979 (emit_case_nodes): Likewise.
3980 (node_has_low_bound): Likewise.
3981 (node_has_high_bound): Likewise.
3982 (node_is_bounded): Likewise.
3983 (case_values_threshold): Likewise.
3984 (reset_out_edges_aux): Likewise.
3985 (compute_cases_per_edge): Likewise.
3986 (add_case_node): Likewise.
3987 (dump_case_nodes): Likewise.
3988 (balance_case_nodes): Likewise.
3989 (expand_switch_as_decision_tree_p): Likewise.
3990 (emit_jump): Likewise.
3991 (emit_case_decision_tree): Likewise.
3992 (try_switch_expansion): Likewise.
3993 (do_jump_if_equal): Likewise.
3994 (emit_cmp_and_jump_insns): Likewise.
3995 (fix_phi_operands_for_edge): New function.
3996 (record_phi_operand_mapping): Likewise.
3997 (class pass_lower_switch): New pass.
3998 (pass_lower_switch::execute): New function.
3999 (make_pass_lower_switch): Likewise.
4000 (conditional_probability):
4001 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
4002 * tree-pass.h: Add make_pass_lower_switch.
4004 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
4007 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
4010 2017-08-29 Richard Biener <rguenther@suse.de>
4012 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
4013 we do not add a DW_AT_inline attribute twice.
4014 (gen_subprogram_die): Remove code setting DW_AT_inline on
4015 DECL_ABSTRACT_P nodes.
4017 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
4019 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
4020 calls to internal functions.
4021 (gimplify_modify_expr): Likewise.
4022 * tree-call-cdce.c (use_internal_fn): Likewise.
4023 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
4024 (convert_to_divmod): Set the nothrow flag.
4025 * tree-if-conv.c (predicate_mem_writes): Likewise.
4026 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4027 (vectorizable_call): Likewise.
4028 (vectorizable_store): Likewise.
4029 (vectorizable_load): Likewise.
4030 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
4031 (vect_recog_mask_conversion_pattern): Likewise.
4033 2017-08-29 Martin Liska <mliska@suse.cz>
4036 * gcc.c (driver_handle_option): Add new argument.
4037 * opts-common.c (handle_option): Pass
4038 target_option_override_hook.
4039 * opts-global.c (lang_handle_option): Add new option.
4040 (set_default_handlers): Add new argument.
4041 (decode_options): Likewise.
4042 * opts.c (target_handle_option): Likewise.
4043 (common_handle_option): Call target_option_override_hook.
4044 * opts.h (struct cl_option_handler_func): Add hook for
4045 target option override.
4046 (struct cl_option_handlers): Likewise.
4047 (set_default_handlers): Add new argument.
4048 (decode_options): Likewise.
4049 (common_handle_option): Likewise.
4050 (target_handle_option): Likewise.
4051 * toplev.c (toplev::main): Pass targetm.target_option.override
4054 2017-08-29 Richard Biener <rguenther@suse.de>
4055 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4057 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
4058 life to the active subtree.
4060 2017-08-28 Jeff Law <law@redhat.com>
4062 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
4063 derive_equivalences.
4064 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
4066 (edge_info::derive_equivalences): New private member function
4068 * tree-ssa-dom.c (class edge_info): Changed from a struct
4069 to a class. Add ctor/dtor, methods and data members.
4070 (edge_info::edge_info): Renamed from allocate_edge_info.
4071 Initialize additional members.
4072 (edge_info::~edge_info): New.
4073 (free_dom_edge_info): Delete the edge info.
4074 (record_edge_info): Use new class & associated member functions.
4075 Tighten forms for testing for edge equivalences.
4076 (record_temporary_equivalences): Iterate over the simple
4077 equivalences rather than assuming there's only one per edge.
4078 (cprop_into_successor_phis): Iterate over the simple
4079 equivalences rather than assuming there's only one per edge.
4080 (optimize_stmt): Use operand_equal_p rather than pointer
4081 equality for mini-DSE code.
4083 2017-08-28 Nathan Sidwell <nathan@acm.org>
4085 * gcc.c (execute): Fold SIGPIPE handling into switch
4086 statement. Adjust internal error message.
4088 2017-08-28 Richard Biener <rguenther@suse.de>
4091 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
4092 Do nothing for removed DIEs.
4094 2017-08-28 Richard Biener <rguenther@suse.de>
4096 PR tree-optimization/81977
4097 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
4100 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
4103 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
4104 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
4105 using targetm.gen_mem_thread_fence.
4107 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
4110 * config/i386/i386.md (*<btsc><mode>): Change operand 2
4111 predicate to register_operand. Reorder operands.
4112 (*btr<mode>): Ditto.
4113 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
4114 (*btr<mode>_mask): Ditto.
4116 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
4118 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
4119 * config/rs6000/xmmintrin.h: New file.
4120 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
4122 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4125 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
4126 parameter and_insn and return it.
4127 (recombine_lvx_pattern): Insert a copy to ensure availability of
4128 the base register of the copied masking operation at the point of
4129 the instruction replacement.
4130 (recombine_stvx_pattern): Likewise.
4132 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
4134 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
4135 undocumented switches.
4136 (-mpower9-dform-vector): Likewise.
4137 (-mpower9-dform): Likewise.
4138 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
4139 comments to delete references to -mpower9-dform* switches.
4140 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
4141 Delete reference to -mpower9-dform* switches, test for
4142 -mpower9-vector instead.
4143 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
4144 (OTHER_P9_VECTOR_MASKS): Likewise.
4145 (POWERPC_MASKS): Likewise.
4146 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
4147 tests against -mpower9-dform* to -mpower9-vector. Delete code
4148 that checked for -mpower9-dform* consistancy with other options.
4149 Add test for -mpower9-misc to enable other power9 switches.
4150 (rs6000_init_hard_regno_mode_ok): Likewise.
4151 (rs6000_option_override_internal): Likewise.
4152 (rs6000_emit_prologue): Likewise.
4153 (rs6000_emit_epilogue): Likewise.
4154 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
4155 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
4156 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
4158 (emit_fusion_p9_store): Likewise.
4159 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
4160 resetting these macros if the assembler does not support ISA 3.0
4162 (TARGET_P9_DFORM_VECTOR): Likewise.
4163 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
4164 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
4166 2017-08-25 Alan Modra <amodra@gmail.com>
4168 PR rtl-optimization/81747
4169 * cse.c (cse_extended_basic_block): Don't attempt to record
4170 equivalences for degenerate conditional branches that branch
4171 to their fall-through.
4173 2017-08-24 Martin Sebor <msebor@redhat.com>
4176 * gimple-fold.c (size_must_be_zero_p): New function.
4177 (gimple_fold_builtin_memory_op): Call it.
4179 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
4181 * config/rs6000/mm_malloc.h: New file.
4183 2017-08-24 Bin Cheng <bin.cheng@arm.com>
4185 PR tree-optimization/81913
4186 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
4187 analysis when either IVs in condition can wrap.
4189 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
4191 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
4192 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
4194 2017-08-24 Richard Biener <rguenther@suse.de>
4197 * targhooks.c (default_target_can_inline_p): Properly
4198 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
4199 is present and always compare.
4200 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
4201 infer -mfpmath=sse from TARGET_SSE_P.
4202 (ix86_can_inline_p): Properly use target_option_default_node when
4203 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
4205 2017-08-24 Richard Biener <rguenther@suse.de>
4208 * dwarf2out.c (output_die): Handle flag_generate_offload like
4210 (output_comp_unit): Likewise.
4211 (gen_array_type_die): Likewise.
4212 (dwarf2out_early_finish): Likewise.
4213 (note_variable_value_in_expr): Likewise.
4214 (dwarf2out_finish): Likewise. Adjust assert.
4215 * cgraphunit.c (symbol_table::compile): Move setting of
4216 flag_generate_offload earlier ...
4217 (symbol_table::finalize_compilation_unit): ... here, before
4218 early debug finalization.
4220 2017-08-24 Richard Biener <rguenther@suse.de>
4222 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
4223 and ipa-fnsummary.h.
4224 (ix86_can_inline_p): When ix86_fpmath flags do not match
4225 check whether the callee uses FP math at all.
4227 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
4230 * tree-ssanames.c (get_nonzero_bits): Use element_precision
4231 instead of TYPE_PRECISION.
4233 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
4234 Alan Hayward <alan.hayward@arm.com>
4235 David Sherwood <david.sherwood@arm.com>
4237 * combine.c (make_extraction): Use subreg_offset_from_lsb.
4239 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
4241 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
4243 (ix86_frame::stack_realign_allocate): New field.
4244 (struct machine_frame_state): Modify comments.
4245 (machine_frame_state::sp_realigned_fp_end): New field.
4246 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
4248 (sp_valid_at): Add assertion to assure no attempt to access invalid
4249 offset of a realigned stack.
4250 (fp_valid_at): Likewise.
4251 (choose_baseaddr): Modify comments.
4252 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
4253 ix86_expand_prologue.
4254 (ix86_expand_prologue): Modify stack realignment and allocation.
4255 (ix86_expand_epilogue): Modify comments.
4256 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
4257 avx2_runtime, avx512f, and avx512f_runtime.
4259 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
4261 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
4262 (mstackrealign): Do not init to -1.
4263 * config/i386/i386.c (ix86_option_override_internal):
4264 Check opts_set, not opts when setting default value of
4265 opts->x_ix86_force_align_arg_pointer.
4267 2017-08-23 Richard Biener <rguenther@suse.de>
4269 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
4270 lang_hooks.decl_printable_name.
4271 * print-rtl-function.c (print_rtx_function): Likewise.
4272 * tree-pretty-print.c (dump_function_header): Likewise.
4274 2017-08-23 Richard Biener <rguenther@suse.de>
4277 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
4278 -g0 at compile-time.
4280 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4283 * doc/sourcebuild.texi (Other hardware attributes):
4286 2017-08-23 Tamar Christina <tamar.christina@arm.com>
4289 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
4290 Fix single-use check.
4292 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
4294 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
4296 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4298 * doc/install.texi: Modify to add more details on running selected
4301 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
4303 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
4304 is combined with -mabi=ms.
4305 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
4308 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4310 PR tree-optimization/81488
4311 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
4312 and cached_basis fields.
4313 (MAX_SPREAD): New constant.
4314 (alloc_cand_and_find_basis): Initialize new fields.
4315 (clear_visited): New function.
4316 (create_phi_basis_1): Rename from create_phi_basis, set visited
4317 and cached_basis fields.
4318 (create_phi_basis): New wrapper function.
4319 (phi_add_costs_1): Rename from phi_add_costs, add spread
4320 parameter, set visited field, short-circuit when limits reached.
4321 (phi_add_costs): New wrapper function.
4322 (record_phi_increments_1): Rename from record_phi_increments, set
4324 (record_phi_increments): New wrapper function.
4325 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
4326 (phi_incr_cost): New wrapper function.
4327 (all_phi_incrs_profitable_1): Rename from
4328 all_phi_incrs_profitable, set visited field.
4329 (all_phi_incrs_profitable): New wrapper function.
4331 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4332 Alan Hayward <alan.hayward@arm.com>
4333 David Sherwood <david.sherwood@arm.com>
4335 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
4336 that takes the outer and inner modes.
4337 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
4338 comparison as the canonical test for a paradoxical subreg.
4339 * combine.c (simplify_set): Use paradoxical_subreg_p.
4340 (make_extraction): Likewise.
4341 (force_to_mode): Likewise.
4342 (rtx_equal_for_field_assignment_p): Likewise.
4343 (gen_lowpart_for_combine): Likewise.
4344 (simplify_comparison): Likewise.
4345 * cse.c (equiv_constant): Likewise.
4346 * expmed.c (store_bit_field_1): Likewise.
4347 * final.c (alter_subreg): Likewise.
4348 * fwprop.c (propagate_rtx): Likewise.
4349 (forward_propagate_subreg): Likewise.
4350 * ira-conflicts.c (ira_build_conflicts): Likewise.
4351 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
4352 * lra-constraints.c (curr_insn_transform): Likewise.
4353 (split_reg): Likewise.
4354 * lra-eliminations.c (move_plus_up): Likewise.
4355 (lra_eliminate_regs_1): Likewise.
4356 * recog.c (general_operand): Likewise.
4357 * ree.c (combine_reaching_defs): Likewise.
4358 * reload.c (push_reload): Likewise.
4359 (find_reloads): Likewise.
4360 * reload1.c (elimination_effects): Likewise.
4361 (compute_reload_subreg_offset): Likewise.
4362 (choose_reload_regs): Likewise.
4363 * rtlanal.c (subreg_lsb_1): Likewise.
4364 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4365 (simplify_subreg): Likewise.
4366 * var-tracking.c (track_loc_p): Likewise.
4367 * emit-rtl.c (byte_lowpart_offset): Likewise.
4368 (paradoxical_subreg_p): Delete out-of-line definition.
4370 2017-08-22 Jeff Law <law@redhat.com>
4372 PR tree-optimization/81741
4373 PR tree-optimization/71947
4374 * tree-ssa-dom.c: Include tree-inline.h.
4375 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
4376 equivalences if one is more expensive to compute than the other.
4377 * tree-ssa-scopedtables.h (class const_or_copies): Make
4378 record_const_or_copy_raw method private.
4379 (class avail_exprs_stack): New method simplify_binary_operation.
4380 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
4381 avail_exprs_stack::simplify_binary_operation as needed.
4382 (avail_exprs_stack::simplify_binary_operation): New function.
4384 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
4386 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
4387 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
4388 (DOT_SYMBOLS): Likewise.
4389 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
4390 (RELOCATABLE_NEEDS_FIXUP): Likewise.
4391 (RS6000_ABI_NAME): Likewise.
4392 (TARGET_CMODEL): Likewise.
4393 (TOC_SECTION_ASM_OP): Likewise.
4394 (SET_CMODEL): New macro.
4395 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
4397 2017-08-22 Richard Biener <rguenther@suse.de>
4399 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
4400 to allow for free-lang-data replacements similar to verify_type_variant.
4402 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
4403 Alan Hayward <alan.hayward@arm.com>
4404 David Sherwood <david.sherwood@arm.com>
4406 * config/aarch64/aarch64.md (casesi): Use DImode rather than
4407 VOIDmode for the LABEL_REF.
4409 2017-08-22 Richard Biener <rguenther@suse.de>
4411 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
4413 2017-08-22 Richard Biener <rguenther@suse.de>
4415 * common.opt (feliminate-dwarf2-dups): Ignore.
4416 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
4417 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
4418 same_die_p_wrap, compute_section_prefix,
4419 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
4420 (comdat_symbol_id, comdat_symbol_number): Likewise.
4421 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
4423 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
4424 (output_die): Mark unreachable path unreachable.
4425 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
4426 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
4427 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
4428 (dwarf2out_early_finish): Likewise.
4430 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
4432 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
4434 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
4437 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
4438 not VAR_P. Filter attribute warnings with OPT_Wattributes.
4439 (avr_attribute_table) <io, io_low, address>: Initialize
4440 .decl_required with true.
4442 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
4444 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
4445 undocumented debugging options.
4446 (-mvsx-scalar-double): Likewise.
4447 (-mallow-df-permute): Likewise.
4448 (-mvectorize-builtins): Likewise.
4449 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
4450 (rs6000_builtin_vectorized_function): Likewise.
4451 (rs6000_builtin_md_vectorized_function): Likewise.
4452 (rs6000_opt_vars): Likewise.
4454 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
4457 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
4458 (*btrq_imm): Rename from *btrq.
4459 (*btcq_imm): Rename from *btcq.
4460 (btsc): New code attribute.
4461 (*<btsc><mode>): New insn pattern.
4462 (*btr<mode>): Ditto.
4463 (*<btsc><mode>_mask): New insn_and_split pattern.
4464 (*btr<mode>_mask): Ditto.
4466 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4467 Alan Hayward <alan.hayward@arm.com>
4468 David Sherwood <david.sherwood@arm.com>
4470 * function.c (pad_below): Simplify padding calculation.
4472 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4473 Alan Hayward <alan.hayward@arm.com>
4474 David Sherwood <david.sherwood@arm.com>
4476 * target.def (function_prologue): Remove frame size argument.
4477 (function_epilogue): Likewise.
4478 * doc/tm.texi: Regenerate.
4479 * final.c (final_start_function): Update call to function_prologue.
4480 (final_end_function): Update call to function_epilogue.
4481 (default_function_pro_epilogue): Remove frame size argument.
4482 * output.h (default_function_pro_epilogue): Likewise.
4483 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
4484 (arm_output_function_prologue): Likewise.
4485 * config/frv/frv.c (frv_function_prologue): Likewise.
4486 (frv_function_epilogue): Likewise.
4487 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
4488 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
4489 (ia64_output_function_epilogue): Likewise.
4490 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
4491 (m32r_output_function_epilogue): Likewise.
4492 * config/microblaze/microblaze.c (microblaze_function_prologue)
4493 (microblaze_function_epilogue): Likewise.
4494 * config/mips/mips.c (mips_output_function_prologue): Likewise.
4495 (mips_output_function_epilogue): Likewise.
4496 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
4497 (mmix_target_asm_function_epilogue): Likewise.
4498 * config/msp430/msp430.c (msp430_start_function): Likewise.
4499 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
4500 (nds32_asm_function_epilogue): Likewise.
4501 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
4502 * config/pa/pa.c (pa_output_function_prologue): Likewise.
4503 (pa_output_function_epilogue): Likewise.
4504 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
4505 (rs6000_output_function_epilogue): Likewise.
4506 * config/rl78/rl78.c (rl78_start_function): Likewise.
4507 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
4508 (rs6000_output_function_epilogue): Likewise.
4509 * config/rx/rx.c (rx_output_function_prologue): Likewise.
4510 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
4511 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
4512 (sparc_asm_function_epilogue): Likewise.
4514 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4516 * tree.h (type_has_mode_precision_p): New function.
4517 * convert.c (convert_to_integer_1): Use it.
4518 * expr.c (expand_expr_real_2): Likewise.
4519 (expand_expr_real_1): Likewise.
4520 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
4521 * match.pd: Likewise.
4522 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
4523 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
4524 * tree-tailcall.c (process_assignment): Likewise.
4525 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4526 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
4527 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
4528 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
4529 (vectorizable_assignment): Likewise.
4530 (vectorizable_shift): Likewise.
4531 (vectorizable_operation): Likewise.
4532 * tree-vrp.c (register_edge_assert_for_2): Likewise.
4534 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
4536 * match.pd: Add pow (C, x) simplification.
4538 2017-08-21 Richard Biener <rguenther@suse.de>
4540 PR tree-optimization/81900
4541 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
4542 for blocks with abnormal predecessors.
4543 (compute_antic): Do not set visited flag prematurely.
4545 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
4548 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
4550 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4552 * stor-layout.h (vector_type_mode): Move to...
4553 * tree.h (vector_type_mode): ...here.
4554 * stor-layout.c (vector_type_mode): Move to...
4555 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
4557 2017-08-21 Richard Biener <rguenther@suse.de>
4559 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
4560 register_external_die hooks.
4561 (debug_false_tree_charstarstar_uhwistar): Declare.
4562 (debug_nothing_tree_charstar_uhwi): Likewise.
4563 * debug.c (do_nothing_debug_hooks): Adjust.
4564 (debug_false_tree_charstarstar_uhwistar): New do nothing.
4565 (debug_nothing_tree_charstar_uhwi): Likewise.
4566 * dbxout.c (dbx_debug_hooks): Adjust.
4567 (xcoff_debug_hooks): Likewise.
4568 * sdbout.c (sdb_debug_hooks): Likewise.
4569 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4570 * dwarf2out.c (macinfo_label_base): New global.
4571 (dwarf2out_register_external_die): New function for the
4572 register_external_die hook.
4573 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
4574 (dwarf2_debug_hooks): Use them.
4575 (dwarf2_lineno_debug_hooks): Adjust.
4576 (struct die_struct): Add with_offset flag.
4577 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
4578 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
4579 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
4580 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
4581 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
4582 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
4583 defining section names for the early LTO debug variants.
4584 (reset_indirect_string): New helper.
4585 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
4586 (print_dw_val): Add support for offsetted symbol references.
4587 (get_ultimate_context): Split out from is_cxx.
4588 (is_cxx): Use get_ultimate_context.
4589 (is_fortran): Add decl overload.
4590 (compute_comp_unit_symbol): Split out worker from
4591 compute_section_prefix.
4592 (compute_section_prefix): Call compute_comp_unit_symbol and
4593 set comdat_type_p here.
4594 (output_die): Skip DIE symbol output for the LTO added one.
4595 Handle DIE symbol references with offset.
4596 (output_comp_unit): Guard section name mangling properly.
4597 For LTO debug sections emit a symbol at the section beginning
4598 which we use to refer to its DIEs.
4599 (add_abstract_origin_attribute): For DIEs registered via
4600 dwarf2out_register_external_die directly refer to the early
4601 DIE rather than indirectly through the shadow one we created.
4602 Remove obsolete call to dwarf2out_abstract_function for
4603 non-function/block origins.
4604 (gen_array_type_die): When generating early LTO debug do
4605 not emit DW_AT_string_length.
4606 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
4607 late when in LTO. As suggested place a gcc_unreachable for
4608 the DECL_ABSTRACT_P case.
4609 (gen_subprogram_die): Avoid another specification DIE
4610 for early built declarations/definitions for the late LTO case.
4611 (gen_variable_die): Add type references for late duplicated VLA dies
4613 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
4614 we have the abstract instance already.
4615 (process_scope_var): Adjust decl DIE contexts in LTO which
4616 first puts them in limbo.
4617 (gen_decl_die): Do not generate type DIEs late apart from
4618 types for VLAs or for decls we do not yet have a DIE. Do not
4619 call dwarf2out_abstract_function late.
4620 (dwarf2out_early_global_decl): Make sure to create DIEs
4621 for abstract instances of a decl first.
4622 (dwarf2out_late_global_decl): Adjust comment.
4623 (output_macinfo_op): With multiple macro sections use
4624 macinfo_label_base to distinguish labels.
4625 (output_macinfo): Likewise. Update macinfo_label_base.
4626 Pass in the line info label.
4627 (note_variable_value_in_expr): When generating LTO resolve
4628 all variable values here by generating DIEs as needed.
4629 (init_sections_and_labels): Add early LTO debug flag parameter
4630 and generate different sections and names if set. Add generation
4631 counter for the labels so we can have multiple of them.
4632 (reset_dies): Helper to allow DIEs to be output multiple times.
4633 (dwarf2out_finish): When outputting DIEs to the fat part of an
4634 LTO object first reset DIEs.
4635 (dwarf2out_early_finish): Output early DIEs when generating LTO.
4636 (modified_type_die): Check for decl_ultimate_origin being self
4638 (gen_type_die_with_usage): Likewise.
4639 (gen_typedef_die): Allow decl_ultimate_origin being self.
4640 (set_decl_abstract_flags): Remove.
4641 (set_block_abstract_flags): Likewise.
4642 (dwarf2out_abstract_function): Treat the early generated DIEs
4643 as the abstract copy and only add DW_AT_inline and
4644 DW_AT_artificial here and call set_decl_origin_self.
4645 If the DIE has an abstract origin don't do anything.
4646 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
4647 if we have none yet (Go fails to build one, PR78628).
4648 (variably_modified_type_p): Prevent endless recursion for Ada
4649 cyclic pointer types.
4650 * lto-streamer-in.c: Include debug.h.
4651 (dref_queue): New global.
4652 (lto_read_tree_1): Stream in DIE references.
4653 (lto_input_tree): Register DIE references.
4654 (input_function): Stream DECL_DEBUG_ARGS.
4655 * lto-streamer-out.c: Include debug.h.
4656 (lto_write_tree_1): Output DIE references.
4657 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
4658 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
4659 (output_function): Stream DECL_DEBUG_ARGS.
4660 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
4661 Stream DECL_ABSTRACT_ORIGIN.
4662 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
4663 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
4664 DECL_CONTEXT for file-scope decls.
4665 * lto-streamer.h (struct dref_entry): Declare.
4666 (dref_queue): Likewise.
4667 * cfgexpand.c (pass_expand::execute): Do not call the
4668 outlining_inline_function hook here.
4669 * lto-wrapper.c (debug_obj): New global.
4670 (tool_cleanup): Unlink it if required.
4671 (debug_objcopy): New function.
4672 (run_gcc): Handle early debug sections in the IL files by
4673 extracting them to separate files, partially linkin them and
4674 feeding the result back as result to the linker.
4675 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
4676 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
4677 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
4678 sections into a separate segment.
4679 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
4681 (darwin_asm_dwarf_section): Likewise.
4682 (darwin_asm_output_dwarf_offset): Likewise.
4683 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
4685 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4686 Alan Hayward <alan.hayward@arm.com>
4687 David Sherwood <david.sherwood@arm.com>
4689 * read-md.h (md_reader::record_potential_iterator_use): Replace
4690 pointer argument with an rtx and an index.
4691 * read-rtl.c (iterator_group::apply_iterator): Likewise.
4692 (apply_mode_iterator): Likewise.
4693 (apply_code_iterator): Likewise.
4694 (apply_int_iterator): Likewise.
4695 (apply_subst_iterator): Likewise.
4696 (record_iterator_use): Likewise.
4697 (record_attribute_use): Likewise.
4698 (md_reader::record_potential_iterator_use): Likewise. Update calls
4699 to record_iterator_use and apply_iterator.
4700 (iterator_use): Replace ptr with x and index.
4701 (attribute_use): Likewise.
4702 (apply_attribute_uses): Update calls to apply_iterator.
4703 (apply_iterators): Likewise. Update initialization of iterator_use.
4704 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
4705 and record_potential_iterator_use.
4706 (rtx_reader::read_rtx_operand): Likewise.
4708 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4709 Alan Hayward <alan.hayward@arm.com>
4710 David Sherwood <david.sherwood@arm.com>
4712 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
4715 2017-08-21 Richard Biener <rguenther@suse.de>
4718 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
4719 at struct end conservatively when comparing common bases.
4721 2017-08-21 Richard Biener <rguenther@suse.de>
4723 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
4724 (mem_ref_in_stmt): Remove.
4725 (determine_max_movement): Use ref index to get at the reference.
4726 (invariantness_dom_walker::before_dom_children): Deal with
4727 lim data already initialized.
4728 (gather_mem_refs_stmt): Initialize lim data and record ref index.
4730 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4732 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
4733 (TARGET_ISA_ROUND): Ditto.
4734 (TARGET_ROUND): Ditto.
4735 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
4736 * config/i386/i386.md: Ditto.
4737 * config/i386/sse.md: Ditto.
4738 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
4739 with OPTION_MASK_ISA_SSE4_1.
4741 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4744 * doc/extend.texi (x86 Built-in Functions): Correct the name of
4745 __builtin_ia32_lzcnt_u16.
4747 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4750 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
4751 (rs6000_set_current_function): Rewrite function to use it.
4753 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
4756 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
4757 and TYPE_WARN_IF_NOT_ALIGN.
4758 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
4759 (handle_warn_if_not_align): New.
4760 (place_union_field): Call handle_warn_if_not_align.
4761 (place_field): Call handle_warn_if_not_align.
4762 Copy TYPE_WARN_IF_NOT_ALIGN.
4763 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
4764 (layout_type): Likewise.
4765 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
4767 (tree_decl_common): Add warn_if_not_align.
4768 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
4769 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
4770 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
4771 (DECL_WARN_IF_NOT_ALIGN): Likewise.
4772 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
4773 * doc/extend.texi: Document warn_if_not_aligned attribute.
4774 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
4776 2017-08-17 Martin Liska <mliska@suse.cz>
4779 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
4780 (get_data_dependence): Use it as pointer type.
4781 (distribute_loop): Likewise.
4783 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4785 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
4786 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
4787 (p8_vmrgow_<mode>_direct): New define_insn.
4788 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
4789 handle endianness for vmrgew and vmrgow permute patterns.
4791 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4793 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
4794 * config/rs6000/rs6000-cpus.def: Remove comment.
4795 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
4796 (POWERPC_MASKS): Likewise.
4797 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
4798 use of TARGET_VSX_TIMODE.
4799 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
4800 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
4801 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
4802 (rs6000_option_override_internal): Remove dead code.
4803 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
4804 (rs6000_legitimize_reload_address): Likewise.
4805 (rs6000_legitimate_address_p): Likewise.
4806 (rs6000_opt_masks): Delete "vsx-timode".
4807 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
4808 from function comment.
4809 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
4810 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
4811 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
4813 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
4814 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
4815 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
4818 (define_peephole2): Likewise.
4820 2017-08-17 Martin Sebor <msebor@redhat.com>
4823 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
4824 past the end of an array.
4825 (test_pp_format): Add test cases.
4827 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
4829 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
4830 missing ECF_NOTHROW flags.
4832 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4835 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
4836 operands residing in integer registers.
4837 (*vsx_le_perm_load_<mode>): Likewise.
4838 (*vsx_le_perm_store_<mode>): Likewise.
4839 (define_peephole2): Add peepholes to optimize the above.
4841 2017-08-17 Marek Polacek <polacek@redhat.com>
4844 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
4845 to mimic what shorten_compare did. Change the return type to bool.
4846 (fold_cond_expr_with_comparison): Update call to
4847 operand_equal_for_comparison_p.
4848 (fold_ternary_loc): Likewise.
4850 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
4852 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
4854 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
4856 2017-08-17 Richard Biener <rguenther@suse.de>
4858 * tree-ssa-structalias.c (solve_graph): When propagating
4859 to successors update the graphs succ edges and avoid duplicate work.
4861 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
4864 * config/i386/i386.c (ix86_option_override_internal): Save target
4865 specific options after ix86_stack_protector_guard_reg was changed.
4867 2017-08-17 Richard Biener <rguenther@suse.de>
4869 PR tree-optimization/81827
4870 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
4871 (new_var_info): Initialize it conservatively.
4872 (get_call_vi): Mark register vars.
4873 (new_scalar_tmp_constraint_exp): Likewise.
4874 (handle_rhs_call): Likewise.
4875 (handle_const_call): Likewise.
4876 (create_function_info_for): Likewise.
4877 (solve_constraints): Sort varinfos to separate register from
4878 non-register vars to pack points-to solution bitmaps during
4881 2017-08-17 Marek Polacek <polacek@redhat.com>
4883 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
4885 2017-08-17 Richard Biener <rguenther@suse.de>
4887 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
4888 to true when overflow is undefined and we saturated the result.
4890 2017-08-17 Alan Modra <amodra@gmail.com>
4893 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
4894 Don't use store multiple if only one reg needs saving.
4895 (interesting_frame_related_regno): New function.
4896 (rs6000_frame_related): Don't emit frame info for regs that
4898 (rs6000_emit_epilogue): Likewise.
4900 2017-08-16 Nathan Sidwell <nathan@acm.org>
4902 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
4903 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
4904 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
4905 * tree.c (free_lang_data_in_type): Use else-if chain. Always
4906 clear TYPE_LANG_1. Remove obsolete member-function stripping.
4907 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
4908 (verify_type): Adjust for TYPE_BINFO move.
4909 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
4910 process TYPE_BINFO directly.
4911 (hash_tree): Likewise.
4912 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
4914 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
4917 2017-08-16 David Malcolm <dmalcolm@redhat.com>
4919 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
4921 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
4924 * config/i386/i386.md (*anddi_1_btr): Change predicates of
4925 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
4926 Add ix86_binary_operator_ok to insn constraint.
4927 (*iordi_1_bts): Ditto.
4928 (*xordi_1_btc): Ditto.
4929 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
4930 Update corresponding peephole2 pattern.
4934 2017-08-16 Bin Cheng <bin.cheng@arm.com>
4936 PR tree-optimization/81832
4937 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
4938 copy loop header which has IFN_LOOP_DIST_ALIAS call.
4940 2017-08-16 Marek Polacek <polacek@redhat.com>
4943 * fold-const.c (fold_indirect_ref_1): Restore original behavior
4944 regarding size_zero_node.
4946 2017-08-16 Martin Liska <mliska@suse.cz>
4949 * config.gcc: Respect previously set extra_objs in case
4952 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
4954 PR tree-optimization/81835
4955 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
4956 the phi SSA_NAME. Check that the condition in a COND_EXPR does
4957 not depend on the phi.
4959 2017-08-16 Alan Modra <amodra@gmail.com>
4961 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
4964 2017-08-16 Alan Modra <amodra@gmail.com>
4966 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
4967 (save_reg_p): ..into this. Update all callers.
4968 (first_reg_to_save): Simplify.
4970 2017-08-16 Alan Modra <amodra@gmail.com>
4972 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
4975 2017-08-15 Joseph Myers <joseph@codesourcery.com>
4979 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
4980 constant count if that count is less than 32.
4982 2017-08-15 Nathan Sidwell <nathan@acm.org>
4984 * gcc.c (execute): Emit friendlier message if inferior is killed
4985 by an external cause.
4987 2017-08-15 Richard Biener <rguenther@suse.de>
4989 PR tree-optimization/81790
4990 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
4991 CONSTRUCTORs from simplifying and VN.
4993 2017-08-14 Martin Sebor <msebor@redhat.com>
4995 * builtin-attrs.def: Add comments.
4997 2017-08-14 Martin Sebor <msebor@redhat.com>
5000 * doc/extend.texi (attribute nonstring): Document new attribute.
5002 2017-08-14 Martin Sebor <msebor@redhat.com>
5005 * tree-diagnostic.c (default_tree_printer): Handle %G.
5006 * gimple-pretty-print.h (percent_G_format): Declare new function.
5007 * gimple-pretty-print.c (percent_G_format): Define.
5008 * tree-pretty-print.c (percent_K_format): Add argument.
5010 2017-08-14 Martin Sebor <msebor@redhat.com>
5012 PR translation/79998
5013 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
5014 Remove a stray space.
5016 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
5019 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
5020 (*iordi_1_bts): Ditto.
5021 (*xordi_1_btc): Ditto.
5023 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5026 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
5027 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5029 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
5030 quoted strings, and make more translator-friendly.
5031 (darwin_rs6000_override_options): Likewise.
5032 (rs6000_option_override_internal): Likewise.
5033 (rs6000_return_in_memory): Fix overlong line.
5034 (init_cmulative_args): Use quoted strings, and make more
5035 translator-friendly.
5036 (rs6000_pass_by_reference): Fix overlong line.
5037 (def_builtin): Use quoted strings.
5038 (altivec_expand_predicate_builtin): Use quoted strings, and make
5039 more translator-friendly.
5040 (htm_expand_builtin): Use quoted strings.
5041 (cpu_expand_builtin): Use quoted strings, and make more
5042 translator-friendly.
5043 (altivec_expand_builtin): Likewise.
5044 (paired_expand_predicate_builtin): Likewise.
5045 (rs6000_invalid_builtin): Likewise.
5046 (builtin_function_type): Use quoted strings.
5047 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
5048 more translator-friendly.
5049 (rs6000_trampoline_init): Likewise.
5050 (rs6000_handle_altivec_attribute): Likewise.
5051 (rs6000_inner_target_options): Use quoted strings.
5052 (rs6000_disable_incompatible_switches): Likewise.
5053 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
5054 strings, and make more translator-friendly.
5055 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
5057 2017-08-14 Bin Cheng <bin.cheng@arm.com>
5059 PR tree-optimization/81799
5060 * tree-loop-distribution.c (version_loop_by_alias_check): Force
5061 cond_expr to simple gimple operand.
5063 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
5066 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
5068 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
5072 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
5073 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
5074 TARGET_GASISR_PROLOGUES.
5075 * config/avr/avr.c (avr_option_override): Same.
5076 (avr_pass_pre_proep::execute): Same.
5078 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
5081 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
5082 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
5084 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
5086 * config/i386/i386.md (*load_tp_<mode>): Redefine as
5087 define_insn_and_split. Split to a memory load from 0 in
5088 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
5089 using PTR mode iterator.
5090 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
5091 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
5092 (*add_tp_<mode>): Redefine as define_insn_and_split.
5093 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
5094 address space. Merge with *add_tp_x32 using PTR mode iterator.
5095 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
5096 Split to an add with a memory load from 0 in
5097 DEFAULT_TLS_SEG_REG address space.
5099 2017-08-12 Andrew Pinski <apinski@cavium.com>
5101 * config/aarch64/aarch64-option-extensions.def (rdma):
5102 Fix feature string to what Linux prints out in /proc/cpuinfo.
5104 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
5107 * dwarf2out.c (modified_type_die): For C typedef types that have
5108 an ultimate origin, process the ultimate origin instead of the
5110 (gen_typedef_die): Assert that input DECLs have no ultimate
5112 (gen_type_die_with_usage): For typedef variants that have an
5113 ultimate origin, just call gen_decl_die on the original DECL.
5114 (process_scope_var): Avoid creating DIEs for local typedefs and
5115 concrete static variables.
5117 2017-08-12 Alan Modra <amodra@gmail.com>
5121 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
5122 match gnu-user.h startfile.
5123 (ENDFILE_LINUX_SPEC): Similarly.
5125 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
5128 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
5130 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
5132 2017-08-11 Tamar Christina <tamar.christina@arm.com>
5133 * config/aarch64/aarch64.md (mov<mode>): Change.
5134 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
5135 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
5136 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
5138 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
5140 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
5141 for storage order barriers.
5143 2017-08-11 Martin Liska <mliska@suse.cz>
5146 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
5147 variables of void type.
5149 2017-08-11 Martin Liska <mliska@suse.cz>
5151 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
5152 TARGET_SUPPORTS_ALIASES.
5153 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
5154 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
5155 (optimize_weakref): Likewise.
5156 * symtab.c (symtab_node::noninterposable_alias): Likewise.
5157 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
5158 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
5160 2017-08-11 Martin Liska <mliska@suse.cz>
5163 * config/i386/i386.c (make_resolver_func): Do complete
5164 refactoring of the function.
5166 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5169 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
5170 * config/i386/i386.c (ix86_stack_protect_guard): Use
5171 ix86_stack_protect_guard_symbol_str to generate varible declaration.
5172 * doc/invoke.texi (x86 Options): Document
5173 -mstack-protector-guard-symbol= option.
5175 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
5177 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
5178 * config/i386/i386.c (ix86_split_stack_guard): New function.
5179 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
5180 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
5181 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
5182 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
5183 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
5184 (split_stack_space_check): Call ix86_split_stack_guard.
5186 2017-08-10 Martin Sebor <msebor@redhat.com>
5188 * print-tree.c (print_node): Print location using the established
5190 Replace spaces with colons.
5191 (debug_raw, debug): Ditto.
5193 2017-08-10 Martin Sebor <msebor@redhat.com>
5196 * pretty-print.c (pp_format): Correct the handling of %s precision.
5198 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
5201 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5203 (ix86_finalize_stack_frame_flags): This. Also clear
5204 frame_pointer_needed if -fno-omit-frame-pointer is used without
5206 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5207 with ix86_finalize_stack_frame_flags.
5208 (ix86_expand_epilogue): Likewise.
5209 (ix86_expand_split_stack_prologue): Likewise.
5210 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
5212 2017-08-10 Martin Liska <mliska@suse.cz>
5215 * c-attribs.c (handle_target_attribute):
5216 Report warning for an empty string argument of target attribute.
5218 2017-08-09 Jakub Jelinek <jakub@redhat.com>
5221 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
5223 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
5224 or DECL_NONLOCAL labels.
5225 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
5226 or DECL_NONLOCAL labels here.
5228 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
5230 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
5231 to indicate when early gimple folding has been disabled.
5232 (rs6000_gimple_fold_builtin): Add debug content.
5233 (rs6000_invalid_builtin): Fix whitespace.
5234 (rs6000_expand_builtin): Fix whitespace.
5235 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
5237 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
5240 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
5241 SAVE_MULTIPLE if not all the registers that saves, should be saved.
5243 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
5245 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
5246 (qdf24xx): Likewise.
5247 * config/aarch64/aarch64.md: Include falkor.md.
5248 * config/aarch64/falkor.md: New.
5250 2017-08-09 Marek Polacek <polacek@redhat.com>
5253 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
5254 * diagnostic.c (emit_diagnostic): Add a comment.
5255 (emit_diagnostic_valist): New function.
5257 2017-08-09 Marek Polacek <polacek@redhat.com>
5260 * input.c (make_location): New overload.
5261 * input.h (make_location): Declare.
5263 2017-08-08 Alan Modra <amodra@gmail.com>
5264 H.J. Lu <hongjiu.lu@intel.com>
5267 * gcc.c (NO_PIE_SPEC): Delete.
5268 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
5270 (LINK_PIE_SPEC): ..to here.
5271 (LINK_COMMAND_SPEC): Support -no-pie.
5272 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
5273 chain of crtbegin*.o selection, update for PIE_SPEC changes and
5275 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
5276 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
5277 (ENDFILE_CRTEND_SPEC): Similarly.
5279 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
5282 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
5283 (mstack-protector-guard-offset=): Ditto.
5284 * config/i386/i386.c (ix86_option_override): Handle
5285 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
5287 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
5288 ix86_stack_protect_guard_offset variables.
5289 (TARGET_STACK_PROTECT_GUARD): Always define.
5290 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
5291 and -mstack-protector-guard-offset= options.
5293 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5295 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
5296 boundary case for the last candidate.
5298 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5300 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
5301 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
5303 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5306 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
5307 * config/aarch64/aarch64-builtins.c
5308 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
5309 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
5310 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
5312 2017-08-08 Tamar Christina <tamar.christina@arm.com>
5313 Andrew Pinski <pinskia@gmail.com>
5316 * internal-fn.def (XORSIGN): New.
5317 * optabs.def (xorsign_optab): New.
5318 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
5319 (convert_expand_mult_copysign): New.
5320 (pass_optimize_widening_mul::execute): Call
5321 convert_expand_mult_copysign.
5323 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5325 PR tree-optimization/81354
5326 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
5327 Insert on edges rather than explicitly creating landing pads.
5328 (analyze_candidates_and_replace): Commit edge inserts.
5330 2017-08-08 Richard Biener <rguenther@suse.de>
5333 * tree-ssa-loop-niter.c: Include tree-dfa.h.
5334 (expand_simple_operations): Also look through ADDR_EXPRs with
5335 MEM_REF bases treating them as POINTER_PLUS_EXPR.
5337 2017-08-08 Richard Biener <rguenther@suse.de>
5339 PR tree-optimization/81723
5340 * tree-vect-slp.c (struct bst_traits): New hash traits.
5341 (bst_fail): New global.
5342 (vect_build_slp_tree_2): New worker, split out from ...
5343 (vect_build_slp_tree): ... this now wrapping it with using
5344 bst_fail set to cache SLP tree build fails. Properly handle
5346 (vect_analyze_slp_instance): Allocate and free bst_fail.
5348 2017-08-08 Martin Liska <mliska@suse.cz>
5351 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
5352 LABEL_DECLs that can be from a different function.
5354 2017-08-08 Bin Cheng <bin.cheng@arm.com>
5356 PR tree-optimization/81744
5357 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
5358 loop's number of iterations.
5360 2017-08-08 Martin Liska <mliska@suse.cz>
5362 * asan.c: Include header files.
5363 * attribs.c (build_decl_attribute_variant): New function moved
5365 (build_type_attribute_qual_variant): Likewise.
5366 (cmp_attrib_identifiers): Likewise.
5367 (simple_cst_list_equal): Likewise.
5368 (omp_declare_simd_clauses_equal): Likewise.
5369 (attribute_value_equal): Likewise.
5370 (comp_type_attributes): Likewise.
5371 (build_type_attribute_variant): Likewise.
5372 (lookup_ident_attribute): Likewise.
5373 (remove_attribute): Likewise.
5374 (merge_attributes): Likewise.
5375 (merge_type_attributes): Likewise.
5376 (merge_decl_attributes): Likewise.
5377 (merge_dllimport_decl_attributes): Likewise.
5378 (handle_dll_attribute): Likewise.
5379 (attribute_list_equal): Likewise.
5380 (attribute_list_contained): Likewise.
5381 * attribs.h (lookup_attribute): New function moved from tree.[ch].
5382 (lookup_attribute_by_prefix): Likewise.
5383 * bb-reorder.c: Include header files.
5384 * builtins.c: Likewise.
5385 * calls.c: Likewise.
5386 * cfgexpand.c: Likewise.
5387 * cgraph.c: Likewise.
5388 * cgraphunit.c: Likewise.
5389 * convert.c: Likewise.
5390 * dwarf2out.c: Likewise.
5391 * final.c: Likewise.
5392 * fold-const.c: Likewise.
5393 * function.c: Likewise.
5394 * gimple-expr.c: Likewise.
5395 * gimple-fold.c: Likewise.
5396 * gimple-pretty-print.c: Likewise.
5397 * gimple.c: Likewise.
5398 * gimplify.c: Likewise.
5399 * hsa-common.c: Likewise.
5400 * hsa-gen.c: Likewise.
5401 * internal-fn.c: Likewise.
5402 * ipa-chkp.c: Likewise.
5403 * ipa-cp.c: Likewise.
5404 * ipa-devirt.c: Likewise.
5405 * ipa-fnsummary.c: Likewise.
5406 * ipa-inline.c: Likewise.
5407 * ipa-visibility.c: Likewise.
5409 * lto-cgraph.c: Likewise.
5410 * omp-expand.c: Likewise.
5411 * omp-general.c: Likewise.
5412 * omp-low.c: Likewise.
5413 * omp-offload.c: Likewise.
5414 * omp-simd-clone.c: Likewise.
5415 * opts-global.c: Likewise.
5416 * passes.c: Likewise.
5417 * predict.c: Likewise.
5418 * sancov.c: Likewise.
5419 * sanopt.c: Likewise.
5420 * symtab.c: Likewise.
5421 * toplev.c: Likewise.
5422 * trans-mem.c: Likewise.
5423 * tree-chkp.c: Likewise.
5424 * tree-eh.c: Likewise.
5425 * tree-into-ssa.c: Likewise.
5426 * tree-object-size.c: Likewise.
5427 * tree-parloops.c: Likewise.
5428 * tree-profile.c: Likewise.
5429 * tree-ssa-ccp.c: Likewise.
5430 * tree-ssa-live.c: Likewise.
5431 * tree-ssa-loop.c: Likewise.
5432 * tree-ssa-sccvn.c: Likewise.
5433 * tree-ssa-structalias.c: Likewise.
5434 * tree-ssa.c: Likewise.
5435 * tree-streamer-in.c: Likewise.
5436 * tree-vectorizer.c: Likewise.
5437 * tree-vrp.c: Likewise.
5439 * ubsan.c: Likewise.
5440 * varasm.c: Likewise.
5441 * varpool.c: Likewise.
5442 * tree.c: Remove functions moved to attribs.[ch].
5444 * config/aarch64/aarch64.c: Add attrs.h header file.
5445 * config/alpha/alpha.c: Likewise.
5446 * config/arc/arc.c: Likewise.
5447 * config/arm/arm.c: Likewise.
5448 * config/avr/avr.c: Likewise.
5449 * config/bfin/bfin.c: Likewise.
5450 * config/c6x/c6x.c: Likewise.
5451 * config/cr16/cr16.c: Likewise.
5452 * config/cris/cris.c: Likewise.
5453 * config/darwin.c: Likewise.
5454 * config/epiphany/epiphany.c: Likewise.
5455 * config/fr30/fr30.c: Likewise.
5456 * config/frv/frv.c: Likewise.
5457 * config/ft32/ft32.c: Likewise.
5458 * config/h8300/h8300.c: Likewise.
5459 * config/i386/winnt.c: Likewise.
5460 * config/ia64/ia64.c: Likewise.
5461 * config/iq2000/iq2000.c: Likewise.
5462 * config/lm32/lm32.c: Likewise.
5463 * config/m32c/m32c.c: Likewise.
5464 * config/m32r/m32r.c: Likewise.
5465 * config/m68k/m68k.c: Likewise.
5466 * config/mcore/mcore.c: Likewise.
5467 * config/microblaze/microblaze.c: Likewise.
5468 * config/mips/mips.c: Likewise.
5469 * config/mmix/mmix.c: Likewise.
5470 * config/mn10300/mn10300.c: Likewise.
5471 * config/moxie/moxie.c: Likewise.
5472 * config/msp430/msp430.c: Likewise.
5473 * config/nds32/nds32-isr.c: Likewise.
5474 * config/nds32/nds32.c: Likewise.
5475 * config/nios2/nios2.c: Likewise.
5476 * config/nvptx/nvptx.c: Likewise.
5477 * config/pa/pa.c: Likewise.
5478 * config/pdp11/pdp11.c: Likewise.
5479 * config/powerpcspe/powerpcspe.c: Likewise.
5480 * config/riscv/riscv.c: Likewise.
5481 * config/rl78/rl78.c: Likewise.
5482 * config/rx/rx.c: Likewise.
5483 * config/s390/s390.c: Likewise.
5484 * config/sh/sh.c: Likewise.
5485 * config/sol2.c: Likewise.
5486 * config/sparc/sparc.c: Likewise.
5487 * config/spu/spu.c: Likewise.
5488 * config/stormy16/stormy16.c: Likewise.
5489 * config/tilegx/tilegx.c: Likewise.
5490 * config/tilepro/tilepro.c: Likewise.
5491 * config/v850/v850.c: Likewise.
5492 * config/vax/vax.c: Likewise.
5493 * config/visium/visium.c: Likewise.
5494 * config/xtensa/xtensa.c: Likewise.
5496 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5499 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
5500 constraints since the -mupper-regs-* switches have been
5502 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
5503 into a vector from a double word element that was extracted from
5504 another vector, and eliminate extra XXPERMDI instructions.
5505 (vsx_concat_<mode>_2): Likewise.
5506 (vsx_concat_<mode>_3): Likewise.
5507 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
5508 concat to allow optimizing inserts from previous extracts.
5510 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
5512 * config/i386/i386.c (ix86_stack_protect_guard): Generate
5513 memory reference to a SSP offset in TLS address space.
5514 (ix86_print_operand) <case '@'>: Remove.
5515 (ix86_print_operand_punct_valid_p): Remove '@' code.
5516 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
5518 (stack_tls_protect_set_<mode>): Remove.
5519 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
5520 (stack_tls_protect_test_<mode>): Remove.
5521 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
5523 2017-08-07 Olivier Hainque <hainque@adacore.com>
5526 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
5528 2017-08-07 Douglas Rupp <rupp@adacore.com>
5530 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
5531 variable was referenced as multidir in command.
5533 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5536 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
5539 2017-08-07 Martin Liska <mliska@suse.cz>
5541 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
5542 * config/rl78/rl78.c: Add include of attribs.h.
5543 * config/sh/sh.c: Likewise.
5544 * config/v850/v850.c: Likewise.
5546 2017-08-07 Tom de Vries <tom@codesourcery.com>
5549 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
5551 2017-08-07 Martin Liska <mliska@suse.cz>
5553 * config/mips/mips.c: Include attribs.h.
5555 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
5558 * doc/invoke.texi: Document change in behvaior for -Ofast for
5561 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
5563 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
5565 (aarch64_pop_regs): Likewise.
5566 (aarch64_gen_load_pair): Likewise.
5567 (aarch64_save_callee_saves): Likewise.
5568 (aarch64_restore_callee_saves): Likewise.
5570 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5572 * config/i386/i386.c: Revert the last change.
5574 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5577 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
5579 (ix86_finalize_stack_frame_flags): This. Also clear
5580 frame_pointer_needed if -fno-omit-frame-pointer is used without
5582 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
5583 with ix86_finalize_stack_frame_flags.
5584 (ix86_expand_epilogue): Likewise.
5585 (ix86_expand_split_stack_prologue): Likewise.
5587 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
5590 * config/i386/i386.c (get_builtin_code_for_version): Set priority
5591 to P_AES for Westmere.
5593 2017-08-07 Jonathan Yong <10walls@gmail.com>
5595 * config/i386/mingw.opt (fset-stack-executable): Removed.
5596 * config/i386/cygming.opt (fset-stack-executable): Moved
5598 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
5600 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
5602 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
5604 2017-08-07 Marek Polacek <polacek@redhat.com>
5607 * fold-const.c (fold_indirect_ref_1): Check type_domain.
5609 2017-08-07 Martin Liska <mliska@suse.cz>
5611 * attribs.h (canonicalize_attr_name): New function.
5612 (cmp_attribs): Move from c-format.c and adjusted.
5613 (is_attribute_p): Moved from tree.h.
5614 * tree-inline.c: Add new includes.
5615 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
5616 (private_is_attribute_p): Remove.
5617 (private_lookup_attribute): Likewise.
5618 (private_lookup_attribute_by_prefix): Simplify.
5619 (remove_attribute): Use is_attribute_p.
5620 * tree.h: Remove removed declarations.
5622 2017-08-07 Jakub Jelinek <jakub@redhat.com>
5625 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
5626 instead of computing it in the function. Formatting fix.
5627 (expand_case): Don't rely on default_edge being the first edge,
5628 clear it if removing it, pass default_edge to
5629 emit_case_dispatch_table.
5630 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
5633 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
5635 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
5636 insn in the function, emit NOP after the insn.
5638 2017-08-06 Tom de Vries <tom@codesourcery.com>
5640 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
5643 2017-08-06 Tom de Vries <tom@codesourcery.com>
5645 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
5648 2017-08-04 Yury Gribov <tetra2005@gmail.com>
5650 PR tree-optimization/57371
5651 * match.pd: New pattern.
5653 2017-08-04 Marek Polacek <polacek@redhat.com>
5656 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
5657 perform the computation in offset_int.
5659 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5661 PR tree-optimization/81136
5662 * tree-vectorizer.h: Include tree-hash-traits.h.
5663 (vec_base_alignments): New typedef.
5664 (vec_info): Add a base_alignments field.
5665 (vect_record_base_alignments): Declare.
5666 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
5668 (DR_IS_CONDITIONAL_IN_STMT): New macro.
5669 (create_data_ref): Add an is_conditional_in_stmt argument.
5670 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
5671 the is_conditional_in_stmt field.
5672 (data_ref_loc): Add an is_conditional_in_stmt field.
5673 (get_references_in_stmt): Set the is_conditional_in_stmt field.
5674 (find_data_references_in_stmt): Update call to create_data_ref.
5675 (graphite_find_data_references_in_stmt): Likewise.
5676 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
5677 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
5678 (vect_record_base_alignment): New function.
5679 (vect_record_base_alignments): Likewise.
5680 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
5681 for nested statements even if we fail to compute a misalignment.
5682 Use pooled base alignments for unconditional references.
5683 (vect_find_same_alignment_drs): Compare base addresses instead
5685 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
5686 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
5688 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5690 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
5691 Add an explicit name for the enum. Use auto_vec for slp_instances
5693 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
5695 (_bb_vec_info): Add a constructor and destructor.
5696 (vinfo_for_stmt): Return NULL for uids of -1 as well.
5697 (destroy_loop_vec_info): Delete.
5698 (vect_destroy_datarefs): Likewise.
5699 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
5700 (vec_info::vec_info): New function.
5701 (vec_info::~vec_info): Likewise.
5702 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
5703 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
5704 destroy_loop_vec_info.
5705 * tree-vect-loop.c (new_loop_vec_info): Replace with...
5706 (_loop_vec_info::_loop_vec_info): ...this.
5707 (destroy_loop_vec_info): Replace with...
5708 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
5709 the stmt_vec_infos. Leave handling of vec_info information to its
5710 destructor. Remove explicit vector releases.
5711 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
5712 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
5713 * tree-vect-slp.c (new_bb_vec_info): Replace with...
5714 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
5715 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
5716 (destroy_bb_vec_info): Replace with...
5717 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
5718 information to its destructor.
5719 (vect_slp_analyze_bb_1): Use new and delete instead of
5720 new_bb_vec_info and destroy_bb_vec_info.
5721 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
5724 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5726 * tree-data-ref.h (subscript): Add access_fn field.
5727 (data_dependence_relation): Add could_be_independent_p.
5728 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
5729 (same_access_functions): Move to tree-data-ref.c.
5730 * tree-data-ref.c (ref_contains_union_access_p): New function.
5731 (access_fn_component_p): Likewise.
5732 (access_fn_components_comparable_p): Likewise.
5733 (dr_analyze_indices): Add a reference to access_fn_component_p.
5734 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
5736 (constant_access_functions): Likewise.
5737 (add_other_self_distances): Likewise.
5738 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
5739 (initialize_data_dependence_relation): Use XCNEW and remove
5740 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
5741 of access functions that have the same type. Allow the
5742 subsequence to end with different bases in some circumstances.
5743 Record the chosen access functions in SUB_ACCESS_FN.
5744 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
5745 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
5746 (subscript_dependence_tester_1): Likewise dra and drb.
5747 (build_classic_dist_vector): Update calls accordingly.
5748 (subscript_dependence_tester): Likewise.
5749 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
5750 DDR_COULD_BE_INDEPENDENT_P.
5751 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
5752 comp_alias_ddrs instead of may_alias_ddrs.
5753 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
5755 (vect_analyze_data_ref_dependence): Use it if
5756 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
5757 distance vectors if that fails.
5758 (dependence_distance_ge_vf): New function.
5759 (vect_prune_runtime_alias_test_list): Use it. Don't clear
5760 LOOP_VINFO_MAY_ALIAS_DDRS.
5762 2017-08-04 Richard Biener <rguenther@suse.de>
5765 * fold-const.c (fold_binary_loc): Properly restrict
5766 minus_var0 && minus_var1 case when associating undefined overflow
5769 2017-08-04 Tom de Vries <tom@codesourcery.com>
5771 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
5773 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5775 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5776 Don't start diagnostic messages with a capital letter.
5777 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5779 (rs6000_invalid_builtin): Likewise.
5780 (rs6000_trampoline_init): Likewise.
5782 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5785 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
5786 after setting changeable df flags.
5788 2017-08-03 Richard Biener <rguenther@suse.de>
5790 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
5791 up if the use is in USE - X.
5793 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
5795 * toplev.c (dumpfile.h): New include.
5796 (internal_error_reentered): New static function. Use it...
5797 (internal_error_function): ...here to handle reentered internal_error.
5799 2017-08-03 Richard Biener <rguenther@suse.de>
5802 * fold-const.c (split_tree): Add minus_var and minus_con
5803 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
5804 here but always use minus_*.
5805 (associate_trees): Assert we never associate with MINUS_EXPR
5806 and NULL first operand. Do not recurse for PLUS_EXPR operands
5807 when associating as MINUS_EXPR either.
5808 (fold_binary_loc): Track minus_var and minus_con.
5810 2017-08-03 Tom de Vries <tom@codesourcery.com>
5813 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
5814 ACCEL_COMPILER, apply finish_options on
5815 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
5817 2017-08-03 Tom de Vries <tom@codesourcery.com>
5820 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
5821 function_entry_patch_area_size > 0.
5823 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5826 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
5827 instead of 10??LU, perform unit multiplication in wide_int,
5828 don't change alloc_object_size_limit if the limit is larger
5831 PR tree-optimization/81655
5832 PR tree-optimization/81588
5833 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
5834 the case when ranges[i].low and high are 1 for unsigned type with
5838 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
5839 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
5841 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5843 * tree-vrp.h: Add include guard.
5845 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
5848 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
5849 (ud2): New insn pattern.
5850 * config/i386/i386.c (ix86_expand_epilogue):
5851 For naked functions, generate ud2 instead of trap insn.
5853 2017-08-02 Marek Polacek <polacek@redhat.com>
5856 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
5858 2017-08-02 Tom de Vries <tom@codesourcery.com>
5859 Cesar Philippidis <cesar@codesourcery.com>
5861 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
5862 Add missing edge probabilities.
5864 2017-08-02 Tamar Christina <tamar.christina@arm.com>
5866 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
5869 2017-08-02 Jakub Jelinek <jakub@redhat.com>
5872 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
5873 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
5874 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
5875 prologue_seq sequences - if any.
5877 2017-08-02 Richard Biener <rguenther@suse.de>
5879 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
5880 via vectors if supported, integer extracts via punning if supported
5881 or otherwise vector extracts.
5883 2017-08-02 Richard Biener <rguenther@suse.de>
5885 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
5887 (bitmap_insert_into_set): ... this.
5889 2017-08-02 Richard Biener <rguenther@suse.de>
5891 PR tree-optimization/81633
5893 2015-08-17 Alan Hayward <alan.hayward@arm.com>
5895 PR tree-optimization/71752
5896 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
5898 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
5900 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
5901 (machine_function::call_ms2sysv_pad_out): Remove field.
5902 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
5903 (ix86_compute_frame_layout): Likewise.
5905 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
5908 * config/i386/i386.c (ix86_set_func_type): Disallow naked
5909 attribute with interrupt attribute.
5911 2017-08-01 Andrew Pinski <apinski@cavium.com>
5913 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
5914 BIT_INSERT_EXPR's operand 1
5915 to see if the types precision matches.
5917 2017-08-01 Martin Liska <mliska@suse.cz>
5920 * builtins.c (expand_builtin_memcpy_args): Remove.
5921 (expand_builtin_memcpy): Call newly added function
5922 expand_builtin_memory_copy_args.
5923 (expand_builtin_memcpy_with_bounds): Likewise.
5924 (expand_builtin_mempcpy): Remove last argument.
5925 (expand_builtin_mempcpy_with_bounds): Likewise.
5926 (expand_builtin_memory_copy_args): New function created from
5927 expand_builtin_mempcpy_args with small modifications.
5928 (expand_builtin_mempcpy_args): Remove.
5929 (expand_builtin_stpcpy): Remove unused argument.
5930 (expand_builtin): Likewise.
5931 (expand_builtin_with_bounds): Likewise.
5933 2017-08-01 Martin Liska <mliska@suse.cz>
5936 Make mempcpy more optimal (PR middle-end/70140).
5938 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5941 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5942 __builtin_vec_cmpne verify both arguments are compatible vectors
5943 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
5944 verify arg1_type is a pointer or array type. For __builtin_vec_st,
5945 move computation of aligned to after checking the argument types.
5949 * config/rs6000/vsx.md (vextract_fp_from_shorth,
5950 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
5953 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
5954 Doug Rupp <rupp@adacore.com>
5955 Olivier Hainque <hainque@adacore.com>
5957 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
5958 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
5960 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
5961 for TARGET_OS_CPP_BUILTIN.
5962 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
5963 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
5965 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
5966 passing required abi options to the assembler for EABI configurations.
5967 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
5968 of .text.hot and .text.unlikely sections for kernel modules when
5969 using ARM style exceptions.
5970 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
5971 options. Add EXTRA_CC1_SPEC.
5972 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
5974 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
5976 (ARM_TARGET2_DWARF_FORMAT): Define.
5977 * config/arm/t-vxworks: Adjust multilib control to removal of the
5978 Diab command line options.
5980 2017-08-01 Martin Liska <mliska@suse.cz>
5982 PR gcov-profile/81561
5983 * gcov.c (unblock): Make unblocking safe as we need to preserve
5984 index correspondence of blocks and block_lists.
5986 2017-08-01 Richard Biener <rguenther@suse.de>
5988 PR tree-optimization/81181
5989 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5990 (compute_antic): ... end of iteration here.
5992 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
5994 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
5995 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
5996 (ftree-slp-vectorize): Likewise.
5997 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
5998 can no longer be set independent of flag_tree_loop_vectorize.
5999 * omp-general.c (emp_max_vf): Likewise.
6000 * opts.c (enable_fdo_optimizations): Remove references to
6001 flag_tree_vectorize, these are now implicit.
6002 (common_handle_option): Remove handling for OPT_ftree_vectorize,
6003 and leave it for the options machinery.
6005 2017-08-01 Martin Liska <mliska@suse.cz>
6008 * builtins.c (expand_builtin_memcpy_args): Remove.
6009 (expand_builtin_memcpy): Call newly added function
6010 expand_builtin_memory_copy_args.
6011 (expand_builtin_memcpy_with_bounds): Likewise.
6012 (expand_builtin_mempcpy): Remove last argument.
6013 (expand_builtin_mempcpy_with_bounds): Likewise.
6014 (expand_builtin_memory_copy_args): New function created from
6015 expand_builtin_mempcpy_args with small modifications.
6016 (expand_builtin_mempcpy_args): Remove.
6017 (expand_builtin_stpcpy): Remove unused argument.
6018 (expand_builtin): Likewise.
6019 (expand_builtin_with_bounds): Likewise.
6021 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6024 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
6025 print "ds:" only for immediates in generic address space.
6027 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
6030 * config/i386/i386.c (ix86_funciton_naked): New prototype.
6031 (ix86_function_ok_for_sibcall): Return false for naked functions.
6033 2017-08-01 Richard Biener <rguenther@suse.de>
6035 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
6036 (compute_antic): Seed worklist with exit block predecessors.
6037 * cfganal.c (dfs_find_deadend): For a cycle return the source
6038 of the edge closing it.
6040 2017-08-01 Tamar Christina <tamar.christina@arm.com>
6042 * config/aarch64/aarch64.c
6043 (aarch64_can_const_movi_rtx_p): Move 0 check.
6045 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6047 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
6048 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
6050 * match.pd: Ditto in address comparison pattern.
6052 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6054 PR tree-optimization/81627
6055 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
6056 closed ssa form for store-store chain.
6058 2017-08-01 Bin Cheng <bin.cheng@arm.com>
6060 PR tree-optimization/81620
6061 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
6062 for store-store chain.
6064 2017-08-01 Jakub Jelinek <jakub@redhat.com>
6066 PR tree-optimization/81588
6067 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
6068 ranges[i].in_p, invert comparison code ccode. For >/>=,
6069 swap rhs1 and rhs2 and comparison code unconditionally,
6070 for </<= don't do that. Don't swap rhs1/rhs2 again if
6071 ranges[i].in_p, instead invert comparison code ccode if
6072 opcode or oe->rank is BIT_IOR_EXPR.
6075 * optabs.def (vec_extract_optab, vec_init_optab): Change from
6076 a direct optab to conversion optab.
6077 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
6078 with GET_MODE_INNER as last argument instead of optab_handler.
6079 * expmed.c (extract_bit_field_1): Likewise. Use vector from
6080 vector extraction if possible and optab is available.
6081 * expr.c (store_constructor): Use convert_optab_handler instead
6082 of optab_handler. Use vector initialization from smaller
6083 vectors if possible and optab is available.
6084 * tree-vect-stmts.c (vectorizable_load): Likewise.
6085 * doc/md.texi (vec_extract, vec_init): Document that the optabs
6087 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
6088 of vec_init from half-sized vectors with the same element mode.
6089 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
6090 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
6091 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
6092 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
6093 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
6094 after mode in gen_vec_extract* calls.
6095 (vec_extract<mode>): Renamed to ...
6096 (vec_extract<mode><ssescalarmodelower>): ... this.
6097 (vec_extract<mode><ssehalfvecmodelower>): New expander.
6098 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
6099 element mode after mode in gen_vec_init* calls.
6100 (VEC_INIT_HALF_MODE): New mode iterator.
6101 (vec_init<mode>): Renamed to ...
6102 (vec_init<mode><ssescalarmodelower>): ... this.
6103 (vec_init<mode><ssehalfvecmodelower>): New expander.
6104 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
6105 (vec_extractv2sfsf): ... this.
6106 (vec_initv2sf): Renamed to ...
6107 (vec_initv2sfsf): ... this.
6108 (vec_extractv2si): Renamed to ...
6109 (vec_extractv2sisi): ... this.
6110 (vec_initv2si): Renamed to ...
6111 (vec_initv2sisi): ... this.
6112 (vec_extractv4hi): Renamed to ...
6113 (vec_extractv4hihi): ... this.
6114 (vec_initv4hi): Renamed to ...
6115 (vec_initv4hihi): ... this.
6116 (vec_extractv8qi): Renamed to ...
6117 (vec_extractv8qiqi): ... this.
6118 (vec_initv8qi): Renamed to ...
6119 (vec_initv8qiqi): ... this.
6120 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
6121 (vec_init<mode>): Renamed to ...
6122 (vec_init<mode><VEC_base_l>): ... this.
6123 (vec_extract<mode>): Renamed to ...
6124 (vec_extract<mode><VEC_base_l>): ... this.
6125 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
6126 (vec_initv2sfsf): ... this.
6127 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
6128 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6129 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
6130 element mode after mode in gen_vec_init* calls.
6131 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
6132 (vec_init<mode><Vel>): ... this.
6133 (vec_extract<mode>): Renamed to ...
6134 (vec_extract<mode><Vel>): ... this.
6135 * config/aarch64/iterators.md (Vel): New mode attribute.
6136 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
6137 Add element mode after mode in gen_vec_extract* calls.
6138 * config/s390/vector.md (non_vec_l): New mode attribute.
6139 (vec_extract<mode>): Renamed to ...
6140 (vec_extract<mode><non_vec_l>): ... this.
6141 (vec_init<mode>): Renamed to ...
6142 (vec_init<mode><non_vec_l>): ... this.
6143 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
6144 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
6146 * config/arm/iterators.md (V_elem_l): New mode attribute.
6147 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
6148 (vec_extract<mode><V_elem_l>): ... this.
6149 (vec_extractv2di): Renamed to ...
6150 (vec_extractv2didi): ... this.
6151 (vec_init<mode>): Renamed to ...
6152 (vec_init<mode><V_elem_l>): ... this.
6153 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
6154 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
6155 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
6156 Add element mode after gen_vec_extract* calls.
6157 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
6158 (vec_init<mode><unitmode>): ... this.
6159 (vec_extract<mode>): Renamed to ...
6160 (vec_extract<mode><unitmode>): ... this.
6161 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
6162 (vec_init<mode><unitmode>): ... this.
6163 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
6164 (vec_initv2sfsf): ... this.
6165 (vec_extractv2sf): Renamed to ...
6166 (vec_extractv2sfsf): ... this.
6167 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
6168 Add element mode after gen_vec_extract* calls.
6169 * config/mips/mips.md (unitmode): New mode iterator.
6170 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
6171 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
6172 * config/spu/spu.md (inner_l): New mode attribute.
6173 (vec_init<mode>): Renamed to ...
6174 (vec_init<mode><inner_l>): ... this.
6175 (vec_extract<mode>): Renamed to ...
6176 (vec_extract<mode><inner_l>): ... this.
6177 * config/sparc/sparc.md (veltmode): New mode iterator.
6178 (vec_init<VMALL:mode>): Renamed to ...
6179 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
6180 * config/ia64/vect.md (vec_initv2si): Renamed to ...
6181 (vec_initv2sisi): ... this.
6182 (vec_initv2sf): Renamed to ...
6183 (vec_initv2sfsf): ... this.
6184 (vec_extractv2sf): Renamed to ...
6185 (vec_extractv2sfsf): ... this.
6186 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
6187 (vec_init<mode>): Renamed to ...
6188 (vec_init<mode><VEC_base_l>): ... this.
6189 (vec_extract<mode>): Renamed to ...
6190 (vec_extract<mode><VEC_base_l>): ... this.
6191 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
6192 (vec_initv2sfsf): ... this.
6193 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
6194 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
6195 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
6196 gen_vec_init* calls.
6198 2017-08-01 Richard Biener <rguenther@suse.de>
6200 PR tree-optimization/81297
6201 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
6202 TREE_OVERFLOW from INTEGER_CSTs.
6204 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
6206 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
6208 2017-07-31 Carl Love <cel@us.ibm.com>
6210 * config/rs6000/rs6000-c: Add support for built-in functions
6211 vector signed char vec_xl_be (signed long long, signed char *);
6212 vector unsigned char vec_xl_be (signed long long, unsigned char *);
6213 vector signed int vec_xl_be (signed long long, signed int *);
6214 vector unsigned int vec_xl_be (signed long long, unsigned int *);
6215 vector signed long long vec_xl_be (signed long long, signed long long *);
6216 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
6217 vector signed short vec_xl_be (signed long long, signed short *);
6218 vector unsigned short vec_xl_be (signed long long, unsigned short *);
6219 vector double vec_xl_be (signed long long, double *);
6220 vector float vec_xl_be (signed long long, float *);
6221 * config/rs6000/altivec.h (vec_xl_be): Add #define.
6222 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
6223 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
6225 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
6226 (altivec_expand_builtin): Add switch statement to call
6227 altivec_expand_xl_be for each builtin.
6228 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
6229 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
6230 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
6231 __builtin_vsx_le_be_v16qi.
6232 * doc/extend.texi: Update the built-in documentation file for the
6233 new built-in functions.
6235 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6238 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
6240 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
6242 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6244 * config.gcc: Add z14.
6245 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
6246 CPU model numbers for z13s and z14.
6247 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
6249 * config/s390/s390-opts.h (enum processor_type): Rename
6250 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6251 * config/s390/s390.c (processor_table): Add field for CPU name to
6252 be passed to Binutils.
6253 (s390_asm_output_machine_for_arch): Use the new field in
6254 processor_table for Binutils.
6255 (s390_expand_builtin): Replace arch12 with z14.
6256 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
6257 (s390_get_sched_attrmask): Likewise.
6258 (s390_get_unit_mask): Likewise.
6259 * config/s390/s390.opt: Add z14 to processor_type enum.
6261 2017-07-31 Martin Jambor <mjambor@suse.cz>
6264 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
6265 regardless of optimization level.
6267 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
6268 Martin Liska <mliska@suse.cz>
6270 * predict.def: Remove old comment and adjust probability.
6271 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
6274 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
6277 * config/i386/i386.c (ix86_function_naked): New function.
6278 (ix86_can_use_return_insn_p): Return false for naked functions.
6279 (ix86_expand_prologue): Skip prologue for naked functions.
6280 (ix86_expand_epilogue): Skip epilogue for naked functions
6281 and emit trap instruction.
6282 (ix86_warn_func_return): New function.
6283 (ix86_attribute_table): Add "naked" attribute specification.
6284 (TARGET_WARN_FUNC_RETURN): Define.
6285 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
6287 2017-07-31 Martin Liska <mliska@suse.cz>
6289 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
6290 (dump_gimple_bb_header): Always dump BB info.
6291 (pp_cfg_jump): Do not append info about BB when dumping a jump.
6293 2017-07-31 Martin Liska <mliska@suse.cz>
6296 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
6297 also with current_function_decl non-null equality.
6299 2017-07-31 Jakub Jelinek <jakub@redhat.com>
6302 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
6303 change type to the element type, instead add eltype variable and
6304 use it where we are interested in the element type.
6306 PR tree-optimization/81603
6307 * ipa-polymorphic-call.c
6308 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
6309 offset arithmetic in offset_int, bail out if the resulting bit offset
6310 doesn't fit into shwi.
6312 2017-07-31 Martin Liska <mliska@suse.cz>
6314 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
6315 (gimplify_save_expr): Fix comment.
6317 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
6320 * config/i386/i386.c (ix86_function_arg): Update arguments for
6322 (ix86_compute_frame_layout): Set the initial stack offset to
6323 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
6324 INCOMING_FRAME_SP_OFFSET.
6325 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
6326 stack before exception handler returns.
6327 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
6328 the 'ERROR_CODE' for exception handler.
6330 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
6332 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
6333 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
6334 (ASM_OUTPUT_REG_POP): Ditto.
6335 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
6336 instead of asm_fprintf to output pure string.
6338 2017-07-29 Jakub Jelinek <jakub@redhat.com>
6340 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
6341 to imported_module_or_decl hook.
6342 (debug_nothing_tree_tree_tree_bool): Remove.
6343 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
6344 * debug.c (do_nothing_debug_hooks): Use
6345 debug_nothing_tree_tree_tree_bool_bool instead of
6346 debug_nothing_tree_tree_tree_bool.
6347 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
6348 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
6349 * sdbout.c (sdb_debug_hooks): Likewise.
6350 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
6351 (gen_namespace_die): Add DW_AT_export_symbols attribute if
6353 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
6354 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
6355 attribute, don't add anything.
6357 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6359 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
6360 (fold_build2_stat_loc): Likewise.
6361 (fold_build3_stat_loc): Likewise.
6362 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
6363 (fold_build1_loc): Remove macro.
6364 (fold_build2_loc): Likewise.
6365 (fold_build3_loc): Likewise.
6367 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6369 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
6370 (gimple_build_debug_bind_source_stat): Likewise.
6371 * gimple.h (gimple_build_debug_bind): Remove macro.
6372 (gimple_build_debug_bind_source): Likewise.
6374 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6376 * bitmap.c (bitmap_alloc): Adjust.
6377 (bitmap_gc_alloc): Likewise.
6378 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
6380 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6382 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
6383 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
6384 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
6385 (bitmap_gc_alloc_stat): Likewise.
6386 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
6388 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6390 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
6391 * rtl.h (shallow_copy_rtx): Remove macro.
6393 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6395 * emit-rtl.c (gen_raw_REG): Adjust.
6396 * gengenrtl.c (gendef): Likewise.
6397 * rtl.c (rtx_alloc_stat): Remove _stat from name.
6398 * rtl.h (rtx_alloc): Remove macro.
6400 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6402 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
6403 (build_tree_list_stat): Likewise.
6404 * tree.h (build_tree_list): Remove macro.
6405 (build_tree_list_vec): Likewise.
6407 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6409 * tree.c (make_vector_stat): Remove _stat from name.
6410 (build_vector_stat): Likewise.
6411 * tree.h (make_vector_stat): Remove macro.
6412 (build_vector_stat): Likewise.
6414 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6416 * tree.h (build_var_debug_value): Remove prototype.
6418 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6420 * tree.c (tree_cons_stat): Remove _stat from name.
6421 * tree.h (tree_cons): Remove macro.
6423 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6425 * tree.c (build_vl_exp_stat): Remove _stat from name.
6426 * tree.h (build_vl_exp): Remove macro.
6428 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6430 * tree.c (build_decl_stat): Remove _stat from name.
6431 * tree.h (build_decl): Remove macro.
6433 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6435 * gimple.c (gimple_build_with_ops_stat): Adjust.
6436 (gimple_alloc_stat): Remove _stat from name.
6437 * gimple.h (gimple_alloc): Remove macro.
6439 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6441 * tree.c (make_tree_vec_stat): Remove _stat from name.
6442 (grow_tree_vec_stat): Likewise.
6443 * tree.h (make_tree_vec_stat): Adjust prototype.
6444 (grow_tree_vec_stat): Likewise.
6445 (make_tree_vec): Remove macro.
6446 (grow_tree_vec): Likewise.
6448 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6450 * fold-const.c (fold_build1_stat_loc): Adjust.
6451 (fold_build2_stat_loc): Likewise.
6452 (fold_build3_stat_loc): Likewise.
6453 * tree.c (build0_stat): Remove _stat from name.
6454 (build1_stat): Likewise.
6455 (build2_stat): Likewise.
6456 (build3_stat): Likewise.
6457 (build4_stat): Likewise.
6458 (build5_stat): Likewise.
6459 * tree.h (build1_loc): Remove macro, and rename _stat function
6461 (build2_loc): Likewise.
6462 (build3_loc): Likewise.
6463 (build4_loc): Likewise.
6464 (build5_loc): Likewise.
6466 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6468 * tree.c (make_int_cst_stat): Remove _stat from name.
6469 * tree.h (make_int_cst_stat): Adjust prototype.
6470 (make_int_cst): Remove macro.
6472 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6474 * tree.c (make_tre_binfo_stat): Remove _stat from name.
6475 * tree.h (make_tree_binfo_stat): Adjust prototype.
6476 (make_tree_binfo): Remove.
6478 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6480 * tree.c (copy_node_stat): Rename to copy_node.
6481 (build_distinct_type_copy): Adjust.
6482 * tree.h (copy_node_stat): Adjust prototype.
6483 (copy_node): Remove macro.
6485 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6487 * tree.c (make_node_stat): rename to make_node.
6488 (build_tree_list_stat): Adjust.
6489 (build0_stat): Likewise.
6490 (build2_stat): Likewise.
6491 (build3_stat): Likewise.
6492 (build4_stat): Likewise.
6493 (build5_stat): Likewise.
6494 (build_decl_stat): Likewise.
6495 * tree.h (make_node_stat): Adjust prototype.
6496 (make_node): remove macro.
6498 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
6500 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
6501 (PPC_FEATURE2_SCV): Likewise.
6502 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
6504 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6506 * config/aarch64/aarch64.c
6507 (aarch64_internal_mov_immediate): Add new special pattern.
6508 * config/aarch64/aarch64.md (*movdi_aarch64):
6509 Add reg/32bit const mov case.
6511 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6512 Richard Sandiford <richard.sandiford@linaro.org>
6514 * config/aarch64/aarch64.md (mov<mode>): Generalize.
6515 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
6516 Add integer and movi cases.
6517 (movi-split-hf-df-sf split, fp16): New.
6518 (enabled): Added TARGET_FP_F16INST.
6519 * config/aarch64/iterators.md (GPF_HF): New.
6520 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
6522 2017-07-28 Tamar Christina <tamar.christina@arm.com>
6524 * config/aarch64/aarch64.c
6525 (aarch64_simd_container_mode): Add prototype.
6526 (aarch64_expand_mov_immediate): Add HI support.
6527 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
6528 (aarch64_can_const_movi_rtx_p): New.
6529 (aarch64_preferred_reload_class):
6530 Remove restrictions of using FP registers for certain SIMD operations.
6531 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
6532 (aarch64_valid_floating_const): Add integer move validation.
6533 (aarch64_simd_imm_scalar_p): Remove.
6534 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
6535 (aarch64_legitimate_constant_p): Expand list of supported cases.
6536 * config/aarch64/aarch64-protos.h
6537 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
6538 (aarch64_reinterpret_float_as_int): New.
6539 (aarch64_simd_imm_scalar_p): Remove.
6540 * config/aarch64/constraints.md (Uvi): New.
6541 (Dd): Split into Ds and new Dd.
6542 * config/aarch64/aarch64.md (*movsi_aarch64):
6544 (*movdi_aarch64): Add SIMD mov case.
6546 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6548 * tree-predcom.c: (struct chain): Handle store-store chain in which
6549 stores for elimination only store loop invariant values.
6550 (execute_pred_commoning_chain): Ditto.
6551 (prepare_initializers_chain_store_elim): Ditto.
6552 (prepare_finalizers): Ditto.
6553 (is_inv_store_elimination_chain): New function.
6554 (initialize_root_vars_store_elim_1): New function.
6556 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6558 * tree-predcom.c: Revise general description of the pass.
6559 (enum chain_type): New enum type for store elimination.
6560 (struct chain): New field supporting store elimination.
6561 (struct component): Ditto.
6562 (dump_chain): Dump store-stores chain.
6563 (release_chain): Release resources.
6564 (split_data_refs_to_components): Compute and create component
6565 contains only stores for elimination.
6566 (get_chain_last_ref_at): New function.
6567 (make_invariant_chain): Initialization.
6568 (make_rooted_chain): Specify chain type in parameter and record it.
6569 (add_looparound_copies): Skip for store-stores chain.
6570 (determine_roots_comp): Compute type of chain and pass it to
6572 (initialize_root_vars_store_elim_2): New function.
6573 (finalize_eliminated_stores): New function.
6574 (remove_stmt): Handle store for elimination.
6575 (execute_pred_commoning_chain): Execute predictive commoning on
6577 (determine_unroll_factor): Skip unroll for store-stores chain.
6578 (prepare_initializers_chain_store_elim): New function.
6579 (prepare_initializers_chain): Hanlde store-store chain.
6580 (prepare_finalizers_chain, prepare_finalizers): New function.
6581 (tree_predictive_commoning_loop): Return integer value indicating
6582 if loop is unrolled or lcssa form is corrupted.
6583 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
6585 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6587 * tree-predcom.c (initialize_root): Delete.
6588 (execute_pred_commoning_chain): Initialize root vars and replace
6589 reference of non-combined chain directly, rather than call above
6592 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6594 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
6595 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
6597 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6599 * tree-predcom.c (struct chain): New field init_seq.
6600 (release_chain): Release init_seq.
6601 (prepare_initializers_chain): Record intialization stmts in above
6603 (insert_init_seqs): New function.
6604 (tree_predictive_commoning_loop): Call insert_init_seqs.
6606 2017-07-28 Bin Cheng <bin.cheng@arm.com>
6608 * tree-predcom.c (determine_roots_comp): Skip trivial components.
6610 2017-07-28 Richard Biener <rguenther@suse.de>
6612 * match.pd: Remove superfluous :c.
6613 * genmatch.c (simplify::id): Add member.
6614 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
6616 (current_id): New global.
6617 (dt_node::parent): Move from ...
6618 (dt_operand::parent): ... here. Add for_id member.
6619 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
6620 (decision_tree::find_node): Relax order requirement when
6621 merging DT_TRUE nodes to ones inbetween the current simplify
6622 and the one we try to merge with. Add diagnostic whenever
6623 we need to enforce pattern order by not merging.
6624 (decision_tree::insert): Set current_id.
6625 (decision_tree::print_node): Dump parent node and for_id.
6626 (parser::last_id): Add member.
6627 (parser::push_simplify): Assign unique id.
6628 (parser::parser): Initialize last_id.
6630 2017-07-28 Martin Liska <mliska@suse.cz>
6633 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
6634 gimple_build_debug_bind.
6636 2017-07-28 Richard Biener <rguenther@suse.de>
6638 PR tree-optimization/81502
6639 * match.pd: Add pattern combining BIT_INSERT_EXPR with
6641 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
6643 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
6644 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
6645 for BIT_FIELD_REF args.
6646 * fold-const.c (make_bit_field_ref): Likewise.
6647 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6649 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6652 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
6653 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
6654 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
6655 Or it into SANITIZER_UNDEFINED.
6656 * ubsan.c: Include gimple-fold.h and varasm.h.
6657 (ubsan_expand_ptr_ifn): New function.
6658 (instrument_pointer_overflow): New function.
6659 (maybe_instrument_pointer_overflow): New function.
6660 (instrument_object_size): Formatting fix.
6661 (pass_ubsan::execute): Call instrument_pointer_overflow
6662 and maybe_instrument_pointer_overflow.
6663 * internal-fn.c (expand_UBSAN_PTR): New function.
6664 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
6665 * sanitizer.def (__ubsan_handle_pointer_overflow,
6666 __ubsan_handle_pointer_overflow_abort): New builtins.
6667 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
6668 * internal-fn.def (UBSAN_PTR): New internal function.
6669 * opts.c (sanitizer_opts): Add pointer-overflow.
6670 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
6671 * fold-const.c (build_range_check): Compute pointer range check in
6672 integral type if pointer arithmetics would be needed. Formatting
6675 2017-07-28 Martin Liska <mliska@suse.cz>
6678 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
6679 parameters that are of a variable-length.
6681 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
6683 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
6685 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
6686 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
6687 (CRT_CALL_STATIC_FUNCTION): Likewise.
6688 (ASM_DEFAULT_SPEC): New define.
6689 (ASM_SPEC32): Likewise.
6690 (ASM_SPEC64): Likewise.
6691 (ASM_SPEC_COMMON): Likewise.
6692 (ASM_SPEC): Likewise.
6693 (INVALID_64BIT): Likewise.
6694 (LINK_OS_DEFAULT_SPEC): Likewise.
6695 (LINK_OS_SPEC32): Likewise.
6696 (LINK_OS_SPEC64): Likewise.
6697 (POWERPC_LINUX): Likewise.
6698 (PTRDIFF_TYPE): Likewise.
6699 (RESTORE_FP_PREFIX): Likewise.
6700 (RESTORE_FP_SUFFIX): Likewise.
6701 (SAVE_FP_PREFIX): Likewise.
6702 (SAVE_FP_SUFFIX): Likewise.
6703 (SIZE_TYPE): Likewise.
6704 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
6705 (TARGET_64BIT): Likewise.
6706 (TARGET_64BIT): Likewise.
6707 (TARGET_AIX): Likewise.
6708 (WCHAR_TYPE_SIZE): Likewise.
6709 (WCHAR_TYPE): Undefine.
6710 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
6711 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
6712 (CPP_OS_RTEMS_SPEC): Delete.
6713 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
6714 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
6716 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
6718 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6720 PR tree-optimization/81578
6721 * tree-parloops.c (build_new_reduction): Bail out if
6722 reduction_code isn't one of the standard OpenMP reductions.
6723 Move the details printing after that decision.
6725 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6727 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
6728 related to reload_in_progress.
6729 (splat_input_operand): Likewise.
6730 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
6732 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
6734 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
6735 (TARGET_INSTANTIATE_DECLS): Likewise.
6736 (legitimate_indexed_address_p): Delete reload_in_progress code.
6737 (rs6000_debug_legitimate_address_p): Likewise.
6738 (rs6000_eliminate_indexed_memrefs): Likewise.
6739 (rs6000_emit_le_vsx_store): Likewise.
6740 (rs6000_emit_move_si_sf_subreg): Likewise.
6741 (rs6000_emit_move): Likewise.
6742 (register_to_reg_type): Likewise.
6743 (rs6000_pre_atomic_barrier): Likewise.
6744 (rs6000_machopic_legitimize_pic_address): Likewise.
6745 (rs6000_allocate_stack_temp): Likewise.
6746 (rs6000_address_for_fpconvert): Likewise.
6747 (rs6000_address_for_altivec): Likewise.
6748 (rs6000_secondary_memory_needed_rtx): Delete function.
6749 (rs6000_check_sdmode): Likewise.
6750 (rs6000_alloc_sdmode_stack_slot): Likewise.
6751 (rs6000_instantiate_decls): Likewise.
6752 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
6753 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
6754 Delete reload_in_progress.
6755 (*vec_reload_and_plus_<mptrsize>): Likewise.
6756 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
6757 (vsx_div_v2di): Likewise.
6758 (vsx_udiv_v2di): Likewise.
6760 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6762 * config/rs6000/rs6000.opt (mlra): Replace with stub.
6763 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
6764 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
6765 (rs6000_debug_reg_global): Delete print of LRA status.
6766 (rs6000_option_override_internal): Delete dead LRA related code.
6767 (rs6000_lra_p): Delete function.
6768 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
6770 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
6772 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
6773 * config/riscv/rtems.h: New file.
6775 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6776 Sudakshina Das <sudi.das@arm.com>
6778 * config/aarch64/aarch64.md
6779 (define_split for and<mode>3nr_compare): Move
6780 non aarch64_logical_operand to a register.
6781 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
6782 register immediate operand to a register.
6783 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
6785 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6788 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
6790 2017-07-27 Richard Biener <rguenther@suse.de>
6792 PR tree-optimization/81573
6793 PR tree-optimization/81494
6794 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
6795 multi defuse cycle case.
6797 2017-07-27 Richard Biener <rguenther@suse.de>
6799 PR tree-optimization/81571
6800 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
6803 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
6805 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
6806 earlier and only if MASK_FPU is set. Adjust formatting.
6808 2017-07-27 Martin Liska <mliska@suse.cz>
6810 * opt-functions.awk: Add validation of value of Init.
6811 * optc-gen.awk: Pass new argument.
6813 2017-07-27 Martin Liska <mliska@suse.cz>
6815 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
6816 Fix wrong condition.
6818 2017-07-27 Martin Liska <mliska@suse.cz>
6820 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
6821 BBs and edges seen by autoFDO.
6823 2017-07-27 Richard Biener <rguenther@suse.de>
6825 PR tree-optimization/81502
6826 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
6827 with incompatible but same sized type.
6828 (execute_update_addresses_taken): Likewise.
6830 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
6832 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
6833 flag_tree_loop_vectorize rather than flag_tree_vectorize.
6835 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6838 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
6839 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
6840 Change s_operand to memory_operand.
6842 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
6844 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
6845 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
6846 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
6847 Emit instructions rather than returning an expression. Handle TFmode
6848 and KFmode by casting to TImode.
6849 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
6850 (rs6000_emit_le_vsx_store): Likewise.
6851 * config/rs6000/vsx.md (VSX_TI): New iterator.
6852 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
6853 (*vsx_le_undo_permute_<mode>): Likewise.
6854 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
6855 emit the split sequence.
6856 (*vsx_le_perm_store_<mode>): Likewise.
6858 2017-07-27 Jakub Jelinek <jakub@redhat.com>
6860 PR tree-optimization/81555
6861 PR tree-optimization/81556
6862 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
6863 if true, force CHANGED for the recursive invocation.
6864 (reassociate_bb): Remember original length of ops array, pass
6865 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
6867 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
6868 attributes for noipa attribute. For naked attribute use
6869 lookup_attribute first before lookup_attribute_spec.
6870 * final.c (rest_of_handle_final): Disable IPA RA for functions with
6872 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
6873 for functions with noipa attribute.
6874 (cgraph_externally_visible_p): Return true for functions with noipa
6876 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
6877 for functions with noipa attribute.
6878 * doc/extend.texi: Document noipa function attribute.
6879 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
6880 also for functions with noipa attribute.
6881 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
6883 2017-07-26 Andrew Pinski <apinski@cavium.com>
6885 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
6886 vec_unalign_load_cost and vec_unalign_store_cost.
6888 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6890 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
6891 -mvsx-small-integer option.
6892 (ISA_3_0_MASKS_IEEE): Likewise.
6893 (OTHER_VSX_VECTOR_MASKS): Likewise.
6894 (POWERPC_MASKS): Likewise.
6895 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
6896 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
6897 code, only testing for DImode being allowed in non-VSX floating
6899 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
6900 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
6902 (rs6000_option_override_internal): Delete -mvsx-small-integer.
6903 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
6904 TARGET_P8_VECTOR test.
6905 (rs6000_secondary_reload_simple_move): Likewise.
6906 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
6907 since TARGET_P9_VECTOR was already tested.
6908 (rs6000_opt_masks): Remove -mvsx-small-integer.
6909 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
6910 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6912 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
6913 test for TARGET_VEXTRACTUB was used, and that uses
6915 (p9 extract splitter): Likewise.
6916 (vsx_extract_<mode>_di_p9): Likewise.
6917 (vsx_extract_<mode>_store_p9): Likewise.
6918 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
6919 for TARGET_P9_VECTOR was used. Delete code that is now dead with
6920 the elimination of TARGET_VSX_SMALL_INTEGER.
6921 (vsx_extract_<mode>_p8): Likewise.
6922 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
6923 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
6924 (vsx_set_<mode>_p9): Likewise.
6925 (vsx_set_v4sf_p9): Likewise.
6926 (vsx_set_v4sf_p9_zero): Likewise.
6927 (vsx_insert_extract_v4sf_p9): Likewise.
6928 (vsx_insert_extract_v4sf_p9_2): Likewise.
6929 * config/rs6000/rs6000.md (sign extend splitter): Change
6930 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
6931 (floatsi<mode>2_lfiwax_mem): Likewise.
6932 (floatunssi<mode>2_lfiwzx_mem): Likewise.
6933 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
6934 since a test for TARGET_P9_VECTOR was used.
6935 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6936 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
6937 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6938 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6939 TARGET_P8_VECTOR test.
6940 (fix_trunc<mode>si2_stfiwx): Likewise.
6941 (fix_trunc<mode>si2_internal): Likewise.
6942 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
6943 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6945 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6946 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6947 TARGET_P8_VECTOR test.
6948 (fixuns_trunc<mode>si2_stfiwx): Likewise.
6949 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
6950 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6952 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6953 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
6954 since a test for TARGET_P9_VECTOR was used.
6955 (splitter for loading small constants): Likewise.
6957 2017-07-26 Andrew Pinski <apinski@cavium.com>
6959 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
6962 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
6965 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
6966 (fp_valid_at): Likewise.
6968 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6970 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
6971 (qdf24xx_addrcost_table): Likewise.
6972 (cortexa57_tunings): Update to use generic_branch_cost.
6973 (cortexa72_tunings): Likewise.
6974 (cortexa73_tunings): Likewise.
6975 (qdf24xx_tunings): Likewise.
6977 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6979 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
6980 (thunderx2t99_branch_cost): Likewise.
6981 (cortexa35_tunings): Update to use generic_branch_cost.
6982 (cortexa53_tunings): Likewise.
6983 (cortexa57_tunings): Likewise.
6984 (cortexa72_tunings): Likewise.
6985 (cortexa73_tunings): Likewise.
6986 (thunderx2t99_tunings): Likewise.
6988 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6990 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
6991 (sparc_option_override): Honour MASK_FSMULD.
6992 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
6993 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
6994 * config/sparc/sparc.opt (mfsmuld): New option.
6995 * doc/invoke.texi (mfsmuld): Document option.
6997 2017-07-26 Marek Polacek <polacek@redhat.com>
7000 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
7002 2017-07-26 Richard Biener <rguenther@suse.de>
7004 * gimple-match-head.c (do_valueize): Return OP if valueize
7006 (get_def): New helper to get at the def stmt of a SSA name
7008 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
7009 do_valueize to get at the def stmt.
7010 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
7012 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
7015 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
7017 2017-07-26 Martin Liska <mliska@suse.cz>
7020 * function.c (expand_function_start): Make expansion of
7021 nonlocal_goto_save_area after parm_birth_insn.
7023 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
7025 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
7026 from all CPU target flags enable members.
7028 2017-07-26 Richard Biener <rguenther@suse.de>
7030 * genmatch.c (dt_simplify::gen): Make iterator vars const.
7031 (decision_tree::gen): Make 'type' const.
7032 (write_predicate): Likewise.
7034 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7036 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
7037 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
7038 (rs6000_option_override_internal): Likewise.
7039 (rs6000_expand_vector_set): Likewise.
7040 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
7041 (TARGET_UPPER_REGS_SF): Likewise.
7042 (TARGET_UPPER_REGS_DI): Likewise.
7043 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
7044 (TARGET_DIRECT_MOVE_64BIT): Likewise.
7045 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
7046 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
7047 (Splitters for DI constants in Altivec registers): Likewise.
7048 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
7049 (vsx_set_v4sf_p9): Likewise.
7050 (vsx_set_v4sf_p9_zero): Likewise.
7051 (vsx_insert_extract_v4sf_p9): Likewise.
7052 (vsx_insert_extract_v4sf_p9_2): Likewise.
7054 2017-07-25 Carl Love <cel@us.ibm.com>
7056 * doc/extend.texi: Update the built-in documentation file for the
7057 existing built-in functions
7058 vector signed char vec_cnttz (vector signed char);
7059 vector unsigned char vec_cnttz (vector unsigned char);
7060 vector signed short vec_cnttz (vector signed short);
7061 vector unsigned short vec_cnttz (vector unsigned short);
7062 vector signed int vec_cnttz (vector signed int);
7063 vector unsigned int vec_cnttz (vector unsigned int);
7064 vector signed long long vec_cnttz (vector signed long long);
7065 vector unsigned long long vec_cnttz (vector unsigned long long);
7067 2017-07-25 Andrew Pinski <apinski@cavium.com>
7069 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
7070 accesses where the use is for the first operand of a BIT_INSERT.
7072 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7075 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7076 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7078 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
7080 * config/i386/gstabs.h: Delete.
7081 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
7083 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
7085 * config/i386/i386.c (ix86_decompose_address): Do not check for
7086 register RTX when looking at index_reg or base_reg.
7087 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
7089 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
7091 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
7092 to update EH info here.
7094 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7096 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
7098 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7100 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
7102 2017-07-25 Torsten Duwe <duwe@suse.de>
7104 * common.opt: Introduce -fpatchable-function-entry
7105 command line option, and its variables function_entry_patch_area_size
7106 and function_entry_patch_area_start.
7107 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
7108 including a two-value parser.
7109 * target.def (print_patchable_function_entry): New target hook.
7110 * targhooks.h (default_print_patchable_function_entry): New function.
7111 * targhooks.c (default_print_patchable_function_entry): Likewise.
7112 * toplev.c (process_options): Switch off IPA-RA if
7113 patchable function entries are being generated.
7114 * varasm.c (assemble_start_function): Look at the
7115 patchable-function-entry command line switch and current
7116 function attributes and maybe generate NOP instructions by
7117 calling the print_patchable_function_entry hook.
7118 * doc/extend.texi: Document patchable_function_entry attribute.
7119 * doc/invoke.texi: Document -fpatchable_function_entry
7120 command line option.
7121 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
7123 * doc/tm.texi: Re-generate.
7125 2017-07-25 Jakub Jelinek <jakub@redhat.com>
7128 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
7129 TARGET_AVX512DQ rather than TARGET_AVX512BW.
7131 2017-07-25 Tamar Christina <tamar.christina@arm.com>
7133 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
7135 2017-07-25 Richard Biener <rguenther@suse.de>
7137 PR tree-optimization/81455
7138 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
7139 not walk in cycles when looking for guards.
7141 2017-07-25 Richard Biener <rguenther@suse.de>
7143 PR tree-optimization/81529
7144 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
7145 when optimizing backedge uses.
7147 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
7149 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
7151 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
7152 to dl_section_ref. On AIX, append an expression to subtract
7153 the size of the section length to dl_section_ref.
7155 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
7157 * configure.ac: If any of the config.* scripts fail, exit 1.
7158 * configure: Regenerate.
7160 2017-07-25 Richard Biener <rguenther@suse.de>
7163 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
7164 of immediate uses, be more verbose on errors.
7166 2017-07-25 Richard Biener <rguenther@suse.de>
7168 PR tree-optimization/81510
7169 * tree-vect-loop.c (vect_is_simple_reduction): When the
7170 reduction stmt is not inside the loop bail out.
7172 2017-07-25 Richard Biener <rguenther@suse.de>
7174 PR tree-optimization/81303
7175 * tree-vect-loop-manip.c (vect_loop_versioning): Build
7176 profitability check against LOOP_VINFO_NITERSM1.
7178 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
7180 * domwalk.c (cmp_bb_postorder): Simplify.
7181 (sort_bbs_postorder): New function. Use it...
7182 (dom_walker::walk): ...here to optimize common cases.
7184 2017-07-25 Martin Liska <mliska@suse.cz>
7187 * ipa-visibility.c (function_and_variable_visibility): Make the
7188 redirection just on target that supports aliasing.
7189 Fix GNU coding style.
7191 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
7194 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
7196 * config/arm/rtems.h: Likewise.
7197 * config/bfin/rtems.h: Likewise.
7198 * config/i386/rtemself.h: Likewise.
7199 * config/lm32/rtems.h: Likewise.
7200 * config/m32c/rtems.h: Likewise.
7201 * config/m68k/rtemself.h: Likewise.
7202 * config/microblaze/rtems.h: Likewise.
7203 * config/mips/rtems.h: Likewise.
7204 * config/moxie/rtems.h: Likewise.
7205 * config/nios2/rtems.h: Likewise.
7206 * config/powerpcspe/rtems.h: Likewise.
7207 * config/rs6000/rtems.h: Likewise.
7208 * config/rtems.h: Likewise.
7209 * config/sh/rtems.h: Likewise.
7210 * config/sh/rtemself.h: Likewise.
7211 * config/sparc/rtemself.h: Likewise.
7213 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7216 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
7217 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
7218 * tree-ssa-structalias.c (alias_get_name): Same.
7220 2017-07-25 Bin Cheng <bin.cheng@arm.com>
7223 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
7224 instructions if no du chain is found.
7226 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
7228 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
7230 2017-07-25 Richard Biener <rguenther@suse.de>
7233 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
7236 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
7238 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
7240 (ISA_2_7_MASKS_SERVER): Likewise.
7241 (ISA_3_0_MASKS_IEEE): Likewise.
7242 (OTHER_P8_VECTOR_MASKS): Likewise.
7243 (OTHER_VSX_VECTOR_MASKS): Likewise.
7244 (POWERPC_MASKS): Likewise.
7245 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
7246 duplicate list of options.
7247 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
7248 explicit -mupper-regs options.
7249 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
7250 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
7251 alias for -mupper-regs-df.
7252 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7253 (rs6000_init_hard_regno_mode_ok): Likewise.
7254 (rs6000_option_override_internal): Likewise.
7255 (rs6000_opt_masks): Likewise.
7256 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
7257 options in terms of whether -mvsx or -mpower8-vector was used.
7258 (TARGET_UPPER_REGS_DI): Likewise.
7259 (TARGET_UPPER_REGS_SF): Likewise.
7260 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
7261 -mupper-regs-* options.
7263 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7265 * passes.c (emergency_dump_function): Print some empty lines and a
7266 header before the RTL dump.
7268 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
7270 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
7272 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
7275 * config/aarch64/aarch64.c (aarch64_classify_symbol):
7276 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
7278 2017-07-24 Carl Love <cel@us.ibm.com>
7280 * config/rs6000/rs6000-c.c: Add support for built-in functions
7281 vector float vec_extract_fp32_from_shorth (vector unsigned short);
7282 vector float vec_extract_fp32_from_shortl (vector unsigned short);
7283 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
7284 vec_extract_fp_from_shortl): Add defines for the two builtins.
7285 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
7286 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
7288 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
7289 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
7290 * doc/extend.texi: Update the built-in documentation file for the
7291 new built-in function.
7293 2017-07-24 Jakub Jelinek <jakub@redhat.com>
7296 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
7298 * doc/generic.texi: Likewise.
7299 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
7300 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
7302 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
7304 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
7305 (aarch64_mls_elt_merge<mode>): Likewise.
7307 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
7309 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7310 having __cxa_atexit.
7312 2017-07-23 Michael Collison <michael.collison@arm.com>
7314 * config/arm/arm.c (arm_option_override): Deprecate
7315 use of -mstructure-size-boundary.
7316 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
7317 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
7319 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7322 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
7323 Reduce cost estimate for direct moves.
7325 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
7328 * config/i386/i386.c (ix86_option_override_internal): Disable
7329 BMI, BMI2 and TBM instructions for -m16.
7331 2017-07-21 Carl Love <cel@us.ibm.com>
7333 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7334 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7335 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7336 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7337 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7338 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7339 VMULOSW): New enum "unspec" values.
7340 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7341 altivec_vmulosw): New patterns.
7342 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7343 VMULOSW): Add definitions.
7345 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
7347 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
7348 (qdf24xx): Likewise.
7349 * config/aarch64/aarch64-options-extensions.def (rdma); New.
7350 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
7351 (AARCH64_FL_V8_1): Renumber.
7352 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
7353 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
7354 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
7355 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
7356 rdma to feature modifiers list.
7358 2017-07-21 Yury Gribov <tetra2005@gmail.com>
7361 * ipa-visibility (function_and_variable_visibility): Convert
7362 recursive PLT call to direct call if appropriate.
7364 2017-07-21 Andrew Pinski <apinski@cavium.com>
7366 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
7367 operand 1 to see if the types precision matches.
7368 * fold-const.c (operand_equal_p): Likewise.
7370 2017-07-21 Richard Biener <rguenther@suse.de>
7372 PR tree-optimization/81303
7373 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
7374 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
7375 (vect_peeling_hash_get_lowest_cost): Adjust.
7376 (vect_enhance_data_refs_alignment): Likewise. Use
7377 vect_get_peeling_costs_all_drs to compute the penalty for no
7378 peeling to match up costs.
7380 2017-07-21 Richard Biener <rguenther@suse.de>
7382 PR tree-optimization/81500
7383 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
7384 we didn't identify a reduction path.
7386 2017-07-21 Tom de Vries <tom@codesourcery.com>
7387 Cesar Philippidis <cesar@codesourcery.com>
7389 PR gcov-profile/81442
7390 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
7393 2017-07-21 Tom de Vries <tom@codesourcery.com>
7396 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
7398 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
7399 nvptx_override_options_after_change.
7401 2017-07-21 Ulrich Drepper <drepper@redhat.com>
7403 * dwarf2out.c (output_file_names): Avoid double testing for
7406 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
7408 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
7410 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7412 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
7414 (try_crossjump_to_edge): Do not punt on partitioned functions.
7416 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
7418 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
7419 Put all BBs reachable only via paths crossing cold region to cold
7421 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
7423 2016-07-21 Richard Biener <rguenther@suse.de>
7425 PR tree-optimization/81303
7426 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
7427 into account prologue and epilogue iterations when raising
7428 min_profitable_iters to sth at least covering one vector iteration.
7430 2017-07-21 Tamar Christina <tamar.christina@arm.com>
7432 * config/arm/arm.c (arm_test_cpu_arch_dat):
7435 2017-07-20 Nathan Sidwell <nathan@acm.org>
7437 Remove TYPE_METHODS.
7438 * tree.h (TYPE_METHODS): Delete.
7439 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
7440 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
7441 (dbxout_type_methods): Scan TYPE_FIELDS.
7442 (dbxout_type): Don't check TYPE_METHODS here.
7443 * function.c (use_register_for_decl): Always ignore register for
7444 class types when not optimizing.
7445 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
7446 * tree.c (free_lang_data_in_type): Stitch out member functions and
7447 templates from TYPE_FIELDS.
7448 (build_distinct_type_copy, verify_type_variant,
7449 verify_type): Member fns are on TYPE_FIELDS.
7450 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
7451 * tree-pretty-print.c (dump_generic_node): Likewise.
7453 2017-07-20 Jakub Jelinek <jakub@redhat.com>
7456 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
7457 V2TImode and V4TImode.
7458 (ix86_expand_vector_extract): Likewise.
7459 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
7460 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
7461 (ssescalarmode): Handle V4TImode and V2TImode.
7462 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
7463 (*vec_extractv2ti, *vec_extractv4ti): New insns.
7464 (VEXTRACTI128_MODE): New mode iterator.
7465 (splitter for *vec_extractv?ti first element): New.
7466 (VEC_INIT_MODE): New mode iterator.
7467 (vec_init<mode>): Consolidate 3 expanders into one using
7468 VEC_INIT_MODE mode iterator.
7470 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7472 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
7473 non_spilled_static_chain_regno_p.
7475 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
7477 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
7479 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
7481 * bb-reorder.c (connect_traces): Allow copying of blocks within
7484 2017-07-20 Richard Biener <rguenther@suse.de>
7486 * gimple.h (gimple_phi_result): Add gphi * overload.
7487 (gimple_phi_result_ptr): Likewise.
7488 (gimple_phi_arg): Likewise. Adjust index assert to only
7489 allow actual argument accesses rather than all slots available
7491 (gimple_phi_arg_def): Add gphi * overload.
7492 * tree-phinodes.c (make_phi_node): Initialize only actual
7494 (resize_phi_node): Clear memory not covered by old node,
7495 do not initialize excess argument slots.
7496 (reserve_phi_args_for_new_edge): Initialize new argument slot
7499 2017-07-20 Bin Cheng <bin.cheng@arm.com>
7501 PR tree-optimization/81388
7503 2016-07-21 Bin Cheng <bin.cheng@arm.com>
7505 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
7506 by removing computation of may_be_zero.
7508 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7509 Tom de Vries <tom@codesourcery.com>
7512 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
7513 when gimple level profile disagrees with what RTL expander did.
7515 2017-07-20 Richard Biener <rguenther@suse.de>
7517 PR tree-optimization/61171
7518 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
7519 (vect_analyze_stmt): Add slp instance parameter.
7520 (vectorizable_reduction): Likewise.
7521 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
7522 (vect_is_simple_reduction): Deal with chains not detected
7523 as SLP reduction chain, specifically not properly associated
7524 chains containing a mix of plus/minus.
7525 (get_reduction_op): Remove.
7526 (get_initial_defs_for_reduction): Simplify, pass in whether
7527 this is a reduction chain, pass in the SLP node for the PHIs.
7528 (vect_create_epilog_for_reduction): Get the SLP instance as
7530 (vectorizable_reduction): Get the SLP instance as arg.
7531 During analysis remember the SLP node with the PHIs in the
7532 instance. Simplify getting at the vectorized reduction PHIs.
7533 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
7534 through SLP instance.
7535 (vect_slp_analyze_operations): Likewise.
7536 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
7537 (vect_transform_stmt): Likewise.
7539 2017-07-20 Tom de Vries <tom@codesourcery.com>
7541 PR tree-optimization/81489
7542 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
7543 read of phi arg location to before loop that modifies phi.
7545 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7547 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
7550 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7553 * except.c (execute): Fix ordering issue.
7555 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7557 PR rtl-optimization/81423
7558 * combine.c (make_compound_operation_int): Don't try to optimize
7559 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
7561 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
7563 PR rtl-optimization/81423
7564 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
7565 with a constant that is -1 in the truncated to mode.
7567 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7569 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
7570 (determine_unlikely_bbs): ... here.
7571 * predict.h (propagate_unlikely_bbs_forward): Declare.
7572 * cfgexpand.c (pass_expand::execute): Use it.
7573 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
7575 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
7576 propagate_unlikely_bbs_forward.
7578 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
7581 * except.c (maybe_add_nop_after_section_switch): New function.
7584 2017-07-19 Tom de Vries <tom@codesourcery.com>
7586 * gimple.h (gimple_phi_set_arg): Make assert more strict.
7588 2017-07-19 Tom de Vries <tom@codesourcery.com>
7590 * gimple.h (gimple_phi_arg): Make assert more strict.
7592 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
7594 * config.gcc (powerpc*-*-*): Add mmintrin.h.
7595 * config/rs6000/mmintrin.h: New file.
7596 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
7598 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7600 PR tree-optimization/81346
7601 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
7603 2017-07-19 Tom de Vries <tom@codesourcery.com>
7605 * config/nvptx/nvptx.md (VECIM): Add V2DI.
7607 2017-07-19 Tom de Vries <tom@codesourcery.com>
7609 * config/nvptx/nvptx-modes.def: Add V2DImode.
7610 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
7611 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
7612 (nvptx_output_mov_insn): Handle lack of mov.b128.
7613 (nvptx_print_operand): Handle 'H' and 'L' codes.
7614 (nvptx_vector_mode_supported): Allow V2DImode.
7615 (nvptx_preferred_simd_mode): New function.
7616 (nvptx_data_alignment): New function.
7617 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
7618 nvptx_preferred_simd_mode.
7619 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
7621 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
7623 2017-07-19 Tom de Vries <tom@codesourcery.com>
7625 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
7626 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
7627 (nvptx_vector_mode_supported): New function. Allow V2SImode.
7628 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
7629 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
7630 (mov<VECIM>_insn): New define_insn.
7631 (define_expand "mov<VECIM>): New define_expand.
7633 2017-07-19 Tom de Vries <tom@codesourcery.com>
7635 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
7637 2017-07-19 Jakub Jelinek <jakub@redhat.com>
7639 PR tree-optimization/81346
7640 * fold-const.h (fold_div_compare, range_check_type): Declare.
7641 * fold-const.c (range_check_type): New function.
7642 (build_range_check): Use range_check_type.
7643 (fold_div_compare): No longer static, rewritten into
7644 a match.pd helper function.
7645 (fold_comparison): Don't call fold_div_compare here.
7646 * match.pd (X / C1 op C2): New optimization using fold_div_compare
7649 2017-07-19 Nathan Sidwell <nathan@acm.org>
7651 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
7652 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
7653 * tree.c (find_decls_types_r, verify_type): Use
7654 TYPE_{MIN,MAX}_VALUE_RAW.
7655 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
7656 (hash_tree): Likewise.
7657 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
7659 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
7662 2017-07-18 Tom de Vries <tom@codesourcery.com>
7665 * omp-expand.c (expand_omp_for_static_chunk): Handle
7666 equal-argument loop exit phi.
7668 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
7671 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
7672 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
7673 operand 2 predicate.
7674 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
7675 operand 2 predicate.
7676 (ror,rol -> rorx splitters): Use const_int_operand as
7677 operand 2 predicate.
7679 2017-06-18 Richard Biener <rguenther@suse.de>
7681 PR tree-optimization/81410
7682 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
7683 the gap in the ! slp_perm SLP case after each group.
7685 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7688 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
7691 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7694 * predict.c (set_even_probabilities): Cleanup; do not affect
7695 probabilities that are already known.
7696 (combine_predictions_for_bb): Call even when count is set.
7698 2017-07-18 Nathan Sidwell <nathan@acm.org>
7700 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
7703 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7706 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
7707 optimization for loop niter analysis.
7709 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
7712 * config/avr/avr.c (avr_optimize_casesi): Don't use
7713 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
7715 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7717 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
7718 body_cost_vec from _vect_peel_extended_info.
7719 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
7720 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
7723 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7725 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
7727 2017-07-18 Richard Biener <rguenther@suse.de>
7729 PR tree-optimization/80620
7730 PR tree-optimization/81403
7731 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
7732 info when re-using a VN table entry.
7734 2017-07-18 Richard Biener <rguenther@suse.de>
7736 PR tree-optimization/81418
7737 * tree-vect-loop.c (vectorizable_reduction): Properly compute
7738 vectype_in. Verify that with lane-reducing reduction operations
7739 we have a single def-use cycle.
7741 2017-07-17 Carl Love <cel@us.ibm.com>
7743 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
7745 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7746 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7747 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7748 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7749 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7750 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7751 VMULOSW): New enum "unspec" values.
7752 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
7753 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
7754 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7755 altivec_vmulosw): New patterns.
7756 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7757 VMULOSW): Add definitions.
7758 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
7760 * config/alpha/alpha.c: Include predict.h.
7762 2017-07-17 Yury Gribov <tetra2005@gmail.com>
7764 * tree-vrp.c (compare_assert_loc): Fix comparison function
7765 to return predictable results.
7767 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7769 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
7772 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
7773 * doc/invoke.texi (mexpand-adddi): Update text.
7775 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7777 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
7778 that also clobbers the CC register. The old expand code is moved
7780 (*arc_clzsi2): ... here.
7781 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
7782 the CC register. The old expand code is moved to ...
7783 (arc_ctzsi2): ... here.
7785 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7787 * config/arc/arc.opt (mindexed-loads): Use initial value
7788 TARGET_INDEXED_LOADS_DEFAULT.
7789 (mauto-modify-reg): Use initial value
7790 TARGET_AUTO_MODIFY_REG_DEFAULT.
7791 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7792 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7793 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7794 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7796 2017-07-17 Martin Liska <mliska@suse.cz>
7799 * opts.c (finish_options): Do not allow -fgnu-tm
7800 w/ -fsanitize={kernel-,}address. Say sorry.
7802 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7805 * tree-loop-distribution.c (classify_partition): Only assert on
7806 numer of iterations.
7807 (merge_dep_scc_partitions): Delete prameter. Update function call.
7808 (distribute_loop): Remove code handling loop with unknown niters.
7809 (pass_loop_distribution::execute): Skip loop with unknown niters.
7811 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7814 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
7815 function sort_partitions_by_post_order.
7817 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7819 PR tree-optimization/81374
7820 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
7821 the max index of basic blocks, rather than number of basic blocks.
7823 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7825 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
7827 (arc_legitimate_pic_operand_p): Likewise.
7828 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
7830 (arc_needs_pcl_p): Likewise.
7831 (arc_legitimate_pc_offset_p): Likewise.
7832 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
7833 function is also used in constrains.md.
7834 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
7835 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
7836 PLUS. Only return true/false in known cases, otherwise assert.
7837 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
7838 is already called in arc_legitimate_constant_p.
7839 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
7841 (LEGITIMATE_PIC_OPERAND_P): Use
7842 arc_raw_symbolic_reference_mentioned_p function.
7843 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
7848 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7849 Andrew Burgess <andrew.burgess@embecosm.com>
7851 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
7852 (arc_return_address_register): New function.
7853 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
7854 (arc_handle_fndecl_attribute): Add naked attribute.
7855 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7856 (TARGET_WARN_FUNC_RETURN): Likewise.
7857 (arc_allocate_stack_slots_for_args): New function.
7858 (arc_warn_func_return): Likewise.
7859 (machine_function): Change type fn_type.
7860 (arc_compute_function_type): Consider new naked function type,
7861 change function return type.
7862 (arc_must_save_register): Adapt to handle new
7863 arc_compute_function_type's return type.
7864 (arc_expand_prologue): Likewise.
7865 (arc_expand_epilogue): Likewise.
7866 (arc_return_address_regs): Delete.
7867 (arc_return_address_register): New function.
7868 (arc_epilogue_uses): Use above function.
7869 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
7870 (arc_function_type): Change encoding, add naked type.
7871 (ARC_INTERRUPT_P): Change to handle the new encoding.
7872 (ARC_FAST_INTERRUPT_P): Likewise.
7873 (ARC_NORMAL_P): Define.
7874 (ARC_NAKED_P): Likewise.
7875 (arc_compute_function_type): Delete prototype.
7876 * config/arc/arc.md (in_ret_delay_slot): Use
7877 arc_return_address_register function.
7878 (simple_return): Likewise.
7879 (p_return_i): Likewise.
7881 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7883 PR tree-optimization/81428
7884 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
7885 can't be built for those types.
7887 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7889 Remove stuff dead since r239246.
7891 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
7892 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
7893 (avr_inform_devices): Remove dead stuff.
7895 2017-07-17 Tamar Christina <tamar.christina@arm.com>
7897 * config/arm/arm_neon.h: Fix softp typo.
7899 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7901 PR tree-optimization/81365
7902 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
7903 aggregate moves onto bb predecessor edges, make sure there are no
7904 loads that could alias the lhs in between the start of bb and the
7907 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7910 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
7911 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
7912 [LSHIFTRT, outer_code = TRUNCATE]: Same.
7914 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7916 PR tree-optimization/81396
7917 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
7918 (init_symbolic_number): Initialize it to 1.
7919 (perform_symbolic_merge): Add n_ops from both operands into the new
7921 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
7922 without base_addr as useless if they need more than one operation.
7923 (bswap_replace): Handle !bswap case for NULL base_addr.
7925 2017-07-17 Tom de Vries <tom@codesourcery.com>
7928 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
7931 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
7933 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
7934 conditional builtin define __FIX_LEON3FT_B2BST.
7936 2017-07-17 Daniel Cederman <cederman@gaisler.com>
7938 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
7939 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
7942 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7944 PR rtl-optimization/81424
7945 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
7946 to remove potential trapping from operands if -fnon-call-exceptions.
7948 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7950 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
7951 profile_proability for scalling.
7952 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
7954 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7956 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
7958 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7960 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
7961 fixpoint arithmetics.
7963 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7965 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
7966 fixpoint arithmetics.
7968 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7970 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
7971 fixpoint arithmetics.
7973 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7975 * profile-count.h (profile_probability::from_reg_br_prob_note,
7976 profile_probability::to_reg_br_prob_note): New functions.
7977 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
7978 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
7979 * predict.c (probability_reliable_p): Update.
7980 (edge_probability_reliable_p): Update.
7981 (br_prob_note_reliable_p): Update.
7982 (invert_br_probabilities): Update.
7983 (add_reg_br_prob_note): New function.
7984 (combine_predictions_for_insn): Update.
7985 * asan.c (asan_clear_shadow): Update.
7986 * cfgbuild.c (compute_outgoing_frequencies): Update.
7987 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
7988 (update_br_prob_note): Update.
7989 (rtl_verify_edges): Update.
7990 (purge_dead_edges): Update.
7991 (fixup_reorder_chain): Update.
7992 * emit-rtl.c (try_split): Update.
7993 * ifcvt.c (cond_exec_process_insns): Update.
7994 (cond_exec_process_if_block): Update.
7995 (dead_or_predicable): Update.
7996 * internal-fn.c (expand_addsub_overflow): Update.
7997 (expand_neg_overflow): Update.
7998 (expand_mul_overflow): Update.
7999 * loop-doloop.c (doloop_modify): Update.
8000 * loop-unroll.c (compare_and_jump_seq): Update.
8001 * optabs.c (emit_cmp_and_jump_insn_1): Update.
8002 * predict.h: Update.
8003 * reorg.c (mostly_true_jump): Update.
8005 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
8006 * config/alpha/alpha.c (emit_unlikely_jump): Update.
8007 * config/arc/arc.c: (emit_unlikely_jump): Update.
8008 * config/arm/arm.c: (emit_unlikely_jump): Update.
8009 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
8010 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
8011 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
8012 (ix86_print_operand): Update.
8013 (ix86_split_fp_branch): Update.
8014 (predict_jump): Update.
8015 * config/ia64/ia64.c (ia64_print_operand): Update.
8016 * config/mmix/mmix.c (mmix_print_operand): Update.
8017 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
8018 (rs6000_expand_split_stack_prologue): Update.
8019 * config/rs6000/rs6000.c: Update.
8020 * config/s390/s390.c (s390_expand_vec_strlen): Update.
8021 (s390_expand_vec_movstr): Update.
8022 (s390_expand_cs_tdsi): Update.
8023 (s390_expand_split_stack_prologue): Update.
8024 * config/sh/sh.c (sh_print_operand): Update.
8025 (expand_cbranchsi4): Update.
8026 (expand_cbranchdi4): Update.
8027 * config/sparc/sparc.c (output_v9branch): Update.
8028 * config/spu/spu.c (get_branch_target): Update.
8029 (ea_load_store_inline): Update.
8030 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
8031 * config/tilepro/tilepro.c: Update.
8033 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
8035 * gimplify.c (mostly_copy_tree_r): Revert latest change.
8036 (gimplify_save_expr): Likewise.
8038 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8040 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
8042 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8044 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
8046 * timevar.def (TV_IPA_FNSUMMARY): Define.
8048 2017-07-16 Daniel Cederman <cederman@gaisler.com>
8050 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
8051 to back store errata sensitive sequence from being generated.
8052 (sqrtdf2_fix): Likewise.
8054 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8056 * tree-ssa-threadupdate.c (compute_path_counts,
8057 update_joiner_offpath_counts): Use profile_probability.
8059 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
8062 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8064 * config/arm/arm-c.c (arm_cpu_builtins): Define
8065 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8067 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
8069 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8070 array entries to represent __ieee128 versions of the
8071 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
8072 scalar_extract_sig, and scalar_insert_exp built-in functions.
8073 (altivec_resolve_overloaded_builtin): Add special case handling
8074 for the __builtin_scalar_insert_exp function, as represented by
8075 the P9V_BUILTIN_VEC_VSIEDP constant.
8076 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
8077 exponent support for __ieee128 argument.
8078 (VSESQP): Add scalar extract signature support for __ieee128
8080 (VSTDCNQP): Add scalar test negative support for __ieee128
8082 (VSIEQP): Add scalar insert exponent support for __int128 argument
8083 with __ieee128 result.
8084 (VSIEQPF): Add scalar insert exponent support for __ieee128
8085 argument with __ieee128 result.
8086 (VSTDCQP): Add scalar test data class support for __ieee128
8088 (VSTDCNQP): Add overload support for scalar test negative with
8090 (VSTDCQP): Add overload support for scalar test data class
8092 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
8094 (UNSPEC_VSX_SIEXPQP): New constant.
8095 (xsxexpqp): New insn for VSX scalar extract exponent quad
8097 (xsxsigqp): New insn for VSX scalar extract significand quad
8099 (xsiexpqpf): New insn for VSX scalar insert exponent quad
8100 precision with floating point argument.
8101 (xststdcqp): New expand for VSX scalar test data class quad
8103 (xststdcnegqp): New expand for VSX scalar test negative quad
8105 (xststdcqp): New insn to match expansions for VSX scalar test data
8106 class quad precision and VSX scalar test negative quad precision.
8107 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
8108 special case operand checking to enforce that second operand of
8109 VSX scalar test data class with quad precision argument is a 7-bit
8111 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
8112 prototypes and descriptions of __ieee128 versions of
8113 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
8114 scalar_test_data_class, and scalar_test_neg built-in functions.
8116 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8118 PR tree-optimization/81162
8119 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
8120 replace a negate with an add.
8122 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
8124 * doc/invoke.texi (arm/-mcpu): Document +crypto.
8126 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8128 * config/arm/arm-c.c (arm_cpu_builtins): Define
8129 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
8131 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8133 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
8134 (armv8-r): Set ARM Cortex-R52 as default CPU.
8135 * config/arm/arm-tables.opt: Regenerate.
8136 * config/arm/arm-tune.md: Regenerate.
8137 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
8139 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
8140 extension for -mcpu=cortex-r52.
8142 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
8144 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
8145 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
8146 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
8147 (fp-armv8): Define it as FP_ARMv8 only.
8148 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
8149 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
8151 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
8152 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
8153 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
8154 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
8155 than TARGET_FPU_ARMV8.
8156 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
8157 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
8158 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
8160 * config/arm/neon.md (neon_vrint): Likewise.
8161 (neon_vcvt): Likewise.
8162 (neon_<fmaxmin_op><mode>): Likewise.
8163 (<fmaxmin><mode>3): Likewise.
8164 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
8165 * config/arm/predicates.md (arm_cond_move_operator): Check against
8166 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
8168 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
8170 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
8173 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8175 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
8176 loop in comment with memset.
8178 2017-07-14 Martin Liska <mliska@suse.cz>
8180 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
8181 * dwarf2out.c (is_java): Remove the function.
8182 (output_pubname): Remove usage of the function.
8183 (lower_bound_default): Remove usage of DW_LANG_Java.
8184 (gen_compile_unit_die): Likewise.
8185 * gcc.c: Remove compiler defaults for .java and .zip files.
8186 * gimple-expr.c (remove_suffix): Change as there's no longer
8187 extension than 4-letter one.
8188 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
8189 (gimplify_save_expr): Likewise.
8190 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
8191 as it's possible even for other languages than Java.
8192 * langhooks.h (struct lang_hooks): Remove Java from a comment.
8193 * lto-opts.c (lto_write_options): Remove reference to Java.
8194 * opts.c (strip_off_ending): Update file extension handling.
8195 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
8196 * tree-eh.c (lower_resx): Likewise.
8197 * tree.c (free_lang_data_in_type): Remove dead code.
8198 (find_decls_types_r): Likewise.
8199 (build_common_builtin_nodes): Remove Java from a comment.
8200 (verify_type): Remove dead code.
8201 * varasm.c (assemble_external): Remove Java from a comment.
8203 2017-07-14 Martin Liska <mliska@suse.cz>
8205 * opts.c (finish_options): Add quotes.
8206 (common_handle_option): Likewise.
8208 2017-07-14 Martin Liska <mliska@suse.cz>
8210 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
8211 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
8213 * dwarf2out.c (lower_bound_default): Do not handle
8215 (gen_compile_unit_die): Likewise.
8216 * gcc.c: Remove default extension binding for GNU Pascal.
8217 * stmt.c: Remove Pascal language from a comment.
8218 * xcoffout.c: Likewise.
8220 2017-07-13 David Malcolm <dmalcolm@redhat.com>
8223 * diagnostic-show-locus.c (fixit_cmp): New function.
8224 (layout::layout): Sort m_fixit_hints.
8225 (column_range::column_range): Assert that the values are valid.
8226 (struct char_span): New struct.
8227 (correction::overwrite): New method.
8228 (struct source_line): New struct.
8229 (line_corrections::add_hint): Add assertions. Reimplement memcpy
8230 calls in terms of classes source_line and char_span, and
8231 correction::overwrite.
8232 (selftest::test_overlapped_fixit_printing_2): New function.
8233 (selftest::diagnostic_show_locus_c_tests): Call it.
8235 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
8237 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
8238 early if there is no lhs.
8240 2017-07-13 Martin Liska <mliska@suse.cz>
8242 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
8243 (gen_reference_type_die): Likewise.
8244 * stor-layout.c: Remove Pascal-related comment.
8246 2017-07-13 Martin Liska <mliska@suse.cz>
8248 * opts.c (finish_options): Add quotes to error messages.
8249 (parse_sanitizer_options): Likewise.
8251 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
8253 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
8255 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
8257 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
8259 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
8261 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
8263 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
8265 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8268 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
8269 provides the hardware capability bits, define the macro
8270 __BUILTIN_CPU_SUPPORTS__.
8271 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
8272 if GLIBC does not provide the hardware capability bits. Add a
8273 gcc_unreachable call if the built-in cpu function is neither
8274 __builtin_cpu_is nor __builtin_cpu_supports.
8275 (rs6000_get_function_versions_dispatcher): Change the warning
8276 that an old GLIBC is used which does not export the capability
8277 bits to be an error.
8278 * doc/extend.texi (target_clones attribute): Document the
8279 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
8280 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
8281 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
8282 the macros defined by GCC if the newer GLIBC is available.
8284 2017-07-12 Jeff Law <law@redhat.com>
8286 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
8287 remaining includes slightly.
8288 * config/riscv/riscv-builtins.c: Include profile-count.h.
8290 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8293 * config/avr/avr.c (avr_set_current_function): In diagnostic
8294 messages: Quote keywords and (parts of) identifiers.
8295 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
8298 2017-07-12 Carl Love <cel@us.ibm.com>
8300 * config/rs6000/rs6000-c.c: Add support for built-in functions
8301 vector bool char vec_revb (vector bool char);
8302 vector bool short vec_revb (vector short char);
8303 vector bool int vec_revb (vector bool int);
8304 vector bool long long vec_revb (vector bool long long);
8305 * doc/extend.texi: Update the built-in documentation file for the
8306 new built-in functions.
8308 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8310 * config/s390/s390.md: Remove movcc splitter.
8312 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8314 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
8315 load/store on condition.
8317 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
8320 * config/avr/avr.c (avr_encode_section_info)
8321 [progmem && !TREE_READONLY]: Error if progmem object needs
8324 2017-07-11 Michael Collison <michael.collison@arm.com>
8326 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
8329 2017-07-11 Carl Love <cel@us.ibm.com>
8331 * config/rs6000/rs6000-c.c: Add support for builtins
8332 vector unsigned int vec_parity_lsbb (vector signed int);
8333 vector unsigned int vec_parity_lsbb (vector unsigned int);
8334 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
8335 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
8336 vector unsigned long long vec_parity_lsbb (vector signed long long);
8337 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
8338 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
8339 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
8340 * doc/extend.texi: Update the built-in documentation file for the
8341 new built-in functions.
8343 2017-07-11 David Malcolm <dmalcolm@redhat.com>
8345 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
8346 (layout::m_primary_loc): New field.
8347 (layout::layout): Initialize new field. Move location filtering
8348 logic from here to...
8349 (layout::maybe_add_location_range): ...this new method. Add
8350 support for filtering to just the lines already specified by other
8352 (layout::will_show_line_p): New method.
8353 (gcc_rich_location::add_location_if_nearby): New method.
8354 (selftest::test_add_location_if_nearby): New test function.
8355 (selftest::diagnostic_show_locus_c_tests): Call it.
8356 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
8359 2017-07-11 Tom de Vries <tom@codesourcery.com>
8361 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
8362 (bb_first_real_insn): New function.
8363 (nvptx_single): Add extra initialization of broadcasted condition
8366 2017-07-11 Nathan Sidwell <nathan@acm.org>
8368 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
8370 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
8372 * doc/extend.texi (AVR Function Attributes): Remove weblink to
8373 Binutils doc as TEXI will mess them up.
8374 * doc/invoke.texi (AVR Options): Same here.
8376 2017-07-11 Daniel Cederman <cederman@gaisler.com>
8378 * config/sparc/sparc.opt (mfix-ut700): New option.
8379 (mfix-gr712rc): Likewise.
8380 (sparc_fix_b2bst): New variable.
8381 * doc/invoke.texi (SPARC options): Document them.
8382 (ARM options): Fix warnings.
8383 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
8384 instructions to prevent sequences that can trigger the store-store
8385 errata for certain LEON3FT processors.
8386 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
8387 (sparc_option_override): Set sparc_fix_b2bst appropriately.
8388 * config/sparc/sparc.md (fix_b2bst): New attribute.
8389 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
8391 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
8394 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
8396 (*rsqrtsf2_sse): Ditto.
8398 (div<mode>3): Macroize insn from divdf3 and divsf3
8399 using MODEF mode iterator.
8401 2017-07-10 Martin Sebor <msebor@redhat.com>
8403 PR tree-optimization/80397
8404 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
8405 instead of testing for equality to INTEGER_TYPE.
8407 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
8409 * config.gcc: Remove uclibc from arc target spec.
8411 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
8413 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
8415 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
8418 * lto-wrapper.c (remove_option): New function.
8419 (merge_and_complain): Merge PIC/PIE options more realistically.
8421 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8423 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
8427 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
8428 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
8429 * config.in: Regenerate.
8430 * configure: Regenerate.
8431 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
8432 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
8433 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
8434 (TARGET_GASISR_PROLOGUES): ...target mask.
8435 * common/config/avr/avr-common.c
8436 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
8437 Set -mgas-isr-prologues.
8438 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
8439 INSERT_PASS_BEFORE for it.
8440 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
8441 * config/avr/avr.c (avr_option_override)
8442 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
8443 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
8444 (avr_attribute_table) <no_gccisr>: Add new function attribute.
8445 (avr_set_current_function) <is_no_gccisr>: Init machine field.
8446 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
8448 (make_avr_pass_pre_proep): New function.
8449 (emit_push_sfr) <treg>: Add argument to function and use it
8451 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
8452 and set machine->gasisr.yes.
8453 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
8454 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
8455 __gcc_isr.n_pushed to .L__stack_usage.
8456 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
8457 (avr_asm_final_postscan_insn): ...this new static function.
8458 * config/avr/avr.h (machine_function)
8459 <is_no_gccisr, use_L__stack_usage>: New fields.
8460 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
8461 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
8462 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
8463 (gasisr, *gasisr): New expander and insn.
8464 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
8465 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
8466 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
8468 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
8470 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
8473 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
8475 Move jump-tables out of .text again.
8478 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
8479 (ASM_OUTPUT_ADDR_VEC): New function.
8480 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
8481 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
8482 INSN_ADDRESSes as asm comment.
8483 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
8484 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
8485 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
8486 * config/avr/avr.md (*tablejump): Adjust comment.
8487 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
8488 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
8490 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
8491 (avr_output_addr_vec): New proto.
8492 (avr_log_t) <insn_addresses>: New field.
8494 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
8497 * config/i386/i386.c (ix86_function_arg_advance): Set
8498 outgoing_args_on_stack to true if there are outgoing arguments
8500 (ix86_function_arg): Likewise.
8501 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
8502 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
8503 * config/i386/i386.h (machine_function): Add
8504 outgoing_args_on_stack.
8506 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
8508 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
8510 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
8512 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
8514 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
8515 (REAL_H): Remove $(MACHMODE_H).
8516 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
8518 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
8519 $(MACHMODE_H) and double-int.h.
8520 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
8522 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
8525 2017-07-07 Andrew Pinski <apinski@cavium.com>
8527 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
8528 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
8530 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8532 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
8533 Add warning if GCC was not configured to link against a GLIBC that
8534 exports the hardware capability bits.
8535 (make_resolver_func): Make resolver function private and not a
8536 COMDAT function. Create the name with clone_function_name instead
8537 of make_unique_name.
8540 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
8541 correct operand in doing the split.
8543 2017-07-07 Carl Love <cel@us.ibm.com>
8545 * config/rs6000/rs6000-c: Add support for built-in function
8546 vector unsigned short vec_pack_to_short_fp32 (vector float,
8548 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
8549 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
8550 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
8551 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
8552 (convert_4f32_8i16): Add define_expand.
8553 * doc/extend.texi: Update the built-in documentation file for the
8554 new built-in function.
8556 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8558 * config/sparc/m8.md: New file.
8559 * config/sparc/sparc.md: Include m8.md.
8561 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8563 * config/sparc/sparc.opt: New option -mvis4b.
8564 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
8565 (sparc_option_override): Handle VIS4B.
8566 (enum sparc_builtins): Define
8567 SPARC_BUILTIN_DICTUNPACK{8,16,32},
8568 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
8569 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
8570 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
8571 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
8572 (check_constant_argument): New function.
8573 (sparc_vis_init_builtins): Define builtins
8574 __builtin_vis_dictunpack{8,16,32},
8575 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
8576 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
8577 __builtin_vis_fpcmpde{8,16,32}shl and
8578 __builtin_vis_fpcmpur{8,16,32}shl.
8579 (sparc_expand_builtin): Check that the constant operands to
8580 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
8581 constant and in range.
8582 * config/sparc/sparc-c.c (sparc_target_macros): Handle
8584 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
8585 (SPARC_IMM5_P): Likewise.
8586 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
8587 (enabled): Handle vis4b.
8588 (UNSPEC_DICTUNPACK): New unspec.
8589 (UNSPEC_FPCMPSHL): Likewise.
8590 (UNSPEC_FPUCMPSHL): Likewise.
8591 (UNSPEC_FPCMPDESHL): Likewise.
8592 (UNSPEC_FPCMPURSHL): Likewise.
8593 (cpu_feature): New CPU feature `vis4b'.
8594 (dictunpack{8,16,32}): New insns.
8595 (FPCSMODE): New mode iterator.
8596 (fpcscond): New code iterator.
8597 (fpcsucond): Likewise.
8598 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
8599 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
8600 (fpcmpde{8,16,32}{si,di}shl): Likewise.
8601 (fpcmpur{8,16,32}{si,di}shl): Likewise.
8602 * config/sparc/constraints.md: Define constraints `q' for unsigned
8603 2-bit integer constants and `t' for unsigned 5-bit integer
8605 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
8607 (imm5_operand_dictunpack16): Likewise.
8608 (imm5_operand_dictunpack32): Likewise.
8609 (imm2_operand): Likewise.
8610 * doc/invoke.texi (SPARC Options): Document -mvis4b.
8611 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
8612 ditunpack* and fpcmp*shl builtins.
8614 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8616 * config.gcc: Handle m8 in --with-{cpu,tune} options.
8617 * config.in: Add HAVE_AS_SPARC6 define.
8618 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
8620 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
8622 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
8623 (CPP_CPU_SPEC): Handle m8.
8624 (ASM_CPU_SPEC): Likewise.
8625 * config/sparc/sparc-opts.h (enum processor_type): Add
8627 * config/sparc/sparc.c (m8_costs): New struct.
8628 (sparc_option_override): Handle TARGET_CPU_m8.
8629 (sparc32_initialize_trampoline): Likewise.
8630 (sparc64_initialize_trampoline): Likewise.
8631 (sparc_issue_rate): Likewise.
8632 (sparc_register_move_cost): Likewise.
8633 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
8634 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
8635 (ASM_CPU64_DEFAULT_SPEC): Likewise.
8636 (CPP_CPU_SPEC): Handle M8.
8637 (ASM_CPU_SPEC): Likewise.
8638 (AS_M8_FLAG): Define.
8639 * config/sparc/sparc.md: Add m8 to the cpu attribute.
8640 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
8641 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
8643 * configure: Regenerate.
8644 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
8647 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8649 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
8651 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
8652 ("*movdi_insn_sp32"): Do not set v3pipe.
8653 ("*movsi_insn"): Likewise.
8654 ("*movdi_insn_sp64"): Likewise.
8655 ("*movsf_insn"): Likewise.
8656 ("*movdf_insn_sp32"): Likewise.
8657 ("*movdf_insn_sp64"): Likewise.
8658 ("*zero_extendsidi2_insn_sp64"): Likewise.
8659 ("*sign_extendsidi2_insn"): Likewise.
8660 ("*mov<VM32:mode>_insn"): Likewise.
8661 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8662 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8663 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8664 ("<vlop:code><VL:mode>3"): Likewise.
8665 ("*not_<vlop:code><VL:mode>3"): Likewise.
8666 ("*nand<VL:mode>_vis"): Likewise.
8667 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
8668 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
8669 ("one_cmpl<VL:mode>2"): Likewise.
8670 ("faligndata<VM64:mode>_vis"): Likewise.
8671 ("alignaddrsi_vis"): Likewise.
8672 ("alignaddrdi_vis"): Likweise.
8673 ("alignaddrlsi_vis"): Likewise.
8674 ("alignaddrldi_vis"): Likewise.
8675 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8676 ("bmaskdi_vis"): Likewise.
8677 ("bmasksi_vis"): Likewise.
8678 ("bshuffle<VM64:mode>_vis"): Likewise.
8679 ("cmask8<P:mode>_vis"): Likewise.
8680 ("cmask16<P:mode>_vis"): Likewise.
8681 ("cmask32<P:mode>_vis"): Likewise.
8682 ("pdistn<P:mode>_vis"): Likewise.
8683 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8685 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8687 * config/sparc/sparc.md ("subtype"): New insn attribute.
8688 ("*wrgsr_sp64"): Set insn subtype.
8689 ("*rdgsr_sp64"): Likewise.
8690 ("alignaddrsi_vis"): Likewise.
8691 ("alignaddrdi_vis"): Likewise.
8692 ("alignaddrlsi_vis"): Likewise.
8693 ("alignaddrldi_vis"): Likewise.
8694 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8695 ("fexpand_vis"): Likewise.
8696 ("fpmerge_vis"): Likewise.
8697 ("faligndata<VM64:mode>_vis"): Likewise.
8698 ("bshuffle<VM64:mode>_vis"): Likewise.
8699 ("cmask8<P:mode>_vis"): Likewise.
8700 ("cmask16<P:mode>_vis"): Likewise.
8701 ("cmask32<P:mode>_vis"): Likewise.
8702 ("fchksm16_vis"): Likewise.
8703 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
8704 ("fmean16_vis"): Likewise.
8705 ("fp<plusminus_insn>64_vis"): Likewise.
8706 ("<plusminus_insn>v8qi3"): Likewise.
8707 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8708 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
8709 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
8710 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
8711 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
8712 ("*movqi_insn"): Likewise.
8713 ("*movhi_insn"): Likewise.
8714 ("*movsi_insn"): Likewise.
8715 ("movsi_pic_gotdata_op"): Likewise.
8716 ("*movdi_insn_sp32"): Likewise.
8717 ("*movdi_insn_sp64"): Likewise.
8718 ("movdi_pic_gotdata_op"): Likewise.
8719 ("*movsf_insn"): Likewise.
8720 ("*movdf_insn_sp32"): Likewise.
8721 ("*movdf_insn_sp64"): Likewise.
8722 ("*zero_extendhisi2_insn"): Likewise.
8723 ("*zero_extendqihi2_insn"): Likewise.
8724 ("*zero_extendqisi2_insn"): Likewise.
8725 ("*zero_extendqidi2_insn"): Likewise.
8726 ("*zero_extendhidi2_insn"): Likewise.
8727 ("*zero_extendsidi2_insn_sp64"): Likewise.
8728 ("ldfsr"): Likewise.
8729 ("prefetch_64"): Likewise.
8730 ("prefetch_32"): Likewise.
8731 ("tie_ld32"): Likewise.
8732 ("tie_ld64"): Likewise.
8733 ("*tldo_ldub_sp32"): Likewise.
8734 ("*tldo_ldub1_sp32"): Likewise.
8735 ("*tldo_ldub2_sp32"): Likewise.
8736 ("*tldo_ldub_sp64"): Likewise.
8737 ("*tldo_ldub1_sp64"): Likewise.
8738 ("*tldo_ldub2_sp64"): Likewise.
8739 ("*tldo_ldub3_sp64"): Likewise.
8740 ("*tldo_lduh_sp32"): Likewise.
8741 ("*tldo_lduh1_sp32"): Likewise.
8742 ("*tldo_lduh_sp64"): Likewise.
8743 ("*tldo_lduh1_sp64"): Likewise.
8744 ("*tldo_lduh2_sp64"): Likewise.
8745 ("*tldo_lduw_sp32"): Likewise.
8746 ("*tldo_lduw_sp64"): Likewise.
8747 ("*tldo_lduw1_sp64"): Likewise.
8748 ("*tldo_ldx_sp64"): Likewise.
8749 ("*mov<VM32:mode>_insn"): Likewise.
8750 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8751 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8753 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8755 * config/sparc/sparc.md ("type"): New insn type viscmp.
8756 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
8758 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
8759 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
8760 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8761 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
8763 ("n7_vis_logical_11cycle"): Likewise.
8764 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
8765 * config/sparc/niagara2.md ("niag3_vis": Likewise.
8766 * config/sparc/niagara.md ("niag_vis"): Likewise.
8767 * config/sparc/ultra3.md ("us3_fga"): Likewise.
8768 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
8770 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8772 * config/sparc/sparc.md: New instruction type `bmask'.
8773 (bmaskdi_vis): Use the `bmask' type.
8774 (bmasksi_vis): Likewise.
8775 * config/sparc/ultra3.md (us3_array): Likewise.
8776 * config/sparc/niagara7.md (n7_array): Likewise.
8777 * config/sparc/niagara4.md (n4_array): Likewise.
8778 * config/sparc/niagara2.md (niag2_vis): Likewise.
8779 (niag3_vis): Likewise.
8780 * config/sparc/niagara.md (niag_vis): Likewise.
8782 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8784 * ipa-comdats.c: Remove optimize check from gate.
8785 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
8786 for functions not optimized.
8787 (ipa_fn_summary_read): Skip optimize check.
8788 (ipa_fn_summary_write): Likewise.
8789 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
8791 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
8793 (can_inline_edge_p): Check flag_pcc_struct_return for match.
8794 (check_callers): Give up on caller which is not optimized.
8795 (inline_small_functions): Likewise.
8796 (ipa_inline): Do not give up when not optimizing.
8797 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
8798 away unoptimizes cdtors.
8799 (whole_program_function_and_variable_visibility): Do
8800 ipa_discover_readonly_nonaddressable_vars in LTO mode.
8801 * ipa.c (process_references): Do not check optimize.
8802 (symbol_table::remove_unreachable_nodes): Update optimize check.
8803 (set_writeonly_bit): Update optimize check.
8804 (pass_ipa_cdtor_merge::gate): Do not check optimize.
8805 (pass_ipa_single_use::gate): Remove.
8807 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8809 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
8810 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
8811 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
8812 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
8813 permute_load, permute_store, adjust_extract, adjust_splat,
8814 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
8815 replace_swap_with_copy, dump_swap_insn_table,
8816 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
8817 recombine_lvx_pattern, recombine_stvx_pattern,
8818 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
8819 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
8820 to file rs6000-p8swap.c.
8821 * config/rs6000/rs6000-p8swap.c: New file.
8822 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
8823 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
8824 and rs6000*-*-* targets.
8826 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8828 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
8830 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8832 * lto-wrapper.c (merge_and_complain): Do not merge
8833 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
8834 fsigned_zeros, ftrapping_math, fwrapv.
8835 (append_compiler_options): Do not track these options.
8836 (append_linker_options): Likewie
8838 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8840 * cgraphunit.c (cgraph_node::finalize_function): When
8841 !flag_toplevel_reorde set no_reorder flag.
8842 (varpool_node::finalize_decl): Likewise.
8843 (symbol_table::compile): Drop no toplevel reorder path.
8845 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8847 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
8848 edges; zero probability is not better than uninitialized.
8850 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8852 * asan.h (asan_sanitize_allocas_p): Declare.
8853 * asan.c (asan_sanitize_allocas_p): New function.
8854 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
8855 (handle_builtin_alloca): Likewise.
8856 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
8857 if !asan_sanitize_allocas_p.
8858 * params.def (asan-instrument-allocas): Add new option.
8859 * params.h (ASAN_PROTECT_ALLOCAS): Define.
8860 * opts.c (common_handle_option): Disable allocas sanitization for
8863 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8865 * asan.c: Include gimple-fold.h.
8866 (get_last_alloca_addr): New function.
8867 (handle_builtin_stackrestore): Likewise.
8868 (handle_builtin_alloca): Likewise.
8869 (asan_emit_allocas_unpoison): Likewise.
8870 (get_mem_refs_of_builtin_call): Add new parameter, remove const
8871 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
8872 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
8873 (instrument_builtin_call): Pass gimple iterator to
8874 get_mem_refs_of_builtin_call.
8875 (last_alloca_addr): New global.
8876 * asan.h (asan_emit_allocas_unpoison): Declare.
8877 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
8878 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
8879 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
8880 if function calls alloca.
8881 * gimple-fold.c (replace_call_with_value): Remove static keyword.
8882 * gimple-fold.h (replace_call_with_value): Declare.
8883 * internal-fn.c: Include asan.h.
8884 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
8885 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
8887 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8889 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
8890 (C_SELFTEST_FLAGS): New.
8891 (CPP_SELFTEST_FLAGS): New.
8892 (SELFTEST_DEPS): New, from deps of s-selftest.
8893 (C_SELFTEST_DEPS): New, from deps of s-selftest.
8894 (CPP_SELFTEST_DEPS): New.
8895 (selftest): Add dependency on s-selftest-c++.
8896 (s-selftest): Rename to...
8897 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
8898 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
8899 than SELFTEST_FLAGS.
8900 (selftest-gdb): Rename to...
8901 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
8903 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
8904 (selftest-valgrind): Rename to...
8905 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
8907 (selftest-valgrind): Reintroduce as an alias for
8908 selftest-c-valgrind.
8909 (s-selftest-c++): New.
8910 (selftest-c++-gdb): New.
8911 (selftest-c++-valgrind): New.
8913 2017-07-06 Olivier Hainque <hainque@adacore.com>
8915 * gcc.c (process_command): When deciding if undefined variables
8916 should be ignored when processing specs, accept "gcc -v" as well.
8918 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8920 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
8921 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
8923 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
8925 * config/arm/arm-cpus.in (armv8-r): Add new entry.
8926 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
8927 * config/arm/arm-tables.opt: Regenerate.
8928 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
8930 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
8932 2017-07-06 Carl Love <cel@us.ibm.com>
8934 * ChangeLog: Clean up from mid air collision
8936 2017-07-06 Carl Love <cel@us.ibm.com>
8938 * config/rs6000/rs6000-c.c: Add support for built-in functions
8939 vector signed int vec_subc (vector signed int, vector signed int);
8940 vector signed __int128 vec_subc (vector signed __int128,
8941 vector signed __int128);
8942 vector unsigned __int128 vec_subc (vector unsigned __int128,
8943 vector unsigned __int128);
8944 vector signed int vec_sube (vector signed int, vector signed int,
8946 vector unsigned int vec_sube (vector unsigned int,
8947 vector unsigned int,
8948 vector unsigned int);
8949 vector signed __int128 vec_sube (vector signed __int128,
8950 vector signed __int128,
8951 vector signed__int128);
8952 vector unsigned __int128 vec_sube (vector unsigned __int128,
8953 vector unsigned __int128,
8954 vector unsigned __int128);
8955 vector signed int vec_subec (vector signed int, vector signed int,
8957 vector unsigned int vec_subec (vector unsigned int,
8958 vector unsigned int,
8959 vector unsigned int);
8960 vector signed __int128 vec_subec (vector signed __int128,
8961 vector signed __int128,
8962 vector signed__int128);
8963 vector unsigned __int128 vec_subec (vector unsigned __int128,
8964 vector unsigned __int128,
8965 vector unsigned __int128);
8966 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
8967 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
8968 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
8969 BU_ALTIVEC_OVERLOAD_X definitions.
8970 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
8971 * doc/extend.texi: Update the built-in documentation file for the new
8974 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8977 * diagnostic-show-locus.c (layout::layout): Use start and finish
8978 spelling location for the start and finish of each range.
8979 * genmatch.c (linemap_client_expand_location_to_spelling_point):
8980 Add unused aspect param.
8981 * input.c (expand_location_1): Add "aspect" param, and use it
8982 to access the correct part of the location.
8983 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
8985 (expand_location_to_spelling_point): Likewise.
8986 (linemap_client_expand_location_to_spelling_point): Add "aspect"
8987 param, and pass it to expand_location_1.
8989 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
8991 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
8992 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
8993 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
8994 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
8995 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
8996 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
8997 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
8998 _mm_maskz_getmant_ss): New intrinsics.
8999 (__builtin_ia32_getexpss128_mask): Changed to ...
9000 __builtin_ia32_getexpss128_round ... this.
9001 (__builtin_ia32_getexpsd128_mask): Changed to ...
9002 __builtin_ia32_getexpsd128_round ... this.
9003 * config/i386/i386-builtin-types.def
9004 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
9005 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
9006 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
9007 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
9008 __builtin_ia32_getmantss_mask_round): New builtins.
9009 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
9010 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
9011 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
9012 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
9013 * config/i386/sse.md
9014 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
9015 avx512f_sgetexp<mode><mask_scalar_name>
9016 <round_saeonly_scalar_name> ... this.
9017 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
9018 %0, %1, %2<round_saeonly_op3>}): Changed to ...
9019 vgetexp<ssescalarmodesuffix>
9020 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9021 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
9022 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
9023 avx512f_vgetmant<mode><mask_scalar_name>
9024 <round_saeonly_scalar_name> ... this.
9025 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
9026 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
9027 vgetmant<ssescalarmodesuffix>
9028 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
9029 %0<mask_scalar_operand4>, %1, %2
9030 <round_saeonly_scalar_mask_op4>, %3} ... this.
9031 * config/i386/subst.md (mask_scalar_operand4,
9032 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
9033 round_saeonly_scalar_nimm_predicate): New subst attributes.
9035 2017-07-06 Julia Koval <julia.koval@intel.com>
9037 * config/i386/i386.c (ix86_erase_embedded_rounding):
9038 Remove code for old rounding pattern.
9040 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
9042 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
9044 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
9046 * doc/sourcebuild.texi (Test Directives, Variants of
9047 dg-require-support): Add documentation for dg-require-stack-check.
9049 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
9051 * config/i386/subst.md (mask_scalar, round_scalar,
9052 round_saeonly_scalar): New meta-templates.
9053 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
9054 round_scalar_mask_operand3, round_scalar_mask_op3,
9055 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
9056 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
9057 round_saeonly_scalar_constraint,
9058 round_saeonly_scalar_prefix): New subst attribute.
9059 * config/i386/sse.md
9060 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
9061 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
9062 <round_scalar_name> ... this.
9063 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
9064 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
9065 <round_scalar_name> ... this.
9066 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
9067 <sse>_vm<code><mode>3<mask_scalar_name>
9068 <round_saeonly_scalar_name> ... this.
9069 (v<plusminus_mnemonic><ssescalarmodesuffix>
9070 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9071 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9072 v<plusminus_mnemonic><ssescalarmodesuffix>
9073 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9074 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9075 (v<multdiv_mnemonic><ssescalarmodesuffix>
9076 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
9077 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
9078 v<multdiv_mnemonic><ssescalarmodesuffix>
9079 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9080 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
9081 (v<maxmin_float><ssescalarmodesuffix>
9082 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
9083 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
9084 v<maxmin_float><ssescalarmodesuffix>
9085 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
9086 %0<mask_scalar_operand3>, %1, %<iptr>2
9087 <round_saeonly_scalar_mask_op3>} ... this.
9089 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
9091 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
9092 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
9094 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
9095 Alan Hayward <alan.hayward@arm.com>
9096 David Sherwood <david.sherwood@arm.com>
9098 * combine.c (simplify_if_then_else): Remove "enum" before
9100 * compare-elim.c (can_eliminate_compare): Likewise.
9101 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9103 (aarch64_lookup_simd_builtin_type): Likewise.
9104 (aarch64_simd_builtin_type): Likewise.
9105 (aarch64_init_simd_builtin_types): Likewise.
9106 (aarch64_simd_expand_args): Likewise.
9107 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
9109 (aarch64_reverse_mask): Likewise.
9110 (aarch64_simd_emit_reg_reg_move): Likewise.
9111 (aarch64_gen_adjusted_ldpstp): Likewise.
9112 (aarch64_ccmp_mode_to_code): Likewise.
9113 (aarch64_operands_ok_for_ldpstp): Likewise.
9114 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9115 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
9117 (aarch64_min_divisions_for_recip_mul): Likewise.
9118 (aarch64_reassociation_width): Likewise.
9119 (aarch64_get_condition_code_1): Likewise.
9120 (aarch64_simd_emit_reg_reg_move): Likewise.
9121 (aarch64_simd_attr_length_rglist): Likewise.
9122 (aarch64_reverse_mask): Likewise.
9123 (aarch64_operands_ok_for_ldpstp): Likewise.
9124 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9125 (aarch64_gen_adjusted_ldpstp): Likewise.
9126 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
9128 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
9129 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9130 (arm_lookup_simd_builtin_type): Likewise.
9131 (arm_simd_builtin_type): Likewise.
9132 (arm_init_simd_builtin_types): Likewise.
9133 (arm_expand_builtin_args): Likewise.
9134 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
9135 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
9136 (ft32_setup_incoming_varargs): Likewise.
9137 (ft32_function_arg): Likewise.
9138 (ft32_function_arg_advance): Likewise.
9139 (ft32_pass_by_reference): Likewise.
9140 (ft32_arg_partial_bytes): Likewise.
9141 (ft32_valid_pointer_mode): Likewise.
9142 (ft32_addr_space_pointer_mode): Likewise.
9143 (ft32_addr_space_legitimate_address_p): Likewise.
9144 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
9146 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
9147 (ix86_emit_outlined_ms2sysv_restore): Likewise.
9148 (iamcu_alignment): Likewise.
9149 (canonicalize_vector_int_perm): Likewise.
9150 (ix86_noce_conversion_profitable_p): Likewise.
9151 (ix86_mpx_bound_mode): Likewise.
9152 (ix86_operands_ok_for_move_multiple): Likewise.
9153 * config/microblaze/microblaze-protos.h
9154 (microblaze_expand_conditional_branch_reg): Likewise.
9155 * config/microblaze/microblaze.c
9156 (microblaze_expand_conditional_branch_reg): Likewise.
9157 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
9159 (rs6000_reassociation_width): Likewise.
9160 (rs6000_invalid_binary_op): Likewise.
9161 (fusion_p9_p): Likewise.
9162 (emit_fusion_p9_load): Likewise.
9163 (emit_fusion_p9_store): Likewise.
9164 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
9166 (riscv_hard_regno_mode_ok_p): Likewise.
9167 (riscv_address_insns): Likewise.
9168 (riscv_split_symbol): Likewise.
9169 (riscv_legitimize_move): Likewise.
9170 (riscv_function_value): Likewise.
9171 (riscv_hard_regno_nregs): Likewise.
9172 (riscv_expand_builtin): Likewise.
9173 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
9174 (riscv_build_integer): Likewise.
9175 (riscv_split_integer): Likewise.
9176 (riscv_legitimate_constant_p): Likewise.
9177 (riscv_cannot_force_const_mem): Likewise.
9178 (riscv_regno_mode_ok_for_base_p): Likewise.
9179 (riscv_valid_base_register_p): Likewise.
9180 (riscv_valid_offset_p): Likewise.
9181 (riscv_valid_lo_sum_p): Likewise.
9182 (riscv_classify_address): Likewise.
9183 (riscv_legitimate_address_p): Likewise.
9184 (riscv_address_insns): Likewise.
9185 (riscv_load_store_insns): Likewise.
9186 (riscv_force_binary): Likewise.
9187 (riscv_split_symbol): Likewise.
9188 (riscv_force_address): Likewise.
9189 (riscv_legitimize_address): Likewise.
9190 (riscv_move_integer): Likewise.
9191 (riscv_legitimize_const_move): Likewise.
9192 (riscv_legitimize_move): Likewise.
9193 (riscv_address_cost): Likewise.
9194 (riscv_subword): Likewise.
9195 (riscv_output_move): Likewise.
9196 (riscv_canonicalize_int_order_test): Likewise.
9197 (riscv_emit_int_order_test): Likewise.
9198 (riscv_function_arg_boundary): Likewise.
9199 (riscv_pass_mode_in_fpr_p): Likewise.
9200 (riscv_pass_fpr_single): Likewise.
9201 (riscv_pass_fpr_pair): Likewise.
9202 (riscv_get_arg_info): Likewise.
9203 (riscv_function_arg): Likewise.
9204 (riscv_function_arg_advance): Likewise.
9205 (riscv_arg_partial_bytes): Likewise.
9206 (riscv_function_value): Likewise.
9207 (riscv_pass_by_reference): Likewise.
9208 (riscv_setup_incoming_varargs): Likewise.
9209 (riscv_print_operand): Likewise.
9210 (riscv_elf_select_rtx_section): Likewise.
9211 (riscv_save_restore_reg): Likewise.
9212 (riscv_for_each_saved_reg): Likewise.
9213 (riscv_register_move_cost): Likewise.
9214 (riscv_hard_regno_mode_ok_p): Likewise.
9215 (riscv_hard_regno_nregs): Likewise.
9216 (riscv_class_max_nregs): Likewise.
9217 (riscv_memory_move_cost): Likewise.
9218 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
9219 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
9220 (rl78_addr_space_address_mode): Likewise.
9221 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9223 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9224 (rs6000_reassociation_width): Likewise.
9225 (rs6000_invalid_binary_op): Likewise.
9226 (fusion_p9_p): Likewise.
9227 (emit_fusion_p9_load): Likewise.
9228 (emit_fusion_p9_store): Likewise.
9229 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
9230 (ok_for_simple_move_operands): Likewise.
9231 (ok_for_simple_move_strict_operands): Likewise.
9232 (ok_for_simple_arith_logic_operands): Likewise.
9233 (visium_legitimize_reload_address): Likewise.
9234 (visium_select_cc_mode): Likewise.
9235 (output_cbranch): Likewise.
9236 (visium_split_double_move): Likewise.
9237 (visium_expand_copysign): Likewise.
9238 (visium_expand_int_cstore): Likewise.
9239 (visium_expand_fp_cstore): Likewise.
9240 * config/visium/visium.c (visium_pass_by_reference): Likewise.
9241 (visium_function_arg): Likewise.
9242 (visium_function_arg_advance): Likewise.
9243 (visium_libcall_value): Likewise.
9244 (visium_setup_incoming_varargs): Likewise.
9245 (visium_legitimate_constant_p): Likewise.
9246 (visium_legitimate_address_p): Likewise.
9247 (visium_legitimize_address): Likewise.
9248 (visium_secondary_reload): Likewise.
9249 (visium_register_move_cost): Likewise.
9250 (visium_memory_move_cost): Likewise.
9251 (prepare_move_operands): Likewise.
9252 (ok_for_simple_move_operands): Likewise.
9253 (ok_for_simple_move_strict_operands): Likewise.
9254 (ok_for_simple_arith_logic_operands): Likewise.
9255 (visium_function_value_1): Likewise.
9256 (rtx_ok_for_offset_p): Likewise.
9257 (visium_legitimize_reload_address): Likewise.
9258 (visium_split_double_move): Likewise.
9259 (visium_expand_copysign): Likewise.
9260 (visium_expand_int_cstore): Likewise.
9261 (visium_expand_fp_cstore): Likewise.
9262 (visium_split_cstore): Likewise.
9263 (visium_select_cc_mode): Likewise.
9264 (visium_split_cbranch): Likewise.
9265 (output_cbranch): Likewise.
9266 (visium_print_operand_address): Likewise.
9267 * expmed.c (flip_storage_order): Likewise.
9268 * expmed.h (emit_cstore): Likewise.
9269 (flip_storage_order): Likewise.
9270 * genrecog.c (validate_pattern): Likewise.
9271 * hsa-gen.c (gen_hsa_addr): Likewise.
9272 * internal-fn.c (expand_arith_overflow): Likewise.
9273 * ira-color.c (allocno_copy_cost_saving): Likewise.
9274 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9275 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
9276 (process_invariant_for_inheritance): Likewise.
9277 * lra-eliminations.c (move_plus_up): Likewise.
9278 * omp-low.c (lower_oacc_reductions): Likewise.
9279 * simplify-rtx.c (simplify_subreg): Likewise.
9280 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
9281 (TARGET_CHKP_BOUND_MODE): Likewise..
9282 * targhooks.c (default_chkp_bound_mode): Likewise.
9283 (default_setup_incoming_vararg_bounds): Likewise.
9284 * targhooks.h (default_chkp_bound_mode): Likewise.
9285 (default_setup_incoming_vararg_bounds): Likewise.
9286 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
9287 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
9288 (have_whole_vector_shift): Likewise.
9289 * tree-vect-stmts.c (vectorizable_load): Likewise.
9290 * doc/tm.texi: Regenerate.
9292 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9294 Graceful degrade if Binutils PR21472 is not available.
9297 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
9298 .rodata in flash test fails.
9299 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
9300 * confgure: Regenerate.
9301 * config.in: Regenerate.
9302 * config/avr/avr.c (avr_asm_named_section)
9303 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
9304 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
9305 (avr_asm_init_sections): Same.
9307 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9309 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
9310 (fma<VH:mode>4_intrinsic): Likewise.
9311 (*fmsub<VCVTF:mode>4): Likewise.
9312 (*fmsub<VH:mode>4_intrinsic): Likewise.
9314 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
9317 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
9318 Don't depend on "optimize > 0".
9319 (out_movhi_r_mr, out_movqi_mr_r): Same.
9320 (out_movhi_mr_r, out_movqi_r_mr): Same.
9321 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
9322 io_address_operand on "optimize > 0".
9324 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9326 * tree-loop-distribution.c: Add general explanantion on the pass.
9327 (generate_loops_for_partition): Mark distributed loop.
9328 (pg_add_dependence_edges): New parameter. Handle alias data
9329 dependence specially and record it in the parameter if asked.
9330 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
9331 (init_partition_graph_vertices, add_partition_graph_edge): New.
9332 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
9333 (free_partition_graph_vdata, build_partition_graph): New.
9334 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
9335 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
9336 (data_ref_segment_size, latch_dominated_by_data_ref): New.
9337 (compute_alias_check_pairs, version_loop_by_alias_check): New.
9338 (version_for_distribution_p, finalize_partitions): New.
9339 (distribute_loop): Handle alias data dependence specially. Factor
9340 out loop fusion code as functions and call these functions.
9342 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9344 * tree-loop-distribution.c (classify_partition): New parameter and
9345 better handle reduction statement.
9346 (rdg_build_partitions): Revise comment.
9347 (distribute_loop): Compute statements in all partitions and pass it
9348 to classify_partition.
9350 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9352 * tree-loop-distribution.c (enum partition_type): New.
9353 (struct partition): New field type.
9354 (partition_merge_into): Add parameter. Update partition type.
9355 (data_dep_in_cycle_p, update_type_for_merge): New functions.
9356 (build_rdg_partition_for_vertex): Compute partition type.
9357 (rdg_build_partitions): Dump partition type.
9358 (distribute_loop): Update calls to partition_merge_into.
9360 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9362 * tree-loop-distribution.c (struct ddr_hasher): New.
9363 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
9365 (classify_partition): Call get_data_dependence.
9366 (pg_add_dependence_edges): Ditto.
9367 (distribute_loop): Release data dependence hash table.
9369 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9371 * tree-loop-distribution.c (ref_base_address): Delete.
9372 (similar_memory_accesses): Rename ...
9373 (share_memory_accesses): ... to this. Check if partitions access
9374 the same memory reference.
9375 (distribute_loop): Call share_memory_accesses.
9377 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9379 * tree-loop-distribution.c (struct partition): New field recording
9381 (partition_alloc, partition_free): Init and release data refs.
9382 (partition_merge_into): Merge data refs.
9383 (build_rdg_partition_for_vertex): Collect data refs for partition.
9384 (pg_add_dependence_edges): Change parameters from vector to bitmap.
9386 (distribute_loop): Remve data refs from vertice data of partition
9389 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9391 * tree-loop-distribution.c (params.h): Include header file.
9392 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
9393 (datarefs_vec): New global var.
9394 (create_rdg_vertices): Use datarefs_vec directly.
9395 (free_rdg): Don't free data references.
9396 (build_rdg): Update use. Don't free data references.
9397 (distribute_loop): Compute global variable for data references.
9398 Bail out if there are too many data references.
9400 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9402 * tree-loop-distribution.c (loop_nest): New global var.
9403 (build_rdg): Use loop directly, rather than loop nest.
9404 (pg_add_dependence_edges): Remove loop nest parameter. Use global
9406 (distribute_loop): Compute global variable loop nest. Update use.
9408 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9410 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
9411 (partition_merge_into): New parameter. Dump reason for fusion.
9412 (distribute_loop): Update use of partition_merge_into.
9414 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9416 * tree-loop-distribution.c (bb_top_order_index): New.
9417 (bb_top_order_index_size, bb_top_order_cmp): New.
9418 (stmts_from_loop): Use topological order.
9419 (pass_loop_distribution::execute): Compute and release topological
9420 order for basic blocks.
9422 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9424 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
9427 2017-07-05 Bin Cheng <bin.cheng@arm.com>
9429 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
9430 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
9431 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
9432 * internal-fn.def (LOOP_DIST_ALIAS): New.
9433 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
9434 (fold_loop_internal_call): ... this.
9435 (vect_loop_dist_alias_call): New function.
9436 (set_uid_loop_bbs): Call fold_loop_internal_call.
9437 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
9440 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9443 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
9444 Require dead FLAGS_REG at the beginning of a peephole.
9446 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
9449 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
9450 arguments in the call to __builtin_ia32_sbb_u32.
9451 (_subborrow_u64): Swap _X and _Y arguments in the call to
9452 __builtin_ia32_sbb_u64.
9454 2017-07-04 Jakub Jelinek <jakub@redhat.com>
9457 * tree-vrp.c (compare_assert_loc): Turn into a function template
9458 with stable template parameter. Only test if a->e is NULL,
9459 !a->e == !b->e has been verified already. Use e == NULL or
9460 e != NULL instead of e or ! e tests. If stable is true, don't use
9461 iterative_hash_expr, on the other side allow a or b or both NULL
9462 and sort the NULLs last.
9463 (process_assert_insertions): Sort using compare_assert_loc<false>
9464 instead of compare_assert_loc, later sort using
9465 compare_assert_loc<true> before calling process_assert_insertions_for
9466 in a loop. Use break instead of continue once seen NULL pointer.
9468 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
9470 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
9471 Cortex-R7 and Cortex-R8 processors.
9473 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9475 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
9476 uninitialized while src is not.
9478 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
9480 * common/config/arm/arm-common.c: Adjust include path for
9482 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
9483 (arm-cpu.h): Create in build directory. Adjust dependency rules.
9484 (arm-cpu-data.h): Likewise.
9485 (arm-cpu-cdata.h): Likewise.
9486 * config/arm/arm-cpu.h: Delete.
9487 * config/arm/arm-cpu-cdata.h: Delete.
9488 * config/arm/arm-cpu-data.h: Delete.
9490 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
9492 * config/arm/arm-cpus.in (cortex-a55): New.
9493 (cortex-a75): Likewise.
9494 (cortex-a75.cortex-a55): Likewise.
9495 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
9497 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
9498 * config/arm/arm-cpu-cdata.h: Regenerate.
9499 * config/arm/arm-cpu-data.h: Regenerate.
9500 * config/arm/arm-cpu.h: Regenerate.
9501 * config/arm/arm-tables.opt: Regenerate.
9502 * config/arm/arm-tune.md: Regenerate.
9504 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9506 * haifa-sched.c (sched_create_recovery_edges): Update profile.
9508 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
9510 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
9513 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
9515 PR tree-optimization/81292
9516 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
9517 full_string_p, also call adjust_related_strinfos if the adjustment
9518 is simple, otherwise invalidate related strinfos.
9520 2017-07-04 Martin Liska <mliska@suse.cz>
9523 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
9524 newly created variable as DECL_IGNORED_P.
9526 2017-07-04 Martin Liska <mliska@suse.cz>
9529 * ipa-inline.c (inline_small_functions):
9530 Use xstrdup_for_dump.
9532 2017-07-04 Tom de Vries <tom@codesourcery.com>
9534 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
9536 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9539 * config/darwin.c (darwin_function_switched_text_sections):
9542 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
9544 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
9546 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9548 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
9550 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9552 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
9553 min_profitable_iters, and th as inclusive lower bounds.
9554 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
9555 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
9556 for min_profitable_iters and min_profitable_estimate.
9557 (vect_transform_loop): Treat th as an inclusive lower bound.
9558 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
9560 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
9563 * config/darwin.c (darwin_function_switched_text_sections):
9564 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
9565 in two pieces, and suppress the use of buf.
9567 2017-07-03 Nathan Sidwell <nathan@acm.org>
9569 * hash-table.h (hash_table_mod1): Fix indentation.
9571 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9574 * predict.c (force_edge_cold): Be more careful about propagation
9576 * profile-count.h (profile_probability::guessed,
9577 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
9579 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
9581 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
9583 * doc/invoke.texi (rcpc architecture extension): Document it.
9585 2017-07-03 Richard Biener <rguenther@suse.de>
9587 PR tree-optimization/60510
9588 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
9589 the scalar reduction PHI and use it.
9590 (vectorizable_reduction): Properly guard the single_defuse_cycle
9591 path for non-SLP reduction chains where we cannot use it.
9592 Rework reduc_def/index and vector type deduction. Rework
9593 vector operand gathering during reduction op code-gen.
9594 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
9595 chains dissolve the chain and leave it to non-SLP reduction
9598 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9600 * tree-data-ref.h (dr_alignment): Declare.
9601 * tree-data-ref.c (dr_alignment): New function.
9602 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
9603 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
9605 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
9607 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9609 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
9610 and base_misalignment fields.
9611 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
9612 * tree-data-ref.c: Include builtins.h.
9613 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
9614 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
9615 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
9616 * tree-vect-data-refs.c: Include tree-cfg.h.
9617 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
9618 fields instead of calculating an alignment here.
9619 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
9620 innermost_loop_behavior fields.
9622 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9624 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
9626 (DR_STEP_ALIGNMENT): New macro.
9627 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
9628 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
9629 (create_data_ref): Print it.
9630 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
9631 to tell whether the step preserves vector (mis)alignment.
9632 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9633 Move the check for an integer step and generalise to all INTEGER_CST.
9634 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
9635 Print the outer step alignment.
9637 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9639 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
9640 with offset_alignment.
9641 (DR_ALIGNED_TO): Delete.
9642 (DR_OFFSET_ALIGNMENT): New macro.
9643 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
9644 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
9645 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
9646 (create_data_ref): Likewise.
9647 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
9648 (vect_analyze_data_refs): Likewise.
9649 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
9650 creating dummy innermost behavior.
9652 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9654 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
9655 with a "innermost_loop_behavior *" and refeence tree.
9656 * tree-data-ref.c (dr_analyze_innermost): Likewise.
9657 (create_data_ref): Update call accordingly.
9658 * tree-predcom.c (find_looparound_phi): Likewise.
9660 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9662 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
9663 fields with dr_wrt_vec_loop.
9664 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
9665 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
9666 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
9667 (vect_dr_behavior): New function.
9668 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9669 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
9670 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
9671 track whether the step preserves the misalignment.
9672 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
9673 Use vect_dr_behavior.
9674 (vect_setup_realignment): Update call accordingly.
9675 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
9676 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
9677 call to vect_create_addr_base_for_vector_ref.
9678 (vect_create_cond_for_align_checks): Likewise.
9679 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
9680 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
9681 (vect_recog_mask_conversion_pattern): Likewise.
9682 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
9683 (new_stmt_vec_info): Remove redundant zeroing.
9685 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9687 * common/config/arm/arm-common.c (arm_be8_option): New function.
9688 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
9689 (ISA_ARMv6): Add isa_bit_be8.
9690 * config/arm/arm.h (arm_be8_option): Add prototype.
9691 (BE8_SPEC_FUNCTION): New define.
9692 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
9693 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
9694 (mlittle-endian): Similarly.
9695 (mbe8, mbe32): New options.
9696 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
9697 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
9699 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9701 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
9703 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9705 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
9706 (cleanup_tree_cfg_bb): Use it.
9707 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
9709 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
9711 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9714 * loop-doloop.c (add_test): Update profile.
9716 2017-07-03 Martin Liska <mliska@suse.cz>
9719 * sanopt.c (rewrite_usage_of_param): New function.
9720 (sanitize_rewrite_addressable_params): Likewise.
9721 (pass_sanopt::execute): Call rewrite_usage_of_param.
9723 2017-07-03 Richard Biener <rguenther@suse.de>
9725 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
9726 back to using VIEW_CONVERT_EXPR.
9728 2017-07-03 Martin Liska <mliska@suse.cz>
9731 * doc/extend.texi: Document when a resolver function is
9732 generated for target_clones.
9734 2017-07-03 Martin Liska <mliska@suse.cz>
9736 * asan.c (asan_emit_stack_protection): Unpoison just red zones
9737 and shadow memory of auto variables which are subject of
9738 use-after-scope sanitization.
9739 (asan_expand_mark_ifn): Add do set only when is_poison.
9741 2016-07-03 Richard Biener <rguenther@suse.de>
9743 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
9745 (vect_force_simple_reduction): Record reduction def -> phi mapping.
9746 (vectorizable_reduction): Perform reduction PHI creation when
9747 visiting a reduction PHI and adjust and simplify code generation
9748 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
9749 (vect_transform_loop): Visit reduction PHIs.
9750 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
9751 defs into the SLP tree.
9752 (vect_build_slp_tree): Reduction defs terminate the recursion.
9753 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
9755 (vect_get_vec_defs_for_stmt_copy): Export.
9756 (vect_get_vec_defs): Likewise.
9757 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
9759 (vect_get_vec_defs_for_stmt_copy): Declare.
9760 (vect_get_vec_defs): Likewise.
9762 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9764 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
9765 parameter with a "loop" parameter and use it instead of the
9766 loop containing DR_STMT. Don't check simple_iv when doing
9767 BB analysis. Describe the two analysis modes in the comment.
9769 2017-07-03 Tom de Vries <tom@codesourcery.com>
9771 PR tree-optimization/69468
9772 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
9773 (find_same_succ_bb): Handle ignore_edge_flags.
9775 2017-07-03 Tom de Vries <tom@codesourcery.com>
9777 PR tree-optimization/81192
9778 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
9780 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
9782 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
9784 2017-07-03 Tom de Vries <tom@codesourcery.com>
9786 PR tree-optimization/81192
9787 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
9788 BB_SAME_SUCC (bb) == NULL.
9790 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9792 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
9795 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9797 * dumpfile.c: Include profile-count.h
9798 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
9800 (insert_cond_bb): Update profile.
9801 * tree-cfg.h (insert_cond_bb): Update prototype.
9802 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
9803 * tree-dump.c: Do not include tree-cfg.
9805 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9807 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
9809 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9811 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
9814 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9816 * tree-complex.c (expand_complex_div_wide): update profile.
9818 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9819 Alan Hayward <alan.hayward@arm.com>
9820 David Sherwood <david.sherwood@arm.com>
9822 * Makefile.in (MACHMODE_H): Remove insn-modes.h
9823 (CORETYPES_H): New define.
9824 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
9825 (insn-modes-inline.h, s-modes-inline-h): New rules.
9826 (generated_files): Add insn-modes-inline.h.
9827 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
9828 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
9829 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
9830 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
9831 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
9832 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
9833 (build/gencodes.o, build/genconditions.o): Likewise.
9834 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
9835 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
9836 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
9837 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
9838 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
9839 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
9840 * coretypes.h: Include everything up to real.h for generators.
9841 Include insn-modes.h first. Include wide-int-print.h after
9842 wide-int.h. Include insn-modes-inline.h and then machmode.h.
9843 * machmode.h: Don't include insn-modes.h here.
9844 * function-tests.c: Remove includes of signop.h, machmode.h,
9845 double-int.h and wide-int.h.
9847 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
9849 * optc-save-gen.awk: Likewise.
9850 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
9851 * godump.c: Remove include of wide-int-print.h.
9852 * pretty-print.h: Likewise.
9853 * wide-int-print.cc: Likewise.
9854 * wide-int.cc: Likewise.
9855 * hash-map-tests.c: Remove include of signop.h.
9856 * hash-set-tests.c: Likewise.
9857 * rtl-tests.c: Likewise.
9858 * mkconfig.sh: Remove include of machmode.h.
9859 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
9861 (emit_insn_modes_inline_h): ...this new function. Emit the code
9862 into an insn-modes-inline.h header file, adding appropriate
9863 include guards and end comments.
9864 (emit_insn_modes_c_header): Remove include of machmode.h.
9865 (emit_min_insn_modes_c_header): Include coretypes.h rather than
9867 (main): Handle -i flag and call emit_insn_modes_inline_h when
9870 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9872 * tree-ssa-strlen.c (strinfo): Rename the length field to
9873 nonzero_chars. Add a full_string_p field.
9874 (compare_nonzero_chars, zero_length_string_p): New functions.
9875 (get_addr_stridx): Add an offset_out parameter.
9876 Use compare_nonzero_chars.
9877 (get_stridx): Update accordingly. Use compare_nonzero_chars.
9878 (new_strinfo): Update after above changes to strinfo.
9879 (set_endptr_and_length): Set full_string_p.
9880 (get_string_length): Update after above changes to strinfo.
9881 (unshare_strinfo): Update call to new_strinfo.
9882 (maybe_invalidate): Likewise.
9883 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
9884 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
9885 as a uhwi instead of an shwi. Update after above changes to
9886 strinfo and new_strinfo.
9887 (zero_length_string): Assert that chainsi contains full strings.
9888 Use zero_length_string_p. Update call to new_strinfo.
9889 (adjust_related_strinfos): Update after above changes to strinfo.
9890 Copy full_string_p from origsi.
9891 (adjust_last_stmt): Use zero_length_string_p.
9892 (handle_builtin_strlen): Update after above changes to strinfo and
9893 new_strinfo. Install the lhs as the string length if the previous
9894 entry didn't describe a full string.
9895 (handle_builtin_strchr): Update after above changes to strinfo
9897 (handle_builtin_strcpy): Likewise.
9898 (handle_builtin_strcat): Likewise.
9899 (handle_builtin_malloc): Likewise.
9900 (handle_pointer_plus): Likewise.
9901 (handle_builtin_memcpy): Likewise. Track nonzero characters
9902 that aren't necessarily followed by a nul terminator.
9903 (handle_char_store): Likewise.
9905 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9907 PR tree-optimization/80769
9908 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
9909 for malloc and calloc. Document the new invariant that all related
9910 strinfos have delayed lengths or none do.
9911 (verify_related_strinfos): Move earlier in file.
9912 (set_endptr_and_length): New function, split out from...
9913 (get_string_length): ...here. Also set the lengths of related
9915 (zero_length_string): Assert that chainsi has known (rather than
9917 (adjust_related_strinfos): Likewise.
9919 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9921 PR tree-optimization/81136
9922 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
9923 assert that two references with the same misalignment have the same
9924 compile-time misalignment if those compile-time misalignments
9927 2017-07-01 Andi Kleen <ak@linux.intel.com>
9929 * print-tree.c (print_node): Print all attributes.
9931 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9933 * cfg.c (scale_bbs_frequencies): New function.
9934 * cfg.h (scale_bbs_frequencies): Declare it.
9935 * cfgloopanal.c (single_likely_exit): Cleanup.
9936 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
9938 (scale_loop_profile): Likewise.
9939 (loop_version): Likewise.
9940 (create_empty_loop_on_edge): Update.
9941 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
9942 scale_loop_frequencies, scale_loop_profile, loopify,
9943 loop_version): Update prototypes.
9944 * modulo-sched.c (sms_schedule): Update.
9945 * predict.c (unlikely_executed_edge_p): Also check probability.
9946 (probably_never_executed_edge_p): Fix typo.
9947 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9948 * tree-parloops.c (gen_parallel_loop): Update.
9949 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
9950 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9951 * tree-ssa-loop-split.c (split_loop): Update.
9952 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9953 * tree-vect-loop-manip.c (vect_do_peeling): Update.
9954 (vect_loop_versioning): Update.
9955 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9957 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9959 * trans-mem.c (split_bb_make_tm_edge): Update profile.
9961 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9963 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
9964 to keep profile consistent.
9966 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9968 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
9969 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
9970 * profile-count.h (max_safe_multiplier): Make unsigned.
9971 (profile_count::guessed_zero): New.
9973 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9975 * bb-reorder.c (fix_up_crossing_landing_pad,
9976 fix_crossing_conditional_branches): Use make_single_succ_edge
9977 to keep profile consistent.
9979 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9981 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
9984 2017-07-01 Jakub Jelinek <jakub@redhat.com>
9987 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
9988 the right scopes, make sure cond_jump isn't preserved between multiple
9989 iterations. Search for fallthru edge whenever there are 3+ edges and
9990 use find_fallthru_edge for it.
9992 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9994 Patch by Alexander Monakov <amonakov@ispras.ru>
9995 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
9996 probabilities consistently.
9998 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10000 * pa.c (pa_expand_compare_and_swap_loop): Update call of
10001 emit_cmp_and_jump_insns.
10003 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10006 * tree-inline.c (expand_call_inline): Combine profile statuses.
10008 2017-06-30 Andrew Pinski <apinski@cavium.com>
10010 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
10011 fold_stmt returned true.
10013 2017-06-30 Nathan Sidwell <nathan@acm.org>
10015 * ggc.h (empty_string): Delete.
10016 * cfgexpand.c (expand_asm_stmt): Use plain "".
10017 * optabs.c (expand_asm_memory_barrier): Likewise.
10018 * stringpool.c (empty_string): Delete.
10019 (digit_vector, digit_string): Delete.
10020 (ggc_alloc_string): Use plain "", don't optimize single digit
10021 strings. Use ggc_alloc_atomic.
10023 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
10025 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
10026 comparison set and one other set, use the cost of the non-comparison
10029 2017-06-30 Nathan Sidwell <nathan@acm.org>
10031 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
10034 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
10036 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
10037 loops. Remove now unneeded calls to gimple_switch_set_label() that
10038 just set removed labels to NULL_TREE.
10040 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
10042 * tree-ssanames.c (set_range_info_raw): Abstract from ...
10043 (set_range_info): ...here. Only call set_range_info_raw if domain
10045 (set_nonzero_bits): Call set_range_info_raw.
10046 * tree-ssanames.h (set_range_info_raw): New.
10048 2017-06-30 Jakub Jelinek <jakub@redhat.com>
10051 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
10052 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
10053 of nonimmediate_operand and <store_mask_constraint> instead of m
10054 for the input operand. For V8FI iterator, always split if input
10055 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
10056 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
10057 <store_mask_predicate> instead of register_operand and
10058 <store_mask_constraint> instead of v for the input operand. Make
10059 sure both operands aren't MEMs for if not <mask_applied>.
10061 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
10063 * lto-wrapper.c (copy_file) Close both file descriptors before
10066 2017-06-30 Martin Liska <mliska@suse.cz>
10069 * multiple_target.c (create_dispatcher_calls): Make ifunc
10070 also for function that don't have calls or are not referenced.
10072 2017-06-30 Richard Biener <rguenther@suse.de>
10074 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
10075 analyze the first scalar stmt. Move vector type computation
10076 for the BB case here from ...
10077 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
10078 live operation processing in the SLP case properly.
10080 2017-06-30 Richard Biener <rguenther@suse.de>
10082 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
10084 2017-06-30 Martin Liska <mliska@suse.cz>
10087 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
10088 before BUILT_IN_UNWIND_RESUME when ASAN is used.
10090 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
10092 * doc/invoke.texi (AArch64): Add missing options and remove redundant
10095 2017-06-30 Richard Biener <rguenther@suse.de>
10097 PR tree-optimization/81249
10098 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
10099 condition reduction result to original scalar type.
10101 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10103 * profile-count.h (enum profile_quality): Fix typos and whitespace
10106 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10108 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
10109 type for branch probabilities.
10111 2017-06-29 Julian Brown <julian@codesourcery.com>
10112 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10114 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
10115 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
10116 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
10117 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
10119 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10121 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
10122 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
10123 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
10124 CC usage from generic code to here.
10125 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
10126 CC usage into the target macros.
10128 2017-06-29 Maya Rashish <coypu@sdf.org>
10130 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
10133 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10135 * arm/arm-builtins.c: Include profile-count.h
10136 * except.c (sjlj_emit_function_enter): Use
10137 profile_probability::unlikely.
10139 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10141 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
10142 and tocrel_offset be pointer args rather than implicitly using
10144 (legitimate_constant_pool_address_p, rs6000_emit_move,
10145 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
10146 tocrel_offset and use in toc_relative_expr_p call.
10147 (print_operand, print_operand_address): Use static tocrel_base_oac
10148 and tocrel_offset_oac.
10149 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
10152 2017-06-29 Maya Rashish <coypu@sdf.org>
10154 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
10156 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
10158 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
10159 objects, take into account only the alignment of 'op0' and 'mode1' if
10162 2017-06-29 Steve Ellcey <sellcey@cavium.com>
10164 * ccmp.c (ccmp_tree_comparison_p): New function.
10165 (ccmp_candidate_p): Update to use above function.
10166 (get_compare_parts): New function.
10167 (expand_ccmp_next): Update to use new functions.
10168 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
10170 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
10171 take mode as argument.
10172 * ccmp.h (expand_ccmp_expr): Add mode as argument.
10173 * expr.c (expand_expr_real_1): Pass mode as argument.
10175 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
10177 * combine.c (combine_instructions): Print insns to dump_file, together
10180 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
10182 * asan.c (asan_emit_stack_protection): Update.
10183 (create_cond_insert_point): Update.
10184 * auto-profile.c (afdo_propagate_circuit): Update.
10185 * basic-block.h (struct edge_def): Turn probability to
10186 profile_probability.
10187 (EDGE_FREQUENCY): Update.
10188 * bb-reorder.c (find_traces_1_round): Update.
10189 (better_edge_p): Update.
10190 (sanitize_hot_paths): Update.
10191 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
10192 (make_single_succ_edge): Update.
10193 (check_bb_profile): Update.
10194 (dump_edge_info): Update.
10195 (update_bb_profile_for_threading): Update.
10196 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
10197 probabilitycount to 0.
10198 * cfgbuild.c (compute_outgoing_frequencies): Update.
10199 * cfgcleanup.c (try_forward_edges): Update.
10200 (outgoing_edges_match): Update.
10201 (try_crossjump_to_edge): Update.
10202 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
10203 (expand_gimple_tailcall): Update.
10204 (construct_init_block): Use make_single_succ_edge.
10205 (construct_exit_block): Use make_single_succ_edge.
10206 * cfghooks.c (verify_flow_info): Update.
10207 (redirect_edge_succ_nodup): Update.
10208 (split_edge): Update.
10209 (account_profile_record): Update.
10210 * cfgloopanal.c (single_likely_exit): Update.
10211 * cfgloopmanip.c (scale_loop_profile): Update.
10212 (set_zero_probability): Remove.
10213 (duplicate_loop_to_header_edge): Update.
10214 * cfgloopmanip.h (loop_version): Update prototype.
10215 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
10216 (force_nonfallthru_and_redirect): Update.
10217 (update_br_prob_note): Update.
10218 (rtl_verify_edges): Update.
10219 (purge_dead_edges): Update.
10220 (rtl_lv_add_condition_to_bb): Update.
10221 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
10222 * cgraphunit.c (init_lowered_empty_function): Update.
10223 (cgraph_node::expand_thunk): Update.
10224 * cilk-common.c: Include profile-count.h
10225 * dojump.c (inv): Remove.
10226 (jumpifnot): Update.
10227 (jumpifnot_1): Update.
10228 (do_jump_1): Update.
10230 (do_jump_by_parts_greater_rtx): Update.
10231 (do_compare_rtx_and_jump): Update.
10232 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
10233 do_jump_1. do_compare_rtx_and_jump): Update prototype.
10234 * dwarf2cfi.c: Include profile-count.h
10235 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
10236 (sjlj_emit_dispatch_table): Likewise.
10237 * explow.c: Include profile-count.h
10238 * expmed.c (emit_store_flag_force): Update.
10239 (do_cmp_and_jump): Update.
10240 * expr.c (compare_by_pieces_d::generate): Update.
10241 (compare_by_pieces_d::finish_mode): Update.
10242 (emit_block_move_via_loop): Update.
10243 (store_expr_with_bounds): Update.
10244 (store_constructor): Update.
10245 (expand_expr_real_2): Update.
10246 (expand_expr_real_1): Update.
10247 * expr.h (try_casesi, try_tablejump): Update prototypes.
10248 * gimple-pretty-print.c (dump_probability): Update.
10249 (dump_profile): New.
10250 (dump_gimple_label): Update.
10251 (dump_gimple_bb_header): Update.
10252 * graph.c (draw_cfg_node_succ_edges): Update.
10253 * hsa-gen.c (convert_switch_statements): Update.
10254 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
10255 (find_if_case_1): Update.
10256 (find_if_case_2): Update.
10257 * internal-fn.c (expand_arith_overflow_result_store): Update.
10258 (expand_addsub_overflow): Update.
10259 (expand_neg_overflow): Update.
10260 (expand_mul_overflow): Update.
10261 (expand_vector_ubsan_overflow): Update.
10262 * ipa-cp.c (good_cloning_opportunity_p): Update.
10263 * ipa-split.c (split_function): Use make_single_succ_edge.
10264 * ipa-utils.c (ipa_merge_profiles): Update.
10265 * loop-doloop.c (add_test): Update.
10266 (doloop_modify): Update.
10267 * loop-unroll.c (compare_and_jump_seq): Update.
10268 (unroll_loop_runtime_iterations): Update.
10269 * lra-constraints.c (lra_inheritance): Update.
10270 * lto-streamer-in.c (input_cfg): Update.
10271 * lto-streamer-out.c (output_cfg): Update.
10272 * mcf.c (adjust_cfg_counts): Update.
10273 * modulo-sched.c (sms_schedule): Update.
10274 * omp-expand.c (expand_omp_for_init_counts): Update.
10275 (extract_omp_for_update_vars): Update.
10276 (expand_omp_ordered_sink): Update.
10277 (expand_omp_for_ordered_loops): Update.
10278 (expand_omp_for_generic): Update.
10279 (expand_omp_for_static_nochunk): Update.
10280 (expand_omp_for_static_chunk): Update.
10281 (expand_cilk_for): Update.
10282 (expand_omp_simd): Update.
10283 (expand_omp_taskloop_for_outer): Update.
10284 (expand_omp_taskloop_for_inner): Update.
10285 * omp-simd-clone.c (simd_clone_adjust): Update.
10286 * optabs.c (expand_doubleword_shift): Update.
10287 (expand_abs): Update.
10288 (emit_cmp_and_jump_insn_1): Update.
10289 (expand_compare_and_swap_loop): Update.
10290 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
10291 * predict.c (predictable_edge_p): Update.
10292 (edge_probability_reliable_p): Update.
10293 (set_even_probabilities): Update.
10294 (combine_predictions_for_insn): Update.
10295 (combine_predictions_for_bb): Update.
10296 (propagate_freq): Update.
10297 (estimate_bb_frequencies): Update.
10298 (force_edge_cold): Update.
10299 * profile-count.c (profile_count::dump): Add missing space into dump.
10300 (profile_count::debug): Add newline.
10301 (profile_count::differs_from_p): Explicitly convert to unsigned.
10302 (profile_count::stream_in): Update.
10303 (profile_probability::dump): New member function.
10304 (profile_probability::debug): New member function.
10305 (profile_probability::differs_from_p): New member function.
10306 (profile_probability::differs_lot_from_p): New member function.
10307 (profile_probability::stream_in): New member function.
10308 (profile_probability::stream_out): New member function.
10309 * profile-count.h (profile_count_quality): Rename to ...
10310 (profile_quality): ... this one.
10311 (profile_probability): New.
10312 (profile_count): Update.
10313 * profile.c (compute_branch_probabilities): Update.
10314 * recog.c (peep2_attempt): Update.
10315 * sched-ebb.c (schedule_ebbs): Update.
10316 * sched-rgn.c (find_single_block_region): Update.
10317 (compute_dom_prob_ps): Update.
10318 (schedule_region): Update.
10319 * sel-sched-ir.c (compute_succs_info): Update.
10320 * stmt.c (struct case_node): Update.
10321 (do_jump_if_equal): Update.
10322 (get_outgoing_edge_probs): Update.
10323 (conditional_probability): Update.
10324 (emit_case_dispatch_table): Update.
10325 (expand_case): Update.
10326 (expand_sjlj_dispatch_table): Update.
10327 (emit_case_nodes): Update.
10328 * targhooks.c: Update.
10329 * tracer.c (better_p): Update.
10330 (find_best_successor): Update.
10331 * trans-mem.c (expand_transaction): Update.
10332 * tree-call-cdce.c: Update.
10333 * tree-cfg.c (gimple_split_edge): Upate.
10334 (move_sese_region_to_fn): Upate.
10335 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
10336 * tree-eh.c (lower_resx): Upate.
10337 (cleanup_empty_eh_move_lp): Upate.
10338 * tree-if-conv.c (version_loop_for_if_conversion): Update.
10339 * tree-inline.c (copy_edges_for_bb): Update.
10340 (copy_cfg_body): Update.
10341 * tree-parloops.c (gen_parallel_loop): Update.
10342 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
10343 (gimple_gen_time_profiler): Update.
10344 * tree-ssa-dce.c (remove_dead_stmt): Update.
10345 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10346 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
10347 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
10348 (unloop_loops): Update.
10349 (try_peel_loop): Update.
10350 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10351 * tree-ssa-loop-split.c (connect_loops): Update.
10352 (split_loop): Update.
10353 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
10354 (hoist_guard): Update.
10355 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
10356 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
10357 (value_replacement): Update.
10358 * tree-ssa-reassoc.c (branch_fixup): Update.
10359 * tree-ssa-tail-merge.c (replace_block_by): Update.
10360 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
10361 (create_edge_and_update_destination_phis): Update.
10362 (compute_path_counts): Update.
10363 (recompute_probabilities): Update.
10364 (update_joiner_offpath_counts): Update.
10365 (freqs_to_counts_path): Update.
10366 (duplicate_thread_path): Update.
10367 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
10368 (struct switch_conv_info): Update.
10369 (gen_inbound_check): Update.
10370 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10371 (vect_do_peeling): Update.
10372 (vect_loop_versioning): Update.
10373 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10374 (optimize_mask_stores): Update.
10375 * ubsan.c (ubsan_expand_null_ifn): Update.
10376 * value-prof.c (gimple_divmod_fixed_value): Update.
10377 (gimple_divmod_fixed_value_transform): Update.
10378 (gimple_mod_pow2): Update.
10379 (gimple_mod_pow2_value_transform): Update.
10380 (gimple_mod_subtract): Update.
10381 (gimple_mod_subtract_transform): Update.
10382 (gimple_ic): Update.
10383 (gimple_stringop_fixed_value): Update.
10384 (gimple_stringops_transform): Update.
10385 * value-prof.h: Update.
10387 2017-06-29 Carl Love <cel@us.ibm.com>
10389 * config/rs6000/rs6000-c.c: Add support for built-in functions
10390 vector signed int vec_signed (vector float);
10391 vector signed long long vec_signed (vector double);
10392 vector signed int vec_signed2 (vector double, vector double);
10393 vector signed int vec_signede (vector double);
10394 vector signed int vec_signedo (vector double);
10395 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
10396 instruction generator.
10397 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10398 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
10399 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
10401 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
10402 vunsignede_v2df): Add define_expands.
10403 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
10404 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
10405 VEC_UNSIGNEDO): Add definitions.
10406 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
10407 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
10408 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
10409 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
10410 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
10411 * config/rs6000/altivec.h (vec_signed, vec_signed2,
10412 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
10413 vec_unsignede, vec_unsignedo): Add builtin defines.
10414 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
10416 * doc/extend.texi: Update the built-in documentation file for the
10417 new built-in functions.
10419 2017-06-29 Richard Biener <rguenther@suse.de>
10421 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
10422 reduction chains to LOOP_VINFO_REDUCTIONS.
10423 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
10424 SLP reductions after processing reduction chains.
10426 2017-06-29 Nathan Sidwell <nathan@acm.org>
10428 * builtins.c (fold_builtin_FUNCTION): Use
10429 lang_hooks.decl_printable_name.
10431 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
10433 PR middle-end/81194
10434 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
10435 with only one label.
10436 * stmt.c (expand_case): Assert NCASES is greater than one.
10438 2017-06-29 Richard Biener <rguenther@suse.de>
10440 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
10442 (group_case_labels): Likewise.
10443 (find_taken_edge): Push sanity checking on val to workers...
10444 (find_taken_edge_cond_expr): ... here
10445 (find_taken_edge_switch_expr): ... and here, handle cases
10446 with just a default label.
10447 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
10448 (group_case_labels): Likewise.
10449 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
10450 group_case_labels does anything cleanup the CFG again.
10452 2017-06-29 Bin Cheng <bin.cheng@arm.com>
10454 PR tree-optimization/81196
10455 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
10456 exit condition comparing two IVs.
10458 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
10460 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
10461 profile to the dummy entry at the end of the list of architectures.
10462 * config/arm/arm-cpu-cdata.h: Regenerated.
10464 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10465 Michael Collison <michael.collison@arm.com>
10468 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
10470 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
10471 (*aarch64_reg_<mode>3_minus_mask): New pattern.
10472 (*aarch64_<optab>_reg_di3_mask2): New pattern.
10473 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
10474 of shift when the shift amount is masked with constant equal to
10475 the size of the mode.
10476 * config/aarch64/predicates.md (subreg_lowpart_operator): New
10479 2017-06-29 Martin Liska <mliska@suse.cz>
10481 * config/i386/i386.opt: Change range from [1,5] to [0,5].
10483 2017-06-29 Yury Gribov <tetra2005@gmail.com>
10486 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
10489 (edge_growth_cache_entry::edge_growth_cache_entry): New
10491 (reset_edge_growth_cache): Update to use constructor.
10493 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10495 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
10496 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10497 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
10499 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
10501 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
10502 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
10504 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
10506 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
10507 (*-linux-uclibc*): Add t-uclibc tmake_file.
10508 * config/t-musl: New.
10509 * config/t-uclibc: New.
10511 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
10513 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
10515 (gen_comm_data): Emit architectural setting of arch_prof.
10516 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
10518 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
10519 (armv8-m.base, armv8-m.main): Likewise.
10520 * arm-protos.h (arm_build_target): Add profile field.
10521 (arch_option): Likewise.
10522 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
10524 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
10525 arm_active_target.profile.
10527 2017-06-28 Richard Biener <rguenther@suse.de>
10529 PR middle-end/81227
10530 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
10531 TYPE_OVERFLOW_WRAPS.
10532 * match.pd (negate_expr_p): Likewise.
10533 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
10534 fold_build2, not fold_binary.
10536 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10538 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
10539 Convert memory address to Pmode.
10540 (aarch64_print_operand): Assert MEM operands are always Pmode.
10542 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
10545 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
10546 Remove redundant if.
10547 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
10548 * config/arm/aarch-common-protos.h
10549 (aarch_forward_to_shift_is_not_shifted_re): Remove.
10550 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
10552 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
10555 * multiple_target.c (create_dispatcher_calls): Set the default
10556 clone to be static, not public.
10558 2017-06-28 Richard Biener <rguenther@suse.de>
10560 * tree-vect-loop.c (vectorizable_reduction): Move special
10561 cond reduction IV var creation ...
10562 (vect_create_epilog_for_reduction): ... here. Remove induction_index
10563 parameter. Use STMT_VINFO_VECTYPE.
10564 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
10567 2017-06-28 Martin Liska <mliska@suse.cz>
10570 * ipa-visibility.c (non_local_p): Handle visibility.
10572 2017-06-28 Martin Liska <mliska@suse.cz>
10575 * common.opt: Add IntegerRange to various options.
10576 * opt-functions.awk (integer_range_info): New function.
10577 * optc-gen.awk: Add integer_range_info to cl_options struct.
10578 * opts-common.c (decode_cmdline_option): Handle
10579 CL_ERR_INT_RANGE_ARG.
10580 (cmdline_handle_error): Likewise.
10581 * opts.c (print_filtered_help): Show valid interval in
10582 when --help is provided.
10583 * opts.h (struct cl_option): Add range_min and range_max fields.
10584 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
10586 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
10588 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
10589 (x * C EQ/NE y * C): New transformation.
10591 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
10593 * genmultilib (combination_space): Accept '+' in option names.
10595 2017-06-28 Martin Liska <mliska@suse.cz>
10598 * asan.c (instrument_derefs): Bail out inner references
10599 that are hard register variables.
10601 2017-06-28 Jakub Jelinek <jakub@redhat.com>
10604 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
10605 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
10607 2017-06-28 Richard Biener <rguenther@suse.de>
10609 * tree-vectorizer.h (vect_get_vec_defs): Remove.
10610 (vect_get_slp_defs): Adjust.
10611 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
10613 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
10615 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10616 get_initial_defs_for_reduction instead of vect_get_vec_defs.
10617 (vectorizable_reduction): Adjust.
10618 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
10620 (vect_get_slp_defs): Likewise.
10621 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
10622 (vectorizable_bswap): Adjust.
10623 (vectorizable_call): Likewise.
10624 (vectorizable_conversion): Likewise.
10625 (vectorizable_assignment): Likewise.
10626 (vectorizable_shift): Likewise.
10627 (vectorizable_operation): Likewise.
10628 (vectorizable_store): Likewise.
10629 (vectorizable_condition): Likewise.
10630 (vectorizable_comparison): Likewise.
10632 2017-06-28 Michael Collison <michael.collison@arm.com>
10635 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
10637 (arm_gen_movmemqi): Removed unused variable 'i'.
10638 Convert 'for' loop into 'while' loop.
10639 (arm_expand_prologue): Remove last unnecessary set of insn.
10640 (thumb_pop): Remove unused variable 'pushed_words'.
10641 (thumb_exit): Remove last unnecessary set of regs_to_pop.
10643 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10645 * config/s390/predicates.md: Use s390_rel_address_ok_p.
10646 * config/s390/s390-protos.h: Add prototype of
10647 s390_rel_address_ok_p.
10648 * config/s390/s390.c (s390_got_symbol): New function.
10649 (s390_rel_address_ok_p): New function.
10650 (legitimize_pic_address): Use s390_rel_address_ok_p.
10651 (s390_load_got): Use s390_got_symbol.
10652 (s390_option_override): Issue error if
10653 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
10654 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
10656 * config/s390/s390.opt: New option mpic-data-is-text-relative.
10658 2017-06-27 Andrew Pinski <apinski@cavium.com>
10660 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
10661 (X * copysign (1.0, X)): New pattern.
10662 (X * copysign (1.0, -X)): New pattern.
10663 (copysign (-1.0, CST)): New pattern.
10665 2017-06-27 Joseph Myers <joseph@codesourcery.com>
10667 * genmultilib (combination_space): Remove variable.
10668 Validate reuse rules against regular expression for any sequence
10669 of multilib options in any order.
10671 2017-06-27 Michael Collison <michael.collison@arm.com>
10673 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
10674 call aarch64_split_simd_combine.
10675 * (aarch64_combine_internal<mode>): Delete pattern.
10676 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
10677 Allow register and subreg operands.
10679 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10681 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
10682 specific need, just fallback on defaults.
10683 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
10685 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10686 Olivier Hainque <hainque@adacore.com>
10688 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
10690 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
10691 targets. Pick a default if no particular attempt applied.
10692 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
10695 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10697 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
10698 (x86_64-wrs-vxworks7): Likewise.
10700 2017-06-27 Marek Polacek <polacek@redhat.com>
10703 * ubsan.c (instrument_null): Check get_base_address's result for null.
10705 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10707 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
10709 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10711 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
10712 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
10713 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
10714 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
10715 New function types.
10716 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
10717 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
10718 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
10719 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
10720 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
10721 BUILT_IN_FEUPDATEENV): New builtins.
10722 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
10723 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
10724 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
10725 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
10727 (builtin_structptr_types): Adjust size.
10728 * tree.c (builtin_structptr_types): Add four entries.
10730 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10731 Olivier Hainque <hainque@adacore.com>
10733 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
10734 (TLS_SYM): New local macro, forcing reference to __tls__ on
10735 link command lines for VxWorks 7 RTPs, triggering initialization
10737 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
10738 OS features TLS support, true for RTPs on VxWorks 7.
10739 * config/vxworks.c (vxworks_override_options): Setup emutls
10742 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10744 * predict.c (test_prediction_value_range): Use -1U instead of -1
10745 to avoid narrowing conversion warning.
10746 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
10747 to avoid narrowing conversion warning.
10748 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
10750 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
10752 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10754 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
10755 64bit configurations.
10756 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
10757 (SIZE_TYPE): Likewise.
10758 * config/vxworks.c (vxworks_emutls_var_fields): Use
10759 long_unsigned_type_node instead of unsigned_type_node as the offset
10760 field type, which is "pointer" mode in emutls.c.
10762 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10765 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
10767 PR middle-end/81207
10768 * gimple-fold.c (replace_call_with_call_and_fold): Handle
10769 gimple_vuse copying separately from gimple_vdef copying.
10771 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10773 * value-prof.c (free_hist): Remove call to memset and the enclosing if
10776 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10777 Olivier Hainque <hainque@adacore.com>
10779 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
10780 for all vxworks7 targets.
10781 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
10782 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
10783 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
10784 variations for VX6/VX7 and 32/64bits later on in ...
10785 (VXWORKS_LIB_SPEC): Leverage new macros.
10786 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
10787 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
10789 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10791 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
10792 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
10794 2017-06-26 Carl Love <cel@us.ibm.com>
10796 * config/rs6000/rs6000-c.c: Add support for built-in functions
10797 vector bool char vec_reve (vector bool char);
10798 vector signed char vec_reve (vector signed char);
10799 vector unsigned char vec_reve (vector unsigned char);
10800 vector bool int vec_reve (vector bool int);
10801 vector signed int vec_reve (vector signed int);
10802 vector unsigned int vec_reve (vector unsigned int);
10803 vector bool long long vec_reve (vector bool long long);
10804 vector signed long long vec_reve (vector signed long long);
10805 vector unsigned long long vec_reve (vector unsigned long long);
10806 vector bool short vec_reve (vector bool short);
10807 vector signed short vec_reve (vector signed short);
10808 vector double vec_reve (vector double);
10809 vector float vec_reve (vector float);
10810 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
10811 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
10812 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
10813 (altivec_vreve): New pattern.
10814 * config/rs6000/altivec.h (vec_reve): New define.
10815 * doc/extend.texi (vec_rev): Update the built-in documentation file
10816 for the new built-in functions.
10818 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10820 PR tree-optimization/71815
10821 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
10823 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
10825 (slsr_process_phi): Likewise.
10826 (replace_uncond_cands_and_profitable_phis): Don't replace a
10827 multiply candidate with a stride of 1 (copy or cast).
10828 (phi_incr_cost): Call uses_consumed_by_stmt rather than
10830 (lowest_cost_path): Likewise.
10831 (total_savings): Likewise.
10833 2017-06-26 Richard Biener <rguenther@suse.de>
10836 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
10837 Use def_builtin_pure for all gather builtins.
10839 2017-06-26 Richard Biener <rguenther@suse.de>
10841 PR tree-optimization/81203
10842 * tree-tailcall.c (find_tail_calls): Do not move stmts into
10843 non-dominating BBs.
10845 2017-06-26 Marek Polacek <polacek@redhat.com>
10848 * doc/invoke.texi: Document -Wmultistatement-macros.
10850 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
10852 * doc/sourcebuild.texi (ARM-specific attributes): Document new
10853 arm_neon_ok_no_float_abi effective target.
10855 2017-06-26 Richard Biener <rguenther@suse.de>
10857 PR tree-optimization/80928
10858 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
10859 (copy_bbs): Set BB_DUPLICATED flag early.
10860 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
10862 (execute_on_shrinking_pred): Likewise.
10863 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
10864 BB_DUPLICATED blocks.
10865 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
10866 iterate over all PHIs considering removal of *gsi.
10868 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
10870 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
10873 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10875 * config/rs6000/rs6000-string.c: (expand_block_clear,
10876 do_load_for_compare, select_block_compare_mode,
10877 compute_current_alignment, expand_block_compare,
10878 expand_strncmp_align_check, expand_strn_compare,
10879 expand_block_move, rs6000_output_load_multiple)
10880 Move functions related to string/block move/compare
10881 to a separate file.
10882 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
10883 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
10884 for this function which is now used in two files.
10885 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
10886 * config.gcc: Add rs6000-string.o to extra_objs for
10887 targets powerpc*-*-* and rs6000*-*-*.
10889 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
10892 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
10893 32-bit, since indexed is not valid for DImode.
10894 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
10895 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
10896 (define_peephole2 for Altivec d-form load): Add 32-bit support.
10897 (define_peephole2 for Altivec d-form store): Likewise.
10900 * multiple_target.c (create_dispatcher_calls): Only create the
10901 dispatcher call if the function is the default clone of a
10902 versioned function.
10904 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
10906 PR middle-end/80902
10907 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
10908 a call, force the call to not be a tail call.
10910 2017-06-23 Jeff Law <law@redhat.com>
10912 * doc/contrib.texi: Add entry for Steven Pemberton's work on
10915 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
10917 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
10918 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
10919 handling for early expansion of vector shifts (sl,sr,sra,rl).
10920 (builtin_function_type): Add vector shift right instructions
10921 to the unsigned argument list.
10923 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
10925 rtl-optimizatoin/79286
10926 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
10927 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
10928 trap. PIC register plus a const unspec without offset can never trap.
10930 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
10932 * tree.h (builtin_structptr_type): New type.
10933 (builtin_structptr_types): Declare new array.
10934 * tree.c (builtin_structptr_types): New array.
10935 (free_lang_data, build_common_tree_nodes): Use it.
10937 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
10940 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
10943 2017-06-22 Matt Turner <mattst88@gmail.com>
10945 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
10946 Lake models to skylake case. Assume skylake for unknown
10947 models with clflushopt.
10949 2017-06-22 Jeff Law <law@redhat.com>
10951 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
10952 frame sizes that do not satisfy aarch64_uimm12_shift.
10954 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
10956 * profile-count.h (apply_probability,
10957 apply_scale, probability_in): Fix checks for zero.
10959 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10961 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
10962 * doc/cppdiropts.texi (-I @var{dir}): Document it.
10964 2016-06-22 Richard Biener <rguenther@suse.de>
10966 * tree-vect-loop.c (vect_model_reduction_cost): Handle
10967 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
10968 REDUC_MAX_EXPR support.
10969 (vectorizable_reduction): Likewise.
10970 (vect_create_epilog_for_reduction): Likewise.
10972 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
10974 * match.pd (A / (1 << B) -> A >> B): New.
10975 * generic-match-head.c: Include optabs-tree.h.
10976 * gimple-match-head.c: Likewise.
10977 * optabs-tree.h (target_supports_op_p): New.
10978 * optabs-tree.c (target_supports_op_p): New.
10980 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10982 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
10983 $gcc_cv_ld --help output.
10984 (gcc_cv_ld_demangle): Likewise.
10985 (gcc_cv_ld_eh_frame_hdr): Likewise.
10986 (gcc_cv_ld_pie): Likewise.
10987 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
10988 (gcc_cv_ld_buildid): Likewise.
10989 (gcc_cv_ld_sysroot): Likewise.
10990 (ld_bndplt_support): Likewise.
10991 (ld_pushpopstate_support): Likewise.
10992 * configure: Regenerate.
10993 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
10995 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10998 * config/i386/sse.md (round<mode>2): Renumber match_dup and
10999 operands indexes to avoid gap between operands and match_dups.
11001 2017-06-21 Andrew Pinski <apinski@cavium.com>
11003 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
11004 Increment Arith_shift and Arith_shift_reg by 1.
11005 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
11007 * config/aarch64/aarch64.c (thunderx_tunings): Enable
11008 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
11009 (aarch64_strip_extend): Add new argument and test for it.
11010 (aarch64_cheap_mult_shift_p): New function.
11011 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
11012 add a cost if it is true.
11013 Update calls to aarch64_strip_extend.
11014 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
11016 2017-06-21 Andrew Pinski <apinski@cavium.com>
11018 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
11020 (thunderxt88): Likewise.
11021 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
11022 (thunderx_prefetch_tune): New variable.
11023 (thunderx2t99_prefetch_tune): Update for the correct values.
11024 (thunderxt88_tunings): New variable.
11025 (thunderx_tunings): Use thunderx_prefetch_tune instead of
11026 generic_prefetch_tune.
11027 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
11029 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11031 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
11032 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
11033 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
11034 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
11035 (aarch64_atomic_cas<mode>, GPI): Likewise.
11037 2017-06-21 Martin Liska <mliska@suse.cz>
11039 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
11040 statements on cold and hot labels.
11041 * predict.c (tree_estimate_probability_bb): Remove the
11042 prediction from this place.
11044 2017-06-21 Martin Liska <mliska@suse.cz>
11046 PR tree-optimization/79489
11047 * gimplify.c (maybe_add_early_return_predict_stmt): New
11049 (gimplify_return_expr): Call the function.
11050 * predict.c (tree_estimate_probability_bb): Remove handling
11052 * predict.def: Update comment about early return predictor.
11053 * gimple-predict.h (is_gimple_predict): New function.
11054 * predict.def: Change default value of early return to 66.
11055 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
11057 * passes.def: Put pass_strip_predict_hints to the beginning of
11060 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
11062 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
11063 FUNCTION_DECL declarations.
11064 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
11066 (dwaf2out_decl): Likewise.
11067 * godump.c (go_early_global_decl): Skip call to the real debug hook
11068 for FUNCTION_DECL declarations.
11069 * passes.c (rest_of_decl_compilation): Skip call to the
11070 early_global_decl debug hook for FUNCTION_DECL declarations, unless
11071 -fdump-go-spec is passed.
11073 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11075 * config/i386/i386.c (struct builtin_isa): New field pure_p.
11076 Reorder for compactness.
11077 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
11078 (def_builtin_pure, def_builtin_pure2): New functions.
11079 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
11081 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
11083 * match.pd (nop_convert): New predicate.
11084 ((A +- CST1) +- CST2): Allow some NOP conversions.
11086 2017-06-21 Jakub Jelinek <jakub@redhat.com>
11089 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
11090 with ctors/dtors if GOVD_SHARED is set.
11092 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11094 * config/aarch64/aarch64.md (movti_aarch64):
11095 Emit mov rather than orr.
11096 (movtf_aarch64): Likewise.
11097 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
11098 Emit mov rather than orr.
11100 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11102 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
11103 Swap alternatives, make integer dup more expensive.
11105 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
11107 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
11108 Return true for non-tls symbols.
11110 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
11112 * config/aarch64/aarch64-cores.def (cortex-a55): New.
11113 (cortex-a75): Likewise.
11114 (cortex-a75.cortex-a55): Likewise.
11115 * config/aarch64/aarch64-tune.md: Regenerate.
11116 * doc/invoke.texi (-mtune): Document new values for -mtune.
11118 2017-06-21 Tom de Vries <tom@codesourcery.com>
11120 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
11121 stack_size feature.
11122 (Effective-Target Keywords, Other attributes): Suggest using
11123 dg-add-options stack_size feature to get stack limit in stack_size
11124 effective target documentation.
11126 2017-06-21 Julian Brown <julian@codesourcery.com>
11127 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11129 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
11130 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
11131 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
11133 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
11134 attribute type list for neon_multiply.
11135 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
11136 attribute type list for neon_multiply.
11137 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
11138 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
11139 attribute type list for neon_multiply.
11140 * config/arm/types.md (crypto_pmull): Add.
11141 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
11142 attribute type list.
11144 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
11146 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
11149 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11151 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
11152 to make sure not to dereference a NULL cost_classes_ptr pointer.
11154 2017-06-20 Carl Love <cel@us.ibm.com>
11156 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11157 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11158 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11159 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11160 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11161 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11162 VMULOSW): New enum "unspec" values.
11163 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11164 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11165 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11166 altivec_vmulosw): New patterns.
11167 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11168 VMULOSW): Add definitions.
11170 2017-06-20 Julia Koval <julia.koval@intel.com>
11172 * config/i386/i386.c: Fix rounding expand for new pattern.
11173 * config/i386/subst.md: Fix pattern (parallel -> unspec).
11175 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11177 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
11178 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
11180 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11182 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
11185 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
11187 * config/aarch64/aarch64-cores.def: Rearrange to sort by
11188 architecture, then by implementer ID.
11189 * config/aarch64/aarch64-tune.md: Regenerate.
11191 2017-06-20 Richard Biener <rguenther@suse.de>
11193 PR middle-end/81097
11194 * fold-const.c (split_tree): Fold to type before negating.
11196 2017-06-20 David Malcolm <dmalcolm@redhat.com>
11198 * diagnostic-show-locus.c
11199 (selftest::test_fixit_deletion_affecting_newline): New function.
11200 (selftest::diagnostic_show_locus_c_tests): Call it.
11202 2017-06-20 Andreas Schwab <schwab@suse.de>
11205 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
11208 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
11210 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
11211 __ARM_FEATURE_COPROC according to support.
11213 2017-06-20 Jakub Jelinek <jakub@redhat.com>
11215 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
11216 Rewritten to avoid overflow for > 32-bit pointers.
11219 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
11220 by removing enum keyword.
11221 (ubsan_type_descriptor): Likewise. Formatting fix.
11224 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
11225 splitter): Require TARGET_SSE2 in the condition.
11227 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
11230 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
11231 for doing vector set of SFmode on ISA 3.0.
11232 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
11233 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
11235 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
11236 SFmode value into a V4SF variable that was extracted from another
11237 V4SF variable without converting the element to double precision
11238 and back to single precision vector format.
11239 (vsx_insert_extract_v4sf_p9_2): Likewise.
11241 2017-06-19 Jakub Jelinek <jakub@redhat.com>
11243 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
11244 in UWHI to avoid undefined overflow.
11247 * ubsan.h (enum ubsan_encode_value_phase): New.
11248 (ubsan_encode_value): Change second argument to
11249 enum ubsan_encode_value_phase with default value of
11250 UBSAN_ENCODE_VALUE_GENERIC.
11251 * ubsan.c (ubsan_encode_value): Change second argument to
11252 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
11253 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
11254 create_tmp_var_raw instead of create_tmp_var and use a
11256 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
11257 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
11258 ubsan_encode_value callers.
11261 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
11262 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
11263 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
11265 2017-06-19 Richard Biener <rguenther@suse.de>
11267 PR middle-end/81118
11268 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
11269 estimates if we changed anything.
11271 2017-06-19 Richard Biener <rguenther@suse.de>
11273 PR tree-optimization/80887
11274 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
11275 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
11276 simplified lookups, then reset mprts_hook.
11277 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
11279 (try_to_simplify): Likewise.
11281 2017-06-19 Martin Liska <mliska@suse.cz>
11284 * gimplify.c (gimplify_switch_expr):
11285 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
11287 2017-06-19 Martin Liska <mliska@suse.cz>
11289 * doc/install.texi: Document that PGO runs in 4 stages.
11291 2017-06-19 Martin Liska <mliska@suse.cz>
11294 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
11295 to dispatcher function name.
11296 * multiple_target.c (replace_function_decl): New function.
11297 (create_dispatcher_calls): Redirect both edges and references.
11299 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
11301 * profile-count.c (profile_count::dump): Dump quality.
11302 (profile_count::differs_from_p): Update for unsigned val.
11303 * profile-count.h (profile_count_quality): New enum.
11304 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
11306 2017-06-19 Richard Biener <rguenther@suse.de>
11308 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
11309 struct function as arg.
11310 (estimate_numbers_of_iterations): Export overload with loop arg.
11311 (free_numbers_of_iterations_estimates_loop): Use an overload of
11312 free_numbers_of_iterations_estimates instead.
11313 * tree-cfg.c (remove_bb): Adjust.
11314 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
11315 * tree-parloops.c (gen_parallel_loop): Likewise.
11316 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
11318 (tree_unroll_loops_completely): Likewise.
11319 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
11320 Use an overload instead and export.
11321 (estimated_loop_iterations): Adjust.
11322 (max_loop_iterations): Likewise.
11323 (likely_max_loop_iterations): Likewise.
11324 (estimate_numbers_of_iterations): Take struct function as arg
11326 (loop_exits_before_overflow): Adjust.
11327 (free_numbers_of_iterations_estimates_loop): Use an overload.
11328 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
11329 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
11331 2017-06-19 Richard Biener <rguenther@suse.de>
11334 * ipa-prop.c (find_constructor_constant_at_offset): Handle
11335 RANGE_EXPR conservatively.
11337 2017-06-16 Carl Love <cel@us.ibm.com>
11339 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11340 definitions for vec_float, vec_float2, vec_floato,
11341 vec_floate built-ins.
11342 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
11343 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
11345 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
11346 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
11347 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
11348 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
11349 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
11350 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
11351 vec_floato): Add builtin defines.
11352 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
11353 Update the built-in documentation file for the new built-in
11356 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11358 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
11359 (mthumb): Mark as the negative of -marm.
11361 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11363 * doc/invoke.texi (ARM Options, -mcpu): Document supported
11365 (ARM Options, -mtune): Document that this accepts the same
11366 extension options as -mcpu.
11367 (ARM Options, -mfpu): Document addition of -mfpu=auto.
11369 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11371 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
11372 permitted extensions.
11374 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11376 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
11377 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
11378 (armv8-m.main): Add option +nodsp.
11379 * config/arm/arm-cpu-cdata.h: Regenerated.
11381 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11383 * config/arm/t-fuchsia: New file.
11384 * config.gcc (arm*-*-fuchsia*): Use it.
11386 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11388 * config/arm/t-symbian: Rewrite for new option infrastructure.
11390 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11392 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
11393 (MULTILIB_REQUIRED): Likewise.
11395 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11397 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
11398 (MULTILIB_RESUE): Likewise.
11399 (MULTILIB_MATCHES): Likewise.
11400 (MULTLIB_REQUIRED): Likewise.
11402 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11404 * config/arm/t-rtems: Rewrite for new option framework.
11406 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11408 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
11409 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
11410 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
11411 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
11412 * config/arm/t-multilib: ... here.
11413 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
11414 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
11415 armv7-a and armv8*-a when A-profile libraries have not been built.
11416 * config/arm/t-rmprofile: Rewrite.
11418 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11420 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
11421 with a backslash. Remove the backslash after substituting unescaped
11423 * doc/fragments.texi (MULTILIB_REUSE): Document it.
11425 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11427 * config.gcc: (arm*-*-*): When building a-profile libraries, force
11428 the driver to pass through the default setting of -mfloat-abi.
11429 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
11431 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
11432 (all_feat_combs): New rule.
11433 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
11435 * config/arm/t-aprofile: Rewrite.
11437 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11439 * config/arm/arm.h (FPUTYPE_AUTO): Define.
11440 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
11441 fpu is not specified by the user/command-line.
11442 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
11443 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
11444 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
11445 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
11446 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
11447 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
11449 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11451 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
11452 * config/arm/t-arm-elf: Rewritten.
11454 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11456 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
11457 have some floating-point instructions.
11458 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
11459 (TARGET_MAYBE_HARD_FLOAT): New macro.
11460 * config/arm/arm-builtins.c (arm_init_builtins): Use
11461 TARGET_MAYBE_HARD_FLOAT.
11462 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
11464 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11466 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
11467 (configargs.h): Include it.
11468 (arm_print_hint_for_fpu_option): New function.
11469 (arm_parse_fpu_option): New function.
11470 (candidate_extension): New class.
11471 (arm_canon_for_multilib): New function.
11472 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
11473 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
11474 (ARCH_CANONICAL_SPECS): New macro.
11475 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
11477 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11479 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
11480 are set after handling multilib fragments. Set target_cpu_default2
11483 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11485 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
11487 (arm*-*-*): Set target_cpu_default2 to a quoted string.
11488 * config/arm/parsecpu.awk (check_cpu): Validate any extension
11490 (check_arch): Likewise.
11491 * config/arm/arm.c (arm_configure_build_target): Handle
11492 TARGET_CPU_DEFAULT being a string constant. Scan any feature
11493 options in the default.
11495 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11497 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
11498 when an option is an alias of another.
11499 * config/arm/parsecpu.awk (optalias): New parser token.
11500 (gen_comm_data): Mark non-alias options as such. Emit entries
11501 for extension aliases.
11502 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
11503 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
11504 (armv6kz, armv6zk, armv6t2): Likewise.
11505 (armv7): Make vfpv3-d16 an alias.
11506 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
11508 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
11509 Sort in canonical order.
11510 (armv8-a): Sort in canonical order.
11511 (armv8.1-a, armv8.2-a): Likewise.
11512 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
11514 (cortex-a9): Sort in canonical order.
11515 * config/arm/arm.c (selftests.h): Include it.
11516 (arm_test_cpu_arch_data): New function.
11517 (arm_run_self_tests): New function.
11518 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
11519 (targetm): Move declaration to the end of the file.
11520 * arm-cpu-cdata.h: Regenerated.
11522 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11524 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
11525 call to target_mode_check describing the type of option passed.
11526 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
11527 (arm_target_thumb_only): Use arm_parse_arch_option_name or
11528 arm_parse_cpu_option_name to match parameters against list of
11530 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
11531 arm_arch_core_flags data structure.
11532 * config/arm/arm-cpu_cdata.h: Regenerated.
11534 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11536 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
11538 (arm_print_hint_for_cpu_option): Likewise.
11539 (arm_print_hint_for_arch_option): Likewise.
11540 (arm_parse_cpu_option_name): Likewise.
11541 (arm_parse_arch_option_name): Likewise.
11542 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
11543 of entries in the all_fpus list.
11544 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
11545 (arm_parse_cpu_option_name): Declare.
11546 (arm_parse_arch_option_name): Declare.
11547 (arm_parse_option_features): Declare.
11548 (arm_intialize_isa): Declare.
11549 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
11551 (gen_comm_data): ... here. Make definitions non-static.
11552 * config/arm/arm-cpu-data.h: Regenerated.
11553 * config/arm/arm-cpu-cdata.h: Regenerated.
11555 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11557 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
11558 (cpu_arch_extension): New structure.
11559 (cpu_arch_option, arch_option, cpu_option): New structures.
11560 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
11561 architecture types.
11562 (gen_data): Generate new format data tables.
11563 * config/arm/arm.c (cpu_tune): New structure.
11564 (cpu_option, processors): Delete.
11565 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
11566 (arm_print_hint_for_cpu_option): ... this and ...
11567 (arm_print_hint_for_arch_option): ... this.
11568 (arm_parse_arch_cpu_name): Delete. Replace with ...
11569 (arm_parse_cpu_option_name): ... this and ...
11570 (arm_parse_arch_option_name): ... this.
11571 (arm_unrecognized_feature): Change type of target parameter to
11573 (arm_parse_arch_cpu_features): Delete. Replace with ...
11574 (arm_parse_option_features): ... this.
11575 (arm_configure_build_target): Rework to use new configuration data
11577 (arm_print_tune_info): Rework for new configuration data tables.
11578 * config/arm/arm-cpu-data.h: Regenerated.
11579 * config/arm/arm-cpu.h: Regenerated.
11581 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11583 * Makefile.in (OBJS): Move sbitmap.o from here ...
11584 (OBJS-libcommon): ... to here.
11586 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11588 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
11589 (ISA_ALL_CRYPTO): New macro.
11590 (ISA_ALL_SIMD): New macro
11591 (ISA_ALL_FP): New macro.
11592 * config/arm/arm.c (fpu_bitlist): Update initializer.
11593 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
11595 (arm9e): Add fpu. Add option for nofp
11596 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
11597 (arm926ej-s, arm1026ej-s): Likewise.
11598 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
11599 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
11600 neon-fp16, neon-vfpv4, nofp and nosimd.
11601 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
11602 (cortex-a8): Add fpu. Add option for nofp.
11603 (cortex-a9): Add fpu. Add options for nosimd and nofp.
11604 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
11605 (cortex-r4f): Add fpu.
11606 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
11607 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
11609 (cortex-r8): Likewise.
11610 (cortex-m4): Add fpu. Add option for nofp.
11611 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
11612 (cortex-a17.cortex-a7): Likewise.
11613 (cortex-a32): Add fpu. Add options for crypto and nofp.
11614 (cortex-a35, cortex-a53): Likewise.
11615 (cortex-a57): Add fpu. Add option for crypto.
11616 (cortex-a72, cortex-a73): Likewise.
11617 (exynos-m1): Likewise.
11618 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
11619 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
11620 (cortex-m33): Add fpu. Add option for nofp.
11621 * config/arm/arm-cpu-cdata.h: Regenerated
11622 * config/arm/arm-cpu-data.h: Regenerated.
11624 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11626 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
11627 (armv5te, armv5tej): Likewise.
11628 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
11629 (armv7): Add options fp and vfpv3-d16.
11630 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
11631 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
11633 (armv7ve): Likewise.
11634 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
11635 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
11636 (armv8-a): Add nocrypto option.
11637 (armv8.1-a, armv8.2-a): Likewise.
11638 (armv8-m.main): add options fp, fp.dp and nofp.
11640 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11642 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
11644 (armv8-a+crc): Delete.
11645 (armv8.1-a): Add options simd, crypto and nofp.
11646 (armv8.2-a): Add options fp16, simd, crypto and nofp.
11647 (armv8.2-a+fp16): Delete.
11648 (armv8-m.main): Add option dsp.
11649 (armv8-m.main+dsp): Delete.
11650 (cortex-a8): Add fpu. Add nofp option.
11651 (cortex-a9): Add fpu. Add nofp and nosimd options.
11652 * config/arm/parsecpu.awk (gen_data): Generate option tables and
11653 link to main cpu and architecture data structures.
11654 (gen_comm_data): Only put isa attributes from the main architecture
11656 (option): New statement for architecture and CPU entries.
11657 * arm.c (struct cpu_option): New structure.
11658 (struct processors): Add entry for options.
11659 (arm_unrecognized_feature): New function.
11660 (arm_parse_arch_cpu_name): Ignore any characters after the first
11662 (arm_parse_arch_cpu_feature): New function.
11663 (arm_configure_build_target): Separate out any CPU and architecture
11664 features and parse separately. Don't error out if -mfpu=auto is
11665 used with only an architecture string.
11666 (arm_print_asm_arch_directives): New function.
11667 (arm_file_start): Call it.
11668 * config/arm/arm-cpu-cdata.h: Regenerated.
11669 * config/arm/arm-cpu-data.h: Likewise.
11670 * config/arm/arm-tables.opt: Likewise.
11672 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11674 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
11675 assembler when it is not -mfpu=auto.
11677 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11679 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
11680 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
11681 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
11682 (ASM_CPU_SPEC): Rewrite.
11683 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
11684 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
11685 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
11686 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
11687 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
11688 copied string is NUL-terminated. Also strip any characters prefixed
11690 (arm_rewrite_selected_arch): New function.
11691 (arm_rewrite_march): New function.
11693 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11695 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
11696 (x_arm_cpu_string, x_arm_tune_string): Likewise.
11697 (march, mcpu, mtune): Convert to string-based options.
11698 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
11699 (arm_parse_arch_cpu_name): New function.
11700 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
11701 identify selected architecture or CPU.
11702 (arm_option_save): New function.
11703 (TARGET_OPTION_SAVE): Redefine.
11704 (arm_option_restore): Restore string options.
11705 (arm_option_print): Print string options.
11707 2017-06-16 Martin Sebor <msebor@redhat.com>
11709 PR tree-optimization/80933
11710 PR tree-optimization/80934
11711 * builtins.c (fold_builtin_3): Do not handle bcmp here.
11712 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
11713 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
11714 (gimple_fold_builtin): Call them.
11716 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11718 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
11719 as unlikely; update profile.
11721 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11723 * predict.c (force_edge_cold): Handle declaring edges impossible
11726 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11728 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
11730 (try_unroll_loop_completely): Fix reporting.
11732 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11734 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
11736 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
11739 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
11740 if given a non-constant argument for an intrinsic which requires a
11743 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11745 * profile.c (compare_freqs): New function.
11746 (branch_prob): Sort edge list.
11747 (find_spanning_tree): Assume that the list is priority sorted.
11749 2017-06-16 Richard Biener <rguenther@suse.de>
11751 PR tree-optimization/81090
11752 * passes.def (pass_record_bounds): Remove.
11753 * tree-pass.h (make_pass_record_bounds): Likewise.
11754 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
11755 make_pass_record_bounds): Likewise.
11756 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
11757 not free niter estimates at the beginning but at the end.
11758 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
11760 2017-06-16 Richard Biener <rguenther@suse.de>
11762 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
11763 initializer to workaround ICE in host GCC 4.8.
11765 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11767 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
11769 (clone_inlined_nodes): Update.
11771 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11773 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
11774 prefetch settings, and enable prefetching by default at -O3.
11776 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11778 * config/aarch64/aarch64.c (aarch64_override_options_internal):
11779 Set flag_prefetch_loop_arrays according to tuning data.
11781 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11783 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
11784 New tune structure.
11785 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
11786 [Unrelated to main purpose of the patch] Place the pointer field last
11787 to enable type checking errors when tune structure are wrongly merged.
11788 * config/aarch64/aarch64.c (generic_prefetch_tune,)
11789 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
11790 (thunderx2t99_prefetch_tune): New tune constants.
11791 (tune_params *_tunings): Update all tunings (no functional change).
11792 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
11793 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
11794 from tunings structures.
11796 2017-06-16 Jakub Jelinek <jakub@redhat.com>
11799 * ubsan.c (instrument_null): Add T argument, use it instead
11800 of computing it based on IS_LHS.
11801 (instrument_object_size): Likewise.
11802 (pass_ubsan::execute): Adjust instrument_null and
11803 instrument_object_size callers to pass gimple_get_lhs or
11804 gimple_assign_rhs1 result to it. Use instrument_null instead of
11805 calling get_base_address and instrument_mem_ref. Handle
11806 aggregate call arguments for object-size sanitization.
11808 2017-06-16 Yury Gribov <tetra2005@gmail.com>
11810 PR tree-optimization/81089
11811 * tree-vrp.c (is_masked_range_test): Validate operands of
11814 2017-06-15 Martin Sebor <msebor@redhat.com>
11817 * dumpfile.c (dump_register): Avoid calling memset to initialize
11818 a class with a default ctor.
11819 * gcc.c (struct compiler): Remove const qualification.
11820 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
11821 * hash-table.h: Ditto.
11822 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
11824 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
11825 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
11827 * params.h (struct param_info): Make struct members non-const.
11828 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
11829 with default initialization.
11830 * vec.h (vec_copy_construct, vec_default_construct): New helper
11832 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
11833 with vec_copy_construct.
11834 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
11835 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
11836 * doc/invoke.texi (-Wclass-memaccess): Document.
11838 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11840 * emit-rtl.h (is_leaf): Update comment about local
11841 register allocator.
11843 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11846 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
11847 for a variable to have a section before checking if the section has a
11849 Set section to.persistent if persistent attribute is set.
11850 Warn if .persistent attribute is used on an automatic variable.
11852 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
11854 PR rtl-optimization/80474
11855 * reorg.c (update_block): Do not ignore instructions in a delay slot.
11857 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
11859 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
11862 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
11864 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
11865 (casesi): Emit bounds checking as RTL.
11866 (casesi_internal_mips16_<mode>): Remove bounds checking.
11868 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
11870 * config/xtensa/xtensa.c (xtensa_option_override): Append
11871 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
11872 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
11873 xtensa_doloop_hooks): Define unconditionally.
11874 (xtensa_reorg_loops): Only call reorg_loops in the presence of
11876 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
11877 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
11878 for it in xtensa_option_override.
11879 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
11880 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
11882 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
11884 * doc/cppopts.texi: Document '-' special value to -MF.
11886 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
11888 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
11889 (cortex_a53_fconst): Likewise.
11890 (cortex_a53_fpmul): Likewise.
11891 (cortex_a53_f_load_64): Likewise.
11892 (cortex_a53_f_load_many): Likewise.
11893 (cortex_a53_advsimd_alu): Likewise.
11894 (cortex_a53_advsimd_alu_q): Likewise.
11895 (cortex_a53_advsimd_mul): Likewise.
11896 (cortex_a53_advsimd_mul_q): Likewise.
11897 (fpmac bypass): Add new bypass for fpmac-fpmac case.
11898 Add missing fmul, r2f_cvt and fconst cases.
11900 2017-06-14 Richard Biener <rguenther@suse.de>
11902 PR middle-end/81088
11903 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
11905 (fold_binary_loc): When associating do not treat pre-existing
11906 TREE_OVERFLOW on literal constants as a reason to allow
11907 TREE_OVERFLOW on associated literal constants.
11909 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11911 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
11912 (MASK_FEATURES): New macro.
11913 * config/sparc/sparc.c (sparc_option_override): Remove the special
11914 handling of -mfpu and generalize it to all MASK_FEATURES switches.
11916 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11918 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
11919 a division of 0 if non-call exceptions are enabled.
11921 2017-06-14 Andrew Pinski <apinski@cavium.com>
11922 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11925 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
11926 Improve vector initialization code gen for only variable case.
11928 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11930 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
11932 2017-06-14 Richard Biener <rguenther@suse.de>
11934 PR tree-optimization/81083
11935 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
11938 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11940 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
11941 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
11942 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
11943 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
11944 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
11945 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
11947 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11949 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
11950 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
11952 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11954 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
11956 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11958 * config/rs6000/t-rtems: Don't handle SPE.
11960 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11962 * config/rs6000/t-linux: Don't handle SPE.
11964 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11966 * config/rs6000/eabispe.h: Delete file.
11968 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11970 * config/rs6000/t-spe: Delete file.
11972 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11974 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
11975 (rs6000_legitimate_offset_address_p): Return false for anything in
11976 V2SImode or V2SFmode.
11978 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11980 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
11981 except V2SF and V2SI. Rearrange the vector modes, and add comments.
11982 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
11984 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
11985 (rs6000_legitimate_offset_address_p): Ditto.
11986 (rs6000_emit_move): Ditto.
11987 (rs6000_init_builtins): Remove V4HI_type_node.
11989 2017-06-13 Martin Liska <mliska@suse.cz>
11992 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
11993 (gate_asan): Likewise.
11994 * asan.h (asan_no_sanitize_address_p): Remove the function.
11995 (sanitize_flags_p): New function.
11996 * builtins.def: Fix coding style.
11997 * common.opt: Use renamed enum value.
11998 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
11999 * doc/extend.texi: Document no_sanitize attribute.
12000 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
12001 to SANITIZE_UNDEFINED_NONDEFAULT.
12002 * gcc.c (sanitize_spec_function): Use the renamed enum value.
12003 * gimple-fold.c (optimize_atomic_compare_exchange_p):
12004 Use sanitize_flags_p.
12005 * gimplify.c (gimplify_function_tree): Likewise.
12006 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
12007 * opts.c (parse_no_sanitize_attribute): New function.
12008 (common_handle_option): Use renamed enum value.
12009 * opts.h (parse_no_sanitize_attribute): Declare.
12010 * tree.c (sanitize_flags_p): New function.
12011 * tree.h: Declared here.
12012 * tsan.c: Use sanitize_flags_p.
12013 * ubsan.c (ubsan_expand_null_ifn): Likewise.
12014 (instrument_mem_ref): Likewise.
12015 (instrument_bool_enum_load): Likewise.
12016 (do_ubsan_in_current_function): Remove the function.
12017 (pass_ubsan::execute): Use sanitize_flags_p.
12018 * ubsan.h: Remove do_ubsan_in_current_function
12019 * tree-cfg.c (print_no_sanitize_attr_value): New function.
12020 (dump_function_to_file): Use it here.
12022 2017-06-13 Martin Jambor <mjambor@suse.cz>
12024 PR tree-optimization/80803
12025 PR tree-optimization/81063
12026 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
12027 (propagate_subaccesses_across_link): Enqueue subtree whenever
12028 necessary instead of relying on the caller.
12030 2017-06-13 Martin Jambor <mjambor@suse.cz>
12032 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
12033 that have a first_link.
12034 (sort_and_splice_var_accesses): Do not check first_link before
12036 (subtree_mark_written_and_enqueue): Likewise.
12037 (propagate_all_subaccesses): Likewise and do not stop at first
12038 parent with a first_link.
12040 2017-06-13 Martin Jambor <mjambor@suse.cz>
12042 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
12045 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12047 * match.pd: New pattern.
12049 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12051 * tree-vrp.c (is_masked_range_test): New function.
12052 (register_edge_assert_for): Determine ranges for
12055 2017-06-13 Yury Gribov <tetra2005@gmail.com>
12057 PR tree-optimization/67328
12058 * fold-const.c (maskable_range_p): New function.
12059 (build_range_check): Generate bittests if possible.
12061 2017-06-13 Martin Liska <mliska@suse.cz>
12063 * gimple-pretty-print.c (dump_probability): Add new argument.
12064 (dump_edge_probability): Dump both probability and count.
12065 (dump_gimple_label): Likewise.
12066 (dump_gimple_bb_header): Likewise.
12068 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
12071 * config/avr/avr-devices.c: Fix indentation.
12072 * config/avr/gen-avr-mmcu-specs.c: Dito.
12074 2017-06-13 Richard Biener <rguenther@suse.de>
12076 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
12077 instead get vector type from stmt_info.
12078 (vectorizable_reduction): Adjust. Remove dead code.
12080 2017-06-13 Richard Biener <rguenther@suse.de>
12082 PR middle-end/81065
12083 * fold-const.c (extract_muldiv_1): Remove bogus distribution
12084 case of C * (x * C2 + C3).
12085 (fold_addr_of_array_ref_difference): Properly fold index difference.
12087 2017-06-12 David S. Miller <davem@davemloft.net>
12090 * config/sparc/sparc.md (return expander): Emit frame blockage if
12091 function uses alloca.
12093 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
12095 * combine.c (make_field_assignment): Check len rather than the mode
12096 precision when calling force_to_mode.
12098 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
12100 Support multilibs and devices that see flash in RAM address range.
12103 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
12104 (avr_mcu_t) <flash_pm_offset>: New field.
12105 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
12106 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
12107 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
12108 (AVR_TINY_PM_OFFSET): Remove macro.
12109 * config/avr/avr.opt (-mshort-calls): New option.
12110 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12111 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
12112 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
12113 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
12114 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
12115 instead of avr_arch->have_jmp_call.
12116 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
12117 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
12118 avr_arch->flash_pm_offset to define.
12119 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
12120 new field flash_pm_offset. Add entry for avrxmega3.
12121 (avr_texinfo): Add entry for avrxmega3.
12122 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
12123 attiny212, attiny214,
12124 attiny412, attiny414, attiny416, attiny417,
12125 attiny814, attiny816, attiny817,
12126 attiny1614, attiny1616, attiny1617,
12127 attiny3214, attiny3216, attiny3217.
12128 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
12129 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
12130 (avr_print_operand_address) [AVR_TINY]: Same.
12131 (avr_asm_init_sections) <readonly_data_section>: Only patch
12132 callback if avr_arch->flash_pm_offset = 0.
12133 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
12134 for rodata if avr_arch->flash_pm_offset != 0.
12135 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
12136 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
12137 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
12138 (m_options): Append opt_rcall.
12139 (m_dirnames): Append dir_rcall.
12140 * config/avr/t-multilib: Regenerate.
12142 * configure.ac [target=avr]: Check whether avrxmega3 default
12143 linker description file works as needed.
12144 * configure: Regenerate.
12145 * doc/avr-mmcu.texi: Regenerate.
12146 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
12147 <__AVR_ARCH__>: Document avrxmega3 and 103.
12148 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
12149 <__AVR_SHORT_CALLS__>: Document it.
12150 <__AVR_PM_BASE_ADDRESS__>: Document it.
12151 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
12152 (AVR Variable Attributes) <progmem>: Document this is
12153 not needed for avrxmega3.
12154 (AVR Named Address Spaces) <__flash>: Dito.
12156 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
12158 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
12160 2017-06-12 Doug Rupp <rupp@adacore.com>
12162 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
12163 Append vxworks-stdint.h to the tm_file list.
12164 * config/vxworks-stdint.h: New file.
12166 2017-06-12 Martin Liska <mliska@suse.cz>
12168 PR tree-optimization/81041
12169 * tree-profile.c (gimple_gen_ic_func_profiler):
12170 Create an extra BB in profile-generate
12171 (gimple_gen_time_profiler): Likewise.
12173 2017-06-12 Jakub Jelinek <jakub@redhat.com>
12175 PR tree-optimization/81003
12176 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
12177 (update_range_test): Use it instead of force_gimple_operand_gsi.
12179 2017-06-12 Richard Biener <rguenther@suse.de>
12181 PR tree-optimization/81053
12182 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
12183 with backedge value not defined in loop. Simplify def stmt
12186 2017-06-11 Tom de Vries <tom@codesourcery.com>
12189 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
12191 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
12192 nvptx_cannot_force_const_mem.
12194 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12196 * opts.c (finish_options): Move test for flag_split_stack after
12197 it has been initialized.
12199 2017-06-11 Jason Merrill <jason@redhat.com>
12201 * tree.h (id_equal): New.
12202 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
12203 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
12204 instead of strcmp of IDENTIFIER_POINTER.
12206 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12208 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
12209 (mark_all_inlined_calls_cdtor): Fix formating.
12210 (inline_transform): Rescale profile before inlining.
12212 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12214 * cgraph.h (cgraph_edge::clone): Update prototype.
12215 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
12216 (cgraph_node::create_clone): Update.
12217 (cgraph_node::create_version_clone): Update.
12218 * tree-inline.c (copy_bb): Update.
12219 (expand_call_inline): Update.
12221 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12223 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
12224 factored out from ...
12225 (rs6000_emit_prologue): ... here.
12227 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
12229 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
12230 factored out from ...
12231 (rs6000_emit_prologue): ... here.
12233 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12235 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
12237 (handle_missing_profiles): Fix computation of tp_first_run.
12238 (counts_to_freqs): Do not touch freqs when count is 0.
12240 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
12242 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
12245 2017-06-10 Tom de Vries <tom@codesourcery.com>
12247 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
12248 attributes): Document signal effective target.
12250 2017-06-10 Tom de Vries <tom@codesourcery.com>
12252 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
12253 Document effective target stack_size.
12255 2017-06-09 David Malcolm <dmalcolm@redhat.com>
12257 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
12258 to the edit_context if they can be auto-applied.
12260 2017-06-9 Ian Lance Taylor <iant@golang.org>
12262 * opts.c (finish_options): If -fsplit-stack, disable implicit
12263 -forder-blocks-and-partition.
12264 * doc/invoke.texi (Optimize Options): Document that when using
12265 -fsplit-stack -forder-blocks-and-partition is not implicitly
12268 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12270 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
12271 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
12272 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
12273 * builtins.def (abort, trap, unreachable): Declare cold.
12274 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
12275 * tree-core.h (ECF_COLD): New.
12276 * tree.c (set_call_expr_flags): Handle ECF_COLD.
12277 (build_common_builtin_nodes): Mark unreachable and abort as cold.
12279 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12281 * predict.c (unlikely_executed_stmt_p): Cleanup.
12283 2017-06-09 Richard Biener <rguenther@suse.de>
12285 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
12286 model if the ref is always written to.
12288 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12290 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
12292 2017-06-09 Tamar Christina <tamar.christina@arm.com>
12294 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
12297 2017-06-09 Tom de Vries <tom@codesourcery.com>
12300 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
12301 "target cannot support label values" when encountering LABEL_REF.
12303 2017-06-09 Martin Liska <mliska@suse.cz>
12305 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
12306 (gimple_gen_ic_func_profiler): Emit direct comparison
12307 of __gcov_indirect_call_callee with NULL.
12308 (gimple_gen_time_profiler): Change probability from
12309 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
12311 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12313 * profile.c (edge_gcov_counts): Turn to pointer.
12314 (compute_branch_probabilities, compute_branch_probabilities): Update.
12315 (branch_prob): Do not clear edge_gcov_count.
12316 * profile.h (edge_gcov_counts): Turn to pointer.
12317 (edge_gcov_count): Update.
12319 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
12321 * gimple.h (gimple_check_failed): Mark cold.
12323 2017-06-09 Richard Biener <rguenther@suse.de>
12325 PR tree-optimization/66623
12326 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
12327 refactor check_reduction into two parts, properly computing
12328 whether we have to check reduction validity for outer loop
12331 2017-06-09 Richard Biener <rguenther@suse.de>
12333 PR tree-optimization/79483
12334 * graphite-scop-detection.c (order): New global.
12335 (get_order): Compute bb to order mapping that satisfies code
12336 generation constraints.
12337 (cmp_pbbs): New helper.
12338 (build_scops): Start domwalk at entry block, sort generated
12341 2017-06-09 Richard Biener <rguenther@suse.de>
12343 PR middle-end/81007
12344 * ipa-polymorphic-call.c
12345 (ipa_polymorphic_call_context::restrict_to_inner_class):
12346 Skip FIELD_DECLs with error_mark_node type.
12347 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
12350 2017-06-09 Martin Liska <mliska@suse.cz>
12352 * predict.c (struct branch_predictor): New struct.
12353 (test_prediction_value_range): New test.
12354 (predict_c_tests): New function.
12355 * selftest-run-tests.c (selftest::run_tests): Run the function.
12356 * selftest.h: Declare new tests.
12358 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
12361 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
12362 gen_add3_insn did not fail.
12363 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
12364 r0, construct that number in a temporary reg and add that reg to r0.
12365 If asked to put the result in r0 as well, fail.
12367 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
12369 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12370 for early expansion of vec_eqv.
12372 2017-06-08 Jakub Jelinek <jakub@redhat.com>
12374 PR middle-end/81005
12375 * ubsan.c (instrument_null): Avoid pointless code temporary.
12376 (pass_ubsan::execute): Instrument aggregate arguments of calls.
12378 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
12382 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
12385 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
12388 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12390 * predict.c (unlikely_executed_edge_p): Move ahead.
12391 (probably_never_executed_edge_p): Use it.
12393 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12395 PR middle-end/79988
12396 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
12397 gimple_call_builtin_p call.
12399 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12401 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
12402 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
12403 rtl_check_failed_type2, rtl_check_failed_code1,
12404 rtl_check_failed_code2, rtl_check_failed_code_mode,
12405 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
12406 rtvec_check_failed_bounds, rtl_check_failed_flag,
12407 _fatal_insn_not_found, _fatal_insn): Likewise.
12408 * tree.h (tree_contains_struct_check_failed,
12409 tree_check_failed, tree_not_check_failed,
12410 tree_class_check_failed, tree_range_check_failed,
12411 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
12412 tree_vec_elt_check_failed, phi_node_elt_check_failed,
12413 tree_operand_check_failed, omp_clause_check_failed,
12414 omp_clause_operand_check_failed, omp_clause_range_check_failed):
12417 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12419 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
12420 flag_branch_probabilities.
12421 * ipa-inline.c (edge_badness): Likewise.
12422 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12423 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
12424 * predict.c (maybe_hot_frequency_p): Likewise.
12425 (probably_never_executed): Likewise.
12426 * sched-ebb.c (schedule_ebbs): Likewise.
12427 * sched-rgn.c (find_single_block_region): Likewise.
12428 * tracer.c (tail_duplicate): Likewise.
12430 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12432 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
12433 longer requires x_flag_profile_use.
12435 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12437 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
12438 instead of flag_reorder_blocks_and_partition.
12439 * dbxout.c (dbxout_function_end): Likewise.
12440 * dwarf2out.c (gen_subprogram_die): Likewise.
12441 * haifa-sched.c (sched_create_recovery_edges): Likewise.
12442 * hw-doloop.c (reorg_loops): Likewise.
12443 * varasm.c (assemble_start_function,
12444 assemble_end_function): Likewise.
12445 (decide_function_section): Do not check for
12446 flag_reorder_blocks_and_partition.
12448 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
12450 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
12452 (chkp_get_hard_register_fake_addr_expr): Ditto.
12453 (chkp_build_addr_expr): Add check for hard reg case.
12454 (chkp_parse_array_and_component_ref): Ditto.
12455 (chkp_find_bounds_1): Ditto.
12456 (chkp_process_stmt): Don't generate bounds store for
12459 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12461 * predict.c (maybe_hot_bb_p): Do not check profile status.
12462 (maybe_hot_edge_p): Likewise.
12463 (probably_never_executed): Check for zero counts even if profile
12465 (unlikely_executed_edge_p): New function.
12466 (unlikely_executed_stmt_p): New function.
12467 (unlikely_executed_bb_p): New function.
12468 (set_even_probabilities): Use unlikely predicates.
12469 (combine_predictions_for_bb): Likewise.
12470 (predict_paths_for_bb): Likewise.
12471 (predict_paths_leading_to_edge): Likewise.
12472 (determine_unlikely_bbs): New function.
12473 (estimate_bb_frequencies): Use it.
12474 (compute_function_frequency): Use zero counts even if profile is
12476 * profile-count.h: Fix typo.
12478 2017-08-08 Julia Koval <julia.koval@intel.com>
12480 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
12481 _mm512_mask_cvtsepi16_storeu_epi8,
12482 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
12483 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
12484 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
12485 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
12486 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
12487 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
12488 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
12489 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
12490 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
12491 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
12492 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
12493 __builtin_ia32_pmovuswb256mem_mask,
12494 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
12495 __builtin_ia32_pmovwb512mem_mask): New builtins.
12497 2017-08-08 Julia Koval <julia.koval@intel.com>
12499 PR target/73350,80862
12500 * config/i386/subst.md (round): Fix round pattern.
12501 * config/i386/i386.c (ix86_erase_embedded_rounding):
12502 Fix erasing rounding for the fixed pattern.
12504 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
12506 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
12508 2017-06-08 Martin Liska <mliska@suse.cz>
12510 PR gcov-profile/80911
12511 * gcov.c (block_info::block_info): New constructor.
12513 2017-06-07 Carl Love <cel@us.ibm.com>
12515 * config/rs6000/rs6000-c: The return type of the following
12516 built-in functions was implemented as int not long long. Fix sign
12517 of return value for the unsigned version of vec_mulo and vec_mule.
12518 vector unsigned long long vec_bperm (vector unsigned long long,
12519 vector unsigned char)
12520 vector signed long long vec_mule (vector signed int,
12522 vector unsigned long long vec_mule (vector unsigned int,
12523 vector unsigned int)
12524 vector signed long long vec_mulo (vector signed int,
12526 vector unsigned long long vec_mulo (vector unsigned int,
12527 vector unsigned int)
12528 * doc/extend.texi: Fix the documentation for the built-in
12531 2017-06-07 Carl Love <cel@us.ibm.com>
12534 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
12537 2017-06-07 Carl Love <cel@us.ibm.com>
12539 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
12540 support, Generate doublehv for signed int/float for BE case only.
12542 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
12544 * doc/invoke.texi (mcx16): Rewrite.
12546 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12548 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
12549 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
12550 *mov<mode>_softfloat, and an anonymous splitter): Use
12551 nonimmediate_operand instead of rs6000_nonimmediate_operand.
12553 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12555 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
12557 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12558 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
12559 (rs6000_debug_reg_global): Adjust.
12560 (rs6000_init_hard_regno_mode_ok): Adjust.
12561 (rs6000_dbx_register_number): Adjust.
12562 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
12563 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12564 Remove SPE_ACC and SPEFSCR.
12565 (REG_ALLOC_ORDER): Ditto.
12566 (FRAME_POINTER_REGNUM): Change to 111.
12567 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
12568 (REG_CLASS_NAMES): Ditto.
12569 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
12570 (REGISTER_NAMES): Ditto.
12571 (ADDITIONAL_REG_NAMES): Ditto.
12572 (rs6000_reg_names): Ditto.
12573 * config/rs6000/rs6000.md: Renumber some register number
12576 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12578 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
12580 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
12581 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
12583 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
12584 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
12585 Delete the SPE high registers.
12586 (REG_ALLOC_ORDER): Ditto.
12587 (enum reg_class): Remove SPE_HIGH_REGS.
12588 (REG_CLASS_NAMES): Ditto.
12589 (REG_CLASS_CONTENTS): Delete the SPE high registers.
12590 (REGISTER_NAMES): Ditto.
12591 (rs6000_reg_names): Ditto.
12592 * doc/tm.texi.in: Remove SPE as example.
12593 * doc/tm.texi: Regenerate.
12595 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12597 * config/rs6000/8540.md (ppc8540_brinc): Delete.
12598 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
12599 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
12600 * config/rs6000/rs6000.md (type): Remove "brinc".
12602 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12604 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
12605 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
12606 * config/rs6000/linuxspe.h: Delete file.
12607 * config/rs6000/rs6000.md: Don't include spe.md.
12608 * config/rs6000/spe.h: Delete file.
12609 * config/rs6000/spe.md: Delete file.
12610 * config/rs6000/t-rs6000: Remove spe.md.
12612 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12614 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
12615 (reg_or_none500mem_operand): Delete.
12616 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
12617 instead of reg_or_none500mem_operand.
12619 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12621 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
12622 handling of SPE flags.
12623 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
12625 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12627 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
12629 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
12630 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
12631 paired_divv2sf3): Similar.
12632 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
12633 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
12634 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
12636 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
12637 Rename the paired_* instruction patterns.
12638 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
12640 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
12641 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
12642 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
12643 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
12644 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
12645 PAIRED_VECTOR_MODE.
12646 (struct machine_function): Delete field spe_insn_chain_scanned_p.
12647 (spe_func_has_64bit_regs_p): Delete.
12648 (spe_expand_predicate_builtin): Delete.
12649 (spe_expand_evsel_builtin): Delete.
12650 (TARGET_DWARF_REGISTER_SPAN): Do not define.
12651 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
12652 (invalid_e500_subreg): Delete.
12653 (rs6000_legitimize_address): Always force_reg op2 as well, for
12654 paired single memory accesses.
12655 (rs6000_member_type_forces_blk): Delete.
12656 (rs6000_spe_function_arg): Delete.
12657 (rs6000_expand_unop_builtin): Delete SPE handling.
12658 (rs6000_expand_binop_builtin): Ditto.
12659 (spe_expand_stv_builtin): Delete.
12660 (bdesc_2arg_spe): Delete.
12661 (spe_expand_builtin): Delete.
12662 (spe_expand_predicate_builtin): Delete.
12663 (spe_expand_evsel_builtin): Delete.
12664 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
12665 (spe_init_builtins): Delete.
12666 (spe_func_has_64bit_regs_p): Delete.
12667 (savres_routine_name): Delete "info" parameter. Adjust callers.
12668 (rs6000_emit_stack_reset): Ditto.
12669 (rs6000_dwarf_register_span): Delete.
12670 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
12671 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
12672 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
12674 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
12676 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
12677 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
12678 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
12679 mulv2sf3, divv2sf3): Delete expanders.
12681 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12683 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
12685 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12687 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
12688 * config/rs6000/rs6000.c: Ditto.
12690 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12692 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
12693 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
12694 comparison_operator.
12696 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12698 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
12699 * config/rs6000/rs6000.opt: Ditto.
12700 * config/rs6000/t-rtems: Ditto.
12702 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12704 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
12705 TARGET_E500_SINGLE by 0, simplify.
12706 * config/rs6000/rs6000.c: Ditto.
12707 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
12708 (spe_build_register_parallel): Delete.
12709 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
12710 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
12711 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
12712 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
12713 (E500_CONVERT): Delete.
12714 * config/rs6000/spe.md: Remove many patterns and all define_constants.
12716 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12718 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
12719 * config/rs6000/dfp.md: Ditto.
12720 (negdd2, *negdd2_fpr): Merge.
12721 (absdd2, *absdd2_fpr): Merge.
12722 (negtd2, *negtd2_fpr): Merge.
12723 (abstd2, *abstd2_fpr): Merge.
12724 * config/rs6000/e500.h: Delete file.
12725 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
12726 TARGET_FPRS by 1 and simplify.
12727 * config/rs6000/rs6000-c.c: Ditto.
12728 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
12729 TARGET_DF_SPE by 0.
12730 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
12732 * config/rs6000/rs6000.md: Ditto.
12733 (floatdidf2, *floatdidf2_fpr): Merge.
12734 (move_from_CR_gt_bit): Delete.
12735 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
12736 (E500_CR_IOR_COMPARE): Delete.
12737 (All patterns that require !TARGET_FPRS): Delete.
12738 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
12740 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12742 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
12744 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12746 * graphds.c (add_edge): Intitialize edge's attached data.
12747 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
12748 pointer parameter. Call pointed function on each edge during
12749 graph traversing. Skip traversing the edge when the function
12751 (graphds_dfs, graphds_scc): Ditto.
12752 (for_each_edge): New parameter. Pass the new parameter to callback
12754 * graphds.h (skip_edge_callback): New function pointer type.
12755 (graphds_dfs, graphds_scc): New function pointer parameter.
12756 (graphds_edge_callback, for_each_edge): New parameter.
12758 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12760 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
12761 out code checking if runtime alias check is possible to below ...
12762 Call the new function.
12763 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
12764 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
12766 2017-06-07 Marek Polacek <polacek@redhat.com>
12769 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
12770 TYPE_OVERFLOW_WRAPS check.
12772 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12774 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
12775 if versioning is required.
12776 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
12777 peeling with the check for versioning.
12779 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12781 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
12782 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
12783 Set true to new parameter if new ssa variable is defined.
12784 (vect_gen_vector_loop_niters): Refactor. Set range information
12785 for the new vector loop bound variable.
12786 (vect_do_peeling): Ditto.
12788 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12790 * tree-affine.c (ssa.h): Include header file.
12791 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
12792 has wrapping overflow behavior.
12794 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12796 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
12798 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12800 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
12801 (tree_to_aff_combination): ... here.
12803 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12805 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
12806 reg_pressure model function.
12807 (ivopts_global_cost_for_size): Delete.
12808 (determine_set_costs, iv_ca_recount_cost): Call new model function
12809 ivopts_estimate_reg_pressure.
12811 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12813 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
12814 expensive than udiv. Remove floating point cases from mod.
12816 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12818 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
12819 Increase idiv cost.
12821 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12823 * config/aarch64/aarch64.md
12824 (copysignsf3): Fix mask generation.
12826 2017-06-07 Jakub Jelinek <jakub@redhat.com>
12828 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
12830 (class dump_manager): Add register_dumps method.
12831 * dumpfile.c: Include langhooks.h.
12832 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
12833 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
12834 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
12835 (dump_manager::dump_register): Start with 512 entries instead of 32.
12836 (dump_manager::register_dumps): New method.
12837 * toplev.c (general_init): Instead of invoking register_dumps
12838 langhook, invoke register_dumps method on the dump manager.
12839 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
12842 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
12844 * doc/md.texi: Clarify the restrictions on a define_insn condition.
12845 Say that # requires an associated define_split to exist, and that
12846 the define_split must be suitable for use after register allocation.
12848 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12850 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
12851 (compute_outgoing_frequencies): Also initialize zero counts.
12852 (find_many_sub_basic_blocks): Do not produce uninitialized profile
12853 around loops; preserve more of profile when nothing changes.
12855 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
12857 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
12859 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
12860 * config/arm/arm-cpu-cdata.h: Regenerate.
12861 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
12862 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
12863 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
12864 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
12865 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
12867 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
12869 * config/arm/t-rmprofile: Likewise.
12870 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
12872 2017-06-06 David S. Miller <davem@davemloft.net>
12875 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
12876 blockage if function uses alloca.
12878 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12880 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
12881 New "uid" fields to hold pretty-print IDs of group and ref.
12882 Memory references are now identified as <group_id>:<ref_id>
12883 instead of using [random] addresses.
12884 (dump_mem_details): Simplify, no functional change.
12885 (dump_mem_ref): Simplify and make output more concise.
12886 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
12887 (find_or_create_group): Initialize group uid.
12888 (record_ref): Initialize ref uid. Improve debug output.
12889 (prune_group_by_reuse, should_issue_prefetch_p,)
12890 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
12891 (mark_nontemporal_store, determine_loop_nest_reuse):
12892 Improve debug output.
12894 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12896 * dbgcnt.def (prefetch): New debug counter.
12897 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
12898 (schedule_prefetches): Stop issueing prefetches if debug counter
12901 2017-06-06 Tom de Vries <tom@codesourcery.com>
12903 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
12904 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
12906 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12908 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
12909 Use aarch64_reg_or_zero predicate for operand 4.
12910 (aarch64_compare_and_swap<mode> define_insn_and_split):
12911 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
12912 (aarch64_store_exclusive<mode>): Likewise for operand 2.
12914 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12916 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
12917 (arm_compute_save_core_reg_mask): This.
12918 (thumb1_compute_save_reg_mask): Rename into ...
12919 (thumb1_compute_save_core_reg_mask): This.
12920 (arm_compute_save_reg0_reg12_mask): Adapt comment.
12921 (arm_compute_frame_layout): Likewise.
12923 2017-06-06 Richard Biener <rguenther@suse.de>
12925 PR tree-optimization/80974
12926 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
12927 keep or clear leaders SSA info.
12929 2017-06-06 Tom de Vries <tom@codesourcery.com>
12931 * config/nvptx/nvptx.c (split_mode_p): New function.
12932 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
12934 2017-06-06 Tom de Vries <tom@codesourcery.com>
12936 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
12938 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12941 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
12944 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12946 * shrink-wrap.c (handle_simple_exit): Update profile.
12947 (try_shrink_wrapping): Upate profile.
12949 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12951 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
12952 (tree_guess_outgoing_edge_probabilities): New.
12953 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
12954 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
12956 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12958 * ipa-split.c (split_function): Initialize return bb profile.
12960 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12962 * profile.c (compute_branch_probabilities): Also initialize
12963 EXIT_BLOCK profile.
12965 2017-06-06 Richard Biener <rguenther@suse.de>
12967 PR tree-optimization/80928
12968 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
12969 (vect_analyze_loop_operations): Properly guard analysis for
12971 (vect_transform_loop): Likewise.
12972 (vect_analyze_loop_2): Also reset SLP type on PHIs.
12973 (vect_model_induction_cost): Do not cost for pure SLP.
12974 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
12975 of induction in inner loop vectorization.
12976 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
12977 (vect_get_and_check_slp_defs): Handle vect_induction_def.
12978 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
12980 (vect_analyze_slp_cost_1): Cost induction.
12981 (vect_detect_hybrid_slp_stmts): Handle PHIs.
12982 (vect_get_slp_vect_defs): Likewise.
12983 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
12984 (vect_transform_stmt): Handle SLP reductions.
12985 * tree-vectorizer.h (vectorizable_induction): Adjust.
12987 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12989 * config/rs6000/rs6000.c (make_resolver_func): Update
12990 init_lowered_empty_function call.
12992 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
12994 * doc/invoke.texi: Document the -fprofile-abs-path option.
12995 * common.opt (fprofile-abs-path): New option.
12996 * gcov-io.h (gcov_write_filename): Declare.
12997 * gcov-io.c (gcov_write_filename): New function.
12998 * coverage.c (coverage_begin_function): Use gcov_write_filename.
12999 * profile.c (output_location): Likewise.
13001 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13003 * shring-wrap.c: Revert accidental commit.
13005 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
13007 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
13009 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13011 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
13013 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
13014 profile in callgraph edge.
13015 * profile-count.h (apply_probability): If THIS is 0, then result is 0
13016 (apply_scale): Likewise.
13017 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
13018 Also scale profile when inlining function with zero profile.
13019 (initialize_cfun): Update exit block profile even when it is zero.
13020 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
13021 when profile is read.
13023 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13025 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
13026 (CLONE_*): New constants to define the processors we can generate
13027 code for with the target_clone attribute.
13028 (rs6000_clone_map): New array to identify which clone processors
13029 the current program is running on.
13030 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
13031 target_clone attribute.
13032 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
13033 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
13034 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
13035 (cpu_expand_builtin): Add support for target_clone attribute.
13036 (rs6000_valid_attribute_p): Allow "default" attribute.
13037 (get_decl_name): New debug function to simplify printing the
13038 current function name in debugging statements.
13039 (rs6000_clone_priority): New functions to support the target_clone
13040 attribute, and be able to generate code to switch between ISA 2.05
13041 through ISA 3.0 (power6 through power9).
13042 (rs6000_compare_version_priority): Likewise.
13043 (rs6000_get_function_versions_dispatcher): Likewise.
13044 (make_resolver_func): Likewise.
13045 (add_condition_to_bb): Likewise.
13046 (dispatch_function_versions): Likewise.
13047 (rs6000_generate_version_dispatcher_body): Likewise.
13048 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
13049 (fusion_gpr_load_p): Fix a spacing issue.
13050 * doc/extend.texi (Common Function Attributes): Document that the
13051 PowerPC supports the target_clone attribute.
13053 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13055 * config/arm/arm.h: explain F symbol found in description of ARM
13056 register allocation in its legend.
13058 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13060 * config/mips/frame-header-opt.c: Include profile-count.h.
13061 * config/riscv/riscv.c: Include profile-count.h
13063 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
13065 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
13067 (sm_set_flag_if_changed): Add bbs field.
13068 (execute_sm_if_changed_flag_set): Pass BBS.
13069 (execute_sm): Update.
13071 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13073 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
13076 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13078 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
13079 (peephole2): New peephole2 to emit the above.
13080 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
13082 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13084 * config/aarch64/aarch64.c (define_peephole2 above
13085 *sub_<shift>_<mode>): New peephole.
13087 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13089 * config/i386/i386.c (make_resolver_func): Update.
13090 * Makefile.in: Add profile-count.h and profile-count.o
13091 * auto-profile.c (afdo_indirect_call): Update to new API.
13092 (afdo_set_bb_count): Update.
13093 (afdo_propagate_edge): Update.
13094 (afdo_propagate_circuit): Update.
13095 (afdo_calculate_branch_prob): Update.
13096 (afdo_annotate_cfg): Update.
13097 * basic-block.h: Include profile-count.h
13098 (struct edge_def): Turn count to profile_count.
13099 (struct basic_block_def): Likewie.
13100 (REG_BR_PROB_BASE): Move to profile-count.h
13101 (RDIV): Move to profile-count.h
13102 * bb-reorder.c (max_entry_count): Turn to profile_count.
13103 (find_traces): Update.
13104 (rotate_loop):Update.
13105 (connect_traces):Update.
13106 (sanitize_hot_paths):Update.
13107 * bt-load.c (migrate_btr_defs): Update.
13108 * cfg.c (RDIV): Remove.
13109 (init_flow): Use alloc_block.
13110 (alloc_block): Uninitialize count.
13111 (unchecked_make_edge): Uninitialize count.
13112 (check_bb_profile): Update.
13113 (dump_edge_info): Update.
13114 (dump_bb_info): Update.
13115 (update_bb_profile_for_threading): Update.
13116 (scale_bbs_frequencies_int): Update.
13117 (scale_bbs_frequencies_gcov_type): Update.
13118 (scale_bbs_frequencies_profile_count): New.
13119 * cfg.h (update_bb_profile_for_threading): Update.
13120 (scale_bbs_frequencies_profile_count): Declare.
13121 * cfgbuild.c (compute_outgoing_frequencies): Update.
13122 (find_many_sub_basic_blocks): Update.
13123 * cfgcleanup.c (try_forward_edges): Update.
13124 (try_crossjump_to_edge): Update.
13125 * cfgexpand.c (expand_gimple_tailcall): Update.
13126 (construct_exit_block): Update.
13127 * cfghooks.c (verify_flow_info): Update.
13128 (dump_bb_for_graph): Update.
13129 (split_edge): Update.
13130 (make_forwarder_block): Update.
13131 (duplicate_block): Update.
13132 (account_profile_record): Update.
13133 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
13134 (get_estimated_loop_iterations): Update.
13135 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
13136 (single_likely_exit): Update.
13137 * cfgloopmanip.c (scale_loop_profile): Update.
13139 (set_zero_probability): Update.
13140 (lv_adjust_loop_entry_edge): Update.
13141 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
13142 (purge_dead_edges): Update.
13143 (rtl_account_profile_record): Update.
13144 * cgraph.c (cgraph_node::create): Uninitialize count.
13145 (symbol_table::create_edge): Uninitialize count.
13146 (cgraph_update_edges_for_call_stmt_node): Update.
13147 (cgraph_edge::dump_edge_flags): Update.
13148 (cgraph_node::dump): Update.
13149 (cgraph_edge::maybe_hot_p): Update.
13150 * cgraph.h: Include profile-count.h
13151 (create_clone), create_edge, create_indirect_edge): Update.
13152 (cgraph_node): Turn count to profile_count.
13153 (cgraph_edge0: Likewise.
13154 (make_speculative, clone): Update.
13155 (create_edge): Update.
13156 (init_lowered_empty_function): Update.
13157 * cgraphclones.c (cgraph_edge::clone): Update.
13158 (duplicate_thunk_for_node): Update.
13159 (cgraph_node::create_clone): Update.
13160 * cgraphunit.c (cgraph_node::analyze): Update.
13161 (cgraph_node::expand_thunk): Update.
13162 * final.c (dump_basic_block_info): Update.
13163 * gimple-streamer-in.c (input_bb): Update.
13164 * gimple-streamer-out.c (output_bb): Update.
13165 * graphite.c (print_global_statistics): Update.
13166 (print_graphite_scop_statistics): Update.
13167 * hsa-brig.c: Include basic-block.h.
13168 * hsa-dump.c: Include basic-block.h.
13169 * hsa-gen.c (T sum_slice): Update.
13170 (convert_switch_statements):Update.
13171 * hsa-regalloc.c: Include basic-block.h.
13172 * ipa-chkp.c (chkp_produce_thunks): Update.
13173 * ipa-cp.c (struct caller_statistics): Update.
13174 (init_caller_stats): Update.
13175 (gather_caller_stats): Update.
13176 (ipcp_cloning_candidate_p): Update.
13177 (good_cloning_opportunity_p): Update.
13178 (get_info_about_necessary_edges): Update.
13179 (dump_profile_updates): Update.
13180 (update_profiling_info): Update.
13181 (update_specialized_profile): Update.
13182 (perhaps_add_new_callers): Update.
13183 (decide_about_value): Update.
13184 (ipa_cp_c_finalize): Update.
13185 * ipa-devirt.c (struct odr_type_warn_count): Update.
13186 (struct decl_warn_count): Update.
13187 (struct final_warning_record): Update.
13188 (possible_polymorphic_call_targets): Update.
13189 (ipa_devirt): Update.
13190 * ipa-fnsummary.c (redirect_to_unreachable): Update.
13191 * ipa-icf.c (sem_function::merge): Update.
13192 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
13193 * ipa-inline.c (compute_uninlined_call_time): Update.
13194 (compute_inlined_call_time): Update.
13195 (want_inline_small_function_p): Update.
13196 (want_inline_self_recursive_call_p): Update.
13197 (edge_badness): Update.
13198 (lookup_recursive_calls): Update.
13199 (recursive_inlining): Update.
13200 (inline_small_functions): Update.
13201 (dump_overall_stats): Update.
13202 (dump_inline_stats): Update.
13203 * ipa-profile.c (ipa_profile_generate_summary): Update.
13204 (ipa_propagate_frequency): Update.
13205 (ipa_profile): Update.
13206 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13207 * ipa-utils.c (ipa_merge_profiles): Update.
13208 * loop-doloop.c (doloop_modify): Update.
13209 * loop-unroll.c (report_unroll): Update.
13210 (unroll_loop_runtime_iterations): Update.
13211 * lto-cgraph.c (lto_output_edge): Update.
13212 (lto_output_node): Update.
13213 (input_node): Update.
13214 (input_edge): Update.
13215 (merge_profile_summaries): Update.
13216 * lto-streamer-in.c (input_cfg): Update.
13217 * lto-streamer-out.c (output_cfg): Update.
13218 * mcf.c (create_fixup_graph): Update.
13219 (adjust_cfg_counts): Update.
13220 (sum_edge_counts): Update.
13221 * modulo-sched.c (sms_schedule): Update.
13222 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
13223 * predict.c (maybe_hot_count_p): Update.
13224 (probably_never_executed): Update.
13225 (dump_prediction): Update.
13226 (combine_predictions_for_bb): Update.
13227 (propagate_freq): Update.
13228 (handle_missing_profiles): Update.
13229 (counts_to_freqs): Update.
13230 (rebuild_frequencies): Update.
13231 (force_edge_cold): Update.
13232 * predict.h: Include profile-count.h
13233 (maybe_hot_count_p, counts_to_freqs): UPdate.
13234 * print-rtl-function.c: Do not include cfg.h
13235 * print-rtl.c: Include basic-block.h
13236 * profile-count.c: New file.
13237 * profile-count.h: New file.
13238 * profile.c (is_edge_inconsistent): Update.
13239 (correct_negative_edge_counts): Update.
13240 (is_inconsistent): Update.
13241 (set_bb_counts): Update.
13242 (read_profile_edge_counts): Update.
13243 (compute_frequency_overlap): Update.
13244 (compute_branch_probabilities): Update; Initialize and deinitialize
13246 (branch_prob): Update.
13247 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
13248 (edge_gcov_count): New.
13249 (bb_gcov_count): New.
13250 * shrink-wrap.c (try_shrink_wrapping): Update.
13251 * tracer.c (better_p): Update.
13252 * trans-mem.c (expand_transaction): Update.
13253 (ipa_tm_insert_irr_call): Update.
13254 (ipa_tm_insert_gettmclone_call): Update.
13255 * tree-call-cdce.c: Update.
13256 * tree-cfg.c (gimple_duplicate_sese_region): Update.
13257 (gimple_duplicate_sese_tail): Update.
13258 (gimple_account_profile_record): Update.
13259 (execute_fixup_cfg): Update.
13260 * tree-inline.c (copy_bb): Update.
13261 (copy_edges_for_bb): Update.
13262 (initialize_cfun): Update.
13263 (freqs_to_counts): Update.
13264 (copy_cfg_body): Update.
13265 (expand_call_inline): Update.
13266 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13267 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
13268 (try_unroll_loop_completely): Update.
13269 (try_peel_loop): Update.
13270 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13271 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
13272 * tree-ssa-loop-split.c (connect_loops): Update.
13273 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
13274 * tree-ssa-reassoc.c (branch_fixup): Update.
13275 * tree-ssa-tail-merge.c (replace_block_by): Update.
13276 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
13277 (compute_path_counts): Update.
13278 (update_profile): Update.
13279 (recompute_probabilities): Update.
13280 (update_joiner_offpath_counts): Update.
13281 (estimated_freqs_path): Update.
13282 (freqs_to_counts_path): Update.
13283 (clear_counts_path): Update.
13284 (ssa_fix_duplicate_block_edges): Update.
13285 (duplicate_thread_path): Update.
13286 * tree-switch-conversion.c (case_bit_test_cmp): Update.
13287 (struct switch_conv_info): Update.
13288 * tree-tailcall.c (decrease_profile): Update.
13289 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13290 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13291 * value-prof.c (check_counter): Update.
13292 (gimple_divmod_fixed_value): Update.
13293 (gimple_mod_pow2): Update.
13294 (gimple_mod_subtract): Update.
13295 (gimple_ic_transform): Update.
13296 (gimple_stringop_fixed_value): Update.
13297 * value-prof.h (gimple_ic): Update.
13299 2017-06-02 Carl Love <cel@us.ibm.com>
13301 * config/rs6000/rs6000-c: Add support for built-in functions
13302 vector double vec_doublee (vector signed int);
13303 vector double vec_doublee (vector unsigned int);
13304 vector double vec_doublee (vector float);
13305 vector double vec_doubleh (vector signed int);
13306 vector double vec_doubleh (vector unsigned int);
13307 vector double vec_doubleh (vector float);
13308 vector double vec_doublel (vector signed int);
13309 vector double vec_doublel (vector unsigned int);
13310 vector double vec_doublel (vector float);
13311 vector double vec_doubleo (vector signed int);
13312 vector double vec_doubleo (vector unsigned int);
13313 vector double vec_doubleo (vector float);.
13314 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
13315 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
13317 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
13318 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
13319 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
13321 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
13322 vec_doublel, vec_doubleh.
13323 * doc/extend.texi: Update the built-in documentation file for the
13324 new built-in functions.
13326 2017-06-02 David Malcolm <dmalcolm@redhat.com>
13329 * ipa-inline-analysis.c (free_growth_caches): Set
13330 edge_removal_hook_holder to NULL after removing it.
13332 2017-06-02 Sudakshina Das <sudi.das@arm.com>
13334 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
13335 comparision with zero.
13337 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13338 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13339 for early expansion of vec_min and vec_max builtins.
13340 (builtin_function_type): Add min/max unsigned variants to those
13341 identified as having unsigned arguments.
13343 2017-06-02 Olivier Hainque <hainque@adacore.com>
13345 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
13347 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13349 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
13350 Use VALL_F16 iterator rather than VALL.
13352 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13354 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
13355 Emit CBNZ inside loop when doing a strong exchange and comparing
13356 against zero. Generate the CC flags after the loop.
13358 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
13360 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
13361 (dl_section_ref): New.
13362 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
13363 On AIX, append an expression to subtract the size of the
13364 section length to dl_section_ref.
13366 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
13368 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13369 for early expansion of vector absolute builtins.
13371 2017-06-02 Richard Biener <rguenther@suse.de>
13373 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
13374 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
13376 2017-06-02 Richard Biener <rguenther@suse.de>
13378 PR tree-optimization/80948
13379 * tree-tailcall.c (find_tail_calls): Track stmts to move in
13380 stmt order as well.
13382 2017-06-02 Richard Biener <rguenther@suse.de>
13384 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
13386 * tree-vect-stmts.c (process_use): Do not mark backedge defs
13387 for inductions as relevant.
13389 2017-06-02 Richard Biener <rguenther@suse.de>
13391 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
13392 (vectorizable_induction): ... this. Remove dead code.
13394 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
13396 * builtins. (expand_builtin_alloca): Remove second parameter and
13397 infer its value from the first parameter instead.
13398 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
13400 2017-06-02 Jakub Jelinek <jakub@redhat.com>
13402 PR rtl-optimization/80903
13403 * loop-doloop.c (add_test): Unshare sequence.
13405 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
13407 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
13409 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
13411 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
13413 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
13414 xlogue_layout::get_instance, logue_layout::xlogue_layout,
13415 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
13416 (xlogue_layout::get_stub_rtx): Make static.
13417 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
13418 (xlogue_layout::compute_stub_managed_regs): Rename to...
13419 (xlogue_layout::count_stub_managed_regs): ...this.
13420 (xlogue_layout::is_stub_managed_reg): New function.
13421 (xlogue_layout::m_stub_names): Rename to...
13422 (xlogue_layout::s_stub_names): ...this, make static.
13423 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
13424 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
13425 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
13426 xlogue_layout::s_stub_names): Instantiate statics.
13427 (stub_managed_regs): Remove.
13428 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
13429 (disable_call_ms2sysv_xlogues): Rename to...
13430 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
13431 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
13433 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
13434 change after reload_completed.
13435 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
13437 (ix86_expand_prologue): Likewise.
13438 (ix86_expand_epilogue): Likewise.
13439 (ix86_expand_split_stack_prologue): Likewise.
13440 (ix86_compute_frame_layout): Remove frame parameter ...
13441 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
13442 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
13444 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
13445 (ix86_frame): Move from here ...
13446 * config/i386/i386.h (ix86_frame): ... to here.
13447 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
13448 complete ix86_frame data structure instead. Remove some_ld_name.
13450 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13452 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
13453 symbols that hold a DECL_VALUE_EXPR.
13455 2017-06-01 Martin Jambor <mjambor@suse.cz>
13457 PR tree-optimization/80898
13458 * tree-sra.c (process_subtree_disqualification): Removed.
13459 (disqualify_candidate): Do not acll
13460 process_subtree_disqualification.
13461 (subtree_mark_written_and_enqueue): New function.
13462 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
13463 RHS has been disqualified and re-queue LHS if necessary. Apart
13464 from that, ignore disqualified RHS.
13466 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13468 * config/s390/s390.c (s390_emit_epilogue): Disable early return
13469 address fetch for z10 or later.
13471 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13473 * config/arc/arc.md (tst_movb): Add guard when splitting.
13475 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13477 * config/arc/arc.c (arc_can_eliminate): Test against
13478 arc_frame_pointer_needed.
13480 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13482 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
13483 to prevent store reordering.
13484 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
13485 (type): Add block type.
13486 (stack_tie): Define special instruction to be used in
13489 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13491 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
13492 constraint. It is not valid for the pattern.
13493 (noncommutative_binary_comparison): Likewise.
13495 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13497 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
13500 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13502 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
13503 be used by the reg-alloc.
13505 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13507 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
13508 reg-alloc when having mul64 or mul32x16 instructions.
13509 (mulsidi3): Likewise.
13510 (umulsidi3): Likewise.
13511 (mulsi32x16): New pattern.
13512 (mulsi64): Likewise.
13513 (mulsidi64): Likewise.
13514 (umulsidi64): Likewise.
13515 (MUL32x16_REG): Define.
13516 (mul64_600): Use MUL32x16_REG.
13517 (mac64_600): Likewise.
13518 (umul64_600): Likewise.
13519 (umac64_600): Likewise.
13521 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
13523 * config/arc/arc.md (mulsi3_700): Make it commutative.
13525 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
13527 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
13529 (*sign_extendsidi2_insn): Likewise for movstosw.
13531 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
13533 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
13534 the type of the input discriminant value. Convert the
13535 discriminant value of signedness vary.
13537 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
13539 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
13540 Add to -Wall section.
13542 2017-06-01 Richard Biener <rguenther@suse.de>
13544 PR middle-end/66313
13545 * fold-const.c (fold_plusminus_mult_expr): If the factored
13546 factor may be zero use a wrapping type for the inner operation.
13547 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
13548 and handle moved defs.
13549 (process_assignment): Properly guard the unary op case. Return a
13550 tri-state indicating that moving the stmt before the call may allow
13551 to continue. Pass through to_move.
13552 (find_tail_calls): Handle moving unrelated defs before
13555 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
13558 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
13559 splitter result in the canonical way.
13561 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
13563 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
13564 also for 32bit target. Update insn attributes.
13565 (zero-extendsidi2 splitter): Allow all registers for operand 1.
13567 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
13569 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
13570 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
13571 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
13572 (_mm_maskz_min_ss): New intrinsics.
13574 2017-05-31 Martin Liska <mliska@suse.cz>
13576 * tree-vect-loop.c (vect_create_epilog_for_reduction):
13577 Change comment style to one we normally use.
13578 (vectorizable_reduction): Likewise.
13579 (vectorizable_induction): Likewise.
13580 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13581 (vectorizable_call): Likewise.
13582 (vectorizable_simd_clone_call): Likewise.
13583 (vectorizable_conversion): Likewise.
13584 (vectorizable_assignment): Likewise.
13585 (vectorizable_shift): Likewise.
13586 (vectorizable_operation): Likewise.
13587 (vectorizable_store): Likewise.
13588 (vectorizable_load): Likewise.
13589 * tree-vectorizer.h: Likewise.
13591 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
13593 * passes.c (emergency_dump_function): New.
13594 * tree-pass.h (emergency_dump_function): Declare.
13595 * plugin.c (plugins_internal_error_function): Remove.
13596 * plugin.h (plugins_internal_error_function): Remove declaration.
13597 * toplev.c (internal_error_function): New static function. Use it...
13598 (general_init): ...here.
13600 2017-05-31 Graham Markall <graham.markall@embecosm.com>
13602 * config/arc/arc.c (arc_print_operand): Handle constant operands.
13603 (arc_rtx_costs): Add costs for new patterns.
13604 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
13605 * config/arc/predicates.md: Add _1_2_3_operand predicate.
13607 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13609 * tree-ssa-strlen.c (get_next_strinfo): New function.
13610 (get_stridx_plus_constant): Use it.
13611 (zero_length_string): Likewise.
13612 (adjust_related_strinfos): Likewise.
13613 (adjust_last_stmt): Likewise.
13615 2017-05-31 Richard Biener <rguenther@suse.de>
13618 * config/i386/i386.c (ix86_expand_builtin): Remove assert
13619 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
13621 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
13623 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
13624 loop_vinfo argument and use of dependence distance vectors.
13625 Check instead whether the two references differ only in their
13626 initial value and assume that they have the same alignment if the
13627 difference is a multiple of the vector alignment.
13628 (vect_analyze_data_refs_alignment): Update call accordingly.
13630 2017-05-31 Martin Liska <mliska@suse.cz>
13633 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
13635 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13637 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
13638 (create_intersect_range_checks): Move from ...
13639 * tree-data-ref.c (create_intersect_range_checks_index)
13640 (create_intersect_range_checks): ... to here.
13641 (create_runtime_alias_checks): New function factored from ...
13642 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
13643 here. Call above function.
13644 * tree-data-ref.h (create_runtime_alias_checks): New function.
13646 2017-05-31 Bin Cheng <bin.cheng@arm.com>
13648 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
13649 segment length for dr_b and compute it in wide_int.
13651 2017-05-31 Richard Biener <rguenther@suse.de>
13653 PR tree-optimization/80906
13654 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
13655 and pass through iv_map.
13656 (copy_bb_and_scalar_dependences): Adjust.
13657 (translate_pending_phi_nodes): Likewise.
13658 (copy_loop_close_phi_args): Handle code-generating IVs instead
13661 2017-05-30 David Malcolm <dmalcolm@redhat.com>
13663 * diagnostic-color.c (color_dict): Add "type-diff".
13664 (parse_gcc_colors): Update comment.
13665 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
13666 -fdiagnostics-show-template-tree and -fno-elide-type.
13667 (GCC_COLORS): Add type-diff to example.
13669 (-fdiagnostics-show-template-tree): New.
13670 (-fno-elide-type): New.
13671 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
13672 the pp_format_decoder callback. Call any m_format_postprocessor's
13674 (pretty_printer::pretty_printer): Initialize
13675 m_format_postprocessor.
13676 (pretty_printer::~pretty_printer): Delete any
13677 m_format_postprocessor.
13678 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
13679 (class format_postprocessor): New class.
13680 (struct pretty_printer::format_decoder): Document the new parameters.
13681 (struct pretty_printer::m_format_postprocessor): New field.
13682 * tree-diagnostic.c (default_tree_printer): Update for new
13683 bool and const char ** params.
13684 * tree-diagnostic.h (default_tree_printer): Likewise.
13686 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
13688 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
13689 (lwa_operand): Delete rs6000_gen_cell_microcode test.
13690 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13691 rs6000_gen_cell_microcode code.
13692 (rs6000_final_prescan_insn): Delete.
13693 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
13694 "warn-cell-microcode" entries.
13695 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
13696 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
13697 throughout. Change cc_reg_not_micro_cr0_operand to
13698 cc_reg_not_cr0_operand throughout.
13699 (*extendhi<mode>2_noload): Delete.
13700 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
13701 (mwarn-cell-microcode): Delete.
13702 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
13703 -mgen-cell-microcode and -mwarn-cell-microcode.
13705 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
13708 * config/i386/constraints.md (Yd): New constraint.
13710 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
13711 and (?Yd, r) alternatives. Update insn attributes.
13712 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
13713 and (?*Yd, r) alternatives. Update insn attributes.
13714 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
13716 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
13718 * gimplify.c (gimplify_modify_expr): Don't create a
13719 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
13722 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
13724 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
13726 2017-05-30 Richard Biener <rguenther@suse.de>
13728 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
13729 and reduc_def fields.
13730 (STMT_VINFO_REDUC_TYPE): New define.
13731 (STMT_VINFO_REDUC_DEF): Likewise.
13732 (vect_force_simple_reduction): Adjust prototype.
13733 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
13734 (vect_is_simple_reduction): Remove check_reduction argument.
13735 (vect_force_simple_reduction): Adjust and set
13736 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13737 (vectorizable_reduction): Do not re-do reduction analysis
13738 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13739 * tree-parloops.c (gather_scalar_reductions): Adjust.
13741 2017-05-30 Richard Biener <rguenther@suse.de>
13743 PR middle-end/80901
13744 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
13747 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
13749 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
13750 Introduce unknown_misalignment parameter and remove vf.
13751 (vect_peeling_hash_get_lowest_cost):
13752 Pass unknown_misalignment parameter.
13753 (vect_enhance_data_refs_alignment):
13754 Fix unsupportable data ref treatment.
13756 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13758 * tree-vect-data-refs.c (vect_get_data_access_cost):
13759 Workaround for SLP handling.
13760 (vect_enhance_data_refs_alignment):
13761 Compute costs for doing no peeling at all, compare to the best
13762 peeling costs so far and avoid peeling if cheaper.
13764 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13766 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
13767 Return peeling info and set costs to zero for unlimited cost
13769 (vect_enhance_data_refs_alignment): Also inspect all datarefs
13770 with unknown misalignment. Compute and costs for unknown
13771 misalignment, compare them to the costs for known misalignment
13772 and choose the cheapest for peeling.
13774 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13776 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
13777 (vect_get_peeling_costs_all_drs): Create function.
13778 (vect_peeling_hash_get_lowest_cost):
13779 Use vect_get_peeling_costs_all_drs.
13780 (vect_peeling_supportable): Create function.
13781 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
13783 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13785 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
13786 DR_HAS_NEGATIVE_STEP.
13787 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
13788 (vect_enhance_data_refs_alignment): Use.
13789 (vect_duplicate_ssa_name_ptr_info): Use.
13790 * tree-vectorizer.h (dr_misalignment): Use.
13791 (known_alignment_for_access_p): Use.
13793 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13796 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
13798 (has_section_name): New function.
13800 2017-05-30 Martin Liska <mliska@suse.cz>
13803 * auto-profile.c (get_function_decl_from_block): Fix
13806 2017-05-30 Richard Biener <rguenther@suse.de>
13808 PR middle-end/80876
13809 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
13811 2017-05-30 Martin Liska <mliska@suse.cz>
13813 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
13814 * dumpfile.h (struct dump_file_info): Remove ctors.
13816 2017-05-30 Martin Liska <mliska@suse.cz>
13818 * predict.def: Fix GNU coding style.
13820 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13822 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13823 Mark 'to' argument with ATTRIBUTE_UNUSED.
13825 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13827 * config/xtensa/xtensa.c (xtensa_emit_call): Use
13828 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
13829 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
13832 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
13834 * doc/install.texi (Options specification): Restore entry of
13835 --enable-sjlj-exceptions.
13837 2017-05-27 Michael Eager <eager@eagercon.com>
13840 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
13842 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
13844 * config/microblaze/microblaze.h
13845 (FIXED_REGISTERS): Update in macro.
13846 (CALL_USED_REGISTERS): Update in macro.
13848 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
13850 * doc/install.texi: Add links to macOS binary distributions.
13852 2017-05-27 Jakub Jelinek <jakub@redhat.com>
13856 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13858 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13860 2017-05-26 Martin Liska <mliska@suse.cz>
13862 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
13864 2017-05-26 Martin Liska <mliska@suse.cz>
13866 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
13867 always leading ';; '.
13868 (dump_bb_info): Likewise.
13869 (brief_dump_cfg): Likewise.
13870 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
13871 * dumpfile.c: Remove usage of TDF_VERBOSE.
13872 * dumpfile.h (enum dump_kind): Likewise.
13873 (dump_gimple_bb_header): Do not use TDF_COMMENT.
13874 * print-tree.c (debug_verbose): Remove.
13875 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
13876 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
13877 * tree-diagnostic.c (default_tree_printer): Replace
13878 TDF_DIAGNOSTIC with TDF_SLIM.
13880 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13882 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
13883 in parameter loop, rather than loop_vinfo.
13884 (create_intersect_range_checks): Ditto.
13885 (vect_create_cond_for_alias_checks): Update call to above functions.
13887 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13889 PR tree-optimization/80815
13890 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
13891 for merging runtime alias checks. Handle negative DR_STEPs.
13893 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13895 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
13897 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
13898 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
13899 out code pruning runtime alias checks.
13900 * tree-data-ref.c (prune_runtime_alias_test_list): New function
13901 factored out from above.
13902 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13904 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13906 (prune_runtime_alias_test_list): New decalaration.
13908 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13910 * tree-vect-data-refs.c (compare_tree): Rename and move ...
13911 * tree-data-ref.c (data_ref_compare_tree): ... to here.
13912 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
13913 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
13914 (operator==, comp_dr_with_seg_len_pair): Ditto.
13915 (vect_prune_runtime_alias_test_list): Ditto.
13917 2017-05-26 Martin Liska <mliska@suse.cz>
13920 * params.def: Bound partial-inlining-entry-probability param.
13922 2017-05-26 Marek Polacek <polacek@redhat.com>
13925 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
13928 2017-05-26 Richard Biener <rguenther@suse.de>
13930 PR tree-optimization/80842
13931 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
13934 2017-05-26 Richard Biener <rguenther@suse.de>
13936 PR tree-optimization/80844
13937 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
13939 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
13941 * doc/md.texi (Machine Constraints): Update x86 family
13942 machine constraints section to match 'config/i386/constraints.md'.
13944 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
13946 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
13948 2017-05-25 Nathan Sidwell <nathan@acm.org>
13950 * doc/invoke.texi (--enable-languages): Update documentation.
13952 2017-05-25 Martin Liska <mliska@suse.cz>
13954 * dumpfile.c: Add TDF_FOLDING.
13955 * dumpfile.h (enum dump_kind): Likewise.
13956 * genmatch.c (dt_simplify::gen_1): Use it.
13958 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13960 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
13962 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13964 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13965 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
13967 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13969 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
13970 * match.pd (X == C): Rewrite it here.
13971 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
13972 with_certain_nonzero_bits2): New predicates.
13973 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
13975 2017-05-24 Nathan Sidwell <nathan@acm.org>
13977 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
13980 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
13983 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13985 * config/powerpcspe: New port. Files are copied from the rs6000
13986 port, with "rs6000" in filenames replaced by "powerpcspe".
13988 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
13990 PR rtl-optimization/80754
13991 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
13993 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
13995 * config/sparc/sparc.md (length): Return the correct value for -mflat
13996 sibcalls to match output_sibcall.
13998 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
14002 * config/rs6000/rs6000.c (struct machine_function): Add new field
14004 (rs6000_get_separate_components): Init that field, use it.
14005 (rs6000_components_for_bb): Use the field.
14007 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
14009 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
14011 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
14013 PR middle-end/80823
14014 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
14016 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14019 * config/s390/s390.c (s390_check_qrst_address): Check incoming
14020 address against address_operand predicate.
14021 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
14023 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
14025 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
14026 parameters passed indirectly.
14028 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
14030 * config/i386/i386.md (*movdi_internal): Remove SSE4
14031 alternative 18 (?r, *v). Update insn attributes.
14032 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
14033 Update insn attributes.
14034 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
14035 Update insn attributes.
14036 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
14037 alternative 1 (r, v). Remove isa attribute.
14038 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
14039 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
14040 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
14042 2017-05-23 Tom de Vries <tom@codesourcery.com>
14044 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
14047 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14049 * cgraphunit.c (symbol_table::process_new_functions): Update.
14050 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
14051 (inline_generate_summary): Rename to ...
14052 (ipa_fn_summary_generate): ... this one.
14053 (inline_read_summary): Rename to ...
14054 (ipa_fn_summary_read): ... this one.
14055 (inline_write_summary): Rename to ...
14056 (ipa_fn_summary_write): ... this one.
14057 (inline_free_summary): Rename to ...
14058 (ipa_free_fn_summary): ... this one.
14059 (pass_data_local_fn_summary, pass_local_fn_summary,
14060 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
14061 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
14062 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
14063 make_pass_ipa_fn_summary): New.
14064 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
14065 inline_write_summary, inline_free_summary): Remove.
14066 (ipa_free_fn_summary) : New.
14067 * ipa-inline.c (ipa_inline): Update.
14068 (pass_ipa_inline): Do not generate summaries.
14069 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
14071 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
14072 and add pass_ipa_fn_summary.
14073 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
14075 (make_pass_inline_parameters): Remove.
14077 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
14079 * omp-low.c (struct omp_context): Remove "default_kind" member.
14082 * omp-offload.c (execute_oacc_device_lower): Remove the
14083 parallelism dimensions function attributes for unparallelized
14084 OpenACC kernels constructs.
14086 2017-05-23 Martin Liska <mliska@suse.cz>
14088 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
14090 (cgraph_edge::make_speculative): Likewise.
14091 (cgraph_edge::resolve_speculation): Likewise.
14092 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
14093 (cgraph_node::dump): Likewise.
14094 * cgraph.h: Likewise.
14095 * cgraphunit.c (analyze_functions): Likewise.
14096 (symbol_table::compile): Likewise.
14097 * ipa-cp.c (print_all_lattices): Likewise.
14098 (determine_versionability): Likewise.
14099 (initialize_node_lattices): Likewise.
14100 (ipcp_verify_propagated_values): Likewise.
14101 (estimate_local_effects): Likewise.
14102 (update_profiling_info): Likewise.
14103 (create_specialized_node): Likewise.
14104 (perhaps_add_new_callers): Likewise.
14105 (decide_about_value): Likewise.
14106 (decide_whether_version_node): Likewise.
14107 (identify_dead_nodes): Likewise.
14108 (ipcp_store_bits_results): Likewise.
14109 * ipa-devirt.c (dump_targets): Likewise.
14110 (ipa_devirt): Likewise.
14111 * ipa-icf.c (sem_item::dump): Likewise.
14112 (sem_function::equals): Likewise.
14113 (sem_variable::equals): Likewise.
14114 (sem_item_optimizer::read_section): Likewise.
14115 (sem_item_optimizer::execute): Likewise.
14116 (congruence_class::dump): Likewise.
14117 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
14118 (dump_inline_summary): Likewise.
14119 (estimate_node_size_and_time): Likewise.
14120 (inline_analyze_function): Likewise.
14121 * ipa-inline-transform.c (inline_call): Likewise.
14122 * ipa-inline.c (report_inline_failed_reason): Likewise.
14123 (want_early_inline_function_p): Likewise.
14124 (edge_badness): Likewise.
14125 (update_edge_key): Likewise.
14126 (inline_small_functions): Likewise.
14127 * ipa-profile.c (ipa_profile): Likewise.
14128 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
14129 (ipa_make_edge_direct_to_target): Likewise.
14130 (remove_described_reference): Likewise.
14131 (ipa_impossible_devirt_target): Likewise.
14132 (propagate_controlled_uses): Likewise.
14133 (ipa_print_node_params): Likewise.
14134 (ipcp_transform_function): Likewise.
14135 * ipa-pure-const.c (pure_const_read_summary): Likewise.
14136 (propagate_pure_const): Likewise.
14137 * ipa-reference.c (generate_summary): Likewise.
14138 (read_write_all_from_decl): Likewise.
14139 (propagate): Likewise.
14140 (ipa_reference_read_optimization_summary): Likewise.
14141 * ipa-utils.c (ipa_merge_profiles): Likewise.
14142 * ipa.c (walk_polymorphic_call_targets): Likewise.
14143 (symbol_table::remove_unreachable_nodes): Likewise.
14144 (ipa_single_use): Likewise.
14145 * passes.c (execute_todo): Likewise.
14146 * predict.c (drop_profile): Likewise.
14147 * symtab.c (symtab_node::get_dump_name): New function.
14148 (symtab_node::dump_name): Likewise.
14149 (symtab_node::dump_asm_name): Likewise.
14150 (symtab_node::dump_references): Likewise.
14151 (symtab_node::dump_referring): Likewise.
14152 (symtab_node::dump_base): Likewise.
14153 (symtab_node::debug_symtab): Likewise.
14154 * tree-sra.c (convert_callers_for_node): Likewise.
14155 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
14156 * value-prof.c (init_node_map): Likewise.
14158 2017-05-23 Martin Liska <mliska@suse.cz>
14160 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
14161 and symtab_node::debug_symtab to symbol_table::debug.
14162 * cgraphunit.c (analyze_functions): Use the renamed function.
14163 (symbol_table::compile): Likewise.
14164 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
14165 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
14166 * passes.c (execute_todo): Likewise.
14167 * symtab.c (symbol_table::dump): New function.
14168 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
14170 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14172 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
14173 that nonconst implies exec.
14175 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
14177 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14178 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14179 (inline_edge_summary_vec): Turn into ...
14180 (ipa_call_summaries): ... this one.
14181 (redirect_to_unreachable, edge_set_predicate,
14182 evaluate_properties_for_edge, inline_summary_alloc,
14183 reset_ipa_call_summary, reset_inline_summary,
14184 inline_summary_t::duplicate): Update.
14185 (inline_edge_duplication_hook): Turn to ...
14186 (ipa_call_summary_t::duplicate): ... this one.
14187 (inline_edge_removal_hook): Turn to ...
14188 (ipa_call_summary_t::remove): ... this one.
14189 (dump_inline_edge_summary): Turn to ...
14190 (dump_ipa_call_summary): ... this one.
14191 (estimate_function_body_sizes): Update.
14192 (inline_update_callee_summaries): Update.
14193 (remap_edge_change_prob): Update.
14194 (remap_edge_summaries): Update.
14195 (inline_merge_summary): Update.
14196 (do_estimate_edge_time): Update.
14197 (inline_generate_summary): Update.
14198 (inline_read_section): Update.
14199 (inline_read_summary): Update.
14200 (inline_free_summary): Update.
14201 * ipa-inline.c (can_inline_edge_p): Update.
14202 (compute_inlined_call_time): Update.
14203 (want_inline_small_function_p): Update.
14204 (edge_badness): Update.
14205 (early_inliner): Update.
14206 * ipa-inline.h (inline_edge_summary): Turn to ...
14207 (ipa_call_summary): ... this one.
14208 (ipa_call_summary_t): New class.
14209 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14210 (ipa_call_summaries): New.
14211 (inline_edge_summary): Remove.
14212 (estimate_edge_growth): Update.
14213 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14214 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14215 * ipa-split.c (execute_split_functions): Update.
14216 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14218 2017-05-23 Tom de Vries <tom@codesourcery.com>
14220 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14221 attributes): Document rdrand effective target.
14223 2017-05-23 Tom de Vries <tom@codesourcery.com>
14225 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
14226 attributes): Sort alphabetically.
14228 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
14230 * config/avr/genmultilib.awk: Use gsub instead of gensub.
14232 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
14235 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
14236 V2DF/V2DI splat into two separate patterns, one that handles
14237 registers, and the other that only handles memory. Drop support
14238 for splatting from a GPR on ISA 2.07 and then splitting the
14239 splat into direct move and splat.
14240 (vsx_splat_<mode>_reg): Likewise.
14241 (vsx_splat_<mode>_mem): Likewise.
14243 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
14245 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
14247 2017-05-22 Jakub Jelinek <jakub@redhat.com>
14249 PR middle-end/80809
14250 * omp-low.c (finish_taskreg_remap): New function.
14251 (finish_taskreg_scan): If unit size of ctx->record_type
14252 is non-constant, unshare the size expression and replace
14253 decls in it with possible outer var refs.
14255 PR middle-end/80809
14256 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
14257 GOVD_SHARED rather than GOVD_PRIVATE with it.
14258 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
14259 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
14261 PR middle-end/80853
14262 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
14263 as last argument to build_outer_var_ref for pointer bases of array
14264 section reductions.
14266 2017-05-19 Martin Sebor <msebor@redhat.com>
14268 * print-tree.c (print_node): Print DECL_READ_P flag.
14270 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14272 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
14273 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
14274 * cgraph.c: Likewise.
14275 * cgraphunit.c: Likewise.
14276 * gengtype.c: Likewise.
14277 * ipa-cp.c: Likewise.
14278 * ipa-devirt.c: Likewise.
14279 * ipa-icf.c: Likewise.
14280 * ipa-predicate.c: Likewise.
14281 * ipa-profile.c: Likewise.
14282 * ipa-prop.c: Likewise.
14283 * ipa-split.c: Likewise.
14285 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
14286 edge_predicate_pool, dump_inline_hints,
14287 inline_summary::account_size_time, redirect_to_unreachable,
14288 edge_set_predicate, set_hint_predicate,
14289 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
14290 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
14291 inline_summary_t::remove, remap_hint_predicate_after_duplication,
14292 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
14293 ipa_call_summary_t::remove, initialize_growth_caches,
14294 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
14295 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
14296 mark_modified, unmodified_parm_1, unmodified_parm,
14297 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
14298 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
14299 compute_bb_predicates, will_be_nonconstant_expr_predicate,
14300 will_be_nonconstant_predicate, record_modified_bb_info,
14301 get_minimal_bb, record_modified, param_change_prob,
14302 phi_result_unknown_predicate, predicate_for_phi_result,
14303 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
14304 estimate_function_body_sizes, compute_inline_parameters,
14305 compute_inline_parameters_for_curren, pass_data_inline_parameters,
14306 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
14307 inline_update_callee_summaries, remap_edge_change_prob,
14308 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
14309 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
14310 inline_analyze_function, inline_summary_t::insert,
14311 inline_generate_summary, read_ipa_call_summary, inline_read_section,
14312 inline_read_summary, write_ipa_call_summary, inline_write_summary,
14313 inline_free_summary): Move to ipa-fnsummary.h
14314 (predicate_t): Remove.
14315 * ipa-fnsummary.c: New file.
14316 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
14317 (enum inline_hints_vals, inline_hints, agg_position_info,
14318 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
14319 inline_summaries, ipa_call_summary, ipa_call_summary_t,
14320 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
14321 dump_inline_summary, dump_inline_hints, inline_generate_summary,
14322 inline_read_summary, inline_write_summary, inline_free_summary,
14323 inline_analyze_function, initialize_inline_failed,
14324 inline_merge_summary, inline_update_overall_summary,
14325 compute_inline_parameters): Move to ipa-fnsummary.h
14326 * ipa-fnsummary.h: New file.
14327 * ipa-inline-transform.h: Include ipa-inline.h.
14328 * ipa-inline.c: LIkewise.
14330 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14332 * ipa-inline.c (edge_badness): Use inlined_time instead of
14333 inline_summaries->get.
14335 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14337 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
14339 2017-05-22 Nathan Sidwell <nathan@acm.org>
14341 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
14342 (fdump-lang): Document 'raw' option.
14343 * dumpfile.h (TDI_tu): Delete.
14344 * dumpfile.c (dump_files): Remove translation-unit.
14345 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
14347 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
14349 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
14350 command option from $(AWK) call.
14351 * config/avr/genmultilib.awk: Simplify and rewrite so that it
14352 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
14353 [FORMAT]: Remove handling of variable.
14354 * config/avr/t-multilib: Regenerate.
14356 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14358 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
14360 (dump_inline_summary): Do not print self_time.
14361 (estimate_function_body_sizes): Do not set self_time.
14362 (compute_inline_parameters): Likewise.
14363 (inline_read_section, inline_write_summary): Do not stream self_time.
14364 * ipa-inline.h (inline_summary): Drop self_time.
14366 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
14368 * ipa-inline-analysis.c (account_size_time): Rename to ...
14369 (inline_summary::account_size_time): ... this one.
14370 (reset_ipa_call_summary): Turn to ...
14371 (ipa_call_summary::reset): ... this one.
14372 (reset_inline_summary): Turn to ...
14373 (inline_summary::reset): ... this one.
14374 (inline_summary_t::remove): Update.
14375 (inline_summary_t::duplicate): Update.
14376 (ipa_call_summary_t::remove): Update.
14377 (dump_inline_summary): Update.
14378 (estimate_function_body_sizes): Update.
14379 (compute_inline_parameters): Update.
14380 (estimate_node_size_and_time): Update.
14381 (inline_merge_summary): Update.
14382 (inline_update_overall_summary): Update.
14383 (inline_read_section): Update.
14384 (inline_write_summary): Update.
14385 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
14386 add account_size_time and reset member functions.
14387 (ipa_call_summary): Add reset function.
14388 * ipa-predicate.h (predicate::operator &): Constify.
14390 2017-05-22 Richard Biener <rguenther@suse.de>
14392 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
14394 2017-05-19 Jason Merrill <jason@redhat.com>
14396 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
14398 2017-05-19 Marek Polacek <polacek@redhat.com>
14401 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
14402 TYPE_OVERFLOW_WRAPS checks.
14404 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
14406 * tree-core.h (enum omp_clause_default_kind): Add
14407 "OMP_CLAUSE_DEFAULT_PRESENT".
14408 * tree-pretty-print.c (dump_omp_clause): Handle it.
14409 * gimplify.c (enum gimplify_omp_var_data): Add
14410 "GOVD_MAP_FORCE_PRESENT".
14411 (gimplify_adjust_omp_clauses_1): Map it to
14412 "GOMP_MAP_FORCE_PRESENT".
14413 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
14415 * gimplify.c (oacc_default_clause): Clarify.
14417 2017-05-19 Nathan Sidwell <nathan@acm.org>
14419 LANG_HOOK_REGISTER_DUMPS
14420 * toplev.c (general_init): Call register dump lang hook.
14421 * doc/invoke.texi: Document -fdump-lang option family.
14422 * dumpfile.c (dump_files): Remove class dump here.
14423 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
14424 * dumpfile.h (tree_dump_index): Remove TDI_class.
14425 * langhooks-def.h (lhd_register_dumps): Declare.
14426 (LANG_HOOKS_REGISTER_DUMPS): Define.
14427 (LANG_HOOKS_INITIALIZER): Add it.
14428 * langhooks.c (lhd_register_dumps): Define.
14429 * langhooks.h (struct lang_hooks): Add register_dumps.
14431 2017-05-19 Nathan Sidwell <nathan@acm.org>
14433 * context.h (context::set_passes): New.
14434 * context.c (context::context): Do not create pass manager.
14435 * toplev.c (general_init): Create pass manager here.
14437 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
14439 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
14440 use this splitter if two add or or instructions would also work for
14441 the constant we want to generate.
14443 2017-05-19 Richard Biener <rguenther@suse.de>
14446 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
14447 predicate evaluation.
14449 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14451 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
14453 * ipa-inline.c (want_inline_small_function_p): Do not cast to
14456 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
14458 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
14459 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
14460 (inline_edge_summary_vec): Turn into ...
14461 (ipa_call_summaries): ... this one.
14462 (redirect_to_unreachable, edge_set_predicate,
14463 evaluate_properties_for_edge, inline_summary_alloc,
14464 reset_ipa_call_summary, reset_inline_summary,
14465 inline_summary_t::duplicate): Update.
14466 (inline_edge_duplication_hook): Turn to ...
14467 (ipa_call_summary_t::duplicate): ... this one.
14468 (inline_edge_removal_hook): Turn to ...
14469 (ipa_call_summary_t::remove): ... this one.
14470 (dump_inline_edge_summary): Turn to ...
14471 (dump_ipa_call_summary): ... this one.
14472 (estimate_function_body_sizes): Update.
14473 (inline_update_callee_summaries): Update.
14474 (remap_edge_change_prob): Update.
14475 (remap_edge_summaries): Update.
14476 (inline_merge_summary): Update.
14477 (do_estimate_edge_time): Update.
14478 (inline_generate_summary): Update.
14479 (inline_read_section): Update.
14480 (inline_read_summary): Update.
14481 (inline_free_summary): Update.
14482 * ipa-inline.c (can_inline_edge_p): Update.
14483 (compute_inlined_call_time): Update.
14484 (want_inline_small_function_p): Update.
14485 (edge_badness): Update.
14486 (early_inliner): Update.
14487 * ipa-inline.h (inline_edge_summary): Turn to ...
14488 (ipa_call_summary): ... this one.
14489 (ipa_call_summary_t): New class.
14490 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
14491 (ipa_call_summaries): New.
14492 (inline_edge_summary): Remove.
14493 (estimate_edge_growth): Update.
14494 * ipa-profile.c (ipa_propagate_frequency_1): Update.
14495 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
14496 * ipa-split.c (execute_split_functions): Update.
14497 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
14499 2017-05-19 Richard Biener <rguenther@suse.de>
14501 PR middle-end/80764
14502 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
14504 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
14506 * config/rs6000/rs6000.c (struct machine_function): Add field
14507 fpr_is_wrapped_separately.
14508 (rs6000_get_separate_components): Use 64 components. Handle the
14509 new FPR components.
14510 (rs6000_components_for_bb): Handle the FPR components.
14511 (rs6000_emit_prologue_components): Handle the FPR components.
14512 (rs6000_emit_epilogue_components): Handle the FPR components.
14513 (rs6000_set_handled_components): Handle the FPR components.
14514 (rs6000_emit_prologue): Don't output prologue code for those FPRs
14515 that are already separately shrink-wrapped.
14516 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
14517 that are already separately shrink-wrapped.
14519 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
14522 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
14525 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
14526 (define_peephole2 for Altivec d-form load): Add peepholes to catch
14527 cases where the register allocator uses a move and an offsettable
14528 memory operation to/from a FPR register on ISA 2.06/2.07.
14529 (define_peephole2 for Altivec d-form store): Likewise.
14531 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
14534 * config/i386/mmx.md (*mov<mode>_internal): Enable
14535 alternatives 11, 12, 13 and 14 also for 32bit targets.
14536 Remove alternatives 15, 16, 17 and 18.
14537 * config/i386/sse.md (vec_concatv2di): Change
14538 alternative (!x, *y) to (x, ?!*Yn).
14540 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
14542 * dumpfile.h (enum dump_kind): Remove stray comma.
14544 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14546 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
14547 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
14548 predicate::num_conditions
14549 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
14550 (CHANGED): turn into predicate::changed.
14551 (agg_position_info): Move to ipa-predicate.h
14552 (add_condition, predicate::add_clause, predicate::operator &=,
14553 predicate::or_with, predicate::evaluate, predicate::probability,
14554 dump_condition, dump_clause, predicate::dump,
14555 predicate::remap_after_duplication, predicate::remap_after_inlining,
14556 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
14557 (evaluate_conditions_for_known_args): Update.
14558 (set_cond_stmt_execution_predicate): Update.
14559 * ipa-inline.h: Include ipa-predicate.h
14560 (condition, inline_param_summary, conditions, agg_position_info,
14561 predicate): Move to ipa-predicate.h
14562 * ipa-predicate.c: New file.
14563 * ipa-predicate.h: New file.
14565 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
14567 * final.c (leaf_function_p): Check we are not in a sequence.
14569 2017-05-18 Martin Liska <mliska@suse.cz>
14571 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
14572 * dumpfile.c (dump_register): Use new enum dump_kind.
14573 (get_dump_file_name): Likewise.
14574 (dump_enable_all): Likewise.
14575 (dump_switch_p_1): Likewise.
14576 (enable_rtl_dump_file): Remove usage of TDF_RTL.
14577 * dumpfile.h (enum dump_kind): New enum type.
14578 (struct dump_file_info): Create constructor and
14579 format fields and comments.
14580 * passes.c (pass_manager::register_one_dump_file):
14582 * statistics.c (statistics_early_init): Likewise.
14583 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
14584 TDF_TREE with TDF_SLIM.
14585 (gather_memory_references_ref): Likewise.
14587 2017-05-18 Martin Liska <mliska@suse.cz>
14589 * vec.h (struct vnull): Use it.
14591 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
14593 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
14594 (true_predicate, false_predicate, true_predicate_p,
14595 false_predicate_p): Remove.
14596 (single_cond_predicate, not_inlined_predicate): Turn to member function
14598 (add_condition): Update.
14599 (add_clause): Turn to...
14600 (predicate::add_clause): ... this one; update; allow passing NULL
14602 (and_predicates): Turn to ...
14603 (predicate::operator &=): ... this one.
14604 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
14605 (or_predicates): Turn to ...
14606 (predicate::or_with): ... this one.
14607 (evaluate_predicate): Turn to ...
14608 (predicate::evaluate): ... this one.
14609 (predicate_probability): Turn to ...
14610 (predicate::probability): ... this one.
14611 (dump_condition): Update.
14612 (dump_predicate): Turn to ...
14613 (predicate::dump): ... this one.
14614 (account_size_time): Update.
14615 (edge_set_predicate): Update.
14616 (set_hint_predicate): UPdate.
14617 (evaluate_conditions_for_known_args): Update.
14618 (evaluate_properties_for_edge): Update.
14619 (remap_predicate_after_duplication): Turn to...
14620 (predicate::remap_after_duplication): ... this one.
14621 (remap_hint_predicate_after_duplication): Update.
14622 (inline_summary_t::duplicate): UPdate.
14623 (dump_inline_edge_summary): Update.
14624 (dump_inline_summary): Update.
14625 (set_cond_stmt_execution_predicate): Update.
14626 (set_switch_stmt_execution_predicate): Update.
14627 (compute_bb_predicates): Update.
14628 (will_be_nonconstant_expr_predicate): Update.
14629 (will_be_nonconstant_predicate): Update.
14630 (phi_result_unknown_predicate): Update.
14631 (predicate_for_phi_result): Update.
14632 (array_index_predicate): Update.
14633 (estimate_function_body_sizes): Update.
14634 (estimate_node_size_and_time): Update.
14635 (estimate_ipcp_clone_size_and_time): Update.
14636 (remap_predicate): Rename to ...
14637 (predicate::remap_after_inlining): ... this one.
14638 (remap_hint_predicate): Update.
14639 (inline_merge_summary): Update.
14640 (inline_update_overall_summary): Update.
14641 (estimate_size_after_inlining): Update.
14642 (read_predicate): Rename to ...
14643 (predicate::stream_in): ... this one.
14644 (read_inline_edge_summary): Update.
14645 (write_predicate): Rename to ...
14646 (predicate::stream_out): ... this one.
14647 (write_inline_edge_summary): Update.
14648 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
14649 (clause_t): Turn to uint32_t
14650 (predicate): Turn to class; implement constructor and operators
14652 (size_time_entry): Update.
14653 (inline_summary): Update.
14654 (inline_edge_summary): Update.
14656 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
14658 * fold-const.c (fold_binary_loc): Move transformation...
14659 * match.pd (C - X CMP X): ... here.
14661 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
14663 * config/sparc/sparc.c (sparc_option_override): Set function
14664 alignment for -mcpu=niagara7 to 64 to match the I$ line.
14665 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
14667 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
14669 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
14671 2017-05-18 Marek Polacek <polacek@redhat.com>
14674 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
14675 (pass_ubsan::execute): Call gimple_assign_single_p instead of
14676 gimple_assign_load_p.
14678 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
14680 PR middle-end/80692
14681 * real.c (do_compare): Give decimal_do_compare preference over
14682 comparing just the signs.
14684 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
14686 * doc/md.texi (Canonicalization of Instructions): Describe the
14687 canonical form of instructions that inherently set a condition
14690 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
14692 PR middle-end/80775
14693 * tree-cfg.c: Move deletion of unreachable case statements to after
14694 the merging of consecutive case labels.
14696 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
14698 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
14699 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
14700 restoring of callee-saved registers.
14702 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
14704 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
14705 * config/visium/visium.c (single_set_and_flags): Likewise.
14706 * config/visium/visium.md (Substitutions): Likewise.
14708 2017-05-17 Martin Liska <mliska@suse.cz>
14710 * cfg.c: Introduce dump_flags_t type and
14711 use it instead of int type.
14713 * cfghooks.c: Likewise.
14714 * cfghooks.h (struct cfg_hooks): Likewise.
14715 * cfgrtl.c: Likewise.
14716 * cfgrtl.h: Likewise.
14717 * cgraph.c (cgraph_node::get_body): Likewise.
14718 * coretypes.h: Likewise.
14719 * domwalk.c: Likewise.
14720 * domwalk.h: Likewise.
14721 * dumpfile.c (struct dump_option_value_info): Likewise.
14722 (dump_enable_all): Likewise.
14723 (dump_switch_p_1): Likewise.
14724 (opt_info_switch_p): Likewise.
14725 * dumpfile.h (enum tree_dump_index): Likewise.
14726 (struct dump_file_info): Likewise.
14727 * genemit.c: Likewise.
14728 * generic-match-head.c: Likewise.
14729 * gengtype.c (open_base_files): Likewise.
14730 * gimple-pretty-print.c: Likewise.
14731 * gimple-pretty-print.h: Likewise.
14732 * graph.c (print_graph_cfg): Likewise.
14733 * graphite-scop-detection.c (dot_all_sese): Likewise.
14734 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
14735 * loop-unroll.c (report_unroll): Likewise.
14736 * passes.c (pass_manager::register_one_dump_file): Likewise.
14737 * print-tree.c: Likewise.
14738 * statistics.c: Likewise.
14739 * tree-cfg.c: Likewise.
14740 * tree-cfg.h: Likewise.
14741 * tree-dfa.c: Likewise.
14742 * tree-dfa.h: Likewise.
14743 * tree-dump.c (dump_function): Likewise.
14744 * tree-dump.h (struct dump_info): Likewise.
14745 * tree-pretty-print.c: Likewise.
14746 * tree-pretty-print.h: Likewise.
14747 * tree-ssa-live.c: Likewise.
14748 * tree-ssa-live.h: Likewise.
14749 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
14750 * tree-vect-loop.c: Likewise.
14751 * tree-vect-slp.c: Likewise.
14753 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
14754 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14756 PR tree-optimization/80457
14757 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
14758 of all arguments to a statement as scalar_to_vec operations.
14759 (vectorizable_call): Adjust call to vect_model_simple_cost for
14761 (vectorizable_conversion): Likewise.
14762 (vectorizable_assignment): Likewise.
14763 (vectorizable_shift): Likewise.
14764 (vectorizable_operation): Likewise.
14765 (vectorizable_comparison): Likewise.
14766 (vect_is_simple_cond): Record the def types for operands.
14767 (vectorizable_condition): Likewise, call vect_model_simple_cost.
14768 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
14769 for statement argument count.
14771 2017-05-16 Carl Love <cel@us.ibm.com>
14773 * config/rs6000/rs6000-c: Add support for built-in functions
14774 vector unsigned long long vec_bperm (vector unsigned long long,
14775 vector unsigned char)
14776 vector signed long long vec_mule (vector signed int,
14778 vector unsigned long long vec_mule (vector unsigned int,
14779 vector unsigned int)
14780 vector signed long long vec_mulo (vector signed int,
14782 vector unsigned long long vec_mulo (vector unsigned int,
14783 vector unsigned int)
14784 vector signed char vec_sldw (vector signed char,
14785 vector signed char,
14787 vector unsigned char vec_sldw (vector unsigned char,
14788 vector unsigned char,
14790 vector signed short vec_sldw (vector signed short,
14791 vector signed short,
14793 vector unsigned short vec_sldw (vector unsigned short,
14794 vector unsigned short,
14796 vector signed int vec_sldw (vector signed int,
14799 vector unsigned int vec_sldw (vector unsigned int,
14800 vector unsigned int,
14802 vector signed long long vec_sldw (vector signed long long,
14803 vector signed long long,
14805 vector unsigned long long vec_sldw (vector unsigned long long,
14806 vector unsigned long long,
14808 * config/rs6000/rs6000-c: Add support for built-in functions
14809 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
14810 * config/rs6000/altivec.h: Add defintion for vec_sldw.
14811 * doc/extend.texi: Update the built-in documentation for the
14812 new built-in functions.
14814 2017-05-16 Marek Polacek <polacek@redhat.com>
14818 * tree.c (save_expr): Don't fold the expression.
14820 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
14822 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
14823 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
14824 and (?*y,m). Update insn attributes.
14826 2017-05-16 Martin Liska <mliska@suse.cz>
14828 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
14829 flags argument of print_gimple_stmt, print_gimple_expr,
14830 print_generic_stmt and print_generic_expr.
14831 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
14832 * coretypes.h: Likewise.
14833 * except.c (dump_eh_tree): Likewise.
14834 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14835 * gimple-pretty-print.h: Likewise.
14836 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
14837 (backprop::push_to_worklist): Likewise.
14838 (backprop::pop_from_worklist): Likewise.
14839 (backprop::process_use): Likewise.
14840 (backprop::intersect_uses): Likewise.
14841 (note_replacement): Likewise.
14842 * gimple-ssa-store-merging.c
14843 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
14844 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
14845 (pass_store_merging::execute): Likewise.
14846 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
14847 (ssa_base_cand_dump_callback): Likewise.
14848 (dump_incr_vec): Likewise.
14849 (replace_refs): Likewise.
14850 (replace_mult_candidate): Likewise.
14851 (create_add_on_incoming_edge): Likewise.
14852 (create_phi_basis): Likewise.
14853 (insert_initializers): Likewise.
14854 (all_phi_incrs_profitable): Likewise.
14855 (introduce_cast_before_cand): Likewise.
14856 (replace_one_candidate): Likewise.
14857 * gimplify.c (gimplify_expr): Likewise.
14858 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
14859 (set_rename): Likewise.
14860 (rename_uses): Likewise.
14861 (copy_loop_phi_nodes): Likewise.
14862 (add_close_phis_to_merge_points): Likewise.
14863 (copy_loop_close_phi_args): Likewise.
14864 (copy_cond_phi_args): Likewise.
14865 (graphite_copy_stmts_from_block): Likewise.
14866 (translate_pending_phi_nodes): Likewise.
14867 * graphite-poly.c (print_pdr): Likewise.
14868 (dump_gbb_cases): Likewise.
14869 (dump_gbb_conditions): Likewise.
14870 (print_scop_params): Likewise.
14871 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
14872 (build_cross_bb_scalars_use): Likewise.
14873 (gather_bbs::before_dom_children): Likewise.
14874 * hsa-dump.c (dump_hsa_immed): Likewise.
14875 * ipa-cp.c (print_ipcp_constant_value): Likewise.
14876 (get_replacement_map): Likewise.
14877 * ipa-inline-analysis.c (dump_condition): Likewise.
14878 (estimate_function_body_sizes): Likewise.
14879 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
14880 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
14881 * ipa-prop.c (ipa_dump_param): Likewise.
14882 (ipa_print_node_jump_functions_for_edge): Likewise.
14883 (ipa_modify_call_arguments): Likewise.
14884 (ipa_modify_expr): Likewise.
14885 (ipa_dump_param_adjustments): Likewise.
14886 (ipa_dump_agg_replacement_values): Likewise.
14887 (ipcp_modif_dom_walker::before_dom_children): Likewise.
14888 * ipa-pure-const.c (check_stmt): Likewise.
14889 (pass_nothrow::execute): Likewise.
14890 * ipa-split.c (execute_split_functions): Likewise.
14891 * omp-offload.c (dump_oacc_loop_part): Likewise.
14892 (dump_oacc_loop): Likewise.
14893 * trans-mem.c (tm_log_emit): Likewise.
14894 (tm_memopt_accumulate_memops): Likewise.
14895 (dump_tm_memopt_set): Likewise.
14896 (dump_tm_memopt_transform): Likewise.
14897 * tree-cfg.c (gimple_verify_flow_info): Likewise.
14898 (print_loop): Likewise.
14899 * tree-chkp-opt.c (chkp_print_addr): Likewise.
14900 (chkp_gather_checks_info): Likewise.
14901 (chkp_get_check_result): Likewise.
14902 (chkp_remove_check_if_pass): Likewise.
14903 (chkp_use_outer_bounds_if_possible): Likewise.
14904 (chkp_reduce_bounds_lifetime): Likewise.
14905 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
14906 (chkp_mark_completed_bounds): Likewise.
14907 (chkp_register_incomplete_bounds): Likewise.
14908 (chkp_mark_invalid_bounds): Likewise.
14909 (chkp_maybe_copy_and_register_bounds): Likewise.
14910 (chkp_build_returned_bound): Likewise.
14911 (chkp_get_bound_for_parm): Likewise.
14912 (chkp_build_bndldx): Likewise.
14913 (chkp_get_bounds_by_definition): Likewise.
14914 (chkp_generate_extern_var_bounds): Likewise.
14915 (chkp_get_bounds_for_decl_addr): Likewise.
14916 * tree-chrec.c (chrec_apply): Likewise.
14917 * tree-data-ref.c (dump_data_reference): Likewise.
14918 (dump_subscript): Likewise.
14919 (dump_data_dependence_relation): Likewise.
14920 (analyze_overlapping_iterations): Likewise.
14921 * tree-inline.c (expand_call_inline): Likewise.
14922 (tree_function_versioning): Likewise.
14923 * tree-into-ssa.c (dump_defs_stack): Likewise.
14924 (dump_currdefs): Likewise.
14925 (dump_names_replaced_by): Likewise.
14926 (dump_update_ssa): Likewise.
14927 (update_ssa): Likewise.
14928 * tree-object-size.c (pass_object_sizes::execute): Likewise.
14929 * tree-parloops.c (build_new_reduction): Likewise.
14930 (try_create_reduction_list): Likewise.
14931 (ref_conflicts_with_region): Likewise.
14932 (oacc_entry_exit_ok_1): Likewise.
14933 (oacc_entry_exit_single_gang): Likewise.
14934 * tree-pretty-print.h: Likewise.
14935 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
14936 (get_scalar_evolution): Likewise.
14937 (add_to_evolution): Likewise.
14938 (get_loop_exit_condition): Likewise.
14939 (analyze_evolution_in_loop): Likewise.
14940 (analyze_initial_condition): Likewise.
14941 (analyze_scalar_evolution): Likewise.
14942 (instantiate_scev): Likewise.
14943 (number_of_latch_executions): Likewise.
14944 (gather_chrec_stats): Likewise.
14945 (final_value_replacement_loop): Likewise.
14946 (scev_const_prop): Likewise.
14947 * tree-sra.c (dump_access): Likewise.
14948 (disqualify_candidate): Likewise.
14949 (create_access): Likewise.
14950 (reject): Likewise.
14951 (maybe_add_sra_candidate): Likewise.
14952 (create_access_replacement): Likewise.
14953 (analyze_access_subtree): Likewise.
14954 (analyze_all_variable_accesses): Likewise.
14955 (sra_modify_assign): Likewise.
14956 (initialize_constant_pool_replacements): Likewise.
14957 (find_param_candidates): Likewise.
14958 (decide_one_param_reduction): Likewise.
14959 (replace_removed_params_ssa_names): Likewise.
14960 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
14961 * tree-ssa-copy.c (dump_copy_of): Likewise.
14962 (copy_prop_visit_cond_stmt): Likewise.
14963 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
14964 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
14965 (record_equivalences_from_stmt): Likewise.
14966 * tree-ssa-dse.c (compute_trims): Likewise.
14967 (delete_dead_call): Likewise.
14968 (delete_dead_assignment): Likewise.
14969 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
14970 (forward_propagate_into_cond): Likewise.
14971 (pass_forwprop::execute): Likewise.
14972 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14973 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
14975 (move_computations_worker): Likewise.
14976 (execute_sm): Likewise.
14977 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
14978 (remove_exits_and_undefined_stmts): Likewise.
14979 (remove_redundant_iv_tests): Likewise.
14980 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
14981 (adjust_iv_update_pos): Likewise.
14982 * tree-ssa-math-opts.c (bswap_replace): Likewise.
14983 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
14984 (value_replacement): Likewise.
14985 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
14986 * tree-ssa-pre.c (print_pre_expr): Likewise.
14987 (get_representative_for): Likewise.
14988 (create_expression_by_pieces): Likewise.
14989 (insert_into_preds_of_block): Likewise.
14990 (eliminate_insert): Likewise.
14991 (eliminate_dom_walker::before_dom_children): Likewise.
14992 (eliminate): Likewise.
14993 (remove_dead_inserted_code): Likewise.
14994 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
14995 * tree-ssa-reassoc.c (get_rank): Likewise.
14996 (eliminate_duplicate_pair): Likewise.
14997 (eliminate_plus_minus_pair): Likewise.
14998 (eliminate_not_pairs): Likewise.
14999 (undistribute_ops_list): Likewise.
15000 (eliminate_redundant_comparison): Likewise.
15001 (update_range_test): Likewise.
15002 (optimize_range_tests_var_bound): Likewise.
15003 (optimize_vec_cond_expr): Likewise.
15004 (rewrite_expr_tree): Likewise.
15005 (rewrite_expr_tree_parallel): Likewise.
15006 (linearize_expr): Likewise.
15007 (break_up_subtract): Likewise.
15008 (linearize_expr_tree): Likewise.
15009 (attempt_builtin_powi): Likewise.
15010 (attempt_builtin_copysign): Likewise.
15011 (transform_stmt_to_copy): Likewise.
15012 (transform_stmt_to_multiply): Likewise.
15013 (dump_ops_vector): Likewise.
15014 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
15015 (print_scc): Likewise.
15016 (set_ssa_val_to): Likewise.
15017 (visit_reference_op_store): Likewise.
15018 (visit_use): Likewise.
15019 (sccvn_dom_walker::before_dom_children): Likewise.
15020 (run_scc_vn): Likewise.
15021 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15023 (expr_hash_elt::print): Likewise.
15024 (const_and_copies::pop_to_marker): Likewise.
15025 (const_and_copies::record_const_or_copy_raw): Likewise.
15026 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
15027 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
15028 (dump_predicates): Likewise.
15029 (find_uninit_use): Likewise.
15030 (warn_uninitialized_phi): Likewise.
15031 (pass_late_warn_uninitialized::execute): Likewise.
15032 * tree-ssa.c (verify_vssa): Likewise.
15033 (verify_ssa): Likewise.
15034 (maybe_optimize_var): Likewise.
15035 * tree-vrp.c (dump_value_range): Likewise.
15036 (dump_all_value_ranges): Likewise.
15037 (dump_asserts_for): Likewise.
15038 (register_edge_assert_for_2): Likewise.
15039 (vrp_visit_cond_stmt): Likewise.
15040 (vrp_visit_switch_stmt): Likewise.
15041 (vrp_visit_stmt): Likewise.
15042 (vrp_visit_phi_node): Likewise.
15043 (simplify_cond_using_ranges_1): Likewise.
15044 (fold_predicate_in): Likewise.
15045 (evrp_dom_walker::before_dom_children): Likewise.
15046 (evrp_dom_walker::push_value_range): Likewise.
15047 (evrp_dom_walker::pop_value_range): Likewise.
15048 (execute_early_vrp): Likewise.
15050 2017-05-16 Richard Biener <rguenther@suse.de>
15052 * dwarf2out.c (loc_list_from_tree_1): Do not create
15053 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
15055 2017-05-16 Richard Biener <rguenther@suse.de>
15057 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
15059 (note_variable_value_in_expr): If we resolved the decl ref
15060 do not push to the stack.
15062 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
15064 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
15065 operations in fast-math mode.
15066 (vaddq_f16): Likewise.
15067 (vmul_f16): Likewise.
15068 (vmulq_f16): Likewise.
15069 (vsub_f16): Likewise.
15070 (vsubq_f16): Likewise.
15071 * config/arm/neon.md (add<mode>3): New.
15073 (fma:<VH:mode>3): New. Also remove outdated comment.
15076 2017-05-16 Martin Liska <mliska@suse.cz>
15080 * ipa-devirt.c (warn_types_mismatch): Fix typo.
15081 (odr_types_equivalent_p): Likewise.
15083 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
15085 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
15087 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
15090 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
15091 non-interunit SSE move alternatives with '?'.
15092 (zero-extendsidi peephole2): New peephole to skip intermediate
15093 general register in SSE zero-extend sequence.
15095 2017-05-15 Jeff Law <law@redhat.com>
15097 * reorg.c (relax_delay_slots): Create a new variable to hold
15098 the temporary target rather than clobbering TARGET_LABEL.
15100 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
15101 missing argument to extract_bit_field call.
15102 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
15104 2017-05-15 Martin Liska <mliska@suse.cz>
15107 * gcc.c (process_command): Do not allow empty argument of -o option.
15109 2017-05-15 Renlin Li <renlin.li@arm.com>
15111 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
15112 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
15113 * config/aarch64/constraints.md (Usf): Add long call check.
15114 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
15115 (call_value): Likewise.
15116 (sibcall): Likewise.
15117 (sibcall_value): Likewise.
15119 (call_value_insn): New.
15120 (sibcall_insn): Update rtx pattern.
15121 (sibcall_value_insn): Likewise.
15122 (call_internal): Remove.
15123 (call_value_internal): Likewise.
15124 (sibcall_internal): Likewise.
15125 (sibcall_value_internal): Likewise.
15126 (call_reg): Likewise.
15127 (call_symbol): Likewise.
15128 (call_value_reg): Likewise.
15129 (call_value_symbol): Likewise.
15131 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
15134 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
15136 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
15138 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
15139 compatible with CCGOCmode and with CCZmode.
15141 2017-05-14 Martin Sebor <msebor@redhat.com>
15143 PR middle-end/77671
15144 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
15145 (gimple_fold_builtin_snprintf): Same.
15146 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
15147 (gimple_fold_builtin_snprintf): Same.
15148 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
15149 of character types.
15150 (is_call_safe): New function.
15151 (try_substitute_return_value): Call it.
15152 (try_simplify_call): New function.
15153 (pass_sprintf_length::handle_gimple_call): Call it.
15155 2017-05-14 Martin Sebor <msebor@redhat.com>
15157 PR middle-end/80669
15158 * builtins.c (expand_builtin_stpncpy): Simplify.
15160 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
15162 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
15163 * config/i386/i386.h
15164 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
15165 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
15166 (struct machine_function): Add new members call_ms2sysv,
15167 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
15168 (struct machine_frame_state): New fields sp_realigned and
15169 sp_realigned_offset.
15170 * config/i386/i386.c
15171 (enum xlogue_stub): New enum.
15172 (enum xlogue_stub_sets): New enum.
15173 (class xlogue_layout): New class.
15174 (struct ix86_frame): New fields stack_realign_allocate_offset,
15175 stack_realign_offset and outlined_save_offset. Modify comments to
15176 detail stack layout when using out-of-line stubs.
15177 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
15178 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
15179 -mcall-ms2sysv-xlogues.
15180 (stub_managed_regs): New static variable.
15181 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
15182 registers managed by out-of-line stub.
15183 (disable_call_ms2sysv_xlogues): New function.
15184 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
15185 m->call_ms2sysv when appropriate and compute frame layout for
15187 (sp_valid_at, fp_valid_at): New inline functions.
15188 (choose_basereg): New function.
15189 (choose_baseaddr): Add align parameter, use choose_basereg and modify
15191 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
15192 Use align parameter of choose_baseaddr to generated aligned SSE movs
15194 (pro_epilogue_adjust_stack): Modify to track
15195 machine_frame_state::sp_realigned.
15196 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
15197 (ix86_nsaved_sseregs): Likewise.
15198 (ix86_emit_save_regs): Likewise.
15199 (ix86_emit_save_regs_using_mov): Likewise.
15200 (ix86_emit_save_sse_regs_using_mov): Likewise.
15201 (get_scratch_register_on_entry): Likewise.
15202 (gen_frame_set): New function.
15203 (gen_frame_load): Likewise.
15204 (gen_frame_store): Likewise.
15205 (emit_outlined_ms2sysv_save): Likewise.
15206 (emit_outlined_ms2sysv_restore): Likewise.
15207 (ix86_expand_prologue): Modify stack re-alignment code and call
15208 emit_outlined_ms2sysv_save when appropriate.
15209 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
15210 parameter rtx_insn *insn, which allows the function to be used to only
15211 generate the notes.
15212 (ix86_expand_epilogue): Modify validity checks of frame and stack
15213 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
15214 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
15215 * config/i386/predicates.md
15216 (save_multiple): New predicate.
15217 (restore_multiple): Likewise.
15218 * config/i386/sse.md
15219 (save_multiple<mode>): New pattern.
15220 (save_multiple_realign<mode>): Likewise.
15221 (restore_multiple<mode>): Likewise.
15222 (restore_multiple_and_return<mode>): Likewise.
15223 (restore_multiple_leave_return<mode>): Likewise.
15224 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
15226 2017-05-14 Julia Koval <julia.koval@intel.com>
15228 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
15229 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
15230 (__builtin_ia32_xsetbv): New builtins.
15231 * config/i386/i386.c (ix86_expand_special_args_builtin):
15233 (ix86_expand_builtin): Special expand for new intrinsics.
15234 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
15235 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
15236 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
15238 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15240 * cfganal.c (inverted_post_order_compute): Change argument type
15242 * cfganal.h (inverted_post_order_compute): Adjust prototype.
15243 * df-core.c (rest_of_handle_df_initialize): Adjust.
15244 (rest_of_handle_df_finish): Likewise.
15245 (df_analyze_1): Likewise.
15246 (df_analyze): Likewise.
15247 (loop_inverted_post_order_compute): Change argument to be a vec *.
15248 (df_analyze_loop): Adjust.
15249 (df_get_n_blocks): Likewise.
15250 (df_get_postorder): Likewise.
15251 * df.h (struct df_d): Change field to be a vec.
15252 * lcm.c (compute_laterin): Adjust.
15253 (compute_available): Likewise.
15254 * lra-lives.c (lra_create_live_ranges_1): Likewise.
15255 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
15256 * tree-ssa-pre.c (compute_antic): Likewise.
15258 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15260 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
15261 (depth_first_search::depth_first_search): Change structure init
15262 function to this constructor.
15263 (depth_first_search::add_bb): Rename function to this member.
15264 (depth_first_search::execute): Likewise.
15265 (flow_dfs_compute_reverse_finish): Adjust.
15267 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15269 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
15270 (longest_simple_path): Likewise.
15271 * shrink-wrap.c (spread_components): Likewise.
15272 (disqualify_problematic_components): Likewise.
15273 (emit_common_heads_for_components): Likewise.
15274 (emit_common_tails_for_components): Likewise.
15275 (insert_prologue_epilogue_for_components): Likewise.
15277 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15279 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
15282 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15284 * df-core.c (df_set_blocks): Start using auto_bitmap.
15285 (df_compact_blocks): Likewise.
15286 * df-problems.c (df_rd_confluence_n): Likewise.
15287 * df-scan.c (df_insn_rescan_all): Likewise.
15288 (df_process_deferred_rescans): Likewise.
15289 (df_update_entry_block_defs): Likewise.
15290 (df_update_exit_block_uses): Likewise.
15291 (df_entry_block_bitmap_verify): Likewise.
15292 (df_exit_block_bitmap_verify): Likewise.
15293 (df_scan_verify): Likewise.
15294 * lra-constraints.c (lra_constraints): Likewise.
15295 (undo_optional_reloads): Likewise.
15296 (lra_undo_inheritance): Likewise.
15297 * lra-remat.c (calculate_gen_cands): Likewise.
15298 (do_remat): Likewise.
15299 * lra-spills.c (assign_spill_hard_regs): Likewise.
15300 (spill_pseudos): Likewise.
15301 * tree-ssa-pre.c (bitmap_set_and): Likewise.
15302 (bitmap_set_subtract_values): Likewise.
15304 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15306 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
15307 management with auto_bitmap.
15308 (fix_inter_tick): Likewise.
15309 (fix_recovery_deps): Likewise.
15310 * ira.c (add_store_equivs): Likewise.
15311 (find_moveable_pseudos): Likewise.
15312 (split_live_ranges_for_shrink_wrap): Likewise.
15313 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
15314 (rtx_reuse_manager::seen_def_p): Likewise.
15315 (rtx_reuse_manager::set_seen_def): Likewise.
15316 * print-rtl.h (class rtx_reuse_manager): Likewise.
15318 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15320 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
15322 (migrate_btr_def): Likewise.
15323 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
15324 * df-core.c (loop_post_order_compute): Likewise.
15325 (loop_inverted_post_order_compute): Likewise.
15326 * hsa-common.h: Likewise.
15327 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
15328 * init-regs.c (initialize_uninitialized_regs): Likewise.
15329 * ipa-inline.c (resolve_noninline_speculation): Likewise.
15330 (inline_small_functions): Likewise.
15331 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
15332 * ira.c (combine_and_move_insns): Likewise.
15333 (build_insn_chain): Likewise.
15334 * loop-invariant.c (find_invariants): Likewise.
15335 * lower-subreg.c (propagate_pseudo_copies): Likewise.
15336 * predict.c (tree_predict_by_opcode): Likewise.
15337 (predict_paths_leading_to): Likewise.
15338 (predict_paths_leading_to_edge): Likewise.
15339 (estimate_loops_at_level): Likewise.
15340 (estimate_loops): Likewise.
15341 * shrink-wrap.c (try_shrink_wrapping): Likewise.
15342 (spread_components): Likewise.
15343 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
15344 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
15345 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
15346 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
15347 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
15348 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
15349 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
15350 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
15351 (mark_threaded_blocks): Likewise.
15352 (thread_through_all_blocks): Likewise.
15353 * tree-ssa.c (verify_ssa): Likewise.
15354 (execute_update_addresses_taken): Likewise.
15355 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
15357 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15359 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
15361 (post_order_compute): Likewise.
15362 (inverted_post_order_compute): Likewise.
15363 (pre_and_rev_post_order_compute_fn): Likewise.
15365 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15367 * genrecog.c (int_set::int_set): Explicitly construct our
15368 auto_vec base class.
15369 * vec.h (auto_vec::auto_vec): New constructor.
15371 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15373 * bitmap.h (class auto_bitmap): New constructor taking
15374 bitmap_obstack * argument.
15376 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15378 * bitmap.h (class auto_bitmap): Change type of m_bits to
15379 bitmap_head, and adjust ctor / dtor and member operators.
15381 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
15383 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
15384 when returned register mode doesn't match original mode.
15386 2017-05-12 Jeff Law <law@redhat.com>
15387 Jakub Jelinek <jakub@redhat.com>
15389 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
15390 we look for cc setter after the compare-elim changes.
15391 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
15392 within the vector to match what compare-elim now expects.
15393 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
15394 (xorsi3_flags, one_cmplsi2_flags): Likewise.
15396 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
15397 after the compare-elim changes.
15398 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
15399 the vector to match what compare-elim now expects.
15400 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
15401 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
15402 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
15403 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
15404 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
15406 * config/visium/visium.c (single_set_and_flags): Fix where
15407 we look for cc setter after the compare-elim changes.
15408 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
15409 with the vector to match what compare-elim now expects.
15410 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
15411 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
15412 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
15413 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
15414 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
15415 (neg<mode>2_insn_set_overflow): Likewise.
15417 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
15419 PR middle-end/79794
15420 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
15421 maybe_expand_insn call, set ops[0].target. If still set after call,
15422 set alt_rtl. Add extra arg to recursive calls.
15423 (extract_bit_field): Add alt_rtl argument. Pass to
15425 * expmed.h (extract_bit_field): Fix prototype.
15426 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
15427 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
15428 to extract_bit_field_calls.
15429 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
15430 Pass alt_rtl to extract_bit_field calls.
15431 * calls.c (store_unaligned_arguments_into_psuedos)
15432 load_register_parameters): Pass extra NULL to extract_bit_field calls.
15433 * optabs.c (maybe_legitimize_operand): Clear op->target when call
15435 * optabs.h (struct expand_operand): Add target bitfield.
15437 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15439 * compare-elim.c (try_eliminate_compare): Canonicalize
15440 operation with embedded compare to
15441 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
15442 (set (reg) (operation)].
15444 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
15446 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
15449 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
15450 cost of adding a carry flag for ADC instruction.
15451 [case MINUS]: Ignore the cost of subtracting a carry flag
15452 for SBB instruction.
15454 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
15456 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
15458 * config/rs6000/bmiintrin.h: New file.
15459 * config/rs6000/bmi2intrin.h: New file.
15460 * config/rs6000/x86intrin.h: New file.
15462 2017-05-12 Jeff Law <law@redhat.com>
15464 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
15467 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
15469 PR middle-end/80707
15470 * tree-cfg.c: Remove cfg edges of unreachable case statements.
15472 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15474 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15475 early expansion of vector divide builtins.
15476 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
15477 builtins identified as having unsigned arguments.
15479 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15481 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
15482 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
15483 expansion of vector logical operations (and, andc, or, xor,
15486 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15488 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
15489 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
15491 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15493 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
15494 early GIMPLE expansion of vector multiplies.
15496 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15498 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
15499 TARGET_HAVE_MOVT conditional.
15500 (movt splitter): Likewise.
15502 2017-05-12 Richard Biener <rguenther@suse.de>
15504 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
15505 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15506 Fold all stmts not inplace.
15508 2017-05-12 Richard Biener <rguenther@suse.de>
15510 PR tree-optimization/80713
15511 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
15512 inserted_exprs bit for not removed stmts.
15514 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
15516 PR middle-end/69921
15517 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
15518 parallelized" attribute for parallelized OpenACC kernels.
15519 * omp-offload.c (execute_oacc_device_lower): Use it.
15521 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
15522 Set "oacc kernels" attribute.
15523 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
15524 parameter. Adjust all users.
15525 (oacc_fn_attrib_kernels_p): Remove function.
15526 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
15527 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
15528 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
15529 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
15530 assert "oacc kernels" attribute is set.
15532 2017-05-11 Carl Love <cel@us.ibm.com>
15534 * config/rs6000/rs6000-c: Add support for built-in functions
15535 vector unsigned char vec_popcnt (vector signed char)
15536 vector unsigned char vec_popcnt (vector unsigned char)
15537 vector unsigned short vec_popcnt (vector signed short)
15538 vector unsigned short vec_popcnt (vector unsigned short)
15539 vector unsigned int vec_popcnt (vector signed int)
15540 vector unsigned int vec_popcnt (vector unsigned int)
15541 vector unsigned long long vec_popcnt (vector signed long long)
15542 vector unsigned long long vec_popcnt (vector unsigned long long)
15543 vector signed long long vec_slo (vector signed long long,
15544 vector signed char)
15545 vector signed long long vec_slo (vector signed long long,
15546 vector unsigned char)
15547 vector unsigned long long vec_slo (vector unsigned long long,
15548 vector signed char)
15549 vector unsigned long long vec_slo (vector unsigned long long,
15550 vector unsigned char)
15551 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
15552 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
15553 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
15554 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
15555 * doc/extend.texi: Update the built-in documentation file for the
15556 new built-in functions.
15558 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
15560 * attribs.h (sorted_attr_string): Move machine independent
15561 functions for target clone support from the i386 port to common
15562 code. Rename ix86_function_versions to common_function_versions.
15563 Rename make_name to make_unique_name.
15564 (common_function_versions): Likewise.
15565 (make_unique_name): Likewise.
15566 (make_dispatcher_decl): Likewise.
15567 (is_function_default_version): Likewise.
15568 * attribs.c (attr_strcmp): Likewise.
15569 (sorted_attr_string): Likewise.
15570 (common_function_versions): Likewise.
15571 (make_unique_name): Likewise.
15572 (make_dispatcher_decl): Likewise.
15573 (is_function_default_version): Likewise.
15574 * config/i386/i386.c (attr_strcmp): Likewise.
15575 (sorted_attr_string): Likewise.
15576 (ix86_function_versions): Likewise.
15577 (make_name): Likewise.
15578 (make_dispatcher_decl): Likewise.
15579 (is_function_default_version): Likewise.
15580 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
15582 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15585 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15586 Account for direct move costs for vec_construct of integer
15589 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
15592 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
15593 (UNSPEC_STX_ATOMIC): Ditto.
15594 (loaddi_via_sse): New insn.
15595 (storedi_via_sse): Ditto.
15596 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
15597 Update corresponding peephole2 patterns.
15598 (atomic_storedi_fpu): Ditto.
15600 2017-05-11 Julia Koval <julia.koval@intel.com>
15602 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
15603 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
15605 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
15606 (__builtin_ia32_rsqrt14ss_mask): New builtins.
15607 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
15609 2017-05-11 Nathan Sidwell <nathan@acm.org>
15611 * graphite-poly.c: Include dumpfile.h.
15613 * dumpfle.h (dump_function): Declare here ...
15614 * tree-dump.h (dump_function): ... not here.
15615 * dumpfile.c: #include tree-cfg.h.
15616 (dump_function): Move here from ...
15617 * tree-dump.c (dump_function): ... here.
15618 * gimplify.c: #include splay-tree.h, not tree-dump.h.
15619 * graphite-poly.c: Don't include tree-dump.h.
15620 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
15621 * print-tree.c: Likewise.
15622 * stor-layout.c: Likewise.
15623 * tree-nested.c: Likewise.
15625 * dumpfile.c (dump_start): Use TDF_FLAGS.
15626 (dump_enable_all): Fix TDF_KIND check thinko.
15628 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
15630 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15631 array entries to represent two legal parameterizations of the
15632 overloaded __builtin_cmpb function, as represented by the
15633 P6_OV_BUILTIN_CMPB constant.
15634 (altivec_resolve_overloaded_builtin): Add special case handling
15635 for the __builtin_cmpb function, as represented by the
15636 P6_OV_BUILTIN_CMPB constant.
15637 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
15638 (BU_P6_64BIT_2): New macro.
15639 (BU_P6_OVERLOAD_2): New macro
15640 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
15641 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
15642 (CMPB): Add overload support to represent both 32-bit and 64-bit
15643 compare-bytes function.
15644 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15645 support for TARGET_CMPB.
15646 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
15647 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15648 documentation of the __builtin_cmpb overloaded built-in function.
15650 2017-05-11 Richard Biener <rguenther@suse.de>
15652 PR tree-optimization/80705
15653 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
15654 bases are not vectorizable.
15656 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15658 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
15659 when counting register pressure.
15661 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15663 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
15664 (iv_ca_more_deps): Renamed to ...
15665 (iv_ca_compare_deps): ... this.
15666 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
15668 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15670 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
15672 (determine_group_iv_costs): ... here.
15673 (find_inv_vars_cb): Record inv var if it's not recorded before.
15675 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15677 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
15678 (get_shiftadd_cost): Ditto.
15680 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15682 * tree-ssa-address.c: Include header file.
15683 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
15685 (add_to_parts): Refactor.
15686 (addr_to_parts): New parameter. Update use of move_hint_to_base.
15687 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
15690 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15692 PR tree-optimization/53090
15693 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
15695 (extract_cond_operands): Detect condition with IV on both sides
15696 and return COMP_IV_EXPR_2.
15697 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
15698 (rewrite_use_compare): Simplify by removing call to function
15699 extract_cond_operands.
15701 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15703 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
15704 (extract_cond_operands): Detect condition comparing against non-
15705 invariant bound and return appropriate enum value.
15706 (find_interesting_uses_cond): Update use of extract_cond_operands.
15707 Handle its return value accordingly.
15708 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
15710 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15712 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
15713 nonlinear iv_use computation in loop invariant sensitive way.
15715 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15717 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
15718 (find_iv_candidates): Call relate_compare_use_with_all_cands.
15720 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15722 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
15723 (dump_cand): Support iv_cand.inv_exprs.
15724 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
15726 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
15729 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15731 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
15733 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
15734 as local function. Include necessary header files.
15735 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
15737 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15739 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
15741 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15743 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
15744 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
15745 RSHIFT_EXPR and BIT_NOT_EXPR.
15747 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15749 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
15750 (adjust_setup_cost): New parameter supporting round up adjustment.
15751 (struct address_cost_data): Delete.
15752 (force_expr_to_var_cost): Don't bound cost with spill_cost.
15753 (split_address_cost, ptr_difference_cost): Delete.
15754 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
15755 (struct ainc_cost_data): New struct.
15756 (get_address_cost_ainc): New function.
15757 (get_address_cost, get_computation_cost): Reimplement.
15758 (determine_group_iv_cost_address): Record inv_expr for all uses of
15760 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
15761 (iv_ca_has_deps): Reimplemented to ...
15762 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
15764 (iv_ca_extend): Call iv_ca_more_deps.
15766 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15768 * tree-ssa-address.c (struct mem_address): Move to header file.
15769 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
15770 * tree-ssa-address.h (struct mem_address): Move from C file.
15771 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
15773 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15775 * tree-affine.h (aff_combination_type): New interface.
15776 (aff_combination_zero_p): Remove static.
15777 (aff_combination_const_p): New interface.
15778 (aff_combination_singleton_var_p): New interfaces.
15780 2017-05-11 Richard Biener <rguenther@suse.de>
15782 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15783 Skip unreachable blocks and destinations.
15784 (eliminate): Move stmt removal and fixup ...
15785 (fini_eliminate): ... here. Skip inserted exprs.
15786 (pass_pre::execute): Move fini_pre after fini_eliminate.
15787 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
15788 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
15789 PRE to get rid of dead code that has invalid SSA form and
15790 split critical edges again.
15792 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15794 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15796 2017-05-11 Richard Biener <rguenther@suse.de>
15798 * passes.c (execute_function_todo): Verify loops if they are
15799 said to be up-to-date.
15800 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
15801 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
15803 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
15806 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
15807 handle calling assemble_external ourself.
15810 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
15811 modes with zero size. Enhance comment.
15813 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15815 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
15816 built-ins for vec_xl and vec_xst with short and char pointer
15819 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15821 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
15822 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
15823 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
15824 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
15825 (_mm_maskz_min_round_ss): New intrinsics.
15826 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
15827 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
15828 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
15829 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
15830 (__builtin_ia32_minss_mask_round): New builtins.
15831 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15832 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15833 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
15835 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
15836 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
15838 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
15841 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15843 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
15844 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
15845 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
15846 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
15847 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
15848 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
15849 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
15850 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15851 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
15852 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
15853 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
15854 (__builtin_ia32_mulss_mask_round): New builtins.
15855 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15856 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15857 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
15859 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
15860 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
15862 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
15865 2017-05-10 Julia Koval <julia.koval@intel.com>
15867 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
15868 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
15869 (_mm256_setr_m128i): New intrinsics.
15871 2017-05-10 Julia Koval <julia.koval@intel.com>
15873 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
15874 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
15875 (_mm_maskz_rcp14_ss): New intrinsics.
15876 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
15877 (__builtin_ia32_rcp14ss_mask): New builtins.
15878 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
15880 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
15882 PR tree-optimization/51513
15883 * tree-cfg.c (gimple_seq_unreachable_p): New function.
15884 (assert_unreachable_fallthru_edge_p): Use it.
15885 (group_case_labels_stmt): Likewise.
15886 * tree-cfg.h: Prototype it.
15887 * stmt.c: Include cfghooks.h and tree-cfg.h.
15888 (emit_case_dispatch_table) <gap_label>: New local variable.
15889 Use it to fill dispatch table gaps.
15890 Test for default_label before updating probabilities.
15891 (expand_case) <default_label>: Remove unneeded initialization.
15892 Test for unreachable default case statement and remove its edge.
15893 Set default_label accordingly.
15894 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
15896 2017-05-10 Carl Love <cel@us.ibm.com>
15898 * config/rs6000/rs6000-c: Add support for built-in functions
15899 vector signed char vec_neg (vector signed char)
15900 vector signed short int vec_neg (vector short int)
15901 vector signed int vec_neg (vector signed int)
15902 vector signed long long vec_neg (vector signed long long)
15903 vector float vec_neg (vector float)
15904 vector double vec_neg (vector double)
15905 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
15907 * config/rs6000/altivec.h: Add define for vec_neg
15908 * doc/extend.texi: Update the built-in documentation for the
15909 new built-in functions.
15911 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15913 PR tree-optimization/77644
15914 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
15916 2017-05-10 Nathan Sidwell <nathan@acm.org>
15918 * dumpfile.h (TDI_lang_all): New.
15919 (TDF_KIND): New. Renumber others
15920 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
15922 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
15924 (get_dump_file_name): Adjust suffix generation.
15925 (dump_enable_all): Use TDF_KIND.
15926 * doc/invoke.texi (-fdump-lang-all): Document.
15928 * dumpfile.h: Tabify.
15930 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
15933 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
15934 Move member access before delete.
15936 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
15938 * tree-inline.c (expand_call_inline): Split block at stmt
15941 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
15944 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
15945 are now unused after splitting mov{sf,sd}_hardfloat.
15946 (f32_lr2): Likewise.
15947 (f32_lm): Likewise.
15948 (f32_lm2): Likewise.
15949 (f32_li): Likewise.
15950 (f32_li2): Likewise.
15951 (f32_lv): Likewise.
15952 (f32_sr): Likewise.
15953 (f32_sr2): Likewise.
15954 (f32_sm): Likewise.
15955 (f32_sm2): Likewise.
15956 (f32_si): Likewise.
15957 (f32_si2): Likewise.
15958 (f32_sv): Likewise.
15959 (f32_dm): Likewise.
15960 (f32_vsx): Likewise.
15961 (f32_av): Likewise.
15962 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
15963 For movsf, order stores so the VSX stores occur before the GPR
15964 store which encourages the register allocator to use a traditional
15965 FPR instead of a GPR. For movsd, order the stores so that the GPR
15966 store comes before the VSX stores to allow the power6 to work.
15967 This is due to the power6 not having a 32-bit integer store
15968 instruction from a FPR.
15969 (movsf_hardfloat): Likewise.
15970 (movsd_hardfloat): Likewise.
15972 2017-05-09 Martin Sebor <msebor@redhat.com>
15974 PR translation/80280
15975 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
15978 PR translation/80280
15979 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
15980 data member added in r247778.
15981 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
15983 2017-05-09 Nathan Sidwell <nathan@acm.org>
15985 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
15987 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
15990 2017-05-09 Marek Polacek <polacek@redhat.com>
15992 * doc/invoke.texi: Fix typo.
15994 2017-05-09 Richard Biener <rguenther@suse.de>
15996 * tree-vrp.c (vrp_val_is_max): Adjust comment.
15997 (vrp_val_is_min): Likewise.
15998 (set_value_range_to_value): Likewise.
15999 (set_value_range_to_nonnegative): Likewise.
16000 (gimple_assign_nonzero_p): Likewise.
16001 (gimple_stmt_nonzero_p): Likewise.
16002 (vrp_int_const_binop): Likewise. Remove unreachable case.
16003 (adjust_range_with_scev): Adjust comments.
16004 (compare_range_with_value): Likewise.
16005 (extract_range_from_phi_node): Likewise.
16006 (test_for_singularity): Likewise.
16008 2017-05-09 Richard Biener <rguenther@suse.de>
16010 * tree-vrp.c (get_single_symbol): Add assert that we don't
16011 get overflowed constants as invariant part.
16012 (compare_values_warnv): Add comment before the TREE_NO_WARNING
16013 checks. Use wi::cmp instead of recursing for integer constants.
16014 (compare_values): Just ignore whether we assumed undefined
16015 overflow instead of failing the compare.
16016 (extract_range_for_var_from_comparison_expr): Add comment before the
16017 TREE_NO_WARNING sets.
16018 (test_for_singularity): Likewise.
16019 (extract_range_from_comparison): Do not disable optimization
16020 when we assumed undefined overflow.
16021 (extract_range_basic): Remove init of unused var.
16023 2017-05-09 Richard Biener <rguenther@suse.de>
16025 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
16026 (extract_range_from_multiplicative_op_1): Adjust.
16027 (extract_range_from_binary_expr_1): Use int_const_binop.
16029 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
16032 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
16033 rs6000_store_data_bypass_p in seven define_bypass directives and
16034 in several comments.
16035 * config/rs6000/rs6000-protos.h: Add prototype for
16036 rs6000_store_data_bypass_p function.
16037 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
16038 function implements slightly different (rs6000-specific) semantics
16039 than store_data_bypass_p, returning false rather than aborting
16040 with assertion error when arguments do not satisfy the
16041 requirements of store data bypass.
16042 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
16043 rs6000_store_data_bypass_p.
16045 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
16047 * config/xtensa/xtensa-protos.h
16048 (xtensa_initial_elimination_offset): New declaration.
16049 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
16050 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
16051 macro definition, add case for FRAME_POINTER_REGNUM when
16052 FRAME_GROWS_DOWNWARD.
16053 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
16054 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
16055 xtensa_initial_elimination_offset.
16057 2017-05-08 Nathan Sidwell <nathan@acm.org>
16059 * doc/invoke.texi: Alphabetize -fdump options.
16061 2017-05-08 Martin Sebor <msebor@redhat.com>
16063 PR translation/80280
16064 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
16066 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
16068 * target.def (compute_frame_layout): New optional target hook.
16069 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
16070 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
16071 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
16073 * reload1.c (verify_initial_elim_offsets): Likewise.
16074 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
16075 (use_simple_return_p): Call arm_compute_frame_layout if needed.
16076 (arm_get_frame_offsets): Split up into this ...
16077 (arm_compute_frame_layout): ... and this function.
16079 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
16081 * config/aarch64/constraints.md (Usa): New constraint.
16082 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
16084 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
16086 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
16087 with_multilib_list after it has been checked.
16089 2017-05-08 Richard Biener <rguenther@suse.de>
16091 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
16092 (bitmap_set_subtract_values): Likewise.
16094 2017-05-08 Richard Biener <rguenther@suse.de>
16096 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
16097 (gimple_assign_nonzero): ... this and remove strict_overflow_p
16099 (gimple_stmt_nonzero_warnv_p): Rename to ...
16100 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
16102 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
16103 (extract_range_basic): Adjust, do not disable propagation on
16104 strict overflow sensitive simplification.
16105 (vrp_visit_cond_stmt): Likewise.
16107 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
16109 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
16110 body size unconditionally.
16112 2017-05-07 Jeff Law <law@redhat.com>
16115 2017-05-06 Jeff Law <law@redhat.com>
16116 PR tree-optimization/78496
16117 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16120 PR tree-optimization/78496
16121 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16122 (simplify_stmt_using_ranges): Call it.
16123 (vrp_dom_walker::before_dom_children): Extract equivalences
16124 from an ASSERT_EXPR with an equality comparison against a
16127 2017-05-06 Jeff Law <law@redhat.com>
16129 PR tree-optimization/78496
16130 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
16133 PR tree-optimization/78496
16134 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
16135 (simplify_stmt_using_ranges): Call it.
16136 (vrp_dom_walker::before_dom_children): Extract equivalences
16137 from an ASSERT_EXPR with an equality comparison against a
16140 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16142 * lra-constraints.c (lra_copy_reg_equiv): New function.
16143 (split_reg): Use it to copy equivalence information from the
16144 original register to the spill register.
16146 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
16148 PR rtl-optimization/75964
16149 * simplify-rtx.c (simplify_const_relational_operation): Remove
16150 invalid handling of comparisons of integer ABS.
16152 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
16154 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
16155 initialize to zero.
16156 (init_regs): Remove declaration.
16157 (function_arg_advance_32): Initialize error_p as boolean variable.
16159 2017-05-05 Nathan Sidwell <nathan@acm.org>
16161 * store-motion.c (remove_reachable_equiv_notes): Reformat long
16162 lines. Use for (;;).
16164 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16166 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
16167 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
16168 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
16169 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
16170 VF=2 that require versioning.
16172 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16174 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
16177 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16179 * diagnostic.h (diagnostic_override_option_index): Convert from
16180 macro to inline function.
16182 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16184 * diagnostic.c (last_module_changed_p): New function.
16185 (set_last_module): New function.
16186 (diagnostic_report_current_module): Convert macro usage to
16187 the above functions.
16188 * diagnostic.h (diagnostic_context::last_module): Strengthen
16189 from const line_map * to const line_map_ordinary *.
16190 (diagnostic_last_module_changed): Delete macro.
16191 (diagnostic_set_last_module): Delete macro.
16193 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16195 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
16196 with diagnostic_report_diagnostic.
16197 (diagnostic_n_impl_richloc): Likewise.
16198 * diagnostic.h (report_diagnostic): Delete macro.
16199 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
16200 with diagnostic_report_diagnostic.
16201 * substring-locations.c (format_warning_va): Likewise.
16203 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16205 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
16206 save/restor of format_spec. Move option-printing code to...
16207 (print_option_information): ...this new function, and
16208 reimplement by simply printing to the pretty_printer,
16209 rather than appending to the format string.
16211 2017-05-05 David Malcolm <dmalcolm@redhat.com>
16213 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
16214 handling logic into...
16215 (update_effective_level_from_pragmas): ...this new function.
16217 2017-05-04 Andrew Waterman <andrew@sifive.com>
16219 * config/riscv/riscv.opt (mstrict-align): New option.
16220 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
16221 (SLOW_UNALIGNED_ACCESS): Define.
16222 (riscv_slow_unaligned_access): Declare.
16223 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
16225 (riscv_slow_unaligned_access): New variable.
16226 (rocket_tune_info): Set slow_unaligned_access to true.
16227 (optimize_size_tune_info): Set slow_unaligned_access to false.
16228 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
16229 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
16230 (riscv_option_override): Set riscv_slow_unaligned_access.
16231 * doc/invoke.texi: Add -mstrict-align to RISC-V.
16233 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
16235 * config/riscv/riscv.md: Unify indentation.
16237 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16242 * config/rs6000/rs6000.md (u code attribute): Add FIX and
16244 (extendsi<mode>2): Add support for doing sign extension via
16245 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
16246 don't have ISA 3.0 instructions.
16247 (extendsi<mode>2 splitter): Likewise.
16248 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
16249 generate the normal insns since SImode can now go in vector
16250 registers. Disallow the special UNSPECs needed for previous
16251 machines to hide SImode being used. Add new insns
16252 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
16253 (fix_trunc<mode>si2_stfiwx): Likewise.
16254 (fix_trunc<mode>si2_internal): Likewise.
16255 (fixuns_trunc<mode>si2): Likewise.
16256 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16257 (fctiw<u>z_<mode>_smallint): Likewise.
16258 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
16259 of floating point to 32-bit integer from doing a direct move to
16260 the GPR registers to do a store.
16261 (fctiwz_<mode>): Break long line.
16263 2017-05-05 Bin Cheng <bin.cheng@arm.com>
16265 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
16266 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
16267 (addr_list, addr_offset_valid_p): New.
16268 (split_address_groups): Check offset validity with above function.
16269 (gt-tree-ssa-loop-ivopts.h): Include header file.
16271 2017-05-05 Nathan Sidwell <nathan@acm.org>
16273 * config.gcc (arm*-*-*): Add missing 'fi'.
16275 2017-05-05 Steve Ellcey <sellcey@cavium.com>
16277 * doc/invoke.texi (-fopt-info): Explicitly say order of options
16278 included in -fopt-info does not matter.
16279 * doc/optinfo.texi (-fopt-info): Fix description of default
16280 behavour. Explicitly say order of options included in -fopt-info
16283 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16285 * config.gcc: Allow combinations of aprofile and rmprofile values for
16286 --with-multilib-list.
16287 * config/arm/t-multilib: New file.
16288 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
16289 variables. Remove setting of ISA and floating-point ABI in
16290 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
16291 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
16292 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
16293 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
16295 * config/arm/t-rmprofile: Likewise except for the matches changes.
16296 * doc/install.texi (--with-multilib-list): Document the combination of
16297 aprofile and rmprofile values and warn about pitfalls in doing that.
16299 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16301 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
16302 (movdi_aarch64): Likewise.
16304 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16306 PR tree-optimization/80632
16307 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
16309 (build_arrays): Initialize it for virtual phis.
16310 (fix_phi_nodes): Use it for virtual phis.
16312 PR tree-optimization/80558
16313 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
16314 [x, y] op z into [x op, y op z] for op & or | if conditions
16317 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
16318 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16321 * config/arm/arm.md (use_literal_pool): Remove.
16322 (64-bit immediate split): No longer takes cost into consideration
16323 if arm_disable_literal_pool is enabled.
16324 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
16325 used when arm_disable_literal_pool is enabled.
16326 (arm_max_const_double_inline_cost): Remove use of
16327 arm_disable_literal_pool.
16328 (push_minipool_fix): Add assert.
16329 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
16330 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
16331 (no_literal_pool_sf_immediate): New.
16333 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16335 PR tree-optimization/80613
16336 * tree-ssa-dce.c (propagate_necessity): Remove cases for
16337 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
16339 2017-05-05 Richard Biener <rguenther@suse.de>
16341 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
16343 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
16345 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
16346 of this flag from insn conditions due to removal from r247495.
16348 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
16350 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
16352 (arm_early_store_addr_dep_ptr): Likewise.
16353 * config/arm/aarch-common-protos.h
16354 (arm_early_load_addr_dep_ptr): Add prototype.
16355 (arm_early_store_addr_dep_ptr): Likewise.
16356 * config/arm/cortex-a53.md: Add new bypasses.
16358 2017-05-05 Jakub Jelinek <jakub@redhat.com>
16360 * tree.c (next_type_uid): Change type to unsigned.
16361 (type_hash_canon): Decrement back next_type_uid if
16362 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
16363 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
16366 2017-05-04 Martin Sebor <msebor@redhat.com>
16368 * builtins.c: Fix a trivial typo in a comment.
16370 PR middle-end/79234
16371 * builtins.c (check_sizes): Adjust to handle reading past the end.
16372 Avoid printing excessive upper bound of ranges. Use %E to print
16373 tree nodes instead of converting them to %wu.
16374 (expand_builtin_memchr): New function.
16375 (compute_dest_size): Rename...
16376 (compute_objsize): ...to this.
16377 (expand_builtin_memcpy): Adjust.
16378 (expand_builtin_mempcpy): Adjust.
16379 (expand_builtin_strcat): Adjust.
16380 (expand_builtin_strcpy): Adjust.
16381 (check_strncat_sizes): Adjust.
16382 (expand_builtin_strncat): Adjust.
16383 (expand_builtin_strncpy): Adjust and simplify.
16384 (expand_builtin_memset): Adjust.
16385 (expand_builtin_bzero): Adjust.
16386 (expand_builtin_memcmp): Adjust.
16387 (expand_builtin): Handle memcmp.
16388 (maybe_emit_chk_warning): Check strncat just once.
16390 2017-05-04 Martin Sebor <msebor@redhat.com>
16392 PR preprocessor/79214
16393 PR middle-end/79222
16394 PR middle-end/79223
16395 * builtins.c (check_sizes): Add inlining context and issue
16396 warnings even when -Wno-system-headers is set.
16397 (check_strncat_sizes): Same.
16398 (expand_builtin_strncat): Same.
16399 (expand_builtin_memmove): New function.
16400 (expand_builtin_stpncpy): Same.
16401 (expand_builtin): Handle memmove and stpncpy.
16403 2017-05-04 Bin Cheng <bin.cheng@arm.com>
16405 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
16406 which is not used any more.
16408 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16410 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
16412 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16414 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
16415 (cortexa53_tunings): Likewise.
16416 (cortexa57_tunings): Likewise.
16417 (cortexa72_tunings): Likewise.
16418 (cortexa73_tunings): Likewise.
16420 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16422 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
16423 Set loop alignment to 8.
16425 2017-05-04 Martin Sebor <msebor@redhat.com>
16427 PR translation/80280
16428 * builtins.c (expand_builtin_object_size): Add missing quoting to
16429 %D and like directives.
16430 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
16431 (hsa_type_for_tree_type): Same.
16432 (verify_function_arguments): Same.
16433 * symtab.c (symbol_table::change_decl_assembler_name): Same.
16434 * varasm.c (get_section): Same.
16437 2017-05-04 Martin Sebor <msebor@redhat.com>
16439 PR translation/80280
16440 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
16442 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
16444 * config/aarch64/aarch64.c (generic_addrcost_table):
16445 Change HI/TI mode setting.
16447 2017-05-04 Martin Jambor <mjambor@suse.cz>
16449 PR tree-optimization/80622
16450 * tree-sra.c (comes_initialized_p): New function.
16451 (build_accesses_from_assign): Only set write lazily when
16452 comes_initialized_p is false.
16453 (analyze_access_subtree): Use comes_initialized_p.
16454 (propagate_subaccesses_across_link): Assert !comes_initialized_p
16455 instead of testing for PARM_DECL.
16457 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16459 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
16460 constraint on operand 0 to allow more general addressing modes.
16461 Adjust output template.
16462 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
16464 * config/aarch64/aarch64-protos.h
16465 (aarch64_address_valid_for_prefetch_p): Declare prototype.
16466 * config/aarch64/constraints.md (Dp): New address constraint.
16467 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
16470 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
16472 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
16473 update use of estimate_ipcp_clone_size_and_time.
16474 (estimate_local_effects): Update use of
16475 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
16476 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
16477 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
16478 Return nonspecialized time.
16480 2017-05-04 Richard Biener <rguenther@suse.de>
16482 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
16483 for the last VUSE which def dominates the PHI. Directly call
16485 (get_continuation_for_phi_1): Remove.
16487 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
16489 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
16490 to explain the use of truncating division. Cap the number of
16491 iterations to the maximum given by nb_iterations_upper_bound,
16494 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16496 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
16497 * configure: Regenerate.
16498 * config.in: Regenerate.
16499 * config/i386/driver-mingw32.c: new file.
16500 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
16501 * config.host: Link driver-mingw32.o on MinGW host.
16502 * doc/install.texi: Document new --enable-mingw-wildcard configure
16505 2017-05-04 Marek Polacek <polacek@redhat.com>
16507 PR tree-optimization/80612
16508 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
16510 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16511 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
16513 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
16514 (movt splitter): Likewise.
16515 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
16516 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
16517 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
16518 block for Thumb-1 with MOVT.
16519 (thumb2_legitimate_address_p): Move code block ...
16520 (can_avoid_literal_pool_for_label_p): ... into this new function.
16521 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
16523 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
16524 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
16525 "M-profile targets with the MOVT instruction".
16527 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16529 * config/arm/arm-builtins.c (arm_init_builtins): Rename
16530 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
16531 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
16533 2017-05-04 Martin Liska <mliska@suse.cz>
16535 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
16536 variable cond_code.
16538 2017-05-04 Richard Biener <rguenther@suse.de>
16540 * tree.c (array_at_struct_end_p): Handle arrays at struct
16541 end with flexarrays more conservatively. Refactor and treat
16542 arrays of arrays or aggregates more strict. Fix
16543 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
16544 * tree.c (array_at_struct_end_p): Adjust prototype.
16545 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
16546 * gimple-fold.c (get_range_strlen): Likewise.
16547 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
16549 2017-05-04 Richard Biener <rguenther@suse.de>
16551 PR tree-optimization/31130
16552 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
16554 (supports_overflow_infinity): Likewise.
16555 (is_negative_overflow_infinity): Likewise.
16556 (is_positive_overflow_infinity): Likewise.
16557 (is_overflow_infinity): Likewise.
16558 (stmt_overflow_infinity): Likewise.
16559 (overflow_infinity_range_p): Likewise.
16560 (usable_range_p): Remove as always returning true.
16561 (make_overflow_infinity): Remove.
16562 (negative_overflow_infinity): Likewise.
16563 (positive_overflow_infinity): Likewise.
16564 (avoid_overflow_infinity): Likewise.
16565 (set_value_range): Adjust accordingly.
16566 (set_value_range_to_nonnegative): Likewise, remove now unused
16567 overflow_infinity arg.
16568 (vrp_operand_equal_p): Adjust.
16569 (update_value_range): Likewise.
16570 (range_int_cst_singleton_p): Likewise.
16571 (operand_less_p): Likewise.
16572 (compare_values_warnv): Likewise.
16573 (extract_range_for_var_from_comparison_expr): Likewise.
16574 (vrp_int_const_binop): Likewise.
16575 (zero_nonzero_bits_from_vr): Likewise.
16576 (extract_range_from_multiplicative_op_1): Likewise.
16577 (extract_range_from_binary_expr_1): Likewise.
16578 (extract_range_from_unary_expr): Likewise.
16579 (extract_range_from_comparison): Likewise.
16580 (extract_range_basic): Likewise.
16581 (adjust_range_with_scev): Likewise.
16582 (compare_ranges): Likewise.
16583 (compare_range_with_value): Likewise.
16584 (dump_value_range): Likewise.
16585 (test_for_singularity): Likewise, remove strict_overflow_p parameter
16587 (simplify_cond_using_ranges): Adjust.
16589 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
16591 * brig-builtins.def: Added a builtin for class_f64.
16592 * builtin-types.def: Added a builtin type needed by class_f64.
16594 2017-05-03 Jason Merrill <jason@redhat.com>
16596 * timevar.def: Add TV_CONSTEXPR.
16598 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16600 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
16602 2017-05-03 Martin Jambor <mjambor@suse.cz>
16604 * ipa-prop.c (ipa_update_after_lto_read): Removed.
16605 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
16606 * ipa-cp.c (ipcp_propagate_stage): Do not call
16607 ipa_update_after_lto_read.
16608 * ipa-inline.c (ipa_inline): Likewise.
16610 2017-05-03 Martin Jambor <mjambor@suse.cz>
16612 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
16613 tag. Added a default constructor and a destructor.
16614 (ipa_edge_args_sum_t): New class;
16615 (ipa_edge_args_sum): Declare.
16616 (ipa_edge_args_vector): Remove declaration.
16617 (IPA_EDGE_REF): Use ipa_edge_args_sum.
16618 (ipa_free_edge_args_substructures): Remove declaration.
16619 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
16620 (ipa_edge_args_info_available_for_edge_p): Likewise.
16621 * ipa-prop.c (ipa_edge_args_vector): Removed.
16622 (edge_removal_hook_holder): Likewise.
16623 (edge_duplication_hook_holder): Likewise.
16624 (ipa_edge_args_sum): New variable.
16625 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
16626 ipa_edge_args_vector.
16627 (ipa_free_edge_args_substructures): Likewise.
16628 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
16629 ipa_edge_args_vector.
16630 (ipa_edge_removal_hook): Turned into method
16631 ipa_edge_args_sum_t::remove.
16632 (ipa_edge_duplication_hook): Turned into method
16633 ipa_edge_args_sum_t::duplicate.
16634 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
16635 registering edge hooks.
16636 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
16637 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
16638 ipa_edge_args_sum instead of ipa_edge_args_vector.
16639 * ipa-profile.c (ipa_profile): Likewise.
16641 2017-05-03 Martin Jambor <mjambor@suse.cz>
16643 * symbol-summary.h (function_summary): New method exists.
16644 (function_summary::symtab_removal): Deallocate through release.
16645 (call_summary): New class.
16646 (gt_ggc_mx): New overload.
16647 (gt_pch_nx): Likewise.
16648 (gt_pch_nx): Likewise.
16650 2017-05-03 Jeff Law <law@redhat.com>
16652 PR tree-optimization/78496
16653 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
16654 from simplify_cond_using_ranges. Split off code to walk
16655 backwards through casts into ...
16656 (simplify_cond_using_ranges_2): New function.
16657 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
16658 (execute_vrp): After identifying jump threads, call
16659 simplify_cond_using_ranges_2.
16661 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
16664 * ipa-inline.h (inline_summary): Add ctor.
16665 (create_ggc): Do not use ggc_cleared_alloc.
16667 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
16668 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16670 * gcc.c (handle_braces): Support escaping in switch matching
16672 * doc/invoke.texi (Spec Files): Document it.
16673 Remove superfluous @code markup in items.
16675 2017-05-03 David Malcolm <dmalcolm@redhat.com>
16677 * diagnostic-show-locus.c (struct column_range): New struct.
16678 (get_affected_columns): New function.
16679 (get_printed_columns): New function.
16680 (struct correction): New struct.
16681 (correction::ensure_capacity): New function.
16682 (correction::ensure_terminated): New function.
16683 (struct line_corrections): New struct.
16684 (line_corrections::~line_corrections): New dtor.
16685 (line_corrections::add_hint): New function.
16686 (layout::print_trailing_fixits): Reimplement in terms of the new
16688 (selftest::test_overlapped_fixit_printing): New function.
16689 (selftest::diagnostic_show_locus_c_tests): Call it.
16691 2017-05-03 Nathan Sidwell <nathan@acm.org>
16693 Canonicalize canonical type hashing
16694 * tree.h (type_hash_canon_hash): Declare.
16695 * tree.c (type_hash_list, attribute_hash_list): Move into
16696 type_hash_canon_hash.
16697 (build_type_attribute_qual_variant): Break out hash code calc into
16698 type_hash_canon_hash.
16699 (type_hash_canon_hash): New. Generic type hash computation.
16700 (build_range_type_1, build_array_type_1, build_function_type,
16701 build_method_type_directly, build_offset_type, build_complex_type,
16702 make_vector_type): Call it.
16704 2017-05-03 Richard Biener <rguenther@suse.de>
16706 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
16707 When all DRs have unknown misaligned do not always peel
16708 when there is a store but apply the same costing model as if
16709 there were only loads.
16711 2017-05-03 Richard Biener <rguenther@suse.de>
16714 PR tree-optimization/80492
16715 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16716 compare_base_decls returning dont-know properly.
16718 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
16720 * config/arm/iterators.md (CCSI): New mode iterator.
16721 (arch): New mode attribute.
16722 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
16723 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
16724 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
16725 code iterator for success result mode.
16726 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
16727 the corresponding new insn generators.
16729 2017-05-03 Bin Cheng <bin.cheng@arm.com>
16732 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16733 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16735 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
16737 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
16738 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
16739 (DDR_A): Wrap DDR argument in brackets.
16740 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
16741 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
16742 (DDR_REVERSED_P): Likewise.
16744 2017-05-03 Jakub Jelinek <jakub@redhat.com>
16746 PR tree-optimization/79472
16747 * tree-switch-conversion.c (struct switch_conv_info): Add
16748 contiguous_range and default_case_nonstandard fields.
16749 (collect_switch_conv_info): Compute contiguous_range and
16750 default_case_nonstandard fields, don't clear final_bb if
16751 contiguous_range and only the default case doesn't have the required
16753 (check_all_empty_except_final): Set default_case_nonstandard instead
16754 of failing if contiguous_range and the default case doesn't have empty
16756 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
16757 and only the default case doesn't have the required constants. Skip
16759 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
16760 if default_case_nonstandard.
16761 (build_constructors): Build constant 1 just once. Assert that default
16762 values aren't inserted in between cases if contiguous_range. Skip
16764 (build_arrays): Skip virtual phis.
16765 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
16766 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
16767 Handle virtual phis.
16768 (gen_inbound_check): Handle default_case_nonstandard case.
16769 (process_switch): Adjust check_final_bb caller. Call
16770 gather_default_values with the first non-default case instead of
16771 default case if default_case_nonstandard.
16773 2017-05-02 Nathan Sidwell <nathan@acm.org>
16775 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
16776 check. Fix formatting.
16778 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
16780 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
16781 errors when comparing specialized and unspecialized times.
16783 2017-05-02 David Malcolm <dmalcolm@redhat.com>
16785 * diagnostic-show-locus.c
16786 (layout::should_print_annotation_line_p): Make private.
16787 (layout::print_annotation_line): Make private.
16788 (layout::annotation_line_showed_range_p): Make private.
16789 (layout::show_ruler): Make private.
16790 (layout::print_source_line): Make private. Pass in line and
16791 line_width, rather than calling location_get_source_line. Drop
16793 (layout::print_leading_fixits): New method.
16794 (layout::print_any_fixits): Rename to...
16795 (layout::print_trailing_fixits): ...this, and make private.
16796 Don't print newline fixits.
16797 (diagnostic_show_locus): Move logic for printing one row into...
16798 (layout::print_line): ...this new function. Move the
16799 location_get_source_line call and error-handling from
16800 print_source_line to here. Call print_leading_fixits, and rename
16801 print_any_fixits to print_trailing_fixits.
16802 (selftest::test_fixit_insert_containing_newline): Update now that
16803 newlines are partially supported.
16804 (selftest::test_fixit_insert_containing_newline_2): New test.
16805 (selftest::test_fixit_replace_containing_newline): Update comments.
16806 (selftest::diagnostic_show_locus_c_tests): Call the new test.
16807 * edit-context.c (class added_line): New class.
16808 (class edited_line): Describe newline handling in comment.
16809 (edited_line::actually_edited_p): New method.
16810 (edited_line::print_content): Delete redundant decl.
16811 (edited_line::m_predecessors): New field.
16812 (edited_file::print_content): Call edited_line::print_content.
16813 (edited_file::print_diff): Update to support newlines.
16814 (edited_file::print_diff_hunk): Likewise.
16815 (edited_file::print_run_of_changed_lines): New function.
16816 (edited_file::print_diff_line): Convert to...
16817 (print_diff_line): ...this.
16818 (edited_file::get_effective_line_count): New function.
16819 (edited_line::edited_line): Initialize new field m_predecessors.
16820 (edited_line::~edited_line): Clean up m_predecessors.
16821 (edited_line::apply_fixit): Handle newlines.
16822 (edited_line::get_effective_line_count): New function.
16823 (edited_line::print_content): New function.
16824 (edited_line::print_diff_lines): New function.
16825 (selftest::test_applying_fixits_insert_containing_newline): New
16827 (selftest::test_applying_fixits_replace_containing_newline): New
16829 (selftest::insert_line): New function.
16830 (selftest::test_applying_fixits_multiple_lines): Add example of
16832 (selftest::edit_context_c_tests): Call the new tests.
16834 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16836 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
16837 parameter cand. Update dump information.
16838 (get_computation_cost): Update uses.
16840 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16842 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
16843 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
16844 (get_computation_at, rewrite_use_address): Update use of
16845 get_computation_aff.
16847 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16849 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
16850 (get_computation): Delete.
16851 (get_computation_cost): Implement like get_computation_cost_at.
16852 Use get_computation_at.
16853 (get_computation_cost_at): Delete.
16854 (rewrite_use_nonlinear_expr): Use get_computation_at.
16855 (rewrite_use_compare, remove_unused_ivs): Ditto.
16857 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16859 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
16861 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16863 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
16864 (ivopts_global_cost_for_size): Rename parameter and update uses.
16865 (iv_ca_recount_cost): Update uses.
16866 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
16867 candidates seperately in n_invs and n_cands.
16868 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
16870 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16872 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
16873 (find_inv_vars_cb): New.
16874 (find_depends): Renamed to ...
16875 (find_inv_vars): ... this.
16876 (add_candidate_1, force_var_cost): Call find_inv_vars.
16877 (split_address_cost, determine_group_iv_cost_cond): Ditto.
16879 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16881 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
16882 inv_vars. Add inv_exprs.
16883 (struct iv_cand): Rename depends_on to inv_vars.
16884 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
16885 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
16886 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
16887 (dump_cand): Dump inv_vars.
16888 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
16889 (record_invariant, find_depends, add_candidate_1): Ditto.
16890 (set_group_iv_cost, force_var_cost): Ditto.
16891 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
16892 (get_computation_cost_at, get_computation_cost): Ditto.
16893 (determine_group_iv_cost_generic): Ditto.
16894 (determine_group_iv_cost_address): Ditto.
16895 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
16896 (determine_group_iv_costs): Ditto.
16897 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
16898 (iv_ca_set_remove_invariants): Renamed to ...
16899 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
16900 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
16901 (iv_ca_set_add_invariants): Renamed to ...
16902 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
16903 (iv_ca_set_cp): Use iv_ca_set_add_invs.
16904 (iv_ca_has_deps): Support inv_vars and inv_exprs.
16905 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
16906 (create_new_ivs): Remove useless dump.
16908 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16910 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
16912 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
16913 (iv_ca_set_no_cp, create_new_iv): Ditto.
16915 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16917 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16919 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16921 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
16922 function tree_check2.
16924 2017-05-02 Martin Liska <mliska@suse.cz>
16926 * doc/gcov.texi: Add missing preposition.
16927 * gcov.c (function_info::function_info): Properly fill up
16928 all member variables.
16930 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16932 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
16934 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16936 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
16938 2017-05-02 Martin Liska <mliska@suse.cz>
16941 * lto-streamer-in.c (lto_read_tree_1): Remove
16942 LTO_STREAMER_DEBUG.
16943 * lto-streamer.c (struct tree_hash_entry): Likewise.
16944 (struct tree_entry_hasher): Likewise.
16945 (tree_entry_hasher::hash): Likewise.
16946 (tree_entry_hasher::equal): Likewise.
16947 (lto_streamer_init): Likewise.
16948 (lto_orig_address_map): Likewise.
16949 (lto_orig_address_get): Likewise.
16950 (lto_orig_address_remove): Likewise.
16951 * lto-streamer.h: Likewise.
16952 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
16953 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
16955 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
16957 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
16958 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
16959 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
16960 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
16961 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
16962 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
16963 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
16964 (mm_maskz_sub_ss): New intrinsics.
16965 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16966 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16967 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
16968 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
16969 (__builtin_ia32_subss_mask_round): New builtins.
16970 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16971 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16972 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
16974 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
16975 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16977 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16980 2017-05-02 Martin Jambor <mjambor@suse.cz>
16982 PR tree-optimization/78687
16983 * tree-sra.c (access): New field parent.
16984 (process_subtree_disqualification): New function.
16985 (disqualify_candidate): Call it.
16986 (build_accesses_from_assign): Reset write flag if creating an
16988 (build_access_subtree): Fill in parent field and also prpagate
16989 down grp_write flag.
16990 (create_artificial_child_access): New parameter set_grp_write, set
16991 grp_write to its value.
16992 (propagate_subaccesses_across_link): Also propagate grp_write flag
16994 (propagate_all_subaccesses): Push the closest parent back to work
16995 queue if add_access_to_work_queue returned true.
16997 2017-05-02 Richard Biener <rguenther@suse.de>
16999 * common.opt (fstrict-overflow): Alias negative to fwrapv.
17000 * doc/invoke.texi (fstrict-overflow): Remove all traces of
17001 -fstrict-overflow documentation.
17002 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
17003 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
17004 flag_strict_overflow.
17005 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
17006 * lto-opts.c (lto_write_options): Do not stream it.
17007 * lto-wrapper.c (merge_and_complain): Do not handle it.
17008 * opts.c (default_options_table): Do not set -fstrict-overflow.
17009 (finish_options): Likewise do not clear it when sanitizing.
17010 * simplify-rtx.c (simplify_const_relational_operation): Do not
17011 test flag_strict_overflow.
17013 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17015 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
17016 using enabled attribute.
17017 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
17018 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
17019 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
17020 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
17021 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
17022 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
17023 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
17024 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
17025 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
17026 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
17028 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
17030 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
17032 2017-05-02 Richard Biener <rguenther@suse.de>
17034 PR tree-optimization/80591
17036 2017-04-10 Richard Biener <rguenther@suse.de>
17038 * tree-ssa-structalias.c (find_func_aliases): Properly handle
17041 2017-05-02 Richard Biener <rguenther@suse.de>
17043 PR tree-optimization/80549
17044 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
17045 (cleanup_tree_cfg_noloop): Create forwarders to known loop
17046 headers if they do not have a preheader.
17048 2017-05-02 Martin Liska <mliska@suse.cz>
17051 * common.opt: Fix typo.
17052 * doc/invoke.texi: Likewise.
17054 2017-05-01 Jan Beulich <jbeulich@suse.com>
17056 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
17057 swapping, add (x,x,m,x,n) alternative.
17059 2017-05-01 Nathan Sidwell <nathan@acm.org>
17061 * calls.c (combine_pending_stack_adjustment_and_call): Remove
17062 unnecessary unadjusted_alignment check.
17064 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
17067 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
17069 * gimplify.c (gimplify_cilk_detach): New function.
17070 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
17071 * tree-core.h: Document EXPR_CILK_SPAWN.
17072 * tree.h (EXPR_CILK_SPAWN): Define.
17074 2017-05-01 David Malcolm <dmalcolm@redhat.com>
17076 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
17077 to use new fixit_hint representation, using the "replace" logic.
17078 (get_line_span_for_fixit_hint): Likewise.
17079 (layout::print_any_fixits): Likewise.
17080 (selftest::test_one_liner_many_fixits): Rename to...
17081 (selftest::test_one_liner_many_fixits_1): ...this, and update
17082 comment and expected output to reflect that the multiple fix-it
17083 hints are now consolidated into one insertion.
17084 (selftest::test_one_liner_many_fixits_2): New test.
17085 (selftest::test_diagnostic_show_locus_one_liner): Update for
17087 (selftest::test_fixit_consolidation): Update for fix-it API
17089 * diagnostic.c (print_parseable_fixits): Likewise.
17090 * edit-context.c (edited_line::m_line_events): Convert from
17091 auto_vec <line_event *> to auto_vec <line_event>.
17092 (class line_event): Convert from abstract base class to a concrete
17093 class, taking over the role of replace_event.
17094 (class insert_event): Delete.
17095 (class replace_event): Rename to class line_event. Convert to
17097 (edit_context::add_fixits): Reimplement.
17098 (edit_context::apply_insert): Delete.
17099 (edit_context::apply_replace): Rename to...
17100 (edit_context::apply_fixit): ...this. Convert to half-open range.
17101 (edited_file::apply_insert): Delete.
17102 (edited_file::apply_replace): Rename to...
17103 (edited_file::apply_fixit): ...this.
17104 (edited_line::~edited_line): Drop deletion of events.
17105 (edited_line::apply_insert): Delete.
17106 (edited_line::apply_replace): Rename to...
17107 (edited_line::apply_fixit): ...this. Convert to half-open range.
17108 Update for change to type of m_line_events.
17109 * edit-context.h (edit_context::apply_insert): Delete.
17110 (edit_context::apply_replace): Rename to...
17111 (edit_context::apply_fixit): ...this.
17113 2017-05-01 Martin Sebor <msebor@redhat.com>
17115 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
17118 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
17121 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
17122 __get_cpuid_max returns 0.
17123 (__get_cpuid_count): Ditto.
17125 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
17127 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
17128 replacement expression is another instance of one of its arguments.
17130 2017-05-01 Jakub Jelinek <jakub@redhat.com>
17133 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
17134 check for stack push/pop autoinc.
17135 * config/i386/i386.c (ix86_agi_dependent): Return false
17136 if the only reason why modified_in_p returned true is that
17137 addr is SP based and set_insn is a push or pop.
17139 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17141 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
17144 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
17147 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
17148 (account_size_time): Use two predicates - exec_pred and
17150 (evaluate_conditions_for_known_args): Compute both clause and
17152 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
17153 (inline_summary_t::duplicate): Update.
17154 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
17156 (compute_inline_parameters): Likewise.
17157 (estimate_edge_size_and_time): Update caluclation of time.
17158 (estimate_node_size_and_time): Compute both time and nonspecialized
17160 (estimate_ipcp_clone_size_and_time): Update.
17161 (inline_merge_summary): Update.
17162 (do_estimate_edge_time): Update.
17163 (do_estimate_edge_size): Update.
17164 (do_estimate_edge_hints): Update.
17165 (inline_read_section, inline_write_summary): Stream both new predicates.
17166 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
17168 (compute_inlined_call_time): Cleanup.
17169 (big_speedup_p): Update.
17170 (edge_badness): Update.
17171 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
17172 (size_time_entry): Replace predicate by exec_predicate and
17173 nonconst_predicate.
17174 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
17175 (estimate_edge_time): Return also nonspec_time.
17176 (reset_edge_growth_cache): Update.
17178 2017-04-29 Jakub Jelinek <jakub@redhat.com>
17180 PR rtl-optimization/80491
17181 * ifcvt.c (noce_process_if_block): When looking for x setter
17182 with missing else_bb, don't check only the insn right before
17183 cond_earliest, but look for the last insn that x is modified in
17184 within the same bb.
17186 PR rtl-optimization/80491
17187 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
17189 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
17191 PR tree-optimization/80487
17192 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
17194 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17196 PR tree-optimization/79697
17197 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
17198 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
17199 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
17201 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
17202 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
17204 2017-04-28 Martin Sebor <msebor@redhat.com>
17206 PR tree-optimization/80523
17207 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
17208 (init_target_to_host_charmap, target_to_host, target_strtol10): New
17210 (maybe_warn, format_directive, parse_directive): Use new functions.
17211 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
17213 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
17215 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
17217 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
17219 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
17220 target_header_dir): Set correctly.
17221 * configure: Regenerated.
17222 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17223 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17224 instead of SYSTEM_HEADER_DIR.
17226 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
17228 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
17229 (estimate_local_effects): Likewise.
17230 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
17231 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
17232 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
17233 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
17234 do_estimate_edge_time, estimate_edge_time): Likewise.
17235 * ipa-inline-analysis.c (estimate_node_size_and_time,
17236 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
17237 (estimate_time_after_inlining): Remove.
17239 2017-04-28 Martin Liska <mliska@suse.cz>
17241 * doc/gcov.texi: Enhance documentation of gcov.
17243 2017-04-28 Martin Liska <mliska@suse.cz>
17245 * doc/gcov.texi: Sort options in alphabetic order.
17246 * doc/gcov-dump.texi: Likewise.
17247 * doc/gcov-tool.texi: Likewise.
17248 * gcov.c (print_usage): Likewise.
17249 * gcov-dump.c (print_usage): Likewise.
17250 * gcov-tool.c (print_merge_usage_message): Likewise.
17251 (print_rewrite_usage_message): Likewise.
17252 (print_overlap_usage_message): Likewise.
17254 2017-04-28 Martin Liska <mliska@suse.cz>
17256 PR gcov-profile/53915
17257 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
17259 2017-04-28 Martin Liska <mliska@suse.cz>
17261 PR gcov-profile/79891
17262 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
17263 is marked by compiler as living on a line.
17264 (get_cycles_count): Remove usage of the union.
17265 (output_intermediate_file): Likewise.
17266 (find_source): Fix GNU coding style.
17267 (accumulate_line_counts): Remove old non-all block mode.
17268 (output_lines): Remove usage of the union.
17269 * profile.c (output_location): Include all BBs, even if
17270 belonging to a same line (and file) as a previous BB.
17272 2017-04-28 Martin Liska <mliska@suse.cz>
17274 * gcov.c (process_args): Handle new argument 'w'.
17275 (read_graph_file): Assign ID to BBs.
17276 (output_branch_count): Display BB # if verbose flag is set.
17277 (output_lines): Likewise for arcs.
17278 (print_usage): Add '--verbose' option help.
17279 * doc/gcov.texi: Document --verbose (-w) option.
17281 2017-04-28 Martin Liska <mliska@suse.cz>
17283 * gcov.c (struct block_location_info): New struct.
17284 (process_file): Fill up the new structure.
17285 (read_graph_file): Replace usage of encoding by the newly added
17287 (add_line_counts): Likewise.
17288 (accumulate_line_counts): Remove usage of the union.
17289 (function_info::function_info): New function.
17290 (function_info::~function_info): Likewise.
17291 (process_file): Call delete instead of release_function.
17292 (release_function): Release the function.
17293 (release_structures): Call delete instead of release_function.
17294 (solve_flow_graph): Replace usage of num_blocks.
17295 (find_exception_blocks): Likewise.
17296 (output_lines): Fix GNU coding style.
17298 2017-04-28 Martin Liska <mliska@suse.cz>
17301 * coverage.c (coverage_remove_note_file): New function.
17302 * coverage.h: Declare the function.
17303 * toplev.c (finalize): Clean if an error has been seen.
17305 2017-04-28 Martin Liska <mliska@suse.cz>
17307 PR gcov-profile/80031
17308 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
17309 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
17310 * gcov.c (read_graph_file): Read just number of blocks.
17311 * profile.c (branch_prob): Do not stream 0 flags per a basic
17314 2017-04-28 Martin Liska <mliska@suse.cz>
17316 * gcov-dump.c (tag_*): Add new argument to declarations.
17317 (dump_gcov_file): Likewise.
17318 (tag_blocks): Add and use new argument depth.
17319 (tag_arcs): Likewise.
17320 (tag_lines): Likewise.
17321 (tag_counters): Likewise.
17322 (tag_summary): Likewise.
17323 (dump_working_sets): Use depth to do a proper indentation.
17325 2017-04-28 Jakub Jelinek <jakub@redhat.com>
17328 * cgraph.h (symtab_node::debug_symtab): No longer inline.
17329 * symtab.c (symtab_node::debug_symtab): Move definition here.
17331 2017-04-28 Richard Biener <rguenther@suse.de>
17333 * lto-streamer.h (LTO_major_version): Bump to 7.
17335 2017-04-28 Richard Biener <rguenther@suse.de>
17337 * tree-vrp.c (assert_info): New struct.
17338 (add_assert_info): New helper.
17339 (register_edge_assert_for_2): Refactor to add asserts to a vector
17341 (register_edge_assert_for_1): Likewise.
17342 (register_edge_assert_for): Likewise.
17343 (finish_register_edge_assert_for): New helper actually registering
17344 asserts where live on edge.
17345 (find_conditional_asserts): Adjust.
17346 (find_switch_asserts): Likewise.
17347 (evrp_dom_walker::try_find_new_range): Generalize.
17348 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
17350 2017-04-27 Marek Polacek <polacek@redhat.com>
17353 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
17354 arg10 and arg11 to itype.
17356 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
17358 * doc/extend.texi (Object Size Checking): Improve grammar.
17360 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
17363 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
17364 that the logic for permitting reciprocal estimates matches that
17367 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17370 * tree.c (type_cache_hasher::equal): Only compare
17371 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
17372 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
17373 non-aggregate element types.
17374 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
17375 about the flag on ARRAY_TYPEs in the comment, formatting fix.
17377 2017-04-27 Richard Biener <rguenther@suse.de>
17379 PR middle-end/80533
17380 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
17381 stripping ARRAY_REFs from MEM_EXPR make sure we're not
17382 keeping a reference to a trailing array.
17384 2017-04-27 Richard Biener <rguenther@suse.de>
17386 PR middle-end/80539
17387 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
17388 being in loop-closed SSA form conservatively.
17389 (chrec_fold_multiply_poly_poly): Likewise.
17391 2017-04-27 Tamar Christina <tamar.christina@arm.com>
17393 PR middle-end/79665
17394 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
17395 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
17397 2017-04-27 Jakub Jelinek <jakub@redhat.com>
17400 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
17401 (aarch64_function_arg_alignment): Return unsigned int again, but still
17402 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
17403 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17404 Don't emit -Wpsabi note.
17405 (aarch64_function_arg_boundary): Likewise.
17406 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17409 2017-04-26 Nathan Sidwell <nathan@acm.org>
17411 * tree.h (crc32_unsigned_n): Declare.
17412 (crc32_unsigned, crc32_unsigned): Make inline.
17413 * tree.c (crc32_unsigned_bits): Replace with ...
17414 (crc32_unsigned_n): ... this.
17415 (crc32_unsigned, crc32_byte): Remove.
17416 (crc32_string): Remove unnecessary braces.
17418 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17420 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
17421 * ipa-inline-analysis.c (MAX_TIME): Remove.
17422 (account_size_time): Use sreal for time.
17423 (dump_inline_summary): Update.
17424 (estimate_function_body_sizes): Update.
17425 (estimate_edge_size_and_time): Update.
17426 (estimate_calls_size_and_time): Update.
17427 (estimate_node_size_and_time): Update.
17428 (inline_merge_summary): Update.
17429 (inline_update_overall_summary): Update.
17430 (estimate_time_after_inlining): Update.
17431 (inline_read_section): Update.
17432 (inline_write_summary): Update.
17433 * ipa-inline.c (compute_uninlined_call_time): Update.
17434 (compute_inlined_call_time): Update.
17435 (recursive_inlining): Update.
17436 (inline_small_functions): Update.
17437 (dump_overall_stats): Update.
17438 * ipa-inline.h: Include sreal.h.
17439 (size_time_entry): Turn time to sreal.
17440 (inline_summary): Turn self_time nad time to sreal.
17442 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
17444 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
17446 (sreal::stream_out, sreal::stream_in): New.
17447 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
17449 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17451 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
17454 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
17457 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
17459 (dimode_scalar_chain::compute_convert_gain): Ditto.
17460 (dimode_scalar_chain::make_vector_copies): Ditto.
17461 (dimode_scalar_chain::convert_reg): Ditto.
17462 (dimode_scalar_chain::convert_insn): Ditto.
17463 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
17464 (VI248_AVX512BW_1): New mode iterator.
17465 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
17466 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
17469 2017-04-25 Martin Sebor <msebor@redhat.com>
17471 PR tree-optimization/80497
17472 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
17473 constants are representable in HOST_WIDE_INT.
17474 (parse_directive): Ditto.
17476 2017-04-25 Martin Sebor <msebor@redhat.com>
17479 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
17480 (new_zero_array): Adjust signature.
17481 (dom_info::dom_init): Used unsigned rather that size_t.
17482 (dom_info::dom_info): Same.
17484 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17485 Jakub Jelinek <jakub@redhat.com>
17488 * config/arm/arm.c: Include gimple.h.
17489 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17490 returns negative, increment ncrn only if it returned positive.
17491 (arm_needs_doubleword_align): Return int instead of bool,
17492 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
17493 members, but if there is any such non-FIELD_DECL
17494 > PARM_BOUNDARY aligned decl, return -1 instead of false.
17495 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
17496 returns negative, increment nregs only if it returned positive.
17497 (arm_setup_incoming_varargs): Likewise.
17498 (arm_function_arg_boundary): Emit -Wpsabi note if
17499 arm_needs_doubleword_align returns negative, return
17500 DOUBLEWORD_ALIGNMENT only if it returned positive.
17502 2017-04-25 Marek Polacek <polacek@redhat.com>
17505 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17506 first argument to type.
17508 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
17511 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
17512 type checks to test for compatibility instead of equality.
17514 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17515 Jakub Jelinek <jakub@redhat.com>
17518 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
17520 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
17521 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
17522 the alignment computation, but return their maximum in warn_alignment.
17523 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
17524 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
17526 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
17527 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
17530 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17532 * config/arc/simdext.md (dmpyh): Fix typo.
17534 2017-04-25 Richard Biener <rguenther@suse.de>
17536 PR tree-optimization/80492
17537 * alias.c (compare_base_decls): Handle registers with asm
17538 specification conservatively.
17539 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
17540 compare_base_decls returning dont-know properly.
17542 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17544 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
17545 (legitimate_offset_address_p): New function.
17546 (arc_legitimate_address_p): Use above function.
17548 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17550 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
17552 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17554 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
17555 ACCH registers whenever they are available.
17557 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17559 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
17560 double regs fix when not used.
17562 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17564 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
17566 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
17567 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
17569 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17571 * config/arc/arc.c (arc_output_addsi): Check for h-register class
17572 when emitting short ADD instructions.
17574 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
17576 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
17578 (cmpsi_cc_c_insn): Likewise.
17579 (cbranchsi4_scratch): Compute proper instruction length using
17580 compact_hreg_operand.
17581 * config/arc/predicates.md (compact_hreg_operand): New predicate.
17583 2017-04-25 Richard Biener <rguenther@suse.de>
17585 PR middle-end/80509
17586 * passes.c (pass_manager::pass_manager): Initialize
17587 m_name_to_pass_map.
17589 2017-04-25 Richard Biener <rguenther@suse.de>
17591 PR tree-optimization/79201
17592 * tree-ssa-sink.c (statement_sink_location): Handle calls.
17594 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17597 * config/s390/vector.md: Split MEM->GPR vector moves for
17598 non-s_operand addresses.
17600 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17603 * config/s390/predicates.md (reload_const_wide_int_operand): New
17605 * config/s390/s390.md ("movti"): Remove d/P alternative.
17606 ("movti_bigconst"): New pattern definition.
17608 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17611 * s390-protos.h (s390_expand_cs_hqi): Removed.
17612 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
17613 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
17614 modes as well as CCZ1mode and CCZmode.
17615 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
17616 signature of s390_emit_compare_and_swap.
17617 (s390_expand_cs_hqi): Likewise, make static.
17618 (s390_expand_cs_tdsi): Generate an explicit compare before trying
17619 compare-and-swap, in some cases.
17620 (s390_expand_cs): Wrapper function.
17621 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
17623 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
17624 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
17625 patterns for small and large integers. Forbid symref memory operands.
17626 Move expander to s390.c. Require cc register.
17627 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
17628 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
17629 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
17630 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
17631 symref memory operands. Remove CC mode and call s390_match_ccmode
17633 ("atomic_exchange<mode>"): Allow and implement all integer modes.
17635 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17637 * config/s390/s390.md (define_peephole2): New peephole to help
17638 combining the load-and-test pattern with volatile memory.
17640 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
17642 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
17643 with CCZmode for TARGET_Z196.
17645 2017-04-25 Jakub Jelinek <jakub@redhat.com>
17647 PR rtl-optimization/80501
17648 * combine.c (make_compound_operation_int): Set subreg_code to SET
17649 even for AND with mask of the sign bit of mode.
17651 PR rtl-optimization/80500
17652 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
17653 sum's initial value.
17655 2017-04-25 Julian Brown <julian@codesourcery.com>
17656 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17658 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
17660 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
17662 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
17664 2017-04-25 Julian Brown <julian@codesourcery.com>
17665 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17667 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
17668 (thunderx2t99_sha): New Reservation.
17670 2017-04-25 Julian Brown <julian@codesourcery.com>
17671 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17673 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
17674 type for 1-element load.
17676 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
17678 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
17680 2017-04-24 Martin Jambor <mjambor@suse.cz>
17682 PR tree-optimization/80293
17683 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
17684 char arrays not totally scalarizable if it is false.
17685 (analyze_all_variable_accesses): Pass correct value in the new
17686 parameter. Add a statistics counter.
17688 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
17690 PR middle-end/79931
17691 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
17693 2017-04-24 Richard Biener <rguenther@suse.de>
17695 PR tree-optimization/80494
17696 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
17697 out for complex types.
17699 2017-04-24 Richard Biener <rguenther@suse.de>
17701 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
17702 * tree-ssa-sccvn.c (print_scc): Print SCC size.
17703 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
17704 (DFS): Adjust and never fail.
17705 (sccvn_dom_walker::fail): Remove.
17706 (sccvn_dom_walker::before_dom_children): Adjust.
17707 (run_scc_vn): Likewise and never fail.
17708 * tree-ssa-pre.c (pass_pre::execute): Adjust.
17709 (pass_fre::execute): Likewise.
17711 2017-04-24 Richard Biener <rguenther@suse.de>
17713 PR tree-optimization/79725
17714 * tree-ssa-sink.c (statement_sink_location): Return whether
17715 failure reason was zero uses. Move that check later.
17716 (sink_code_in_bb): Deal with zero uses by removing the stmt
17719 2017-04-24 Richard Biener <rguenther@suse.de>
17722 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
17723 pointer-based references.
17725 2017-04-24 Richard Biener <rguenther@suse.de>
17728 * pass_manager.h (pass_manager::operator new): Remove.
17729 (pass_manager::operator delete): Likewise.
17730 * passes.c (pass_manager::operator new): Remove.
17731 (pass_manager::operator delete): Likewise.
17732 (pass_manager::pass_manager): Zero individual pass members.
17734 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
17737 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
17738 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
17739 Check "XEXP (src, 1)" operand here.
17740 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
17741 Check "XEXP (src, 1)" operand here.
17742 (dimode_scalar_chain::make_vector_copies): Detect count register
17743 of a shift instruction. Zero extend count register from QImode
17744 to DImode to satisfy vector shift pattern count operand predicate.
17745 Substitute vector shift count operand with a DImode copy.
17746 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
17749 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
17751 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
17752 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17753 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
17754 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17755 (UNSPEC_NOREX_MEM): Remove definition.
17757 2017-04-21 Richard Biener <rguenther@suse.de>
17759 PR tree-optimization/79547
17760 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17761 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
17762 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
17763 without any constraints.
17765 2017-04-21 Richard Biener <rguenther@suse.de>
17767 PR tree-optimization/78847
17768 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
17770 2017-04-21 Richard Biener <rguenther@suse.de>
17772 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
17773 (build_distinct_type_copy): Likewise.
17774 (build_variant_type_copy): Likewise.
17775 * tree.c (build_qualified_type): Pass down mem-stat info.
17776 (build_distinct_type_copy): Likewise.
17777 (build_variant_type_copy): Likewise.
17779 2017-04-21 Richard Biener <rguenther@suse.de>
17781 PR tree-optimization/80237
17782 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
17784 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
17785 for a simplified result.
17787 2016-04-21 Richard Biener <rguenther@suse.de>
17789 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
17790 sth as strict as a simple_iv but a chrec without symbols and an
17791 operand defined in the loop we are peeling (and not some subloop).
17792 (propagate_constants_for_unrolling): Propagate all constants.
17794 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17797 * config/i386/i386.c (print_reg): Remove assert for disalowed
17798 regno values, call output_operand_lossage instead.
17800 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17803 * config/i386/constraints.md (Yc): New register constraint.
17804 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
17805 Use Yc constraint for alternative 2 of operand 0. Remove
17806 preferred_for_speed attribute.
17808 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17810 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
17811 lastprivate clauses in SIMT case.
17813 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
17815 * doc/invoke.texi (-Wextra-semi): Document new warning option.
17817 2017-04-20 Richard Biener <rguenther@suse.de>
17819 PR tree-optimization/57796
17820 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
17821 as N scalar stores.
17822 (vect_model_load_cost): Cost gathers as N scalar loads.
17824 2017-04-20 Richard Biener <rguenther@suse.de>
17826 * ggc-page.c (ggc_allocated_p): Rename to ...
17827 (safe_lookup_page_table_entry): ... this and return the lookup
17829 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
17831 2017-04-20 Richard Biener <rguenther@suse.de>
17833 PR tree-optimization/80453
17834 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
17835 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
17836 from the conditions.
17837 (vn_phi_eq): Pass them down.
17838 (vn_phi_lookup): Record them.
17839 (vn_phi_insert): Likewise.
17841 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
17843 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
17844 uninitialized variable warning to avoid buffer overrun.
17846 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17849 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
17850 is suppressed for '{ 0 }' in C.
17852 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17854 * BASE-VER: Set to 8.0.0.
17856 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
17858 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
17859 priority .init_array and .fini_array section with SECTION_NOTYPE
17862 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17864 PR middle-end/80423
17865 * tree.h (build_array_type): Add typeless_storage default argument.
17866 * tree.c (type_cache_hasher::equal): Also compare
17867 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
17868 (build_array_type): Add typeless_storage argument, set
17869 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
17871 (build_nonshared_array_type): Adjust build_array_type_1 caller.
17872 (build_array_type): Likewise. Add typeless_storage argument.
17874 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17875 Jakub Jelinek <jakub@redhat.com>
17877 PR tree-optimization/80426
17878 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
17879 operation on symbolic operands, also compute the overflow for the
17880 invariant part when the operation degenerates into a negation.
17882 2017-04-19 Jakub Jelinek <jakub@redhat.com>
17885 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
17886 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
17889 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
17891 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
17894 * config/avr/avr.c (tree.h): Include it.
17895 (cgraph.h): Include it.
17896 (avr_encode_section_info): Don't warn for uninitialized progmem
17897 variable if it's just an alias.
17899 2017-04-19 Richard Biener <rguenther@suse.de>
17902 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
17903 when needed by AutoPGO.
17905 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
17908 * doc/lto.texi: Remove an extra 'that'.
17910 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
17912 PR rtl-optimization/80429
17913 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
17914 are only used in debug insns.
17916 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17917 Vladimir Makarov <vmakarov@redhat.com>
17919 * config/sparc/predicates.md (input_operand): Add comment. Return
17920 true for any memory operand when LRA is in progress.
17921 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
17923 2017-04-18 Jeff Law <law@redhat.com>
17926 * mips.md ({return,simple_return}_internal): Do not overwrite
17929 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17931 PR tree-optimization/80443
17932 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
17933 instead of adding 1, subtract -1 and similarly instead of subtracting
17936 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
17938 PR rtl-optimization/80357
17939 * haifa-sched.c (tmp_bitmap): New variable.
17940 (model_recompute): Handle duplicate use records.
17941 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
17942 (free_global_sched_pressure_data): Free it.
17944 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
17947 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
17948 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17949 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17950 instead of SYSTEM_HEADER_DIR.
17952 2017-04-18 Jeff Law <law@redhat.com>
17954 PR middle-end/80422
17955 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
17956 predecessors after walking up the insn chain.
17958 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17961 * dwarf2out.c (modified_type_die): Try harder not to emit internal
17962 sizetype type into debug info.
17964 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17967 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
17968 unneeded test for TARGET_UPPER_REGS_SF.
17969 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
17971 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17974 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
17975 instead of gsi_after_labels.
17977 2017-04-18 Jeff Law <law@redhat.com>
17979 * regcprop.c (maybe_mode_change): Avoid creating copies of the
17983 2017-04-13 Jeff Law <law@redhat.com>
17984 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17985 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17987 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
17990 * config/avr/avr.c (intl.h): Include it.
17991 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
17993 2017-04-18 Martin Liska <mliska@suse.cz>
17995 PR gcov-profile/78783
17996 * gcov-tool.c (gcov_output_files): Validate that destination
17997 file is either removed by the tool or by a user.
17999 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18000 Guy Benyei <guybe@mellanox.com>
18002 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
18003 block, and do not negate it, the stored id is already negative.
18005 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
18007 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
18009 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
18012 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
18013 masks of options that should be turned off if the VSX vector
18014 options are turned off.
18015 (OTHER_P8_VECTOR_MASKS): Likewise.
18016 (OTHER_VSX_VECTOR_MASKS): Likewise.
18017 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
18018 rs6000_disable_incompatible_switches to validate no type switches
18020 (rs6000_incompatible_switch): New function to disallow turning on
18021 other vector options if -mno-vsx, -mno-power8-vector, or
18022 -mno-power9-vector are specified.
18024 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18026 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
18028 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18030 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
18031 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
18032 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
18033 (ARG_POINTER_CFA_OFFSET): Likewise.
18035 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
18037 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
18038 conditions to take advantage of various optimizations.
18040 2017-04-13 Jeff Law <law@redhat.com>
18042 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
18043 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
18044 (zero_extendsidi2_dext): Likewise.
18046 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18049 * fold-const.c (fold_ternary_loc): Revert
18050 use op0 instead of fold_convert_loc (loc, type, arg0) part of
18053 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
18055 PR rtl-optimization/80343
18056 * lra-remat.c (update_scratch_ops): Assign original hard reg to
18057 new scratch pseudo.
18059 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
18062 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
18063 to ubsan_encode_value.
18065 2017-04-13 Jeff Law <law@redhat.com>
18067 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
18068 appearing in DEBUG_INSNs.
18070 2017-04-13 Martin Liska <mliska@suse.cz>
18072 PR gcov-profile/80413
18073 * gcov-io.c (gcov_write_string): Copy to buffer just when
18074 allocated size is greater than zero.
18076 2017-04-13 Jakub Jelinek <jakub@redhat.com>
18079 * dwarf2out.c (decls_for_scope): Ignore declarations of
18080 current_function_decl in BLOCK_NONLOCALIZED_VARS.
18082 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
18085 * ipa-visibility.c (non_local_p): Fix typos.
18086 (localize_node): When localizing symbol in same comdat group,
18087 dissolve the group only when we know external symbols are going
18089 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
18091 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18093 PR tree-optimization/79390
18094 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
18095 order does not result in usable sequence, retry with reversed operand
18101 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
18102 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
18103 op0 instead of fold_convert_loc (loc, type, arg0).
18105 2017-04-12 Jeff Law <law@redhat.com>
18107 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
18108 has a delay slot in the generated code.
18110 * config/cris/cris.md (cris_preferred_reload_class): Return
18111 GENNONACR_REGS rather than GENERAL_REGS.
18113 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18116 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
18117 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
18118 signedness of the result type.
18120 2017-04-12 Richard Biener <rguenther@suse.de>
18121 Jeff Law <law@redhat.com>
18123 PR tree-optimization/80359
18124 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
18125 trim stores to TARGET_MEM_REFs.
18127 2017-04-12 Richard Biener <rguenther@suse.de>
18129 PR tree-optimization/79390
18130 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
18131 threading case even more.
18133 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
18136 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
18137 for quad_address_p for TImode, instead of just not indexed_address.
18139 2017-04-12 Richard Biener <rguenther@suse.de>
18140 Bernd Edlinger <bernd.edlinger@hotmail.de>
18142 PR middle-end/79671
18143 * alias.c (component_uses_parent_alias_set_from): Handle
18144 TYPE_TYPELESS_STORAGE.
18145 (get_alias_set): Likewise.
18146 * tree-core.h (tree_type_common): Add typeless_storage flag.
18147 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
18148 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
18149 for types containing members with TYPE_TYPELESS_STORAGE.
18150 (place_field): Likewise.
18151 (layout_type): Likewise for ARRAY_TYPE.
18152 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
18153 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
18154 TYPE_TYPELESS_STORAGE.
18155 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
18157 2017-04-12 Jakub Jelinek <jakub@redhat.com>
18160 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
18161 first argument to type.
18163 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18167 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
18168 CONST0_RTX (mode) rather than const0_rtx where appropriate.
18169 (rs6000_expand_binop_builtin): Likewise.
18170 (rs6000_expand_ternop_builtin): Likewise; also add missing
18171 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
18172 vshasigma built-ins.
18173 * doc/extend.texi: Document that vec_xxpermdi's third argument
18174 must be a constant.
18176 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
18178 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
18179 Use shift_const cost parameter when calculating gain of STV shifts.
18181 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
18183 PR rtl-optimization/70478
18184 * lra-constraints.c (process_alt_operands): Check memory for
18185 disfavoring memory insn operand.
18187 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18189 PR middle-end/80100
18190 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
18191 left shift in unsigned HOST_WIDE_INT type.
18193 PR rtl-optimization/80385
18194 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
18195 (not (neg X)) into (plus X -1) for complex or non-integral modes.
18198 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
18199 if they have any depend clauses.
18201 2017-04-11 Martin Liska <mliska@suse.cz>
18204 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
18205 * ipa-split.c (split_function): Create a local comdat symbol
18206 if caller is in a comdat group.
18208 2017-04-11 Martin Liska <mliska@suse.cz>
18211 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
18214 2017-04-11 Martin Sebor <msebor@redhat.com>
18216 PR middle-end/80364
18217 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
18218 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
18220 (directive::set_width, directive::set_precision, format_character):
18222 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
18225 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
18228 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
18229 conflict, set target->arch_name instead of target->cpu_name.
18231 2017-04-11 Richard Biener <rguenther@suse.de>
18233 PR tree-optimization/80374
18234 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
18235 build_zero_cst, remove fold_convertible_p check again.
18237 2017-04-11 Martin Liska <mliska@suse.cz>
18240 * ubsan.c (instrument_object_size): Do not instrument register
18243 2017-04-11 Jakub Jelinek <jakub@redhat.com>
18246 * config/i386/i386-builtin-types.def
18247 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
18248 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
18249 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
18250 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
18251 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
18252 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
18253 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
18254 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
18255 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
18256 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
18257 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
18258 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
18259 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
18260 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
18261 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
18262 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
18263 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
18264 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
18265 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
18266 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
18267 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
18268 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
18269 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
18270 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
18271 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
18272 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
18273 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
18274 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
18275 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
18276 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
18277 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
18278 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
18279 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
18280 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
18281 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
18282 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
18283 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
18284 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
18285 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
18286 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
18287 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
18288 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
18289 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
18290 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
18291 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
18292 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
18294 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
18295 flag to second_arg_count, handle 4 argument function type _COUNT
18296 aliases, handle second_arg_count on second argument rather than last.
18298 2017-04-10 Jeff Law <law@redhat.com>
18300 PR tree-optimization/80374
18301 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
18302 record anything if we can not convert integer_zero_node to the
18305 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18308 * config/rs6000/rs6000.c (rs6000_option_override_internal):
18309 Enhance special handling given to the TARGET_P9_MINMAX option in
18310 relation to certain other options.
18312 2017-04-10 Bin Cheng <bin.cheng@arm.com>
18314 PR tree-optimization/80153
18315 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
18316 remove POINTER_PLUS_EXPR's base part directly, rather than through
18319 2017-04-10 Richard Biener <rguenther@suse.de>
18320 Bin Cheng <bin.cheng@arm.com>
18322 PR tree-optimization/80153
18323 * tree-affine.c (aff_combination_to_tree): Get base pointer from
18324 the first element of pointer type aff_tree. Build result expr in
18326 (add_elt_to_tree): Convert to type unconditionally. Remove other
18327 fold_convert calls.
18328 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
18329 (rewrite_use_nonlinear_expr): Check invariant using iv information.
18331 2017-04-10 Richard Biener <rguenther@suse.de>
18333 * tree-ssa-structalias.c (find_func_aliases): Properly handle
18336 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
18338 PR rtl-optimization/70478
18339 * lra-constraints.c (curr_small_class_check): New.
18340 (update_and_check_small_class_inputs): New.
18341 (process_alt_operands): Update curr_small_class_check. Disfavor
18342 alternative insn memory operands. Check available regs for small
18345 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18348 * config/mips/mips.opt (-mvirt): Update description.
18349 * doc/invoke.texi (-mvirt): Likewise.
18351 2017-04-10 Richard Biener <rguenther@suse.de>
18353 PR middle-end/80362
18354 * fold-const.c (fold_binary_loc): Look at unstripped ops when
18355 looking for NEGATE_EXPR in -A / -B to A / B folding.
18357 2017-04-10 Martin Liska <mliska@suse.cz>
18359 PR gcov-profile/80224
18360 * gcov.c (print_usage): Fix usage string.
18361 (get_gcov_intermediate_filename): Remove.
18362 (output_gcov_file): Use both for normal and intermediate format.
18363 (generate_results): Do not initialize special file for
18364 intermediate format.
18366 2017-04-10 Richard Biener <rguenther@suse.de>
18368 PR tree-optimization/80304
18369 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
18372 2017-04-10 Nathan Sidwell <nathan@acm.org>
18375 * config/rs6000/rs6000.c (rs6000_vector_type): New.
18376 (rs6000_init_builtins): Use it.
18378 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18380 * config/arm/arm.md (<mrc>): Add mode to SET source.
18381 (<mrrc>): Likewise.
18383 2017-04-10 Richard Biener <rguenther@suse.de>
18385 PR middle-end/80344
18386 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
18388 2017-04-10 Jakub Jelinek <jakub@redhat.com>
18391 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
18392 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
18393 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
18394 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
18395 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
18396 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
18397 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
18398 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
18399 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
18400 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
18401 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
18402 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
18403 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
18404 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
18405 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
18406 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
18407 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
18408 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
18409 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
18410 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
18411 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
18412 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
18413 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
18415 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
18417 PR rtl-optimization/70478
18418 * lra-constraints.c: Reverse the last patch.
18420 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
18422 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
18423 Add comment for WCHAR_T.
18425 2017-04-08 Martin Liska <mliska@suse.cz>
18428 2017-04-07 Martin Liska <mliska@suse.cz>
18431 * ipa-split.c (split_function): Add function part to a same comdat
18434 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18437 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
18439 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
18441 * rs6000/rs6000.c (vec_load_pendulum): Rename...
18442 (vec_pairing): ...to this.
18443 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
18444 (rs6000_sched_init): Adjust for name change.
18445 (struct rs6000_sched_context): Likewise.
18446 (rs6000_init_sched_context): Likewise.
18447 (rs6000_set_sched_context): Likewise.
18449 2017-04-07 Jakub Jelinek <jakub@redhat.com>
18455 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
18457 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
18458 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
18459 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
18461 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
18463 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
18465 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18467 PR rtl-optimization/70703
18468 * ira-color.c (update_conflict_hard_regno_costs): Use
18469 int64_t instead of HOST_WIDE_INT.
18471 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
18473 PR rtl-optimization/70478
18474 * lra-constraints.c (process_alt_operands): Disfavor alternative
18475 insn memory operands.
18477 2017-04-07 Jeff Law <law@redhat.com>
18479 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
18480 CALL and NOTE_INSN_CALL_ARG_LOCATION.
18482 2017-04-07 Martin Liska <mliska@suse.cz>
18485 * config/aarch64/aarch64.c (aarch64_process_target_attr):
18486 Show error message instead of an ICE.
18488 2017-04-07 Martin Liska <mliska@suse.cz>
18491 * ipa-split.c (split_function): Add function part to a same comdat
18494 2017-04-07 Richard Biener <rguenther@suse.de>
18496 PR middle-end/80341
18497 * tree.c (get_unwidened): Also handle ! for_type case for
18499 * convert.c (do_narrow): Split out from ...
18500 (convert_to_integer_1): ... here. Do not pass final truncation
18501 type to get_unwidened for TRUNC_DIV_EXPR.
18503 2017-04-07 Richard Biener <rguenther@suse.de>
18505 * tree-affine.c (wide_int_ext_for_comb): Take type rather
18507 (aff_combination_const): Adjust.
18508 (aff_combination_scale): Likewise.
18509 (aff_combination_add_elt): Likewise.
18510 (aff_combination_add_cst): Likewise.
18511 (aff_combination_convert): Likewise.
18512 (add_elt_to_tree): Likewise. Remove unused argument.
18513 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
18515 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
18517 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
18519 * config/arm/arm.c (arm_default_short_enums): Use
18520 ARM_DEFAULT_SHORT_ENUMS.
18521 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
18523 2017-04-06 Jakub Jelinek <jakub@redhat.com>
18526 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
18527 members with redundant out-of-class redeclaration.
18529 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18532 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
18533 * config/i386/i386.md (*zero_extendsidi2):
18534 Add (?*x,*x) and (?*v,*v) alternatives.
18536 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
18539 * config/i386/i386.c (ix86_expand_builtin)
18540 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
18541 mode from insn data. Convert operands to insn operand mode.
18542 Copy operands that don't satisfy insn predicate to a register.
18544 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
18546 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
18549 2017-04-06 Richard Biener <rguenther@suse.de>
18551 PR tree-optimization/80334
18552 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
18553 preserve alignment of accesses.
18555 2017-04-06 Richard Biener <rguenther@suse.de>
18557 PR tree-optimization/80262
18558 * tree-sra.c (build_ref_for_offset): Preserve address-space
18560 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
18561 Drop useless address-space information on MEM_REF offsets.
18563 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
18565 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
18567 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18569 PR rtl-optimization/70703
18570 * ira-color.c (update_conflict_hard_regno_costs): Use
18571 HOST_WIDE_INT instead of long.
18573 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
18576 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
18577 not defined for x86_64 target. Add -mmmx target option when __SSE2__
18579 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
18580 for x86_64 target. Handle -m3dnowa option.
18582 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
18584 PR rtl-optimization/70703
18585 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
18586 (update_conflict_hard_regno_costs): Use long instead of unsigned
18587 arithmetic for cost calculation.
18589 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18590 Bernd Edlinger <bernd.edlinger@hotmail.de>
18593 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
18596 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
18599 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
18600 ptr_mode with Pmode throughout.
18601 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
18602 into probe_stack_range and use DImode.
18604 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
18607 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
18608 call_eh_return is true.
18610 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18612 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
18613 Initialize last_match_fntype_index.
18615 2017-04-05 Jakub Jelinek <jakub@redhat.com>
18618 * tree-nvr.c: Include internal-fn.h.
18619 (pass_return_slot::execute): Ignore internal calls without
18622 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18623 Richard Biener <rguenther@suse.de>
18626 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
18627 captures used multiple times, except for the last use.
18628 * generic-match-head.c: Include gimplify.h.
18630 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18632 PR tree-optimization/79390
18633 * target.h (struct noce_if_info): Declare.
18634 * targhooks.h (default_noce_conversion_profitable_p): Declare.
18635 * target.def (noce_conversion_profitable_p): New target hook.
18636 * ifcvt.h (struct noce_if_info): New type, moved from ...
18637 * ifcvt.c (struct noce_if_info): ... here.
18638 (noce_conversion_profitable_p): Renamed to ...
18639 (default_noce_conversion_profitable_p): ... this. No longer
18641 (noce_try_store_flag_constants, noce_try_addcc,
18642 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
18643 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
18644 instead of noce_conversion_profitable_p.
18645 * config/i386/i386.c: Include ifcvt.h.
18646 (ix86_option_override_internal): Don't override
18647 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
18648 (ix86_noce_conversion_profitable_p): New function.
18649 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
18650 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
18651 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
18652 * doc/tm.texi: Regenerated.
18654 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18656 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
18659 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18662 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
18663 instructions for small multiply cores.
18665 2017-04-04 Jeff Law <law@redhat.com>
18667 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
18669 (mips_expand_vec_perm_const): Initialize elements in orig_perm
18670 that are not set by the loop over the elements.
18672 2017-04-04 Jakub Jelinek <jakub@redhat.com>
18675 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
18676 int mode, convert_modes it to mode as unsigned, otherwise use
18677 lowpart_subreg to mode rather than SImode.
18678 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
18679 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
18680 Use DImode instead of SImode for the shift count operand.
18681 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
18684 2017-04-04 Richard Biener <rguenther@suse.de>
18686 PR middle-end/80281
18687 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
18688 arithmetic done for the negate or the plus. Simplify.
18689 (A - (-B) -> A + B): Likewise.
18690 * fold-const.c (split_tree): Make sure to not negate pointers.
18692 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
18694 PR rtl-optimization/60818
18695 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
18696 a compare of comparisons with the thing compared if this results
18697 in a different machine mode.
18699 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
18701 * alias.c (base_alias_check): Fix typo in comment.
18702 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
18703 * cgraphunit.c (symbol_table::compile): Likewise.
18704 * collect2.c (maybe_run_lto_and_relink): Likewise.
18705 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
18706 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
18707 * config/avr/avr.c (avr_map_op_t): Likewise.
18708 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
18709 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
18710 * config/epiphany/epiphany.md (movcc): Likewise.
18711 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
18712 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
18714 * config/mips/mips.c (mips_save_restore_reg): Likewise.
18715 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
18716 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
18717 * config/sh/sh.c (sh_rtx_costs): Likewise.
18718 * fold-const.c (fold_truth_andor): Likewise.
18719 * genautomata.c (collapse_flag): Likewise.
18720 * gengtype.h (struct type::u::s): Likewise.
18721 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
18722 * input.c (FORMAT_AMOUNT): Likewise.
18723 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
18724 (known_aggs_to_agg_replacement_list): Likewise.
18725 * ipa-inline-analysis.c: Likewise.
18726 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
18727 * ipa-polymorphic-call.c
18728 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
18729 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
18730 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
18732 * modulo-sched.c (apply_reg_moves): Likewise.
18733 * omp-expand.c (build_omp_regions_1): Likewise.
18734 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
18735 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
18736 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
18737 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
18738 * value-prof.c: Likewise.
18739 * var-tracking.c (val_reset): Likewise.
18741 2017-04-03 Richard Biener <rguenther@suse.de>
18743 PR tree-optimization/80275
18744 * fold-const.c (split_address_to_core_and_offset): Handle
18747 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
18749 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
18750 descriptors is at least equal to that of functions.
18752 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18754 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
18756 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18759 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
18760 (mov<IMOD4:mode>): New expander.
18761 (*mov<IMOD4:mode>_internal): New insn and split pattern.
18763 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
18765 PR rtl-optimization/79405
18766 * fwprop.c (propagations_left): New variable.
18767 (forward_propagate_into): Decrement it.
18768 (fwprop_init): Initialize it.
18769 (fw_prop): If the variable has reached zero, stop propagating.
18770 (fwprop_addr): Ditto.
18772 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18775 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
18776 a FUNCTION_DECL, pass it as decl instead of origin to
18779 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
18781 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
18784 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
18787 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
18788 TARGET_VSX_SMALL_INTEGER.
18790 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18792 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18793 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
18795 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18797 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
18798 extraction from odd-numbered MSA register.
18800 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18802 PR middle-end/80173
18803 * expmed.c (store_bit_field_1): Don't attempt to create
18804 a word subreg out of hard registers wider than word if they
18805 have HARD_REGNO_NREGS of 1 for their mode.
18807 PR middle-end/80163
18808 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
18809 conversions to integer types wider than word and pointer.
18812 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
18813 (rtx_equal_for_cselib_p): Pass 0 to it.
18814 * cselib.c (cselib_hasher::equal): Likewise.
18815 (rtx_equal_for_cselib_1): Add depth argument. If depth
18816 is 128, don't look up VALUE locs and punt. Increment
18817 depth in recursive calls when walking VALUE locs.
18819 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
18821 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
18822 (make_gcov_file_name): Use the canonical path name for generating
18824 (read_line): Fix handling of files with ascii null bytes.
18826 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
18828 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
18829 to initialise a vector register instead
18830 of using a const_int.
18832 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18834 PR translation/80189
18835 * gimplify.c (omp_default_clause): Use %qs instead of %s in
18836 diagnostic messages.
18838 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
18841 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
18842 (dfp_diex_<mode>): Update mode of operand 1.
18843 * doc/extend.texi (dxex, dxexq): Document change to return type.
18844 (diex, diexq): Document change to argument type.
18846 2017-03-30 Martin Jambor <mjambor@suse.cz>
18849 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
18850 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
18851 it reflects the signature changes performed at the callee side.
18852 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
18853 to cgraph_build_function_type_skip_args.
18854 (build_function_decl_skip_args): Adjust call to the above function.
18856 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18859 * config/i386/sse.md
18860 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
18861 register as dest whenever it is a MEM not rtx_equal_p to the
18862 corresponding dup operand, and when forcing into reg move the
18863 reg into the memory afterwards.
18864 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
18865 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
18866 for the force_reg mode.
18867 (avx512vl_vextractf128<mode>): Use register as dest either
18868 always when a MEM, or when it is a MEM not rtx_equal_p to the
18869 corresponding dup operand, or even not when it is a CONST_VECTOR
18870 depending on the mode and lo vs. hi.
18871 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
18873 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
18874 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
18875 Likewise. Require that operands[2] is even.
18876 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
18877 Remove extraneous parens. Require that operands[2] is a multiple
18879 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
18880 operands[0] is a MEM if <mask_applied>, the predicates/constraints
18881 disallow memory then.
18883 2017-03-30 Richard Biener <rguenther@suse.de>
18885 PR tree-optimization/77498
18886 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
18887 to non-constants over backedges.
18889 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
18891 PR rtl-optimization/80233
18892 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
18893 as last_combined_insn. Do not test for BARRIER_P separately.
18895 2017-03-29 Andreas Schwab <schwab@suse.de>
18898 * calls.c (prepare_call_address): Convert funexp to Pmode before
18899 copying to temp reg.
18901 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18903 PR tree-optimization/80158
18904 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
18905 Handle possible future case of more than one alternate
18907 (replace_rhs_if_not_dup): Likewise.
18908 (replace_one_candidate): Likewise.
18910 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
18912 PR rtl-optimization/80193
18913 * ira.c (ira): Do not check allocation for LRA.
18915 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
18917 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
18918 (nvptx_output_simt_exit): Declare.
18919 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
18920 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
18921 (init_softstack_frame): Move initialization of crtl->is_leaf to...
18922 (nvptx_declare_function_name): ...here. Emit declaration of local
18923 memory space buffer for omp_simt_enter insn.
18924 (nvptx_output_unisimt_switch): New.
18925 (nvptx_output_softstack_switch): New.
18926 (nvptx_output_simt_enter): New.
18927 (nvptx_output_simt_exit): New.
18928 * config/nvptx/nvptx.h (struct machine_function): New fields
18929 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
18930 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
18931 (UNSPECV_SIMT_EXIT): Ditto.
18932 (omp_simt_enter_insn): New insn.
18933 (omp_simt_enter): New expansion.
18934 (omp_simt_exit): New insn.
18935 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
18937 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
18938 (expand_GOMP_SIMT_ENTER_ALLOC): New.
18939 (expand_GOMP_SIMT_EXIT): New.
18940 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
18941 (GOMP_SIMT_ENTER_ALLOC): Ditto.
18942 (GOMP_SIMT_EXIT): Ditto.
18943 * target-insns.def (omp_simt_enter): New insn.
18944 (omp_simt_exit): Ditto.
18945 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
18947 (lower_rec_simd_input_clauses): Implement SIMT privatization.
18948 (lower_rec_input_clauses): Likewise.
18949 (lower_lastprivate_clauses): Handle SIMT privatization.
18951 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
18952 (ompdevlow_adjust_simt_enter): New.
18953 (find_simtpriv_var_op): New.
18954 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
18955 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
18957 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
18958 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
18959 (copy_decl_for_dup_finish): Ditto.
18961 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
18963 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
18966 * config/i386/i386.c (ix86_option_override_internal): Always
18967 allow -mpreferred-stack-boundary=3 for 64-bit targets.
18969 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18971 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
18973 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18975 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
18976 mark new edge's irreducible flag accordign to it.
18977 (vect_do_peeling): Check loop preheader edge's irreducible flag
18978 and pass it to function slpeel_add_loop_guard.
18980 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
18982 PR tree-optimization/80218
18983 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18984 Update block frequencies and counts.
18986 2017-03-28 Richard Biener <rguenther@suse.de>
18988 PR tree-optimization/78644
18989 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
18990 of a simplification result we may not use it at all.
18992 2017-03-28 Richard Biener <rguenther@suse.de>
18995 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
18996 without arguments, generate default definition of a SSA name.
18998 2017-03-28 Richard Biener <rguenther@suse.de>
19000 PR middle-end/80222
19001 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
19002 TYPE_REF_CAN_ALIAS_ALL references.
19003 * fold-const.c (fold_indirect_ref_1): Likewise.
19005 2017-03-28 Martin Liska <mliska@suse.cz>
19008 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
19009 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
19011 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19012 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
19014 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
19015 (EXTRA_SPECS): Define.
19016 (SUBTARGET_EXTRA_SPECS): Likewise.
19017 (SUBTARGET_CPP_SPEC): Likewise.
19018 * config/arc/elf.h (EXTRA_SPECS): Renamed to
19019 SUBTARGET_EXTRA_SPECS.
19020 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
19022 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
19024 * config/arc/simdext.md (vst64_insn): Update pattern.
19025 (vld32wh_insn): Likewise.
19026 (vld32wl_insn): Likewise.
19027 (vld64_insn): Likewise.
19028 (vld32_insn): Likewise.
19030 2017-03-28 Marek Polacek <polacek@redhat.com>
19033 * fold-const.c (fold_comparison): Use protected_set_expr_location
19034 instead of SET_EXPR_LOCATION.
19036 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
19038 * tree.c (add_expr): Avoid name lookup warning.
19040 2017-03-27 Jeff Law <law@redhat.com>
19042 PR tree-optimization/80216
19043 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
19044 function name. Limit recursion depth.
19045 (record_temporary_equivalences): Corresponding changes.
19047 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
19049 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
19052 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19055 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
19057 * cfgcleanup.c (reg_note_cfa_p): New array.
19058 (insns_have_identical_cfa_notes): New function.
19059 (old_insns_match_p): Don't cross-jump in between /f
19060 and non-/f instructions. If both i1 and i2 are frame related,
19061 verify all CFA notes, their order and content.
19063 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
19066 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
19067 HImode and SImode with zero extend to DImode to one insn.
19068 (bswap<mode>2_extenddi): Likewise.
19069 (bswapsi2_extenddi): Likewise.
19070 (bswaphi2_extendsi): Likewise.
19071 (bswaphi2): Combine bswap HImode and SImode into one insn.
19072 Separate memory insns from swapping register.
19073 (bswapsi2): Likewise.
19074 (bswap<mode>2): Likewise.
19075 (bswaphi2_internal): Delete, no longer used.
19076 (bswapsi2_internal): Likewise.
19077 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
19078 store, and gpr<-gpr swap insns.
19079 (bswap<mode>2_store): Likewise.
19080 (bswaphi2_reg): Register only splitter, combine with the splitter.
19081 (bswaphi2 splitter): Likewise.
19082 (bswapsi2_reg): Likewise.
19083 (bswapsi2 splitter): Likewise.
19084 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
19085 the insns into load, store, and register/register insns.
19086 (bswapdi2_ldbrx): Likewise.
19087 (bswapdi2_load): Likewise.
19088 (bswapdi2_store): Likewise.
19089 (bswapdi2_reg): Likewise.
19091 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
19093 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
19094 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
19096 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
19099 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
19101 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19102 special handling for target option conflicts between dform
19103 options (-mpower9-dform, -mpower9-dform-vector,
19104 -mpower9-dform-scalar) and -mno-direct-move.
19106 2017-03-27 Richard Biener <rguenther@suse.de>
19108 PR tree-optimization/80181
19109 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
19111 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19113 * config/arc/predicates.md (move_double_src_operand): Replace the
19114 call to move_double_src_operand with a call to address_operand.
19116 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19118 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
19119 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
19120 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
19122 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19124 * config/arc/predicates.md (long_immediate_loadstore_operand):
19125 Consider scaled addresses cases.
19127 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
19129 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
19130 restored when in interrupt.
19131 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
19132 doesn't have delay slot.
19134 2017-03-27 Richard Biener <rguenther@suse.de>
19137 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
19138 inlined thunk clones.
19140 2017-03-27 Jakub Jelinek <jakub@redhat.com>
19143 * asan.c (instrument_derefs): Copy over last operand from
19144 original COMPONENT_REF to the new COMPONENT_REF with
19145 DECL_BIT_FIELD_REPRESENTATIVE.
19146 * ubsan.c (instrument_object_size): Likewise.
19148 2017-03-27 Richard Biener <rguenther@suse.de>
19150 PR tree-optimization/80170
19151 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
19152 sure DR/SCEV didnt fold in constants we do not see when looking
19153 at the reference base alignment.
19155 2017-03-27 Richard Biener <rguenther@suse.de>
19157 PR middle-end/80171
19158 * gimple-fold.c (fold_ctor_reference): Properly guard against
19159 NULL return value from canonicalize_constructor_val.
19161 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
19164 * config/i386/i386.c (ix86_expand_builtin)
19165 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
19166 flags reg setting and flags reg using instructions.
19167 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
19168 clobbering instructions to zero extend op2.
19170 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
19172 * doc/install.texi (Configuration) <--with-aix-soname>:
19173 Update link to AIX ld.
19175 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
19177 PR rtl-optimization/80160
19178 PR rtl-optimization/80159
19179 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
19180 reg_alternate_class into account.
19182 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
19185 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
19186 to consider in curr_insn_transform.
19188 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19190 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
19191 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
19192 and emit_mode_inner.
19194 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19196 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
19197 argument to the overloaded builtin variants. Use the new flag to
19198 deprecate certain builtin variants.
19199 * config/s390/s390-builtin-types.def: Add new builtin types.
19200 * config/s390/s390-builtins.h: Support new flags field for
19201 overloaded builtins.
19202 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
19203 (s390_macro_to_expand): Enable vector float data type.
19204 (s390_cpu_cpp_builtins_internal): Indicate support of the new
19205 builtins by incrementing the __VEC__ version number.
19206 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
19208 (s390_resolve_overloaded_builtin): Emit error messages depending
19209 on the builtin flags.
19210 * config/s390/s390.c (s390_expand_builtin): Support additional
19211 flags argument. Change error message to match the messages
19212 emitted in s390-c.c.
19213 * config/s390/s390.md: New UNSPEC_* constants.
19214 (op_type): Add new instruction types.
19215 * config/s390/vecintrin.h: Add new builtins and test data class
19217 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
19218 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
19219 (VEC_INEXACT, VEC_NOINEXACT): New constants.
19220 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
19221 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
19222 ("vec_mergel<mode>"): V_HW -> VEC_HW.
19224 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
19225 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
19226 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
19227 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
19229 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
19230 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
19231 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
19232 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
19234 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
19235 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
19236 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
19237 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
19238 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
19239 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
19240 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
19242 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
19243 ("vec_scatter_element<V_HW_4:mode>_DI")
19244 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
19245 ("vec_fpint<mode>", "vflls")
19246 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
19247 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
19248 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
19249 ("*vec_cmphe<mode>_cc"): ... these.
19251 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
19252 mode constant instead of magic value.
19254 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19256 * config/s390/s390.c (s390_expand_vec_compare): Support other
19257 vector floating point modes than just V2DF.
19258 (s390_expand_vcond): Likewise.
19259 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
19260 (s390_cannot_change_mode_class): Prevent mode changes between TF
19261 and V1TF in vector registers.
19262 * config/s390/s390.md (DF, SF): New mode attributes.
19263 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
19264 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
19265 SFmode support for VRs.
19266 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
19268 (VFT, VF_HW): New mode iterators.
19269 (vw, sdx): New mode attributes.
19270 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
19271 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
19272 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
19273 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
19274 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
19275 also the new vector floating point modes. Renaming to ...
19277 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
19278 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
19279 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
19280 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
19281 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
19282 ("vec_unordered<mode>"): ... these.
19284 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
19285 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
19286 ("*vec_extendv2df"): New insn definitions.
19288 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19290 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
19291 ("mulditi3_2", "*muldi3_sign"): New patterns.
19292 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
19293 rename the pattern definition.
19295 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19297 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
19299 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
19301 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19303 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
19304 instruction if possible.
19305 * config/s390/vector.md (vec_halfnumelts): New mode
19307 ("*vec_vllezlf<mode>"): New pattern.
19309 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19311 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
19312 ("popcountv4si2", "popcountv2di2"): Rename to ...
19313 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
19314 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
19316 ("popcount<mode>2_vxe"): New pattern.
19318 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19320 * common/config/s390/s390-common.c (processor_flags_table): Add
19322 * config.gcc: Add arch12.
19323 * config/s390/driver-native.c (s390_host_detect_local_cpu):
19324 Default to arch12 for unknown CPU model numbers.
19325 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
19326 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
19327 PROCESSOR_max sanity check.
19328 * config/s390/s390-opts.h (enum processor_type): Add
19330 * config/s390/s390.c (processor_table): Add arch12.
19331 (s390_expand_builtin): Add check for B_VXE flag.
19332 (s390_issue_rate): Add PROCESSOR_ARCH12.
19333 (s390_get_sched_attrmask): Likewise.
19334 (s390_get_unit_mask): Likewise.
19335 (s390_sched_score): Enable z13 scheduling for arch12.
19336 (s390_sched_reorder): Likewise.
19337 (s390_sched_variable_issue): Likewise.
19338 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
19340 (s390_tune_attr): Use z13 scheduling also for arch12.
19341 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
19342 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
19343 (TARGET_VXE_P): New macros.
19344 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
19345 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
19346 * config/s390/s390.opt: Add arch12 as processor_type.
19348 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19350 * config/s390/s390.md
19351 ("fixuns_truncdddi2", "fixuns_trunctddi2")
19352 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
19353 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
19355 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
19356 Rename expanders to ...
19358 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
19359 ("fixuns_truncdddi2_emu"): ... these.
19361 ("fixuns_trunc<mode>si2_emu"): New expander.
19363 ("*fixuns_truncdfdi2_z13"): Rename to ...
19364 ("*fixuns_truncdfdi2_vx"): ... this.
19366 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19368 * config/s390/2964.md: Remove the single element vector compare
19369 instructions which are no longer used.
19370 * config/s390/s390.c (s390_select_ccmode): Remove handling of
19372 (s390_canonicalize_comparison): Remove handling of DFmode
19374 (s390_expand_vec_compare_scalar): Remove function.
19375 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
19376 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
19378 ("*cmp<mode>_ccs"): Add wfcdb instruction.
19380 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19382 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
19384 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
19385 will anyway by matched by mov<mode>_64dfp.
19387 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19389 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
19390 vlef/vstef. Add missing operand to vleif.
19392 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19394 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
19395 pair for all vector types with 64 bit elements.
19396 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
19397 * config/s390/vector.md (V_HW_64): ... here.
19398 (V_128_NOSINGLE): New mode iterator.
19399 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
19400 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
19401 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
19402 ("*vec_load_pairv2di"): Change to ...
19403 ("*vec_load_pair<mode>"): ... this one.
19405 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19407 * config/s390/constraints.md: Add comments.
19408 (jKK): Reject element sizes > 8 bytes.
19409 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
19411 * config/s390/s390.md: Add the s_operand checks formerly in
19412 s390_split_ok_p to various splitters where they are still
19414 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
19415 for 128 bit vectors. Plus two splitters.
19417 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19419 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
19422 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19425 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
19426 error if the boundary argument is not constant.
19428 2017-03-24 Jakub Jelinek <jakub@redhat.com>
19430 PR rtl-optimization/80112
19431 * loop-doloop.c (doloop_condition_get): Don't check condition
19432 if cmp isn't SET with IF_THEN_ELSE src.
19434 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19436 PR tree-optimization/80158
19437 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
19438 replacing a candidate statement, also replace it for the
19439 candidate's alternate interpretation.
19440 (replace_rhs_if_not_dup): Likewise.
19441 (replace_one_candidate): Likewise.
19443 2017-03-24 Richard Biener <rguenther@suse.de>
19445 PR tree-optimization/80167
19446 * graphite-isl-ast-to-gimple.c
19447 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
19449 (translate_isl_ast_to_gimple::get_rename): Likewise.
19451 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
19453 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
19454 handling of certain combinations of target options, including the
19455 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
19456 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
19458 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19461 * config/arm/arm.md (*load_multiple): Add reload_completed to
19462 matching condition.
19464 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19465 Richard Biener <rguenth@suse.de>
19467 PR tree-optimization/79908
19468 PR tree-optimization/80136
19469 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19470 been cast away, gimplify_and_add suffices.
19472 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
19474 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
19476 2017-03-23 Richard Biener <rguenther@suse.de>
19478 PR tree-optimization/80032
19479 * gimplify.c (gimple_push_cleanup): Forced unconditional
19480 cleanups still have to go to the conditional_cleanups
19483 2017-03-22 Jakub Jelinek <jakub@redhat.com>
19485 PR tree-optimization/80072
19486 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
19488 (next_operand_entry_id): Change type to unsigned int.
19489 (sort_by_operand_rank): Make sure to return the right return value
19490 even if unsigned fields are bigger than INT_MAX.
19491 (struct oecount): Change cnt and id type to unsigned int.
19492 (oecount_hasher::equal): Formatting fix.
19493 (oecount_cmp): Make sure to return the right return value
19494 even if unsigned fields are bigger than INT_MAX.
19495 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
19498 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
19499 TREE_READONLY on result if writing it more than once.
19502 * doc/invoke.texi (-fsanitize=thread): Document that with
19503 -fnon-call-exceptions atomics are not able to throw
19507 * tsan.c: Include tree-eh.h.
19508 (instrument_builtin_call): Call maybe_clean_eh_stmt or
19509 maybe_clean_or_replace_eh_stmt where needed.
19510 (instrument_memory_accesses): Add cfg_changed argument.
19511 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
19512 if it returned true.
19513 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
19515 PR rtl-optimization/63191
19516 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
19517 wrapper function, moved the whole old content into ...
19518 (ix86_delegitimize_address_1): ... this. New inline function.
19519 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
19520 true as last argument instead of ix86_delegitimize_address.
19522 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19524 * config/aarch64/aarch64.c (generic_branch_cost): Copy
19525 cortexa57_branch_cost.
19527 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
19529 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
19531 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19534 * doc/md.texi (Constraints): Document wA constraint.
19535 * config/rs6000/constraints.md (wA): New.
19536 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
19537 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
19538 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
19539 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
19541 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
19544 * gimplify.c (is_oacc_declared): New function.
19545 (oacc_default_clause): Use it to set default flags for acc declared
19546 variables inside parallel regions.
19547 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
19548 declared variables.
19549 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
19550 declare attribute to any decl as necessary.
19552 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
19555 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
19556 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
19557 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
19558 (arm_arch_lpae): This.
19559 * config/arm/arm.c (arm_arch7ve): Rename into ...
19560 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
19561 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
19564 2017-03-22 Martin Liska <mliska@suse.cz>
19567 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
19568 error message instead of an ICE.
19570 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19572 * doc/extend.texi (6.11 Additional Floating Types): Revise.
19574 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
19576 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19578 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19581 2017-03-21 Martin Sebor <msebor@redhat.com>
19583 * doc/extend.texi: Use "cannot" instead of "can't."
19584 * doc/hostconfig.texi: Same.
19585 * doc/install.texi: Same.
19586 * doc/invoke.texi: Same.
19587 * doc/loop.texi: Same.
19588 * doc/md.texi: Same.
19589 * doc/objc.texi: Same.
19590 * doc/rtl.texi: Same.
19591 * doc/tm.texi: Same.
19592 * doc/tm.texi.in: Same.
19593 * doc/trouble.texi: Same.
19595 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
19598 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
19599 (collect_checksum_attributes): Set it.
19600 (die_checksum_ordered): Use it.
19602 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19604 PR tree-optimization/79908
19605 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
19606 change: For a VA_ARG whose LHS has been cast away, use
19607 force_gimple_operand to construct the side effects.
19609 2017-03-21 David Malcolm <dmalcolm@redhat.com>
19611 PR translation/80001
19612 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
19613 more amenable to translation.
19614 (oacc_loop_auto_partitions): Likewise.
19616 2017-03-21 Marek Polacek <polacek@redhat.com>
19617 Martin Sebor <msebor@redhat.com>
19619 PR tree-optimization/80109
19620 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
19621 on INTEGRAL_TYPE_P.
19623 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19624 Segher Boessenkool <segher@kernel.crashing.org>
19627 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
19628 check reg_used_between_p between insn and one of succ or succ2
19629 depending on if succ is artificial insn not inserted into insn
19632 2017-03-21 Martin Liska <mliska@suse.cz>
19634 PR gcov-profile/80081
19635 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
19636 * doc/gcc.texi: Include gcov-dump stuff.
19637 * doc/gcov-dump.texi: New file.
19639 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
19641 PR rtl-optimization/79150
19642 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
19643 conditional jump, if the jump is the last insn of the loop.
19645 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19646 Richard Biener <rguenth@suse.de>
19648 PR tree-optimization/79908
19649 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
19650 been cast away, use force_gimple_operand to construct the side
19653 2017-03-21 Martin Liska <mliska@suse.cz>
19655 PR libfortran/79956
19656 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
19659 2017-03-21 Brad Spengler <spender@grsecurity.net>
19662 * plugin.c (htab_hash_plugin): New function.
19663 (add_new_plugin): Use it and adjust.
19664 (parse_plugin_arg_opt): Adjust.
19665 (init_one_plugin): Likewise.
19667 2017-03-21 Richard Biener <rguenther@suse.de>
19669 PR tree-optimization/80032
19670 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
19671 if set force the cleanup to happen unconditionally.
19672 (gimplify_target_expr): Push inserted clobbers with force_uncond
19673 to avoid them being removed by control-dependent DCE.
19675 2017-03-21 Richard Biener <rguenther@suse.de>
19677 PR tree-optimization/80122
19678 * tree-inline.c (copy_bb): Do not expans va-arg packs or
19679 va_arg_pack_len when the inlined call stmt requires pack
19681 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
19683 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19686 * tsan.c (instrument_builtin_call): If the memory model argument
19687 is not a constant, assume it is valid.
19690 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
19693 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19695 PR rtl-optimization/79910
19696 * combine.c (can_combine_p): Do not allow combining an I0 or I1
19697 if its dest is used by an insn before I2 (other than the combined
19698 insns themselves, which are properly handled already).
19700 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19703 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19705 * combine.c (record_used_regs): New static function.
19706 (try_combine): Handle situations where there is an additional
19707 instruction between I2 and I3 which needs to have a LOG_LINK
19711 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19713 * combine.c (try_combine): Delete redundant i1 test. Call
19714 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19716 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19719 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
19720 alternatives 13/14.
19722 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19724 PR tree-optimization/80054
19725 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
19726 the optimization if a PHI or any of its arguments is not dominated
19727 by the candidate's basis. Use gphi* rather than gimple* as
19729 (replace_profitable_candidates): Clean up a gimple* variable that
19730 should be a gphi* variable.
19732 2017-03-20 Martin Sebor <msebor@redhat.com>
19735 * doc/extend.texi (attribute constructor): Document present limitation.
19737 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
19740 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
19741 __POWER9_VECTOR__ #ifdef control, change template definition to
19742 use Power9-specific built-in function.
19743 (vec_any_eq): Likewise.
19744 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
19745 to control outcomes from this test.
19746 (vector_ae_<mode>p): For VEC_F modes, likewise.
19748 2017-03-20 Ian Lance Taylor <iant@google.com>
19750 * config/i386/i386.c (ix86_function_regparm): Save an extra
19751 register for -fsplit-stack with DECL_STATIC_CHAIN.
19753 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19756 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
19757 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
19759 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19761 * config/riscv/riscv.c (riscv_print_operand): Use "fence
19763 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
19766 2017-03-20 Marek Polacek <polacek@redhat.com>
19769 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
19771 2017-03-20 Richard Biener <rguenther@suse.de>
19773 PR tree-optimization/80113
19774 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
19775 allocate extra SSA name for PHI def.
19776 (add_close_phis_to_outer_loops): Likewise.
19777 (add_close_phis_to_merge_points): Likewise.
19778 (copy_loop_close_phi_args): Likewise.
19779 (copy_cond_phi_nodes): Likewise.
19781 2017-03-20 Martin Liska <mliska@suse.cz>
19783 PR middle-end/79753
19784 * tree-chkp.c (chkp_build_returned_bound): Do not build
19785 returned bounds for a LHS that's not a BOUNDED_P type.
19787 2017-03-20 Martin Liska <mliska@suse.cz>
19791 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
19792 COMPLEX_CST and VECTOR_CST.
19794 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19797 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
19798 target operand. A new splitter adds the clobber statement in case
19799 the target operand is dead anyway.
19801 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
19803 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
19804 to age-old versions of binutils and glibc.
19806 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
19808 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
19810 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19812 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
19814 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19816 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
19817 requirement for binutils 2.13.
19819 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19821 * combine.c (try_combine): Delete redundant i1 test. Call
19822 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19824 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
19826 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
19827 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
19829 <riscv64-*-elf>: Re-arrange section
19830 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
19831 <riscv32-*-linux>: Likewise.
19832 <riscv64-*-elf>: Likewise
19833 <riscv64-*-linux>: Likewise.
19835 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
19838 * aarch64.opt(verbose-cost-dump): Fix typo.
19840 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
19843 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
19844 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
19846 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19848 * reload.c (find_reloads): When reloading a nonoffsettable address,
19849 use RELOAD_OTHER for it and its address reloads.
19851 PR rtl-optimization/79910
19852 * combine.c (record_used_regs): New static function.
19853 (try_combine): Handle situations where there is an additional
19854 instruction between I2 and I3 which needs to have a LOG_LINK
19857 2017-03-17 Jeff Law <law@redhat.com>
19859 PR tree-optimization/71437
19860 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
19861 conditional in the hash table first.
19862 (vrp_dom_walker::before_dom_children): Extract condition from
19863 ASSERT_EXPR. Record condition, its inverion and any implied
19864 conditions as well.
19866 2017-03-17 Marek Polacek <polacek@redhat.com>
19867 Markus Trippelsdorf <markus@trippelsdorf.de>
19869 PR tree-optimization/80079
19870 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
19873 2017-03-17 Richard Biener <rguenther@suse.de>
19875 PR middle-end/80075
19876 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
19877 Properly verify the LHS before the RHS possibly claims to be
19879 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
19882 2017-03-17 Martin Jambor <mjambor@suse.cz>
19884 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
19885 (List of -O2 options): Likewise.
19886 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
19889 2017-03-17 Tom de Vries <tom@codesourcery.com>
19891 * gcov-dump.c (print_usage): Print bug_report_url.
19893 2017-03-17 Richard Biener <rguenther@suse.de>
19895 PR middle-end/80050
19896 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
19897 (parser::peek): Likewise.
19899 2017-03-17 Richard Biener <rguenther@suse.de>
19901 PR tree-optimization/80048
19902 * sese.c (free_sese_info): Properly release rename_map and
19903 copied_bb_map elements.
19905 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
19907 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
19908 Add linked-list forward and backlinks. Insert on
19909 construction, remove on destruction.
19910 (class pass_store_merging): Add m_stores_head field.
19911 (pass_store_merging::terminate_and_process_all_chains):
19912 Iterate over m_stores_head list.
19913 (pass_store_merging::terminate_all_aliasing_chains):
19915 (pass_store_merging::execute): Check for debug stmts first.
19916 Push new chains onto the m_stores_head stack.
19918 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
19921 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
19922 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
19923 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
19925 2017-03-16 Jeff Law <law@redhat.com>
19927 PR tree-optimization/71437
19928 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
19929 member function. Implementation moved into after_dom_children
19930 member function and into the threader's thread_outgoing_edges
19932 (dom_opt_dom_walker::after_dom_children): Simplify by moving
19933 some code into new thread_outgoing_edges.
19934 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
19935 definition. Simplify marker handling (do it here). Assume we always
19936 have the available expression and the const/copies tables.
19937 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
19939 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
19940 * tree-vrp.c (equiv_stack): No longer file scoped.
19941 (vrp_dom_walker): New class.
19942 (vrp_dom_walker::before_dom_children): New member function.
19943 (vrp_dom_walker::after_dom_children): Likewise.
19944 (identify_jump_threads): Setup domwalker. Use it rather than
19945 walking edges in a random order by hand. Simplify setup/finalization.
19946 (finalize_jump_threads): Remove.
19947 (vrp_finalize): Do not call identify_jump_threads here.
19948 (execute_vrp): Do it here instead and call thread_through_all_blocks
19951 PR tree-optimization/71437
19952 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
19954 (simplify_stmt_for_jump_threading): Add basic_block argument. All
19956 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
19957 (dom_opt_dom_walker::thread_across_edge): Remove
19958 handle_dominating_asserts argument. All callers changed.
19959 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
19960 changes. Remove calls to lhs_of_dominating_assert. Other
19961 uses of handle_dominating_asserts turn into unconditional code
19962 (simplify_control_stmt_condition_1): Likewise.
19963 (simplify_control_stmt_condition): Likewise.
19964 (thread_through_normal_block, thread_across_edge): Likewise.
19965 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
19966 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
19967 object if it is not an SSA_NAME.
19968 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
19969 before calling into the VRP specific simplifiers.
19970 (identify_jump_threads): Remove handle_dominating_asserts
19973 2017-03-16 Jakub Jelinek <jakub@redhat.com>
19976 * tree-diagnostic.c (default_tree_printer): No longer static.
19977 * tree-diagnostic.h (default_tree_printer): New prototype.
19979 2017-03-16 Tamar Christina <tamar.christina@arm.com>
19981 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
19982 Change ins into fmov.
19984 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19986 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
19987 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
19988 Use h_con constraint for operand 1.
19989 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
19990 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
19992 2017-03-15 Jeff Law <law@redhat.com>
19994 PR tree-optimization/71437
19995 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
19996 (record_temporary_equivalences): Use it.
19998 PR tree-optimization/71437
19999 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
20000 tree-ssa-scopedtables.
20001 (lookup_avail_expr, build_and_record_new_cond): Likewise.
20002 (record_conditions, record_cond, vuse_eq): Likewise.
20003 (record_edge_info): Adjust to API tweak of record_conditions.
20004 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
20005 (record_temporary_equivalences, optimize_stmt): Likewise.
20006 (eliminate_redundant_computations): Likewise.
20007 (record_equivalences_from_stmt): Likewise.
20008 * tree-ssa-scopedtables.c: Include options.h and params.h.
20009 (vuse_eq): New function, moved from tree-ssa-dom.c
20010 (build_and_record_new_cond): Likewise.
20011 (record_conditions): Likewise. Accept vector of conditions rather
20012 than edge_equivalence structure for first argument.
20013 for the first argument.
20014 (avail_exprs_stack::lookup_avail_expr): New member function, moved
20015 from tree-ssa-dom.c.
20016 (avail_exprs_stack::record_cond): Likewise.
20017 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
20018 from tree-ssa-dom.c.
20019 (avail_exprs_stack): Add new member functions lookup_avail_expr
20021 (record_conditions): Declare.
20023 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
20026 * lra-constraints.c (process_alt_operands): Increase reject for
20027 reloading an input/output operand.
20029 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
20032 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
20033 insns to convert from signed/unsigned char/short to IEEE 128-bit
20035 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
20037 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
20040 * config/i386/i386.c (ix86_vector_duplicate_value): Create
20041 subreg of inner mode for values already in registers.
20043 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
20045 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
20046 iteration reg is used after the loop.
20048 2017-03-14 Martin Sebor <msebor@redhat.com>
20050 PR tree-optimization/79800
20051 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
20052 precision in negative-positive range.
20053 (format_floating): Call non-const overload with adjusted precision.
20055 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
20058 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
20061 2017-03-14 Martin Sebor <msebor@redhat.com>
20063 PR middle-end/80020
20064 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
20065 * builtins.def (aligned_alloc): Use it.
20068 * Makefile.in (GTFILES): Add calls.c.
20069 * calls.c: Include "gt-calls.h".
20071 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
20073 PR rtl-optimization/79728
20074 * regs.h (struct target_regs): New field
20075 x_contains_allocatable_regs_of_mode.
20076 (contains_allocatable_regs_of_mode): New macro.
20077 * reginfo.c (init_reg_sets_1): Initialize it, and change
20078 contains_reg_of_mode so it includes global regs as well.
20079 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
20080 rather than contains_regs_of_mode.
20082 2017-03-14 Martin Liska <mliska@suse.cz>
20084 * doc/invoke.texi: Document options that can't be combined with
20085 -fcheck-pointer-bounds.
20087 2017-03-14 Martin Liska <mliska@suse.cz>
20089 PR middle-end/79831
20090 * doc/invoke.texi (-Wchkp): Document the option.
20092 2017-03-14 Martin Liska <mliska@suse.cz>
20094 * Makefile.in: Install gcov-dump.
20096 2017-03-14 Martin Liska <mliska@suse.cz>
20098 * multiple_target.c (expand_target_clones): Bail out for
20099 an invalid attribute.
20101 2017-03-14 Richard Biener <rguenther@suse.de>
20103 * alias.c (struct alias_set_entry): Pack properly.
20104 * cfgloop.h (struct loop): Likewise.
20105 * cse.c (struct set): Likewise.
20106 * ipa-utils.c (struct searchc_env): Likewise.
20107 * loop-invariant.c (struct invariant): Likewise.
20108 * lra-remat.c (struct cand): Likewise.
20109 * recog.c (struct change_t): Likewise.
20110 * rtl.h (struct address_info): Likewise.
20111 * symbol-summary.h (function_summary): Likewise.
20112 * tree-loop-distribution.c (struct partition): Likewise.
20113 * tree-object-size.c (struct object_size_info): Likewise.
20114 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
20115 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
20116 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
20117 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
20118 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
20119 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
20120 (struct _stmt_vec_info): Likewise.
20122 2017-03-14 Martin Liska <mliska@suse.cz>
20125 * multiple_target.c (create_dispatcher_calls): Check that
20126 a target can create a function dispatcher.
20128 2017-03-14 Martin Liska <mliska@suse.cz>
20131 * multiple_target.c (expand_target_clones): Drop local.local
20132 flag for default implementation.
20134 2017-03-14 Richard Biener <rguenther@suse.de>
20136 PR tree-optimization/80030
20137 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
20139 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
20141 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
20142 gcc_fallthrough() instead of __attribute__((fallthrough));
20144 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20146 * doc/gcc.texi: Remove "up" link to (DIR).
20147 * doc/gccint.texi: Ditto.
20149 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
20151 * doc/install.texi (Specific) <avr>: Remove reference to
20154 2017-03-13 Jeff Law <law@redhat.com>
20156 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
20157 attribute rather than comments.
20159 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
20160 match_scratch operand is highest.
20162 2017-03-13 Martin Liska <mliska@suse.cz>
20164 PR middle-end/78339
20165 * ipa-pure-const.c (warn_function_noreturn): If the declarations
20166 is a CHKP clone, use original declaration.
20168 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20170 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
20171 (arc_conditional_register_usage): Use a different allocation order
20172 when optimizing for size.
20173 * common/config/arc/arc-common.c (arc_option_optimization_table):
20174 Section anchors default on when optimizing for size.
20176 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20178 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
20180 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20182 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
20183 * config/arc/arc.md (cpu_facility): Add cd variant.
20184 (*movqi_insn): Add code density variant.
20185 (*movhi_insn): Likewise.
20186 (*movqi_insn): Likewise.
20187 (*addsi3_mixed): Likewise.
20188 (subsi3_insn): Likewise.
20190 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20192 * config/arc/arc.md (movsi_cond_exec): Update constraint.
20194 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
20196 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
20197 expressions with MINUS and UNARY ops.
20199 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20202 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
20204 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
20205 between vec_select and vector argument.
20206 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
20207 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
20208 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
20209 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
20210 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
20211 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
20213 2017-03-13 Richard Biener <rguenther@suse.de>
20216 * params.def (vect-max-peeling-for-alignment): Fix typo.
20218 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20220 * doc/install.texi (Specific) <mips-*-*>: Remove description of
20221 issue that only occurred with binutils below 2.18.
20223 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20225 * doc/install.texi (Specific) <cris-axis-elf>: No longer
20226 refer to binutils 2.11/2.12 minimum.
20228 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
20230 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
20231 ftp.kernel.org and simplify binutils requirement.
20233 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
20235 * doc/invoke.texi (Warning Options): Fix spelling of link-time
20237 (Optimize Options): Ditto. Also remove redundancy.
20239 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20241 PR translation/79848
20242 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
20244 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
20245 to G_ to avoid double translation.
20247 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20249 PR translation/79923
20250 * auto-profile.c (get_combined_location): Convert leading
20251 character of diagnostics to lower case and remove trailing period.
20252 (read_profile): Likewise for various diagnostics.
20253 * config/arm/arm.c (arm_option_override): Remove trailing period
20254 from various diagnostics.
20255 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
20256 (msp430_expand_delay_cycles): Likewise.
20258 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20261 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
20262 full command-line argument, rather than just "str".
20263 (aarch64_validate_march): Likewise.
20264 (aarch64_validate_mtune): Likewise.
20266 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
20268 PR rtl-optimization/78911
20269 * lra-assigns.c (must_not_spill_p): New function.
20270 (spill_for): Use it.
20272 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20274 PR tree-optimization/79981
20275 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
20276 ATOMIC_COMPARE_EXCHANGE ifn result.
20277 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
20278 IFN_ATOMIC_COMPARE_EXCHANGE.
20280 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20283 * opts.c (parse_sanitizer_options): Add missing question mark to
20284 "did you mean" message.
20286 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20288 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
20290 (VMULEUH_UNS): Likewise.
20291 (VMULOUB_UNS): Likewise.
20292 (VMULOUH_UNS): Likewise.
20293 * config/rs6000/rs6000.c (builtin_function_type): Remove
20294 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
20296 2017-03-10 David Malcolm <dmalcolm@redhat.com>
20299 * read-rtl-function.c (function_reader::read_rtx_operand): Update
20300 x with result of extra_parsing_for_operand_code_0.
20301 (function_reader::extra_parsing_for_operand_code_0): Convert
20302 return type from void to rtx, returning x. When reading
20303 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
20304 larger size containing struct block_symbol.
20306 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
20308 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
20309 -mfloat128-hardware without -m64.
20311 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
20314 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
20315 entries to the case statement that marks unsigned arguments to
20316 overloaded functions.
20318 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
20320 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
20321 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
20323 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
20326 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
20327 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
20329 2017-03-10 Martin Liska <mliska@suse.cz>
20333 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
20334 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
20337 2017-03-10 Olivier Hainque <hainque@adacore.com>
20339 * tree-switch-conversion (array_value_type): Start by resetting
20340 candidate type to it's main variant.
20342 2017-03-10 Jakub Jelinek <jakub@redhat.com>
20344 PR rtl-optimization/79909
20345 * combine.c (try_combine): Use simplify_replace_rtx on individual
20346 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
20347 of the whole CALL_INSN_FUNCTION_USAGE.
20349 PR tree-optimization/79972
20350 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
20351 get_range_info on SSA_NAMEs. Formatting fixes.
20353 2017-03-10 Richard Biener <rguenther@suse.de>
20354 Jakub Jelinek <jakub@redhat.com>
20356 PR tree-optimization/77975
20357 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
20358 edge to be constant.
20359 (get_val_for): For constant x return it. Formatting fix.
20360 (loop_niter_by_eval): Avoid pointless looping if the next iteration
20361 would use the same bases as the current one.
20363 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20365 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
20366 instead of vec_select for V1TImode.
20367 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
20369 (VSX_LE_128): Add V1TI to this mode iterator.
20370 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
20371 (*vsx_le_perm_store_<mode>): Likewise.
20372 (pre-reload splitter for VSX stores): Likewise.
20373 (post-reload splitter for VSX stores): Likewise.
20374 (*vsx_xxpermdi2_le_<mode>): Likewise.
20375 (*vsx_lxvd2x2_le_<mode>): Likewise.
20376 (*vsx_stxvd2x2_le_<mode>): Likewise.
20378 2017-03-09 Michael Eager <eager@eagercon.com>
20380 Correct failures with --enable-checking=yes,rtl.
20382 * config/microblaze/microblaze.c (microblaze_expand_shift):
20383 Replace GET_CODE test with CONST_INT_P and INTVAL test with
20384 test for const0_rtx.
20385 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
20386 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
20388 2017-03-09 Richard Biener <rguenther@suse.de>
20390 PR tree-optimization/79977
20391 * graphite-scop-detection.c (scop_detection::merge_sese):
20392 Handle the case of extra exits to blocks dominating the entry.
20394 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
20396 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20399 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
20401 PR rtl-optimization/79949
20402 * lra-constraints.c (process_alt_operands): Check memory when
20403 trying to predict a cycle. Print about the overall increase.
20405 2017-03-09 Richard Biener <rguenther@suse.de>
20407 PR middle-end/79971
20408 * gimple-expr.c (useless_type_conversion_p): Preserve
20409 TYPE_SATURATING for fixed-point types.
20411 2017-03-09 Richard Biener <rguenther@suse.de>
20414 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
20415 alignment of BLKmode params.
20417 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20420 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
20421 (VALL_NO_V2Q): Likewise.
20423 * config/aarch64/aarch64-simd.md
20424 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
20426 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
20427 VALL_NO_V2Q mode iterator.
20428 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
20430 2017-03-09 Martin Liska <mliska@suse.cz>
20432 PR tree-optimization/79631
20433 * tree-chkp-opt.c (chkp_is_constant_addr): Call
20434 tree_int_cst_sign_bit just for INTEGER constants.
20436 2017-03-09 Martin Liska <mliska@suse.cz>
20440 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
20443 2017-03-09 Marek Polacek <polacek@redhat.com>
20446 * tree.c (inchash::add_expr): Handle TREE_VEC.
20448 2017-03-09 Martin Liska <mliska@suse.cz>
20451 (chkp_narrow_size_and_offset): New function.
20452 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
20453 (void chkp_parse_bit_field_ref): New function.
20454 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
20455 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
20457 2017-03-09 Martin Liska <mliska@suse.cz>
20460 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
20461 (chkp_find_bounds_1): Remove gcc_unreachable.
20463 2017-03-09 Jakub Jelinek <jakub@redhat.com>
20466 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
20467 BUILT_IN_SYNC*, determine the access type from the size suffix and
20468 always build a MEM_REF with that type. Handle forgotten
20469 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
20472 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
20473 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
20474 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
20475 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
20476 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
20477 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
20478 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
20479 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
20480 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
20481 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
20482 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
20483 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
20484 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
20485 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
20486 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
20487 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
20488 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
20489 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
20490 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
20491 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
20492 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
20493 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
20494 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
20495 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
20496 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
20497 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
20498 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
20499 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
20500 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
20501 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
20502 definitions outside of __OPTIMIZE__ guarded section.
20505 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
20506 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
20507 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
20508 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
20511 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20513 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
20514 ("vfenez<mode>"): Add missing constraints.
20516 2017-03-08 Martin Sebor <msebor@redhat.com>
20519 * config/nds32/nds32.c (nds32_option_override):
20520 Fix misspelled diagnostic.
20522 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20525 * gimplify.c (gimplify_omp_for): Replace index var in outer
20526 taskloop statement with an artificial variable and add
20527 OMP_CLAUSE_PRIVATE clause for it.
20529 2017-03-08 Richard Biener <rguenther@suse.de>
20531 PR tree-optimization/79955
20532 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
20533 for accesses that are completely outside of the variable.
20535 2017-03-08 Andrew Haley <aph@redhat.com>
20537 PR tree-optimization/79943
20538 * tree-ssa-loop-split.c (compute_new_first_bound): When
20539 calculating the new upper bound, (END-BEG) should be added, not
20542 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20544 * config/avr/avr.md (setmemhi): Make sure match_dup
20545 operand number comes before match_scratch.
20547 2017-03-08 Richard Biener <rguenther@suse.de>
20549 PR tree-optimization/79920
20550 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
20551 with ncopies == 1 to ...
20552 (vect_transform_slp_perm_load): ... here. Properly compute
20553 all element loads by iterating VF times over the group. Do
20554 not handle ncopies (computed in a broken way) in
20555 vect_create_mask_and_perm.
20557 2017-03-08 Jakub Jelinek <jakub@redhat.com>
20560 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
20561 is a uniform vector, use uniform_vector_p return value instead of
20562 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
20564 2017-03-07 Marek Polacek <polacek@redhat.com>
20566 PR middle-end/79809
20567 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
20568 (alloca_call_type): Likewise.
20570 2017-03-07 Martin Liska <mliska@suse.cz>
20572 * gcov.c (process_args): Put comment to correct location.
20574 2017-03-07 Martin Liska <mliska@suse.cz>
20576 PR middle-end/68270
20577 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
20578 Use array_at_struct_end_p instead of DECL_CHAIN (field).
20579 (chkp_narrow_bounds_for_field): Likewise.
20580 (chkp_parse_array_and_component_ref): Pass one more argument to
20583 2017-03-07 Richard Biener <rguenther@suse.de>
20585 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
20588 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
20590 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
20591 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
20593 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20596 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
20597 to end of description.
20598 (PARAM_MAX_STORES_TO_MERGE): Likewise.
20600 2017-03-07 Jakub Jelinek <jakub@redhat.com>
20602 PR rtl-optimization/79901
20603 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
20605 (*avx512f_<code><mode>3<mask_name>): ... this.
20606 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
20607 iterator instead of VI8_AVX2_AVX512BW.
20609 PR rtl-optimization/79901
20610 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
20611 min/max expander, expand it using expand_vec_cond_expr.
20614 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
20617 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20620 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
20621 to void * for PCH reasons.
20622 * dwarf2out.c (output_loc_operands, output_die): Cast
20623 v.val_vec.array to unsigned char *.
20625 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
20628 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
20631 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
20633 PR rtl-optimization/79571
20634 * lra-constraints.c (process_alt_operands): Calculate static
20635 reject and subtract it from overall when only addresses will be
20638 2017-03-06 Julia Koval <julia.koval@intel.com>
20641 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
20642 incoming stack boundary to 128 for 64-bit targets.
20644 2017-03-06 Richard Biener <rguenther@suse.de>
20646 PR tree-optimization/79894
20647 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
20648 to NULL after folding it.
20650 2017-03-06 Richard Biener <rguenther@suse.de>
20652 PR tree-optimization/79824
20653 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
20654 check disabling peeling for gaps.
20656 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
20658 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20659 attributes): Document gettimeofday.
20661 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20663 * config/s390/s390.c (s390_option_override_internal): Set
20664 PARAM_MIN_VECT_LOOP_BOUND
20666 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
20668 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
20669 * config/s390/s390.md: Likewise.
20671 2017-03-06 Jakub Jelinek <jakub@redhat.com>
20674 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
20675 (<avx2_avx512>_perm<mode>): Rename to ...
20676 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
20678 (<avx512>_perm<mode>_mask): Rename to ...
20679 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
20681 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
20682 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
20683 instead of VI8F_256_512.
20684 (avx512f_perm<mode>): New define_expand.
20685 (avx512f_perm<mode>_mask): Likewise.
20686 (avx512f_perm<mode>_1<mask_name>): New define_insn.
20687 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
20689 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20691 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
20692 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
20694 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
20696 2017-03-06 Martin Liska <mliska@suse.cz>
20699 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
20700 when having a SSA NAME w/o VAR_DECL assigned to it.
20702 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20704 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
20705 msa_dpsub_<su>_d): Fix MODE for vec_select.
20707 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20709 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
20711 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
20713 2017-03-06 Richard Biener <rguenther@suse.de>
20715 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
20716 * plugin.c (register_plugin_info): Likewise.
20717 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
20719 2017-03-05 Jakub Jelinek <jakub@redhat.com>
20721 * config/i386/sse.md (sse_storehps, sse_storelps,
20722 avx_<castmode><avxsizesuffix>_<castmode>,
20723 avx512f_<castmode><avxsizesuffix>_<castmode>,
20724 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
20725 in condition that at least one operand is not a MEM.
20727 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20729 PR middle-end/79805
20730 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
20731 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
20733 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
20734 gimple_call_nothrow_p flag based on whether original builtin can throw.
20735 If it can, emit following stmts on the fallthrough edge.
20736 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
20737 don't create new bb if inserting just debug stmts on the edge, try to
20738 insert them on the fallthru bb or just reset debug stmts.
20740 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
20743 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
20744 restore recog_data (including the operand rtxes inside it) around
20745 the call to get_insn_template.
20747 2017-03-03 Martin Sebor <msebor@redhat.com>
20749 PR tree-optimization/79699
20750 * context.c (context::~context): Free MPFR caches to avoid
20751 a memory leak on program exit.
20753 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20755 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
20756 Use wide_int::ulow () instead of .elt (0).
20758 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20760 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
20761 (*pushxf): Limit oF constraint to 32bit targets and add oC
20762 constraint for 64bit targets.
20763 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
20764 (*pushdf): Change rmF constraint to rmC.
20766 2017-03-03 Martin Liska <mliska@suse.cz>
20768 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
20769 Remove unused variable.
20771 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20774 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
20775 is a memory operand, increase num_memory.
20776 (ix86_expand_args_builtin): Likewise.
20778 2017-03-03 Jan Hubicka <jh@suse.cz>
20781 * ipa-devirt.c (maybe_record_node): Properly handle
20782 __cxa_pure_virtual visibility.
20784 2017-03-03 Martin Liska <mliska@suse.cz>
20786 PR tree-optimization/79803
20787 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
20789 (pass_loop_prefetch::execute): Disabled optimization if an
20790 assumption about L1 cache size is not met.
20792 2017-03-03 Martin Liska <mliska@suse.cz>
20794 PR rtl-optimization/79574
20795 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
20796 (hash_scan_set): Likewise.
20797 (dump_hash_table): Likewise.
20798 (hoist_code): Likewise.
20800 2017-03-03 Richard Biener <rguenther@suse.de>
20802 * fixed-value.c (fixed_from_string): Restore use of elt (1)
20803 in place of uhigh ().
20804 (fixed_convert_from_real): Likewise.
20806 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20809 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
20811 2017-03-03 Richard Biener <rguenther@suse.de>
20813 PR middle-end/79818
20814 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
20815 TYPE_OVERFLOW_UNDEFINED check.
20817 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20819 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
20821 (vector_ae_<mode>_p): Likewise.
20822 (vector_nez_<mode>_p): Likewise.
20823 (vector_ne_v2di_p): Likewise.
20824 (vector_ae_v2di_p): Likewise.
20825 (vector_ne_<mode>_p): Likewise.
20826 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
20828 (vsx_tsqrt<mode>2_fe): Likewise.
20830 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
20833 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
20835 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20837 PR rtl-optimization/79780
20838 * cprop.c (one_cprop_pass): When second and further conditional trap
20839 in a single basic block is turned into an unconditional trap, turn it
20840 into a deleted note to avoid RTL verification failures.
20842 2017-03-02 Richard Biener <rguenther@suse.de>
20844 * fold-const.c (const_binop): Use ulow () instead of elt (0).
20846 2017-03-02 Richard Biener <rguenther@suse.de>
20848 PR tree-optimization/79345
20850 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
20851 param and abort the walk, returning -1 if it is hit.
20852 (walk_aliased_vdefs): Take a limit param and pass it on.
20853 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
20854 defaulting to 0 and return a signed int.
20855 * tree-ssa-uninit.c (struct check_defs_data): New struct.
20856 (check_defs): New helper.
20857 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
20858 about uninitialized memory.
20859 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
20860 bogus uninitialized warning.
20861 (fixed_convert_from_real): Likewise.
20863 2017-03-02 Bin Cheng <bin.cheng@arm.com>
20865 PR tree-optimization/66768
20866 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
20867 iv_use if base object can't be determined.
20869 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20871 PR tree-optimization/79345
20872 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
20873 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
20874 (get_pattern_stats): Initialize it.
20875 * genemit.c (gen_expand): Verify match_scratch numbers come after
20876 match_operand/match_dup numbers.
20877 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
20878 match_scratch numbers.
20879 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
20881 * config/s390/s390.md (trunctdsd2): Likewise.
20883 2017-03-02 Richard Biener <rguenther@suse.de>
20885 * wide-int.h (wide_int_storage::operator=): Implement in terms
20888 2017-03-02 Richard Biener <rguenther@suse.de>
20890 PR tree-optimization/79777
20891 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
20892 the to insert expression to sth existing.
20894 2017-03-01 Martin Sebor <msebor@redhat.com>
20896 PR middle-end/79692
20897 * gimple-ssa-sprintf.c
20898 (directive::known_width_and_precision): New function.
20899 (format_integer): Use it.
20900 (get_mpfr_format_length): Consider the full range of precision
20901 when computing %g output with the # flag. Set the likely byte
20902 count to 3 rather than 1 when precision is indeterminate.
20903 (format_floating): Correct the lower bound of precision.
20905 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20907 * doc/invoke.texi: Document default code model for 64-bit Linux.
20909 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20912 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
20913 udiv rather than div since input pattern is unsigned.
20915 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
20917 * config/i386/i386.c (print_reg): Warn for values of
20918 unsupported size in integer register.
20920 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
20923 * config/rs6000/predicates.md (current_file_function_operand): Do
20924 not allow self calls to be local if the function is replaceable.
20926 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
20929 * config/rs6000/altivec.h (vec_ctz and others): Change the
20930 preprocessor macro that controls conditional compilation from
20931 _ARCH_PWR9 to __POWER9_VECTOR__.
20932 (vec_all_ne): Change parameterization of __altivec_scalar_pred
20933 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
20934 control (instead of _ARCH_PWR9 control) so that template
20935 definition uses power9-specific function.
20936 (vec_any_eq): Likewise.
20937 (vec_all_ne): Change macro definition to use a power9-specific
20938 expansion under #ifdef __POWER9_VECTOR__ control (instead of
20939 _ARCH_PWR9 control).
20940 (vec_any_eq) Likewise.
20941 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
20942 expansion for CMPNEF to remove support for xvcmpnesp instruction.
20943 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
20944 support for xvcmpnedp instruction.
20945 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
20946 macro expansion so that Power9 implementation of vec_all_ne does
20947 not use the AltiVec predicate framework.
20948 (VCMPNEH_P): Likewise.
20949 (VCMPNEW_P): Likewise.
20950 (VCMPNED_P): Likewise.
20951 (VCMPNEFP_P): Likewise.
20952 (VCMPNEDP_P): Likewise.
20953 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
20954 implementation of vec_any_eq to not use AltiVec predicate
20956 (VCMPAEH_P): Likewise.
20957 (VCMPAEW_P): Likewise.
20958 (VCMPAED_P): Likewise.
20959 (VCMPAEFP_P): Likewise.
20960 (VCMPAEDP_P): Likewise.
20961 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
20962 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
20963 not use the AltiVec predicate framework.
20964 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
20965 of vec_any_eq to not use AltiVec predicate framework.
20966 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20967 support for predefined __POWER9_VECTOR__ macro to indicate that
20968 Power9 instruction selection is enabled.
20969 (altivec_overloaded_builtins): Remove extraneous
20970 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
20971 function argument types RS6000_BTI_bool_V16QI and
20972 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
20973 entry for overloaded function argument types RS6000_BTI_bool_V4SI
20974 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
20975 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
20976 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
20977 Power9 for implementations of vec_cmpne. Change the signature for
20978 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
20979 (representing vec_all_ne) to remove the previously described first
20980 argument of type RS6000_BTI_INTSI, as this was an artifact of
20981 reliance on the AltiVec predicate framework, which is no longer
20982 used in the implementation of these functions. Add
20983 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
20984 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
20985 since, unlike the AltiVec predicate framework implementation, we
20986 do not share function descriptors between vec_alle and vec_anyeq.
20987 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
20988 set of modes that receive special treatment even when
20989 TARGET_P9_VECTOR is true. The special treatment emits code that
20990 does not depend on Power9 instructions.
20991 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
20992 define_expand to not rely on AltiVec predicate framework.
20993 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20995 (vector_ne_v2di_p): Change this define_expand to not rely on
20996 AltiVec predicate framework.
20997 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
20999 (vector_ne_<mode>_p): Change this define_expand to not rely on
21000 AltiVec predicate framework.
21001 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
21003 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
21004 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
21005 define_insn pattern.
21006 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
21007 define_insn pattern because the xvcmpne<VSs>. instruction is not
21009 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
21010 instruction is not supported.
21012 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21014 * config/nvptx/nvptx.c: Include intl.h.
21016 2017-03-01 Martin Jambor <mjambor@suse.cz>
21019 * ipa-prop.h (ipa_bits): Removed field known.
21020 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
21021 to pointers. Adjusted their comments to warn about their sharing.
21022 (ipcp_transformation_summary): Change bits to a vector of pointers.
21023 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
21024 (ipa_get_ipa_bits_for_value): Declare.
21025 * tree-vrp.h (value_range): Mark as GTY((for_user)).
21026 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
21027 (ipa_bits_hash_table): Likewise.
21028 (ipa_vr_ggc_hash_traits): Likewise.
21029 (ipa_vr_hash_table): Likewise.
21030 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
21031 being pointers and vr_known being removed.
21032 (ipa_set_jf_unknown): Likewise.
21033 (ipa_get_ipa_bits_for_value): New function.
21034 (ipa_set_jfunc_bits): Likewise.
21035 (ipa_get_value_range): New overloaded functions.
21036 (ipa_set_jfunc_vr): Likewise.
21037 (ipa_compute_jump_functions_for_edge): Use the above functions to
21038 construct bits and vr parts of jump functions.
21039 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
21040 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21042 (ipcp_grow_transformations_if_necessary): Also allocate
21043 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
21045 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
21046 them. Fix too long lines.
21047 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
21048 vr_known being removed.
21049 (ipa_read_jump_function): Use new setter functions to construct bits
21050 and vr parts of jump functions or set them to NULL.
21051 (write_ipcp_transformation_info): Adjust for bits being pointers.
21052 (read_ipcp_transformation_info): Likewise.
21053 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
21055 Include gt-ipa-prop.h.
21056 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
21058 (ipcp_store_bits_results): Likewise.
21059 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
21060 Do not write to existing jump functions but use a temporary instead.
21062 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21065 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
21066 attempt to use its first operand as BIT_FIELD_REF base.
21068 2017-03-01 Richard Biener <rguenther@suse.de>
21070 PR middle-end/79721
21071 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
21072 interpolating formula in wrapping arithmetic.
21073 (chrec_apply): Convert chrec_evaluate return value to wanted type.
21075 2017-03-01 Jakub Jelinek <jakub@redhat.com>
21077 PR tree-optimization/79734
21078 * tree-vect-generic.c (expand_vector_condition): Optimize
21079 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
21080 Handle VEC_COND_EXPR where comparison has different inner width from
21081 type's inner width.
21083 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
21085 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
21086 markup, and similar issues. Remove @opindex entries for things
21087 that aren't options. Add missing -mmpy-option entries.
21089 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21091 PR tree-optimization/79737
21092 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
21093 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
21094 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
21095 instead of byte_size. Formatting fix.
21096 (shift_bytes_in_array_right): Formatting fix.
21098 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
21101 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
21102 condition on optimize for the leaf function test.
21104 2017-02-28 Martin Liska <mliska@suse.cz>
21107 * read-rtl-function.c (function_reader::handle_unknown_directive):
21108 Bail out when one uses -flto.
21110 2017-02-28 Martin Liska <mliska@suse.cz>
21112 * common.opt: Replace space with tabular for options of <number>
21114 * config/i386/i386.opt: Show <number> value for
21115 -mlarge-data-threshold.
21116 * opts.c (print_filtered_help): Do not display number in hexadecimal
21119 2017-02-28 Martin Liska <mliska@suse.cz>
21121 * common.opt: Fix --help=option -Q for options which are of
21124 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
21126 * config/i386/i386.c (print_reg): Error out for values
21127 of 8-bit size in invalid integer register.
21129 2017-02-28 Martin Sebor <msebor@redhat.com>
21131 PR tree-optimization/79691
21132 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
21134 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21137 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
21138 gcc_unreachable with output_operand_lossage.
21140 2017-02-28 Richard Biener <rguenther@suse.de>
21142 PR tree-optimization/79740
21143 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
21145 (visit_nary_op): Insert the nary into the hashtable if we
21146 pattern-matched sth.
21147 * tree-ssa-pre.c (eliminate_insert): Robustify.
21149 2017-02-28 Richard Biener <rguenther@suse.de>
21151 PR middle-end/79731
21152 * fold-const.c (decode_field_reference): Reject out-of-bound
21155 2017-02-28 Jakub Jelinek <jakub@redhat.com>
21157 * config/i386/i386.c: Include intl.h.
21158 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
21159 instead of just cond ? "..." : "...".
21160 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
21161 * coverage.c (read_counts_file): Likewise.
21162 * omp-offload.c: Include intl.h.
21163 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
21164 of just cond ? "..." : "...".
21165 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
21166 of just cond ? "..." : "...".
21168 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
21171 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
21173 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
21174 'tune for' CPU name.
21175 * config/arm/arm-cpu-data.h: Regenerated.
21177 2017-02-28 Richard Biener <rguenther@suse.de>
21179 PR tree-optimization/79732
21180 * tree-inline.c (expand_call_inline): Do not shadow var.
21182 2017-02-28 Richard Biener <rguenther@suse.de>
21184 PR tree-optimization/79723
21185 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
21186 address-space properly.
21188 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
21190 * doc/optinfo.texi (Optimization groups): Fix option used for
21192 * doc/invoke.texi (-fopt-info): Document "omp".
21193 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
21194 (OPTGROUP_ALL): Add OPTGROUP_OMP.
21195 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
21196 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
21197 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
21199 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
21201 * dumpfile.c (optgroup_options): Instead of "openmp", associate
21202 OPTGROUP_OMP with "omp".
21204 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
21207 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
21208 for arithmetic shift of unsigned V2DI.
21210 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
21212 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
21213 arc/linux.h headers.
21214 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
21215 (LINK_SPEC): Likewise.
21216 (ARC_TLS_EXTRA_START_SPEC): Likewise.
21217 (EXTRA_SPECS): Likewise.
21218 (STARTFILE_SPEC): Likewise.
21219 (ENDFILE_SPEC): Likewise.
21220 (LIB_SPEC): Likewise.
21221 (TARGET_SDATA_DEFAULT): Likewise.
21222 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
21223 (MULTILIB_DEFAULTS): Likewise.
21224 (DWARF2_UNWIND_INFO): Likewise.
21225 * config/arc/big.h: New file.
21226 * config/arc/elf.h: Likewise.
21227 * config/arc/linux.h: Likewise.
21228 * config/arc/t-uClibc: Remove.
21230 2017-02-27 Bin Cheng <bin.cheng@arm.com>
21232 PR tree-optimization/77536
21233 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
21234 (tree_transform_and_unroll_loop): Use above function to compute the
21235 estimated niter of unrolled loop and use it when scaling profile.
21236 Also use count info rather than frequency if it's non-zero.
21237 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
21238 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
21239 (vect_transform_loop): Call above function.
21241 2017-02-27 Richard Biener <rguenther@suse.de>
21243 PR tree-optimization/45397
21244 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
21245 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
21246 (visit_nary_op): Add pattern matching for CSEing sign-changed
21247 or truncated operations with wider ones.
21249 2017-02-27 Richard Biener <rguenther@suse.de>
21251 PR tree-optimization/79690
21252 * tree-vect-stmts.c (vectorizable_store): Use vector type
21253 built from the DR with address-space.
21255 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
21257 * doc/invoke.texi (Optimize Options): Refine the description
21258 of asan-use-after-return.
21260 2017-02-25 Alan Modra <amodra@gmail.com>
21262 PR rtl-optimization/79584
21263 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
21264 base, not ad->base_term, the reg within base. Remove assertion
21265 that ad->base == ad->base_term. Replace gen_int_mode using
21266 bogus mode with const0_rtx.
21268 2017-02-25 Jakub Jelinek <jakub@redhat.com>
21270 PR middle-end/79396
21271 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
21272 FMA_EXPR like tcc_binary or tcc_unary.
21274 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
21277 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
21279 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
21280 (output_loc_operands): Handle DW_OP_call_ref and
21281 DW_OP_GNU_variable_value.
21282 (struct variable_value_struct): New type.
21283 (struct variable_value_hasher): Likewise.
21284 (variable_value_hash): New variable.
21285 (string_types): Remove.
21286 (copy_loc_descr): New function.
21287 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
21288 (prepend_loc_descr_to_each): New function.
21289 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
21290 instead of add_loc_descr_to_each if the first argument is single
21291 location list and the second has multiple.
21292 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
21293 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
21294 when looking for variable value which doesn't have other location info.
21295 (loc_list_from_tree): Formatting fix.
21296 (gen_array_type_die): Simplify DW_AT_string_length handling.
21297 (adjust_string_types): Remove.
21298 (gen_subprogram_die): Don't call adjust_string_types nor test/set
21299 string_types. Call resolve_variable_values.
21300 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
21301 (resolve_addr_in_expr): Likewise. Add A argument.
21302 (copy_deref_exprloc): Remove deref argument. Adjust for the
21303 original expression being DW_OP_GNU_variable_value with optionally
21304 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
21305 optionally after it.
21306 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
21307 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
21308 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
21309 (variable_value_hasher::hash, variable_value_hasher::equal): New
21311 (resolve_variable_value_in_expr, resolve_variable_value,
21312 resolve_variable_values, note_variable_value_in_expr,
21313 note_variable_value): New functions.
21314 (dwarf2out_early_finish): Call note_variable_value on all toplevel
21317 2017-02-24 Jakub Jelinek <jakub@redhat.com>
21320 * opts.h (handle_generated_option): Add GENERATED_P argument.
21321 * opts-common.c (handle_option): Adjust function comment.
21322 (handle_generated_option): Add GENERATED_P argument, pass it to
21324 (control_warning_option): Pass false to handle_generated_option
21326 * opts.c (maybe_default_option): Pass true to handle_generated_option
21328 * optc-gen.awk: Likewise.
21330 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21332 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
21333 a REG, look at the REG it is a SUBREG of.
21334 (splitter for cmpeqsi_t): Ditto.
21336 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21338 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
21339 the special USEs with the pattern of the insn, not the insn itself.
21341 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
21344 * doc/invoke.texi: Document -mload-store-pairs.
21346 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21347 Sandra Loosemore <sandra@codesourcery.com>
21349 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
21350 argument isn't a CONST_INT.
21351 (nios2_alternate_compare_const): Assert op is a CONST_INT.
21352 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
21353 (nios2_validate_compare): Bypass alternate compare logic if *op2
21354 is not a CONST_INT.
21355 (ldstwm_operation_p): Return false if first_base is not a REG or
21356 if first_offset is not a CONST_INT.
21358 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21360 * config/cris/cris.md: Use correct operand in a define_peephole2.
21362 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21364 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
21366 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
21368 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
21369 this_insn if it is an INSN or JUMP_INSN.
21370 (force_offsettable): Look at base, not at addr.
21371 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
21372 on things that aren't necessarily CONST_INTs.
21374 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
21376 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
21377 -mfpmath=sse is the default also for x86-32 targets with SSE2
21378 instruction set when @option{-ffast-math} is enabled
21380 2017-02-24 Jeff Law <law@redhat.com>
21382 PR rtl-optimizatoin/79286
21383 * ira.c (update_equiv_regs): Drop may_trap_p exception to
21386 2017-02-24 Richard Biener <rguenther@suse.de>
21388 PR tree-optimization/79389
21389 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
21392 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
21394 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
21395 function comment to reflect reality.
21396 (loop_exits_before_overflow): Fix typo in function description.
21398 2017-02-24 Richard Biener <rguenther@suse.de>
21400 PR tree-optimization/79389
21401 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
21402 properly that a threading opportunity exists. Detect conditional
21403 copy/constant propagation opportunities.
21405 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
21407 * config/visium/visium.md (type): Add trap.
21408 (b): New mode attribute.
21409 (*btst): Rename into...
21410 (*btst<mode>): ...this and adjust.
21411 (*cbranchsi4_btst_insn): Rename into...
21412 (*cbranch<mode>4_btst_insn): ...this and adjust.
21413 (trap): New define_insn.
21415 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21417 PR tree-optimization/79389
21418 * ifcvt.c (struct noce_if_info): Add rev_cond field.
21419 (noce_reversed_cond_code): New function.
21420 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
21421 reversed_comparison_code. Formatting fix.
21422 (noce_try_store_flag): Test rev_cond != NULL in addition to
21423 reversed_comparison_code.
21424 (noce_try_store_flag_constants): Likewise.
21425 (noce_try_store_flag_mask): Likewise.
21426 (noce_try_addcc): Use rev_cond if non-NULL instead of
21427 reversed_comparison_code.
21428 (noce_try_cmove_arith): Likewise. Formatting fixes.
21429 (noce_try_minmax, noce_try_abs): Clear rev_cond.
21430 (noce_find_if_block): Initialize rev_cond.
21431 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
21432 instead of false as last argument never attempt to reverse it
21435 2017-02-23 Bin Cheng <bin.cheng@arm.com>
21437 PR tree-optimization/79663
21438 * tree-predcom.c (combine_chains): Process refs in reverse order
21439 only for ZERO length chains, and add explaining comment.
21441 2017-02-23 Jeff Law <law@redhat.com>
21443 PR tree-optimization/79578
21444 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
21445 in call to operand_equal_p.
21447 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
21450 * config/i386/cpuid.h: Fix another undefined behavior.
21452 2017-02-23 Richard Biener <rguenther@suse.de>
21454 PR tree-optimization/79683
21455 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
21456 vector types for data-refs.
21458 2017-02-23 Martin Liska <mliska@suse.cz>
21460 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
21462 2017-02-23 Jakub Jelinek <jakub@redhat.com>
21464 PR middle-end/79665
21465 * internal-fn.c (get_range_pos_neg): Moved to ...
21466 * tree.c (get_range_pos_neg): ... here. No longer static.
21467 * tree.h (get_range_pos_neg): New prototype.
21468 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
21469 are known to be in between 0 and signed maximum inclusive, try to
21470 expand both unsigned and signed divmod and use the cheaper one from
21473 2017-02-22 Jeff Law <law@redhat.com>
21475 PR tree-optimization/79578
21476 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
21477 to compare base operands.
21479 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
21482 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
21483 gpc_reg_operand instead of fpr_reg_operand.
21485 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
21487 * config/mips/mips.c (mips_return_in_memory): Force FP
21488 vector types to be returned in memory for o32 ABI.
21490 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21492 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
21493 instead of DW_TAG_member for static data member declarations and don't
21494 set no_linkage_name for static inline data members.
21495 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
21498 2017-02-22 Martin Liska <mliska@suse.cz>
21500 * doc/invoke.texi: Replace inequality signs with square brackets
21503 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21505 PR tree-optimization/68644
21506 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
21508 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
21511 * lra-constraints.c (simplify_operand_subreg): Handle
21512 WORD_REGISTER_OPERATIONS targets.
21514 2017-02-22 Jakub Jelinek <jakub@redhat.com>
21517 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
21518 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
21519 elimination by swapping fld*.
21521 2017-02-22 Richard Biener <rguenther@suse.de>
21523 PR tree-optimization/79673
21524 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
21525 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
21526 irrelevant address-space qualifiers and avoiding a
21527 ADDR_SPACE_CONVERT_EXPR from fold_convert.
21529 2017-02-22 Richard Biener <rguenther@suse.de>
21531 PR tree-optimization/79666
21532 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
21533 to not symbolically negate if that may introduce undefined
21536 2017-02-22 Martin Liska <mliska@suse.cz>
21539 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
21540 * data-streamer-out.c (streamer_write_gcov_count_stream):
21542 * value-prof.c (stream_out_histogram_value): Make assert more
21543 precise based on type of counter.
21545 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
21548 * config/i386/i386.md (standard_x87sse_constant_load splitter):
21549 Use nonimmediate_operand instead of memory_operand for operand 1.
21550 (float-extend standard_x87sse_constant_load splitter): Ditto.
21552 2017-02-21 Jeff Law <law@redhat.com>
21554 PR tree-optimization/79621
21555 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
21556 blocks with edges to themselves.
21558 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21561 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
21562 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
21563 Use gimple_call_builtin_p.
21566 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
21567 on temporarily removed DEBUG_INSNs.
21569 PR tree-optimization/79649
21570 * tree-loop-distribution.c (classify_partition): Give up on
21571 non-generic address space loads/stores.
21573 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
21575 * doc/loop.texi (Loop manipulation): Remove nonexistent
21576 tree_ssa_loop_version from the documentation.
21577 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
21579 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21582 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
21583 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
21584 * config/rs6000/rs6000.c: Include except.h.
21585 (rs6000_expand_split_stack_prologue): Call
21586 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
21588 2017-02-21 Martin Jambor <mjambor@suse.cz>
21591 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
21592 have been analyzed.
21594 2017-02-21 Martin Jambor <mjambor@suse.cz>
21596 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
21597 for backward compatibility only.
21598 * doc/invoke.texi (Option Summary): Remove all references to
21599 -fipa-cp-alignment.
21601 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
21605 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21607 * lra-constraints.c (curr_insn_transform): Handle
21608 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21610 2017-02-21 Martin Liska <mliska@suse.cz>
21612 * config/i386/i386.opt: Replace -masm-dialect with -masm.
21614 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
21616 PR translation/79638
21617 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
21619 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
21622 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
21623 (arm_function_ok_for_sibcall): Return false for an indirect call by
21624 descriptor if all the argument registers are used.
21625 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
21626 alignment of the function.
21628 2017-02-21 Jakub Jelinek <jakub@redhat.com>
21630 PR tree-optimization/61441
21631 * simplify-rtx.c (simplify_const_unary_operation): For
21632 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
21633 the sNaN unmodified.
21635 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21637 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21638 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21639 instead of SYSTEM_HEADER_DIR.
21641 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
21642 Martin Liška <mliska@suse.cz>
21644 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
21645 Fix typos and grammar, use active voice, and clarify.
21647 2017-02-20 Marek Polacek <polacek@redhat.com>
21649 PR middle-end/79537
21650 * gimplify.c (gimplify_expr): Handle unused *&&L;.
21653 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
21655 2017-02-20 Jakub Jelinek <jakub@redhat.com>
21658 * config/i386/i386.c (ix86_expand_builtin): Handle
21659 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
21660 ix86_builtins_isa[fcode].isa as a requirement of those
21661 flags and any other flag in the bitmask.
21662 (ix86_init_mmx_sse_builtins): Use 0 instead of
21663 ~OPTION_MASK_ISA_64BIT as mask.
21664 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
21665 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
21666 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
21667 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
21669 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21672 * lra-constraints.c (split_reg): Check requested split mode
21673 is supported by the register.
21675 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21677 * lra-constraints.c (simplify_operand_subreg): Remove early
21680 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21683 * lra-constraints.c (curr_insn_transform): Tighten condition
21684 for converting SUBREG reloads from OP_OUT to OP_INOUT.
21686 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21689 * lra-constraints.c (curr_insn_transform): Handle
21690 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21692 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
21695 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
21697 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
21699 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
21702 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
21705 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21707 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
21710 2017-02-18 Jakub Jelinek <jakub@redhat.com>
21712 * final.c (last_columnnum, override_columnnum): New variables.
21713 (final_start_function): Set last_columnnum, pass it to begin_prologue
21714 hook and pass 0 to dwarf2out_begin_prologue.
21715 (final_scan_insn): Update override_columnnum. Pass last_columnnum
21716 to source_line debug hook.
21717 (notice_source_line): Compute last_columnnum and for debug_column_info
21718 return true on column changes.
21719 * debug.h (struct gcc_debug_hooks): Add column argument to
21720 source_line and begin_prologue hooks.
21721 (debug_nothing_int_charstar_int_bool): Remove prototype.
21722 (debug_nothing_int_int_charstar,
21723 debug_nothing_int_int_charstar_int_bool): New prototypes.
21724 (dwarf2out_begin_prologue): Add column argument.
21725 * debug.c (do_nothing_debug_hooks): Adjust source_line and
21726 begin_prologue hooks.
21727 (debug_nothing_int_charstar_int_bool): Remove.
21728 (debug_nothing_int_int_charstar,
21729 debug_nothing_int_int_charstar_int_bool): New functions.
21730 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
21731 through to dwarf2out_source_line.
21732 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
21733 (dwarf2out_source_line): Add column argument, emit it if requested.
21734 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
21736 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21737 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21738 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
21739 through to dwarf2out_begin_prologue.
21740 (vmsdbgout_source_line): Add column argument, pass it through to
21741 dwarf2out_source_line.
21742 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
21743 dbxout_source_line caller.
21744 (dbxout_source_line): Add column argument.
21746 * common.opt (gno-column-info, gcolumn-info): New options.
21747 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
21748 (check_die): Also test for multiple DW_AT_decl_column attributes.
21749 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
21750 DW_AT_decl_column if requested.
21751 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
21753 (gen_variable_die): Likewise.
21754 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
21755 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
21758 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
21759 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
21760 (ix86_handle_option): Handle OPT_m3dnowa.
21761 * doc/invoke.texi (-m3dnowa): Document.
21762 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
21763 -m3dnowa instead of -m3dnow -march=athlon.
21766 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
21767 instead of gcc_assert for K, r and R code checks. Formatting fixes.
21769 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21772 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
21773 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
21774 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
21775 generator for vsx_xxpermdi_<mode>_be.
21776 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
21777 force big-endian semantics.
21778 (vsx_xxpermdi_<mode>_be): New define_expand with same
21779 implementation as previous version of vsx_xxpermdi_<mode>.
21781 2017-02-17 Jakub Jelinek <jakub@redhat.com>
21783 PR tree-optimization/79327
21784 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
21785 variable, its initialization and use.
21787 2017-02-17 Julia Koval <julia.koval@intel.com>
21789 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
21790 (OPTION_MASK_ISA_PKU_UNSET): New.
21791 (ix86_handle_option): Handle -mrdpid.
21792 * config/i386/cpuid.h (bit_RDPID): New.
21793 * config/i386/driver-i386.c (host_detect_local_cpu):
21794 Detect RDPID feature.
21795 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
21796 * config/i386/i386-c.c (ix86_target_macros_internal):
21798 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
21799 (ix86_valid_target_attribute_inner_p): Add "rdpid".
21800 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
21801 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
21802 * config/i386/i386.md (define_insn "rdpid"): New.
21803 * config/i386/i386.opt Add -mrdpid.
21804 * config/i386/immintrin.h (_rdpid_u32): New.
21806 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
21808 PR rtl-optimization/79541
21809 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
21810 instead of transforming it into USE.
21812 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
21814 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
21815 If HONOR_SNANS (SFmode) force the input to a register.
21816 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
21817 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
21818 an frsp or similar insn.
21820 2017-02-17 Martin Liska <mliska@suse.cz>
21822 PR rtl-optimization/79577
21823 * params.def (selsched-max-sched-times): Increase minimum to 1.
21825 2017-02-17 Martin Liska <mliska@suse.cz>
21827 PR rtl-optimization/79574
21828 * gcse.c (want_to_gcse_p): Prevent integer overflow.
21830 2017-02-17 Martin Liska <mliska@suse.cz>
21832 PR tree-optimization/79529
21833 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
21834 ssa_defined_default_def_p to handle cases which are implicitly
21836 * tree-ssa.c (ssa_defined_default_def_p): New function.
21837 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
21838 which are implicitly defined.
21839 * tree-ssa.h (ssa_defined_default_def_p): Declare.
21841 2017-02-17 Richard Biener <rguenther@suse.de>
21843 PR middle-end/79576
21844 * params.def (max-ssa-name-query-depth): Limit to 10.
21846 2017-02-17 Richard Biener <rguenther@suse.de>
21848 PR tree-optimization/79552
21849 * tree-ssa-structalias.c (visit_loadstore): Properly verify
21852 2017-02-17 Richard Biener <rguenther@suse.de>
21855 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
21857 2017-02-17 Marek Polacek <polacek@redhat.com>
21859 PR middle-end/79536
21860 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
21861 (fold_negate_expr): New wrapper.
21863 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
21865 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
21866 Correct terminology and de-emphasize pre-standard behavior.
21868 2017-02-16 Alan Modra <amodra@gmail.com>
21870 PR rtl-optimization/79286
21871 * ira.c (def_dominates_uses): New function.
21872 (update_equiv_regs): Don't create an equivalence for insns that
21873 may trap where the register def does not dominate the use.
21875 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
21877 PR rtl-optimization/78127
21878 * lra.c (lra): Call lra_eliminate before finish the loop after
21881 2017-02-16 Richard Biener <rguenther@suse.de>
21883 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
21885 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
21886 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
21887 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
21888 (isl_val_int_from_wi): New function.
21889 (extract_affine_gmp): Rename to ...
21890 (extract_affine_wi): ... this, take a widest_int.
21891 (extract_affine_int): Just wrap extract_affine_wi.
21892 (add_param_constraints): Use isl_val_int_from_wi.
21893 (add_loop_constraints): Likewise, and extract_affine_wi.
21895 2017-02-15 Jeff Law <law@redhat.com>
21897 PR middle-end/79521
21898 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
21899 ira_init_register_move_cost_if_necessary.
21901 2017-02-15 Martin Sebor <msebor@redhat.com>
21903 PR middle-end/32003
21904 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
21905 removed in a prior commit.
21907 2017-02-15 Bin Cheng <bin.cheng@arm.com>
21909 PR tree-optimization/79347
21910 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
21911 counters during peeling.
21913 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
21915 * Makefile.in (site.exp): Remove "set ISLVER".
21917 2017-02-15 Jakub Jelinek <jakub@redhat.com>
21920 * real.c (real_from_integer): Call real_convert even for decimal.
21922 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
21925 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
21927 2017-02-14 Andrew Pinski <apinski@cavium.com>
21929 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
21930 cores and change the partno/implementer to be correct.
21931 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
21932 the 'B" as the implementer.
21933 * config/aarch64/aarch64-tune.md: Regenerate.
21935 2017-02-14 Carl Love <cel@us.ibm.com>
21937 * config/rs6000/rs6000.c: Add case statement entry to make the
21938 xvcvuxdsp built-in argument unsigned.
21939 * config/rs6000/vsx.md: Fix the source and return operand types so they
21940 match the instruction definitions from the ISA document. Fix typo
21941 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
21944 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
21947 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
21948 member early_clobber_alts.
21949 * lra-lives.c (reg_early_clobber_p): New.
21950 (process_bb_lives): Use it.
21951 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
21952 (debug_operand_data): Initialize early_clobber_alts.
21953 (setup_operand_alternative): Set up early_clobber_alts.
21954 (collect_non_operand_hard_regs): Ditto. Pass early clobber
21955 alternatives to new_insn_reg.
21956 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
21958 (lra_update_insn_regno_info): Pass the new arg.
21960 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21962 PR middle-end/79505
21963 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
21964 (new_oacc_loop_raw): Don't clear already cleared fields.
21967 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
21968 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
21969 _mm512_prefetch_i64gather_ps): New inline functions and macros.
21971 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
21974 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
21976 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
21979 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
21980 the extra instruction to the right place to store 128-bit constant
21983 2017-02-14 Martin Sebor <msebor@redhat.com>
21985 PR middle-end/79448
21986 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
21987 warning for strings of unknown length.
21989 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
21991 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
21993 2017-02-14 Jeff Law <law@redhat.com>
21996 * ira-costs.c (scan_one_insn): Initialize register move costs
21997 for pseudos seen in USE/CLOBBER insns.
21999 PR tree-optimization/79095
22000 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
22001 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
22002 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
22003 if the operands are known to be not equal, then the resulting range
22005 (intersect_ranges): If the new range is ~[0,0] and the old range is
22006 wide, then prefer ~[0,0].
22007 * tree-vrp.c (overflow_comparison_p_1): New function.
22008 (overflow_comparison_p): New function.
22009 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
22010 if NAME is used in an overflow test.
22011 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
22012 overflow check that can be expressed as an equality test, then adjust
22013 ops to be that equality test.
22015 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22017 * config/s390/s390-builtin-types.def: Remove flags argument.
22018 * config/s390/s390.c (s390_init_builtins): Likewise.
22020 2017-02-14 Martin Liska <mliska@suse.cz>
22022 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
22023 vector. Fix trailing white spaces.
22025 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
22027 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
22030 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22032 PR rtl-optimization/68664
22033 * config/arm/arm.c (arm_sched_can_speculate_insn):
22034 New function. Declare prototype.
22035 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22037 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22039 PR rtl-optimization/68664
22040 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
22042 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
22044 2017-02-14 Amit Pawar <amit.pawar@amd.com>
22046 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
22047 max skip bytes for function, loop and jump.
22049 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22051 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
22052 ABS_EXPR for gimple dump.
22054 2017-02-14 Jakub Jelinek <jakub@redhat.com>
22057 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
22059 PR tree-optimization/79408
22060 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
22061 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
22062 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
22063 also if rhs1 is INTEGER_CST.
22065 2017-02-14 Richard Biener <rguenther@suse.de>
22067 PR middle-end/79432
22068 * tree-into-ssa.c (insert_phi_nodes): When the function can
22069 have abnormal edges rewrite SSA names with broken use-def
22070 dominance out of SSA and register them for PHI insertion.
22072 2017-02-13 Martin Sebor <msebor@redhat.com>
22074 PR middle-end/79496
22075 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
22076 clearing info.nowrite flag when snprintf size argument is a range.
22078 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22080 * cprop.c (cprop_jump): Add missing space in string literal.
22081 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
22082 (get_constraint_for_component_ref): Likewise.
22083 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
22084 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
22085 * lra-constraints.c (process_alt_operands): Likewise.
22086 * ipa-inline.c (inline_small_functions): Likewise.
22087 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
22088 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22089 * trans-mem.c (diagnose_tm_1_op): Likewise.
22090 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
22091 (grid_parallel_clauses_gridifiable): Likewise.
22093 * config/nvptx/mkoffload.c (process): Add space in between
22096 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
22097 "MOD4_SSE_REGS" and "ALL_REGS".
22099 * spellcheck.c (test_data): Add , in between "foo" and "food".
22101 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22104 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
22105 boundary crossing check and subsequent code generation agree.
22107 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22109 * config/aarch64/aarch64.c (has_memory_op): Delete.
22110 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
22113 2017-02-13 Jakub Jelinek <jakub@redhat.com>
22115 PR rtl-optimization/79388
22116 PR rtl-optimization/79450
22117 * combine.c (distribute_notes): When removing TEM_INSN for which
22118 corresponding dest has last value recorded, invalidate that last
22121 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22123 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
22124 of explicit '@'. Add missing assembly comment marker on branch costs
22127 2017-02-13 Nathan Sidwell <nathan@acm.org>
22129 * gengtype-lex.l (<in_struct>): Add '/'.
22131 2017-02-13 Martin Liska <mliska@suse.cz>
22134 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
22136 2017-02-13 Richard Biener <rguenther@suse.de>
22138 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
22140 * configure: Re-generate.
22141 * config.in: Likewise.
22142 * graphite-dependences.c: Simplify as if
22143 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
22144 * graphite-isl-ast-to-gimple.c: Likewise.
22145 * graphite-optimize-isl.c: Likewise.
22146 * graphite-poly.c: Likewise.
22147 * graphite-sese-to-poly.c: Likewise.
22148 * graphite.h: Likewise.
22149 * toplev.c: Include isl/version.h and use isl_version () for
22150 printing the ISL version.
22151 * doc/install.texi: Update ISL requirement.
22153 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22155 * doc/standards.texi (Standards): Update reference to
22158 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
22160 * doc/extend.texi (Named Address Spaces): sourceware.org now
22162 * doc/install.texi (Binaries): Ditto.
22165 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22167 * doc/cpp.texi: Replace "stringify"/"stringification" with C
22168 standard terminology "stringize"/"stringizing" throughout.
22169 * doc/cppinternals.texi: Likewise.
22171 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
22173 * doc/extend.texi: Fix some spelling mistakes and typos.
22174 * doc/invoke.texi: Likewise.
22176 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22179 * params.def (inline-min-speedup) Change from 10 to 8.
22181 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22183 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
22186 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
22189 * ipa-inline-analysis.c (get_minimal_bb): New function.
22190 (record_modified): Use it.
22191 (remap_edge_change_prob): Handle also ancestor functions.
22193 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
22195 * doc/contrib.texi (Contributors): Remove broken link into
22196 the Mauve CVS repository.
22198 2017-02-11 Jakub Jelinek <jakub@redhat.com>
22200 PR middle-end/79454
22201 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
22202 result computation whenever lhs doesn't have vector mode, not
22203 just when it has BLKmode.
22205 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22207 * doc/makefile.texi (profiledbootstrap): Refer to the
22208 installation instructions only in textual form.
22210 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22213 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
22215 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
22217 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
22218 (Specific): Update mingw-w64 reference.
22220 (Specific): Remove broken link to Renesas RX processor.
22222 2017-02-10 Richard Biener <rguenther@suse.de>
22224 * toplev.c (process_options): Do not mention obsolete graphite
22225 options when printing sorry message about missing graphite support.
22226 Mention -floop-nest-optimize.
22228 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
22230 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
22231 (vtst_p16): Likewise.
22232 (vtstq_p8): Likewise.
22233 (vtstq_p16): Likewise.
22235 (vtstq_p64): Likewise.
22236 * config/arm/arm_neon.h (vgetq_lane_p64): New.
22237 (vset_lane_p64): New.
22238 (vsetq_lane_p64): New.
22240 2017-02-10 Jakub Jelinek <jakub@redhat.com>
22242 PR tree-optimization/79411
22243 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
22244 stmt operands are SSA_NAMEs used in abnormal phis.
22245 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
22248 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22251 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
22254 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
22256 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
22258 2017-02-09 Jakub Jelinek <jakub@redhat.com>
22260 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
22264 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
22265 not arbitrary TREE_CONSTANT.
22268 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
22269 "omp declare target link" attribute unless is_global_var.
22270 * omp-offload.c (find_link_var_op): Likewise.
22272 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
22273 Chung-Lin Tang <cltang@codesourcery.com>
22275 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
22277 (gimplify_adjust_omp_clauses): Don't delete TILE.
22278 (gimplify_omp_for): Deal with TILE.
22279 * internal-fn.c (expand_GOACC_TILE): New function.
22280 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
22282 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
22283 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
22285 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
22286 avoid DIV for outermost collapse var.
22287 (expand_oacc_for): Insert tile element loop as needed. Adjust.
22288 Remove out of date comments, fix whitespace.
22289 * omp-general.c (omp_extract_for_data): Deal with tiling.
22290 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
22291 adjust OLF_DIM_BASE value.
22292 (struct omp_for_data): Add tiling field.
22293 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
22294 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
22295 for auto loops. Remove default auto determining, moved to
22296 oacc_loop_fixed_partitions.
22297 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
22298 stmts, add e_mask field.
22299 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
22300 (oacc_thread_numbers): Use oacc_dim_call.
22301 (oacc_xform_tile): New.
22302 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
22303 (finish_oacc_loop): Adjust for ifns vector.
22304 (oacc_loop_discover_walk): Append loop abstraction sites to list,
22305 add case for GOACC_TILE fns.
22306 (oacc_loop_xform_loop): Delete.
22307 (oacc_loop_process): Iterate over call list directly, and add
22308 handling for GOACC_TILE fns.
22309 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
22311 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
22312 vector partitioning to outer loops. Assign 2 partitions to loops
22313 when available. Add TILE handling.
22314 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
22315 (execite_oacc_device_lower): Process GOACC_TILE fns,
22316 ignore unknown specs.
22317 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
22318 * tree.c (omp_clause_num_ops): Adjust TILE ops.
22319 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
22321 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
22323 * configure.ac (ACX_BUGURL): Update.
22324 * configure: Regenerate.
22326 2017-02-09 Richard Biener <rguenther@suse.de>
22328 PR tree-optimization/69823
22329 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
22330 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
22332 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22334 * config/arc/arc-c.def: Add __NPS400__ definition.
22335 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
22336 (TARGET_NPS400): Define.
22338 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
22340 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
22342 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
22343 pointer, arch_info.
22344 (arc_cpu_types): Fill the arch_info field with a pointer into the
22345 arc_arch_types table.
22346 (arc_selected_cpu): Declare.
22347 * config/arc/arc.c (arc_selected_cpu): Make global.
22348 (arc_selected_arch): Delete.
22349 (arc_base_cpu): Delete.
22350 (arc_override_options): Remove references to deleted variables,
22351 update access to arch information.
22352 (ARC_OPT): Update access to arch information.
22353 (ARC_OPTX): Likewise.
22354 * config/arc/arc.h (arc_base_cpu): Remove declaration.
22355 (TARGET_ARC600): Update access to arch information.
22356 (TARGET_ARC601): Likewise.
22357 (TARGET_ARC700): Likewise.
22358 (TARGET_EM): Likewise.
22359 (TARGET_HS): Likewise.
22360 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
22363 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
22366 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
22367 condition/operands for integer GE/LE/GEU/LEU operations.
22369 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
22371 PR translation/79397
22372 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
22375 2017-02-08 Martin Jambor <mjambor@suse.cz>
22378 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
22379 whether allocation happened.
22380 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
22381 nothing was allocated.
22383 2017-02-08 Jakub Jelinek <jakub@redhat.com>
22385 PR tree-optimization/79408
22386 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
22387 constant, but SSA_NAME with a known integer range, use the minimum
22388 of that range instead of op1 to determine if modulo can be replaced
22389 with its first operand.
22391 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22393 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
22395 2017-02-08 Richard Biener <rguenther@suse.de>
22397 PR tree-optimization/71824
22398 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22399 Check all loops contained in the merged region.
22401 2017-02-07 Andrew Pinski <apinski@cavium.com>
22403 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
22405 2017-02-07 Andrew Pinski <apinski@cavium.com>
22407 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
22408 (thunderxt88): Likewise.
22409 (thunderxt81): Disable LSE and change v8.1 to v8.
22410 (thunderxt83): Likewise.
22412 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22413 Richard Biener <rguenther@suse.de>
22415 PR middle-end/79399
22416 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
22417 type from int to size_t.
22418 * ira-costs.c (struct_costs_size): Change type from int to size_t.
22420 2017-02-07 Jakub Jelinek <jakub@redhat.com>
22422 PR rtl-optimization/79386
22423 * cprop.c (bypass_conditional_jumps): Initialize
22424 bypass_last_basic_block already before splitting bbs after
22425 unconditional traps...
22426 (bypass_conditional_jumps): ... rather than here.
22429 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
22430 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
22431 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
22432 fix -masm=intel patterns.
22434 2017-02-07 Richard Biener <rguenther@suse.de>
22436 PR tree-optimization/79256
22437 PR middle-end/79278
22438 * builtins.c (get_object_alignment_2): Use min_align_of_type
22439 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
22440 and ADJUST_FIELD_ALIGN.
22442 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
22444 * doc/tm.texi: Regenerate.
22445 * stor-layout.c (layout_decl): Adjust.
22446 (update_alignment_for_field): Likewise.
22447 (place_field): Likewise.
22448 (min_align_of_type): Likewise.
22449 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
22450 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
22451 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
22452 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
22453 * config/frv/frv.c (frv_adjust_field_align): Likewise.
22454 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
22455 * config/i386/i386.c (x86_field_alignment): Likewise.
22456 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
22457 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
22458 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
22459 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
22460 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
22461 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
22465 2017-01-30 Richard Biener <rguenther@suse.de>
22467 PR tree-optimization/79256
22468 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22469 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22472 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
22474 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
22475 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
22476 builtins to SImode and emit a zero-extend, if necessary.
22478 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22480 * docs/invoke.texi (RISC-V Options): Alphabetize.
22482 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22484 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
22487 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
22489 * config/riscv/riscv.c: New file.
22490 * common/config/riscv/riscv-common.c: Likewise.
22491 * config.gcc: Likewise.
22492 * config/riscv/constraints.md: Likewise.
22493 * config/riscv/elf.h: Likewise.
22494 * config/riscv/generic.md: Likewise.
22495 * config/riscv/linux.h: Likewise.
22496 * config/riscv/multilib-generator: Likewise.
22497 * config/riscv/peephole.md: Likewise.
22498 * config/riscv/pic.md: Likewise.
22499 * config/riscv/predicates.md: Likewise.
22500 * config/riscv/riscv-builtins.c: Likewise.
22501 * config/riscv/riscv-c.c: Likewise.
22502 * config/riscv/riscv-ftypes.def: Likewise.
22503 * config/riscv/riscv-modes.def: Likewise.
22504 * config/riscv/riscv-opts.h: Likewise.
22505 * config/riscv/riscv-protos.h: Likewise.
22506 * config/riscv/riscv.h: Likewise.
22507 * config/riscv/riscv.md: Likewise.
22508 * config/riscv/riscv.opt: Likewise.
22509 * config/riscv/sync.md: Likewise.
22510 * config/riscv/t-elf-multilib: Likewise.
22511 * config/riscv/t-linux: Likewise.
22512 * config/riscv/t-linux-multilib: Likewise.
22513 * config/riscv/t-riscv: Likewise.
22514 * configure.ac: Likewise.
22515 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
22516 Waterman as RISC-V maintainers.
22517 * doc/install.texi: Add RISC-V entries.
22518 * doc/invoke.texi: Add RISC-V options section.
22519 * doc/md.texi: Add RISC-V constraints section.
22520 * configure: Regenerated.
22522 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
22525 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
22526 false values to be constant vectors with all 0 or all 1 bits set.
22527 (vcondu<mode><mode>): Likewise.
22528 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
22530 (fpmask_comparison_operator): Update comment.
22531 (vecint_comparison_operator): New predicate.
22532 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
22533 vector conditionals when the true and false values are constant
22534 vectors with all 0 bits or all 1 bits set.
22536 2017-02-06 Martin Sebor <msebor@redhat.com>
22538 PR tree-optimization/79376
22539 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
22541 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
22543 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
22544 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
22545 to simplify split condition.
22547 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22549 * omp-expand.c (oxpand_omp_atomic_fetch_op,
22550 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
22553 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
22555 PR rtl-optimization/68664
22556 * target.def (can_speculate_insn): New hook.
22557 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
22558 * doc/tm.texi: Regenerate.
22559 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
22560 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
22561 (rs6000_sched_can_speculate_insn): New function.
22563 2017-02-06 Jakub Jelinek <jakub@redhat.com>
22565 PR tree-optimization/79284
22566 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
22567 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
22568 vectorizable_mask_load_store, vectorizable_operation,
22569 vect_is_simple_cond, get_same_sized_vectype): Use it instead
22570 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
22571 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
22572 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
22573 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
22574 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
22575 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
22576 is_gimple_assign (stmt). Replace another such test with
22577 is_gimple_assign (stmt).
22579 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22582 * config/avr/avr.c (rtl-iter.h): Include it.
22583 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
22584 (avr_legitimate_combined_insn): ...and implementation.
22586 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
22588 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
22589 * config/s390/s390.c (s390_const_operand_ok)
22590 (s390_canonicalize_comparison, s390_extract_part)
22591 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
22592 (s390_contiguous_bitmask_p, s390_rtx_costs)
22593 (legitimize_pic_address): Likewise.
22594 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
22595 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
22596 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
22597 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
22598 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
22600 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
22602 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
22603 REGNO($0) == REGNO($1).
22605 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22607 * config/s390/linux.h(SIZE_TYPE): Add comment.
22609 2017-02-06 Julian Brown <julian@codesourcery.com>
22610 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22611 Virendra Pathak <virendra.pathak@broadcom.com>
22613 * config/aarch64/aarch64-cores.def: Change the scheduler
22615 * config/aarch64/aarch64.md: Include thunderx2t99.md.
22616 * config/aarch64/thunderx2t99.md: New file.
22618 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22620 * doc/standards.texi (Go Language): Update link to language
22623 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22625 * tree-eh.c (lower_resx): Sanitize profile.
22626 (cleanup_empty_eh_move_lp): Likewise.
22628 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
22631 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
22633 * cfgloopmanip.h (loop_version): Update prototype.
22634 * modulo-sched.c (sms_schedule): Update call of loop_version.
22635 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
22636 * tree-parloops.c (gen_parallel_loop): Likewise.
22637 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
22638 * tree-ssa-loop-split.c (split_loop): Likewise.
22639 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
22640 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
22642 2017-02-05 Martin Liska <mliska@suse.cz>
22645 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
22647 (print_operand_address): Initialize a struct to zero.
22649 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22651 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
22652 garbage collector only in textual form.
22654 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
22656 * doc/extend.texi (x86 specific memory model extensions for
22657 transactional memory): Simplify a phrase.
22659 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
22662 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
22663 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
22664 (atomic_storedi_1): Likewise.
22666 2017-02-04 Jakub Jelinek <jakub@redhat.com>
22668 PR tree-optimization/79338
22669 * tree-parloops.c (gather_scalar_reductions): Don't call
22670 vect_analyze_loop_form for loop->inner before destroying loop's
22673 2017-02-03 Martin Sebor <msebor@redhat.com>
22675 PR tree-optimization/79327
22676 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
22677 when precision has resulted in leading zeros.
22678 (format_integer): Adjust the likely counter to assume an unknown
22679 argument that may be zero is non-zero.
22681 2017-02-03 Jason Merrill <jason@redhat.com>
22684 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
22685 avoid copying non-taken branch.
22687 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22689 PR tree-optimization/79340
22690 * tree-vect-loop.c (vectorizable_reduction): Release
22691 vec_defs elements after safe_splicing them into other vectors.
22694 PR tree-optimization/79327
22695 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
22696 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
22698 (format_integer): Use wide_int_to_tree instead of build_int_cst
22699 + to_?hwi. If argmin is NULL, just set argmin and argmax to
22700 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
22701 of shortest and longest sequence.
22703 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
22705 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
22706 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
22708 2017-02-03 Walter Lee <walt@tilera.com>
22711 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
22712 after initial stackframe link reg save.
22713 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
22715 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22718 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
22719 wu for stxssp alternative.
22721 2017-02-03 Martin Sebor <msebor@redhat.com>
22723 PR tree-optimization/79352
22724 * gimple-fold.c (get_range_strlen): Add argument.
22725 (get_range_strlen): Change return type to bool.
22726 (get_maxval_strlen): Pass in a dummy argument.
22727 * gimple-fold.h (get_range_strlen): Change return type to bool.
22728 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
22729 * tree.h (array_at_struct_end_p): Add argument.
22730 * tree.c (array_at_struct_end_p): Handle it.
22732 2017-02-03 Martin Liska <mliska@suse.cz>
22735 * multiple_target.c (create_dispatcher_calls): Redirect edge
22736 from a caller of a dispatcher.
22737 (expand_target_clones): Make the clones local.
22738 (ipa_target_clone): Do both target clones and resolvers.
22739 (ipa_dispatcher_calls): Remove the pass.
22740 (pass_dispatcher_calls::gate): Likewise.
22741 (make_pass_dispatcher_calls): Likewise.
22742 * passes.def (pass_target_clone): Put as very first IPA early
22745 2017-02-03 Martin Liska <mliska@suse.cz>
22747 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
22748 in case of a function with ifunc attribute.
22750 2017-02-03 Martin Liska <mliska@suse.cz>
22752 * cgraph.c (cgraph_node::dump): Dump function version info.
22753 * symtab.c (symtab_node::dump_base): Add missing new line.
22755 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22757 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
22758 (ifcombine_ifandif): Use it.
22760 2017-02-03 Martin Liska <mliska@suse.cz>
22762 * doc/invoke.texi: Document default value for
22763 use-after-scope-direct-emission-threshold.
22765 2017-02-03 Martin Liska <mliska@suse.cz>
22767 PR tree-optimization/79339
22768 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
22769 (format_floating): Likewise.
22771 2017-02-03 Martin Liska <mliska@suse.cz>
22774 * ipa-prop.c (ipa_node_params_t::insert): Remove current
22776 (ipa_node_params_t::remove): Likewise.
22777 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
22778 initialization from removed ipa_node_params_t::insert.
22779 (ipa_node_params::~ipa_node_params): Move from removed
22780 ipa_node_params_t::release.
22781 * symbol-summary.h (symbol_summary::m_released): New member.
22782 Do not release a summary twice. Do not allow to call finalizer
22783 for types of a summary that live in GGC memory.
22785 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22787 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
22790 2017-02-02 Martin Sebor <msebor@redhat.com>
22792 PR middle-end/79275
22793 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
22794 (format_string): Tighten up the range of output for non-constant
22795 strings and correct the expected range for wide non-constant strings.
22797 2017-02-02 Martin Sebor <msebor@redhat.com>
22799 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
22801 PR middle-end/32003
22802 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
22804 (-fdump-tree-@var): Add to index and document how to come up
22805 with pass-specific option and dump file names.
22806 (-fdump-passes): Clarify where to look for output.
22808 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22810 PR middle-end/77445
22811 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
22812 statistics of the analyzed path; allow threading for speed when
22813 any of BBs along the path are optimized for speed.
22815 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
22817 PR middle-end/78468
22818 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
22819 settings of the virtual registers.
22822 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
22824 * explow.c (get_dynamic_stack_size): Take known alignment of stack
22825 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
22828 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22830 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
22831 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
22833 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22835 * config/s390/s390.md: Add missing comments with the expanded
22837 * config/s390/vector.md: Likewise.
22838 * config/s390/vx-builtins.md: Likewise.
22840 2017-02-02 Jakub Jelinek <jakub@redhat.com>
22843 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
22844 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
22845 conditions on a single line.
22847 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22849 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22850 __S390_VX__ to __VX__.
22852 2017-02-01 Andrew Pinski <apinski@cavium.com>
22854 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
22855 stmt_info to record_stmt_cost.
22856 (vect_get_known_peeling_cost): Pass stmt_info if known to
22858 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
22859 cpu_vector_cost field into
22860 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
22861 field into vec_int_stmt_cost and vec_fp_stmt_cost.
22862 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
22863 splitting of scalar_stmt_cost and vec_stmt_cost.
22864 (thunderx_vector_cost): Likewise.
22865 (cortexa57_vector_cost): LIkewise.
22866 (exynosm1_vector_cost): Likewise.
22867 (xgene1_vector_cost): Likewise.
22868 (thunderx2t99_vector_cost): Improve after the splitting of the two
22870 (aarch64_builtin_vectorization_cost): Update for the splitting of
22871 scalar_stmt_cost and vec_stmt_cost.
22873 2017-02-01 Torvald Riegel <triegel@redhat.com>
22874 Richard Henderson <rth@redhat.com>
22876 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
22877 conditional on existance of a fast atomic load.
22878 * optabs-query.c (can_atomic_load_p): New function.
22879 * optabs-query.h (can_atomic_load_p): Declare it.
22880 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
22881 no fast atomic load is available for the particular size of access.
22882 (expand_atomic_compare_and_swap): Likewise.
22883 (expand_atomic_load): Likewise.
22884 (expand_atomic_store): Likewise.
22885 (expand_atomic_fetch_op): Likewise.
22886 * testsuite/lib/target-supports.exp
22887 (check_effective_target_sync_int_128): Remove x86 because it provides
22888 no fast atomic load.
22889 (check_effective_target_sync_int_128_runtime): Likewise.
22891 2017-02-01 Richard Biener <rguenther@suse.de>
22893 * graphite.c: Include tree-vectorizer.h for find_loop_location.
22894 (graphite_transform_loops): Provide opt-info for optimized nests.
22895 * tree-parloop.c (parallelize_loops): Provide opt-info for
22896 parallelized loops.
22898 2017-02-01 Richard Biener <rguenther@suse.de>
22900 PR middle-end/79315
22901 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
22902 was not set before.
22904 2017-02-01 Richard Biener <rguenther@suse.de>
22906 PR tree-optimization/71824
22907 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22908 Verify the loops are valid in the merged SESE region.
22909 (scop_detection::can_represent_loop_1): Check analyzing the
22910 evolution of the number of iterations in the region succeeds.
22912 2017-01-31 Ian Lance Taylor <iant@golang.org>
22914 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
22915 REG_ARGS_SIZE note to 32-bit push insns and call insn.
22917 2017-01-31 David Malcolm <dmalcolm@redhat.com>
22919 PR preprocessor/79210
22920 * input.c (get_substring_ranges_for_loc): Replace line_width
22921 assertion with error-handling.
22923 2017-01-31 Richard Biener <rguenther@suse.de>
22925 PR tree-optimization/77318
22926 * graphite-sese-to-poly.c (extract_affine): Fix assert.
22927 (create_pw_aff_from_tree): Take loop parameter.
22928 (add_condition_to_pbb): Pass loop of the condition to
22929 create_pw_aff_from_tree.
22931 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22933 * config/s390/s390.c (s390_asan_shadow_offset): New function.
22934 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
22936 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
22940 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
22942 (convert_int_to_float128): Likewise.
22943 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
22944 (convert_int_to_float128): Likewise.
22945 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
22946 (UNSPEC_IEEE128_CONVERT): Likewise.
22947 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
22948 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
22949 Use local variables for IBM extended format.
22950 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
22951 (fix_trunc<mode>si2_fprs): Likewise.
22952 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
22953 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
22954 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
22955 to know that we can now have integers of all sizes in vector
22957 (fix<uns>_<mode>di2_hw): Likewise.
22958 (float<uns>_<mode>si2_hw): Likewise.
22959 (fix_<mode>si2_hw): Likewise.
22960 (fixuns_<mode>si2_hw): Likewise.
22961 (float<uns>_<mode>di2_hw): Likewise.
22962 (float_<mode>di2_hw): Likewise.
22963 (float_<mode>si2_hw): Likewise.
22964 (floatuns_<mode>di2_hw): Likewise.
22965 (floatuns_<mode>si2_hw): Likewise.
22966 (xscvqp<su>wz_<mode>): Delete, no longer used.
22967 (xscvqp<su>dz_<mode>): Likewise.
22968 (xscv<su>dqp_<mode>): Likewise.
22969 (ieee128_mfvsrd_64bit): Likewise.
22970 (ieee128_mfvsrd_32bit): Likewise.
22971 (ieee128_mfvsrwz): Likewise.
22972 (ieee128_mtvsrw): Likewise.
22973 (ieee128_mtvsrd_64bit): Likewise.
22974 (ieee128_mtvsrd_32bit): Likewise.
22976 2017-01-31 Martin Liska <mliska@suse.cz>
22979 * ipa-prop.c (ipa_free_all_node_params): Call release method
22980 instead of ~sumbol_summary to not to trigger double times
22983 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
22985 PR tree-optimization/71691
22986 * bitmap.h (class auto_bitmap): New.
22987 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
22988 is_maybe_undefined instead of ssa_undefined_value_p.
22990 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22992 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22993 __S390_ARCH_LEVEL__ to __ARCH__.
22995 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22997 PR tree-optimization/79267
22998 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
22999 if should_remove_lhs_p is true.
23001 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
23004 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
23005 (add_alignment_attribute): New.
23006 (base_type_die): Add alignment attribute.
23007 (subrange_type_die): Likewise.
23008 (modified_type_die): Likewise.
23009 (gen_array_type_die): Likewise.
23010 (gen_descr_array_type_die: Likewise.
23011 (gen_enumeration_type_die): Likewise.
23012 (gen_subprogram_die): Likewise.
23013 (gen_variable_die): Likewise.
23014 (gen_field_die): Likewise.
23015 (gen_ptr_to_mbr_type_die): Likewise.
23016 (gen_struct_or_union_type_die): Likewise.
23017 (gen_subroutine_type_die): Likewise.
23018 (gen_typedef_die): Likewise.
23019 (base_type_cmp): Compare alignment attribute.
23021 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23024 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
23025 (setb_unsigned) New pattern for setb with CCUNS.
23026 * config/rs6000/rs6000.c (expand_block_compare): Use a different
23027 subfc./subfe sequence to avoid overflow problems. Generate a
23028 shorter sequence with cmpld/setb for power9.
23029 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
23030 for generating subfc. instruction.
23031 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
23032 now uses this instruction.
23034 2017-01-30 Ian Lance Taylor <iant@google.com>
23037 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
23038 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
23040 2017-01-30 Martin Sebor <msebor@redhat.com>
23042 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
23043 Move constant to the right of a relational operator.
23044 (get_mpfr_format_length, format_character, format_string): Ditto.
23045 (should_warn_p, maybe_warn): Same.
23047 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
23049 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
23052 * asan.c (get_translation_unit_decl): Remove function.
23053 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
23055 2017-01-30 Martin Liska <mliska@suse.cz>
23057 PR gcov-profile/79259
23058 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
23059 -fprofile-generate.
23061 2017-01-30 Martin Liska <mliska@suse.cz>
23064 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
23065 Initialize variables with NULL value.
23067 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
23070 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
23072 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
23074 2017-01-30 Richard Biener <rguenther@suse.de>
23076 PR tree-optimization/79276
23077 * tree-vrp.c (process_assert_insertions): Properly adjust common
23078 when removing a duplicate.
23080 * gcc.dg/torture/pr79276.c: New testcase.
23082 2017-01-30 Richard Biener <rguenther@suse.de>
23084 PR tree-optimization/79256
23085 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
23086 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
23088 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
23090 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
23093 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
23094 ("*r<noxa>sbg_<mode>_sll_bitmask")
23095 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
23096 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
23097 Use contiguous_bitmask_nowrap_operand.
23099 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23102 * config/rs6000/altivec.h (vec_xl): Revise #define.
23103 (vec_xst): Likewise.
23105 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
23107 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
23109 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
23111 PR rtl-optimization/79194
23112 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
23113 traps before call to bypass_conditional_jumps.
23115 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23117 PR tree-optimization/71374
23118 * lra-constraints.c (check_conflict_input_operands): New.
23119 (match_reload): Use it.
23121 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
23124 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
23125 account to calculate conflict_set.
23127 2017-01-27 Bin Cheng <bin.cheng@arm.com>
23129 PR rtl-optimization/78559
23130 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
23131 other_insn in combine.
23133 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
23135 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
23136 uint16_type_node for BT_UINT16.
23138 2017-01-27 David Malcolm <dmalcolm@redhat.com>
23140 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
23141 "RTL Tests" to menu.
23142 (GIMPLE Tests): New node.
23143 (RTL Tests): New node.
23145 2017-01-27 Richard Biener <rguenther@suse.de>
23147 PR tree-optimization/79245
23148 * tree-loop-distribution.c (distribute_loop): Apply cost
23149 modeling also to detected patterns.
23151 2017-01-27 Richard Biener <rguenther@suse.de>
23153 PR tree-optimization/71433
23154 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
23155 (compare_assert_loc): New function.
23156 (process_assert_insertions): Sort and optimize assert locations
23157 to remove duplicates and push down identical assertions on
23158 edges to their destination block.
23160 2017-01-27 Richard Biener <rguenther@suse.de>
23162 PR tree-optimization/79244
23163 * tree-vrp.c (remove_range_assertions): Forcefully propagate
23164 out SSA names even if abnormal.
23166 2017-01-27 Jakub Jelinek <jakub@redhat.com>
23168 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
23169 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
23170 instead of MPFR_RNDN.
23172 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
23175 * arm.c (arm_option_override): Don't call build_target_option_node
23176 until after doing all option overrides.
23177 (arm_valid_target_attribute_tree): Likewise.
23179 2017-01-27 Martin Liska <mliska@suse.cz>
23181 * doc/invoke.texi (-fprofile-arcs): Document profiling support
23182 for {cd}tors and C++ {cd}tors.
23184 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
23186 * config/s390/s390.md ("*setmem_long_and")
23187 ("*setmem_long_and_31z"): Use zero_extend instead of and.
23189 2017-01-26 Martin Sebor <msebor@redhat.com>
23191 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
23194 2017-01-26 Martin Sebor <msebor@redhat.com>
23196 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
23197 HAVE_DFmode before using XFmode or DFmode.
23198 (parse_directive): Avoid using the z length modifier to avoid
23199 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
23201 PR middle-end/78703
23202 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
23203 to accept adjustment as an array.
23204 (get_int_range): New function.
23205 (struct directive): Make width and prec arrays.
23206 (directive::set_width, directive::set_precision): Call get_int_range.
23207 (format_integer, format_floating): Handle width and precision ranges.
23208 (format_string, parse_directive): Same.
23210 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23213 * dwarf2out.c (generate_skeleton_bottom_up): For children with
23214 comdat_type_p set, just clone them, but keep the children in the
23218 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
23219 which have direct callers with -fvar-tracking-assignments enabled
23221 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
23222 inside of type units.
23224 2017-01-26 Martin Sebor <msebor@redhat.com>
23226 PR middle-end/78703
23227 * gimple-ssa-sprintf.c (struct result_range): Add likely and
23229 (struct format_result): Replace number_chars, number_chars_min,
23230 and number_chars_max with a single member of struct result_range.
23232 (format_result::operator+=): Adjust.
23233 (struct fmtresult): Remove bounded. Handle likely and unlikely
23235 (fmtresult::adjust_for_width_or_precision): New function.
23236 (fmtresult:type_max_digits): New function.
23237 (bytes_remaining): Handle likely and unlikely counters.
23238 (min_bytes_remaining): Remove.
23239 (format_percent): Simplify.
23240 (format_integer, format_floating): Set likely and unlikely counters.
23241 (get_string_length, format_character, format_string): Same.
23242 (format_plain, should_warn_p): New function.
23243 (maybe_warn): Call should_warn_p. Update diagnostic messages
23244 and handle those for all directives, including plain strings.
23245 (format_directive): Handle likely and unlikely counters.
23246 Remove unnecessary quoting from diagnostics. Add an informational
23248 (add_bytes): Remove.
23249 (pass_sprintf_length::compute_format_length): Simplify.
23250 (try_substitute_return_value): Handle likely and unlikely counters.
23252 2017-01-26 Carl Love <cel@us.ibm.com>
23254 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
23255 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
23257 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
23260 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
23261 endianess for subregs into account.
23262 * lra-constraints.c (lra_constraints): Do risky transformations
23263 always on the first iteration.
23264 * lra-lives.c (check_pseudos_live_through_calls): Add arg
23265 last_call_used_reg_set.
23266 (process_bb_lives): Define and use last_call_used_reg_set.
23267 * lra.c (lra): Always continue after lra_constraints on the first
23270 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
23272 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
23274 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
23276 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23278 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
23279 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
23280 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
23281 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
23282 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
23283 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
23284 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
23285 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
23286 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
23288 2017-01-26 Marek Polacek <polacek@redhat.com>
23291 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
23292 for the third operand.
23294 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23296 PR middle-end/79236
23297 * omp-low.c (struct omp_context): Add simt_stmt field.
23298 (scan_omp_for): Return omp_context *.
23299 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
23300 context to the _simt_ SIMD stmt.
23301 (lower_omp_for): For combined SIMD with sibling _simt_
23302 SIMD, make sure to use the same decls in _looptemp_
23303 clauses as in the sibling.
23305 2017-01-26 David Sherwood <david.sherwood@arm.com>
23307 PR middle-end/79212
23308 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
23311 2017-01-26 Jakub Jelinek <jakub@redhat.com>
23314 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
23315 emit fld b; fld a; if possible.
23317 * brig-builtins.def: Update copyright years.
23318 * config/arm/arm_acle_builtins.def: Update copyright years.
23320 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
23323 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
23324 constraint instead of o for the stxsd instruction.
23326 2017-01-25 Carl Love <cel@us.ibm.com>
23328 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
23329 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
23331 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
23333 * doc/invoke.texi (C++ Dialect Options): Fix typo.
23335 2017-01-25 Richard Biener <rguenther@suse.de>
23337 PR tree-optimization/69264
23338 * target.def (vector_alignment_reachable): Improve documentation.
23339 * doc/tm.texi: Regenerate.
23340 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
23342 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
23343 earlier changes with respect to TYPE_USER_ALIGN.
23344 (vector_alignment_reachable_p): Likewise. Improve dumping.
23346 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23349 * config/arm/arm.md (xordi3): Force constant operand into a register
23352 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23354 * doc/invoke.texi (-fstore-merging): Correct default optimization
23355 levels at which it is enabled.
23356 (-O): Move -fstore-merging from list to...
23359 2017-01-25 Richard Biener <rguenther@suse.de>
23362 * omp-expand.c: Include debug.h.
23363 (expand_omp_taskreg): Make sure to generate early debug before
23364 outlining anything from a function.
23365 (expand_omp_target): Likewise.
23366 (grid_expand_target_grid_body): Likewise.
23368 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
23371 * asan.c (get_translation_unit_decl): New function.
23372 (asan_add_global): Extract modules file name from globals
23373 TRANSLATION_UNIT_DECL name.
23375 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
23378 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
23379 for long calls with APCS frame and VFP.
23381 2017-01-24 David Malcolm <dmalcolm@redhat.com>
23383 * cfg.c (original_copy_tables_initialized_p): New function.
23384 * cfg.h (original_copy_tables_initialized_p): New decl.
23385 * cfgrtl.c (relink_block_chain): Guard the call to
23386 free_original_copy_tables with a call to
23387 original_copy_tables_initialized_p.
23388 * cgraph.h (symtab_node::native_rtl_p): New decl.
23389 * cgraphunit.c (symtab_node::native_rtl_p): New function.
23390 (symtab_node::needed_p): Don't assert for early assembly output
23391 for __RTL functions.
23392 (cgraph_node::finalize_function): Set "force_output" for __RTL
23394 (cgraph_node::analyze): Bail out early for __RTL functions.
23395 (analyze_functions): Update assertion to support __RTL functions.
23396 (cgraph_node::expand): Bail out early for __RTL functions.
23397 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
23399 * function.h (struct function): Update comment for field
23401 * gimple-expr.c: Include "tree-pass.h".
23402 (gimple_has_body_p): Return false for __RTL functions.
23403 * Makefile.in (OBJS): Add run-rtl-passes.o.
23404 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
23406 (gcc::pass_manager::get_clean_slate): New accessor.
23407 * passes.c: Include "insn-addr.h".
23408 (should_skip_pass_p): Add logging. Update logic for running
23409 "expand" to be compatible with both __GIMPLE and __RTL. Guard
23410 property-provider override so it is only done for gimple passes.
23412 (skip_pass): New function.
23413 (execute_one_pass): Call skip_pass when skipping passes.
23414 * read-md.c (md_reader::read_char): Support filtering
23415 the input to a subset of line numbers.
23416 (md_reader::md_reader): Initialize fields
23417 m_first_line and m_last_line.
23418 (md_reader::read_file_fragment): New function.
23419 * read-md.h (md_reader::read_file_fragment): New decl.
23420 (md_reader::m_first_line): New field.
23421 (md_reader::m_last_line): New field.
23422 * read-rtl-function.c (function_reader::create_function): Only
23423 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
23424 curr_properties. Set DECL_INITIAL to a dummy block.
23425 (read_rtl_function_body_from_file_range): New function.
23426 * read-rtl-function.h (read_rtl_function_body_from_file_range):
23428 * run-rtl-passes.c: New file.
23429 * run-rtl-passes.h: New file.
23431 2017-01-24 Jeff Law <law@redhat.com>
23433 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
23436 2017-01-24 Bin Cheng <bin.cheng@arm.com>
23438 PR tree-optimization/79159
23439 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
23440 (record_nonwrapping_iv): Improve boundary using above function if no
23441 value range information.
23443 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
23444 Martin Jambor <mjambor@suse.cz>
23446 * brig-builtins.def: New file.
23447 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
23448 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
23449 (DEF_HSAIL_SAT_BUILTIN): Likewise.
23450 (DEF_HSAIL_INTR_BUILTIN): Likewise.
23451 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
23452 * builtin-types.def (BT_INT8): New.
23453 (BT_INT16): Likewise.
23454 (BT_UINT8): Likewise.
23455 (BT_UINT16): Likewise.
23456 (BT_FN_ULONG): Likewise.
23457 (BT_FN_UINT_INT): Likewise.
23458 (BT_FN_UINT_ULONG): Likewise.
23459 (BT_FN_UINT_LONG): Likewise.
23460 (BT_FN_UINT_PTR): Likewise.
23461 (BT_FN_ULONG_PTR): Likewise.
23462 (BT_FN_INT8_FLOAT): Likewise.
23463 (BT_FN_INT16_FLOAT): Likewise.
23464 (BT_FN_UINT32_FLOAT): Likewise.
23465 (BT_FN_UINT16_FLOAT): Likewise.
23466 (BT_FN_UINT8_FLOAT): Likewise.
23467 (BT_FN_UINT64_FLOAT): Likewise.
23468 (BT_FN_UINT16_UINT32): Likewise.
23469 (BT_FN_UINT32_UINT16): Likewise.
23470 (BT_FN_UINT16_UINT16_UINT16): Likewise.
23471 (BT_FN_INT_PTR_INT): Likewise.
23472 (BT_FN_UINT_PTR_UINT): Likewise.
23473 (BT_FN_LONG_PTR_LONG): Likewise.
23474 (BT_FN_ULONG_PTR_ULONG): Likewise.
23475 (BT_FN_VOID_UINT64_UINT64): Likewise.
23476 (BT_FN_UINT8_UINT8_UINT8): Likewise.
23477 (BT_FN_INT8_INT8_INT8): Likewise.
23478 (BT_FN_INT16_INT16_INT16): Likewise.
23479 (BT_FN_INT_INT_INT): Likewise.
23480 (BT_FN_UINT_FLOAT_UINT): Likewise.
23481 (BT_FN_FLOAT_UINT_UINT): Likewise.
23482 (BT_FN_ULONG_UINT_UINT): Likewise.
23483 (BT_FN_ULONG_UINT_PTR): Likewise.
23484 (BT_FN_ULONG_ULONG_ULONG): Likewise.
23485 (BT_FN_UINT_UINT_UINT): Likewise.
23486 (BT_FN_VOID_UINT_PTR): Likewise.
23487 (BT_FN_UINT_UINT_PTR: Likewise.
23488 (BT_FN_UINT32_UINT64_PTR): Likewise.
23489 (BT_FN_INT_INT_UINT_UINT): Likewise.
23490 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
23491 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
23492 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
23493 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
23494 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
23495 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
23496 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
23497 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
23498 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
23499 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
23500 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
23501 * doc/frontends.texi: List BRIG FE.
23502 * doc/install.texi (Testing): Add BRIG tesring requirements.
23503 * doc/invoke.texi (Overall Options): Mention BRIG.
23504 * doc/standards.texi (Standards): Doucment BRIG HSA version.
23506 2017-01-24 Richard Biener <rguenther@suse.de>
23508 PR translation/79208
23509 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
23511 2017-01-24 Martin Jambor <mjambor@suse.cz>
23514 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
23515 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
23516 and known_contexts.
23518 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
23520 PR middle-end/79123
23521 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
23522 casts from signed to unsigned really don't have a range.
23524 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
23526 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
23527 GMP_RNDx for compatiblity.
23529 2017-01-24 Martin Liska <mliska@suse.cz>
23532 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
23533 that would prevent us to call alloca with -1 as argument.
23535 2017-01-24 Jakub Jelinek <jakub@redhat.com>
23537 * dwarf2out.c (output_compilation_unit_header, output_file_names):
23538 Avoid -Wformat-security warning.
23540 2017-01-23 Andrew Pinski <apinski@cavium.com>
23542 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
23545 2017-01-23 Martin Sebor <msebor@redhat.com>
23547 PR middle-end/78703
23548 * gimple-ssa-sprintf.c (warn_level): New global.
23549 (format_integer): Use it here and throughout the rest of the file.
23550 Use the same switch to compute sign as base.
23551 (maybe_warn): New function.
23552 (format_directive): Factor out warnings into maybe_warn.
23553 Add debugging output. Use warn_level.
23554 (add_bytes): Use warn_level.
23555 (pass_sprintf_length::compute_format_length): Add debugging output.
23556 (try_substitute_return_value): Same.
23557 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
23559 PR middle-end/78703
23560 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
23561 (struct fmtresult, format_integer, format_floating): Adjust.
23562 (fmtresult::fmtresult): Set max correctly in two argument ctor.
23563 (get_string_length, format_string,format_directive): Same.
23564 (pass_sprintf_length::compute_format_length): Same.
23565 (try_substitute_return_value): Simplify slightly.
23567 PR middle-end/78703
23568 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
23569 (fmtresult::operator+=): Outlined.
23570 (struct fmtresult): Add ctors.
23571 (struct conversion_spec): Rename...
23572 (struct directive): ...to this. Add and remove data members.
23573 (directive::set_width, directive::set_precision): New functions.
23574 (format_percent): Use fmtresult ctor.
23575 (get_width_and_precision): Remove.
23576 (format_integer): Make naming changes. Avoid computing width and
23578 (format_floating): Same. Adjust indentation.
23579 (format_character, format_none): New functions.
23580 (format_string): Moved character handling to format_character.
23581 (format_directive): Remove arguments, change return type.
23582 (parse_directive): New function.
23583 (pass_sprintf_length::compute_format_length): Move directive
23584 parsing to parse_directive.
23586 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23588 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
23589 (assign_assembler_name_if_needed): ... this.
23590 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
23591 (assign_assembler_name_if_needed): ... this.
23592 (free_lang_data_in_cgraph): Adjust callers.
23593 * cgraphunit.c (cgraph_node::analyze): Likewise.
23594 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
23597 2017-01-23 Richard Biener <rguenther@suse.de>
23599 PR tree-optimization/79088
23600 PR tree-optimization/79188
23601 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
23602 resetting loop bounds after last path deletion. Reset loop
23603 bounds of the target loop, make code match the comments.
23604 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
23605 Make sure loops need no fixups.
23607 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23609 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
23610 exponent support with double type for first argument.
23611 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
23612 type returned by __builtin_vec_extract_sig,
23613 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
23614 functions from "vector int" to "vector unsigned int" or from
23615 "vector long long int" to "vector unsigned long long int".
23616 Changed type returned by __builtin_vec_extract_exp,
23617 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
23618 functions from "vector int" to "vector unsigned int" or from
23619 "vector long long int" to "vector unsigned long long int".
23620 Changed return type of __builtin_vec_test_data_class,
23621 __builtin_vec_test_data_class_sp, and
23622 __builtin_vec_test_data_class_dp from "vector int" to
23623 "vector bool int" or from "vector long long int" to "vector bool
23624 long long int" and changed second argument type from "unsigned
23625 int" to "int". Added new overloaded function forms "vector float
23626 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
23627 "vector float __builtin_vec_insert_exp_sp (vector float, vector
23628 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
23629 double, vector unsigned long long int)" and "vector double
23630 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
23631 long int)". Changed return type of
23632 __builtin_scalar_test_data_class and
23633 __builtin_scalar_test_data_class_sp and
23634 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
23635 int" and changed second argument from "unsigned int" to "int".
23636 Changed type returned by __builtin_scalar_test_neg,
23637 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
23638 from "int" to "bool int". Added new overloaded function form
23639 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
23640 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
23641 exponent double-precision with floating point first argument.
23642 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
23643 documentation of scalar_test_data_class, scalar_test_neg,
23644 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
23645 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
23646 vec_test_data_class built-in functions to reflect refinements in
23647 their type signatures.
23649 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
23651 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
23653 (aarch64_elf_asm_destructor): Likewise.
23655 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
23657 PR rtl-optimization/78634
23658 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
23659 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
23660 * ifcvt.c (noce_try_cmove): Add missing cost check.
23662 PR rtl-optimization/71724
23663 * combine.c (if_then_else_cond): Look for situations where it is
23664 beneficial to undo the work of one of the recursive calls.
23666 2017-01-23 Bin Cheng <bin.cheng@arm.com>
23668 PR tree-optimization/70754
23669 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
23670 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
23671 combined stmt before it if not NULL.
23672 (combine_chains): Process refs reversely and compute dominance point
23675 2017-01-23 Martin Liska <mliska@suse.cz>
23677 PR tree-optimization/79196
23678 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
23679 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
23681 (strlen_optimize_stmt): Call the renamed function.
23683 2017-01-23 Michael Matz <matz@suse.de>
23685 PR tree-optimization/78384
23686 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
23688 2017-01-23 Richard Biener <rguenther@suse.de>
23690 PR tree-optimization/79186
23691 * tree-vrp.c (register_new_assert_for): Make sure we've seen
23692 both incoming edges before moving an assert.
23694 2017-01-23 Martin Jambor <mjambor@suse.cz>
23696 * ipa-prop.c (load_from_param_1): Removed.
23697 (load_from_unmodified_param): Bits from load_from_param_1 put back
23699 (load_from_param): Removed.
23700 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
23701 with stmt. Reverted back to use of load_from_unmodified_param.
23703 2017-01-23 Martin Jambor <mjambor@suse.cz>
23706 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
23707 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
23708 field a pointer to garbage collected vector, mark lattices and
23709 ipcp_orig_node with GTY((skip)).
23710 (ipa_get_param_count): Adjust to descriptors being a pointer.
23711 (ipa_get_param): Likewise.
23712 (ipa_get_type): Likewise.
23713 (ipa_get_param_move_cost): Likewise.
23714 (ipa_set_param_used): Likewise.
23715 (ipa_get_controlled_uses): Likewise.
23716 (ipa_set_controlled_uses): Likewise.
23717 (ipa_is_param_used): Likewise.
23718 (ipa_node_params_t): Move into garbage collector. New methods insert
23720 (ipa_node_params_sum): Annotate wth GTY(()).
23721 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
23723 (ipa_load_from_parm_agg): Adjust declaration.
23724 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
23725 * ipa-profile.c (ipa_profile): Likewise.
23726 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
23727 (ipa_populate_param_decls): Make descriptors parameter garbage
23729 (ipa_dump_param): Adjust to descriptors being a pointer.
23730 (ipa_alloc_node_params): Likewise.
23731 (ipa_initialize_node_params): Likewise.
23732 (load_from_param_1): Make descriptors parameter garbage collected.
23733 (load_from_unmodified_param): Likewise.
23734 (load_from_param): Likewise.
23735 (ipa_load_from_parm_agg): Likewise.
23736 (ipa_node_params::~ipa_node_params): Removed.
23737 (ipa_free_all_node_params): Remove call to delete operator.
23738 (ipa_node_params_t::insert): New.
23739 (ipa_node_params_t::remove): Likewise.
23740 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
23741 copy known_csts and known_contexts vectors.
23742 (ipa_read_node_info): Adjust to descriptors being a pointer.
23743 (ipcp_modif_dom_walker): Make m_descriptors field garbage
23745 (ipcp_transform_function): Make descriptors variable garbage
23748 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
23750 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
23751 * config/i386/avx512dqintrin.h: Ditto.
23752 * config/i386/avx512fintrin.h: Ditto.
23753 * config/i386/i386.c: Handle new builtins.
23754 * config/i386/i386-builtin.def: Add new builtins.
23755 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
23756 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
23758 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23759 Martin Liska <mliska@suse.cz>
23761 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
23762 * asan.c (asan_expand_poison_ifn): Support stores and use
23763 appropriate ASAN report function.
23764 * internal-fn.c (expand_ASAN_POISON_USE): New function.
23765 * internal-fn.def (ASAN_POISON_USE): Declare.
23766 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
23767 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
23768 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
23769 ASAN_POISON calls w/o LHS.
23770 * tree-ssa.c (execute_update_addresses_taken): Create clobber
23771 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
23772 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
23773 * gimplify.c (asan_poison_variables): Add attribute
23774 use_after_scope_memory to variables that really needs to live
23776 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
23777 having the attribute.
23779 2017-01-23 Martin Liska <mliska@suse.cz>
23781 * asan.c (create_asan_shadow_var): New function.
23782 (asan_expand_poison_ifn): Likewise.
23783 * asan.h (asan_expand_poison_ifn): New declaration.
23784 * internal-fn.c (expand_ASAN_POISON): Likewise.
23785 * internal-fn.def (ASAN_POISON): New builtin.
23786 * sanopt.c (pass_sanopt::execute): Expand
23787 asan_expand_poison_ifn.
23788 * tree-inline.c (copy_decl_for_dup_finish): Make function
23790 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
23791 * tree-ssa.c (is_asan_mark_p): New function.
23792 (execute_update_addresses_taken): Rewrite local variables
23793 (identified just by use-after-scope as addressable) into SSA.
23795 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23797 * doc/install.texi (Specific): opensource.apple.com uses https
23798 now. Remove trailing slash.
23800 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23802 * README.Portability: Remove note on an Irix compatibility issue.
23804 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
23806 * gcov.c (INCLUDE_ALGORITHM): Define.
23807 (INCLUDE_VECTOR): Define.
23808 No longer include <vector> and <algorithm> directly.
23810 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
23812 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
23814 * doc/invoke.texi (Code Gen Options): Ditto.
23816 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
23819 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
23821 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
23823 rtl-optimization/79125
23824 * cprop.c (local_cprop_pass): Handle cases where we make an
23825 unconditional trap.
23827 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
23831 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
23832 read from, for big endian.
23834 2017-01-20 Jiong Wang <jiong.wang@arm.com>
23836 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
23837 register pauth builtins for LP64 only.
23839 2017-01-20 Marek Polacek <polacek@redhat.com>
23842 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
23845 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23847 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
23849 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
23850 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
23851 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
23853 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23856 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
23857 in big-endian mode when they are not a single duplicated value.
23859 2017-01-20 Richard Biener <rguenther@suse.de>
23861 * BASE-VER: Bump to 7.0.1.
23863 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23865 * omp-low.c (omplow_simd_context): New struct. Use it...
23866 (lower_rec_simd_input_clauses): ...here and...
23867 (lower_rec_input_clauses): ...here to hold common data. Adjust all
23868 references to idx, lane, max_vf, is_simt.
23870 2017-01-20 Graham Markall <graham.markall@embecosm.com>
23872 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
23875 2017-01-20 Martin Jambor <mjambor@suse.cz>
23877 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
23878 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
23880 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
23881 (GTFILES): Rename hsa.c to hsa-common.c.
23882 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
23883 * hsa-dump.c: Likewise.
23884 * hsa-gen.c: Likewise.
23885 * hsa-regalloc.c: Likewise.
23886 * ipa-hsa.c: Likewise.
23887 * omp-expand.c: Likewise.
23888 * omp-low.c: Likewise.
23889 * toplev.c: Likewise.
23891 2017-01-20 Marek Polacek <polacek@redhat.com>
23894 * doc/invoke.texi: Document -Wduplicated-branches.
23895 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
23896 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
23897 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
23898 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
23899 return 0 only when not OEP_LEXICOGRAPHIC.
23900 (fold_build_cleanup_point_expr): Use the expression
23901 location when building CLEANUP_POINT_EXPR.
23902 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
23903 * tree.c (add_expr): Handle error_mark_node.
23905 2017-01-20 Martin Liska <mliska@suse.cz>
23908 * tree-profile.c (init_ic_make_global_vars): Do not call
23910 (gimple_init_gcov_profiler): Likewise.
23912 2017-01-20 Martin Liska <mliska@suse.cz>
23915 * cgraph.h (maybe_create_reference): Remove argument and
23917 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
23919 * ipa-cp.c (create_specialized_node): Likewise.
23920 * symtab.c (symtab_node::maybe_create_reference): Handle
23921 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
23923 2017-01-20 Martin Liska <mliska@suse.cz>
23925 * read-rtl-function.c (function_reader::create_function): Use
23926 build_decl instread of build_decl_stat.
23928 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
23930 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
23931 * config/i386/avx512dqintrin.h: Ditto.
23932 * config/i386/avx512fintrin.h: Ditto.
23933 * config/i386/i386-builtin-types.def: Add new types.
23934 * config/i386/i386.c: Handle new types.
23935 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
23936 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
23937 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
23938 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
23939 (__builtin_ia32_kshiftridi): New.
23940 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
23942 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
23946 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
23947 define to rs6000_init_stack_protect_guard.
23948 (rs6000_init_stack_protect_guard): New function.
23950 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23951 Yunqiang Su <yunqiang.su@imgtec.com>
23953 * config.gcc (supported_defaults): Add madd4.
23954 (with_madd4): Add validation.
23955 (all_defaults): Add madd4.
23956 * config/mips/mips.opt (mmadd4): New option.
23957 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23959 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23961 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
23962 (ISA_HAS_FUSED_MADD4): Likewise.
23963 * doc/invoke.texi (-mmadd4): Document the new option.
23964 * doc/install.texi (--with-madd4): Document the new option.
23966 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23968 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
23969 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
23970 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
23971 (aarch64_init_pauth_hint_builtins): New.
23972 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
23973 (aarch64_expand_builtin): Expand new builtins.
23975 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23977 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
23978 * combine-stack-adj.c (no_unhandled_cfa): Handle
23979 REG_CFA_TOGGLE_RA_MANGLE.
23980 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
23981 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
23982 info for return address signing.
23983 (aarch64_expand_epilogue): Likewise.
23985 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23987 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
23988 * config/aarch64/aarch64-protos.h
23989 (aarch64_return_address_signing_enabled): New declaration.
23990 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
23992 (aarch64_expand_prologue): Sign return address before it's pushed onto
23994 (aarch64_expand_epilogue): Authenticate return address fetched from
23996 (aarch64_override_options): Sanity check for ILP32 and ISA level.
23997 (aarch64_attributes): New function attributes for "sign-return-address".
23998 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
23999 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
24000 ("*do_return"): Generate combined instructions according to key index.
24001 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
24002 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
24004 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
24005 * config/aarch64/aarch64.opt (msign-return-address=): New.
24006 * doc/extend.texi (AArch64 Function Attributes): Documents
24007 "sign-return-address=".
24008 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
24010 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
24012 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
24013 overall option summary.
24015 2017-01-19 Jiong Wang <jiong.wang@arm.com>
24017 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
24018 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
24019 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
24020 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
24022 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
24024 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
24025 -mpower9-minmax by default for -mcpu=power9.
24026 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
24027 128-bit floating point.
24029 2017-01-20 Alan Modra <amodra@gmail.com>
24031 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
24032 optimizing for size.
24034 2017-01-20 Alan Modra <amodra@gmail.com>
24037 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
24038 for strcmp and strncmp from corresponding builtin decl.
24040 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24042 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
24043 instead of i386/rtems-64.h.
24044 * config/i386/rtems-64.h: Remove.
24046 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
24050 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
24052 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
24054 2017-01-19 Tamar Christina <tamar.christina@arm.com>
24056 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
24057 Change int to HOST_WIDE_INT.
24058 * config/aarch64/aarch64-protos.h
24059 (aarch64_simd_gen_const_vector_dup): Likewise.
24060 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
24062 2017-01-19 David Malcolm <dmalcolm@redhat.com>
24064 * langhooks-def.h (lhd_type_for_size): New decl.
24065 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
24066 * langhooks.c (lhd_type_for_size): New function, taken from
24069 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
24071 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
24072 define_bypass for CR latency.
24073 (power9-cracked-alu): Update bypass latency and remove power9-branch.
24074 (power9-alu2): Add define_bypass for CR latency.
24076 (power9-mul): Update insn latency.
24077 (power9-mul-compare): Update insn latency, bypass latency and remove
24080 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24082 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
24084 * config/aarch64/aarch64.md
24085 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
24086 aarch64_nopcrelative_literal_loads.
24087 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
24089 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
24091 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
24092 TARGET_LOONGSON_3A.
24093 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
24095 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
24098 * config.gcc (supported_defaults): Add lxc1-sxc1.
24099 (with_lxc1_sxc1): Add validation.
24100 (all_defaults): Add lxc1-sxc1.
24101 * config/mips/mips.opt (mlxc1-sxc1): New option.
24102 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
24104 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
24105 __mips_no_lxc1_sxc1.
24106 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
24107 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
24108 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
24110 2017-01-19 Richard Biener <rguenther@suse.de>
24112 PR tree-optimization/72488
24113 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
24114 sure to restore SSA info.
24115 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
24117 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
24119 PR rtl-optimization/79121
24120 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
24121 of the inner type when shifting an extended value.
24123 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24126 * symtab.c (symtab_node::equal_address_to): Fix comparing of
24127 interposable aliases.
24129 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
24132 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
24133 Use the evmergelohi instruction.
24134 (mov_si<mode>_e500_subreg4_2_le): Likewise.
24135 (mov_sitf_e500_subreg8_2_be): Likewise.
24136 (mov_sitf_e500_subreg12_2_le): Likewise.
24137 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
24138 (mov_si<mode>_e500_subreg4_2_be): Likewise.
24139 (mov_sitf_e500_subreg8_2_le): Likewise.
24140 (mov_sitf_e500_subreg12_2_be): Likewise.
24142 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24144 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
24145 attribute from vecsimple to vecperm.
24146 (altivec_vbpermq2): Likewise.
24148 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24151 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
24153 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24154 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
24155 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
24156 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
24157 case where N arg is SIZE_MAX.
24158 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
24159 (cmpstrsi): Add pattern.
24161 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24163 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24164 __builtin_vec_revb builtins.
24165 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
24166 built-in functions to support generation of the ISA 3.0 XXBR<x>
24167 vector byte reverse instructions.
24168 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
24169 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
24170 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
24171 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
24172 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
24173 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
24174 (P9V_BUILTIN_VEC_REVB): Likewise.
24175 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
24176 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
24177 (p9_xxbrq_v16qi): Likewise.
24178 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
24179 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
24180 (p9_xxbrh_v8hi): Likewise.
24181 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
24182 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
24183 vec_revb built-in functions.
24185 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
24187 PR rtl-optimization/78952
24188 * config/i386/i386.md (any_extract): New code iterator.
24189 (*insvqi_2): Use any_extract for source operand.
24190 (*insvqi_3): Use any_shiftrt for source operand.
24192 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
24194 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
24196 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
24198 2017-01-18 Matthias Klose <doko@ubuntu.com>
24200 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
24202 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24204 * config/rs6000/altivec.h (vec_bperm): Change #define.
24205 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
24206 (altivec_vbpermq2): New define_insn.
24207 (altivec_vbpermd): Likewise.
24208 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
24209 function interface.
24210 (VBPERMD): Likewise.
24211 (VBPERM): New polymorphic function interface.
24212 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
24213 Add entries for P9V_BUILTIN_VEC_VBPERM.
24214 * doc/extend.texi: Add interfaces for vec_bperm.
24216 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24218 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
24219 first letter of error messages.
24220 (s390_resolve_overloaded_builtin): Likewise.
24221 * config/s390/s390.c (s390_expand_builtin): Likewise.
24222 (s390_invalid_arg_for_unprototyped_fn): Likewise.
24223 (s390_valid_target_attribute_inner_p): Likewise.
24224 * config/s390/s390.md ("tabort"): Likewise.
24226 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
24228 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
24229 (ISA_AVOID_DIV_HILO): New macro.
24230 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
24231 (ISA_HAS_DDIV): Likewise.
24233 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24235 * doc/invoke.texi (fabi-version): Correct number of occurrences.
24237 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24239 * doc/invoke.texi (fabi-version): Spelling fix.
24241 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24244 * doc/invoke.texi (fabi-version): Mention mangling fix for
24247 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
24250 * doc/invoke.texi (fabi-version): Document discriminator mangling.
24252 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
24255 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
24256 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
24258 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
24260 (stack_protect_test): Ditto.
24261 * config/rs6000/rs6000.opt (mstack-protector-guard=,
24262 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
24264 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
24265 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
24266 -mstack-protector-guard-offset=.
24267 (RS/6000 and PowerPC Options): Ditto.
24269 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24271 * config/i386/i386.h (MASK_CLASS_P): New define.
24272 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
24273 there are no registers from different register sets also when
24274 mask registers are used. Update function comment.
24275 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
24276 to (*k/*r) and (*k/*km) alternatives.
24278 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
24280 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
24281 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
24282 (EH_RETURN_HANDLER_RTX): New define.
24283 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
24284 Force frame pointer in EH return functions.
24285 (aarch64_expand_epilogue): Add barrier for eh_return.
24286 (aarch64_final_eh_return_addr): Remove.
24287 (aarch64_eh_return_handler_rtx): New function.
24288 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
24290 (aarch64_eh_return_handler_rtx): New prototype.
24292 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24294 * config/rs6000/altivec.h (vec_rlmi): New #define.
24295 (vec_vrlnm): Likewise.
24296 (vec_rlnm): Likewise.
24297 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
24298 (UNSPEC_VRLNM): Likewise.
24299 (VIlong): New mode iterator.
24300 (altivec_vrl<VI_char>mi): New define_insn.
24301 (altivec_vrl<VI_char>nm): Likewise.
24302 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
24304 (VRLDNM): Likewise.
24305 (RLNM): New polymorphic function entry.
24306 (VRLWMI): New monomorphic function entry.
24307 (VRLDMI): Likewise.
24308 (RLMI): New polymorphic function entry.
24309 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
24310 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
24311 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
24314 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24317 * dwarf2out.c (field_byte_offset): Restore the
24318 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
24319 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
24320 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
24323 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24326 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
24328 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24331 * dwarf2out.c (add_data_member_location_attribute): For constant
24332 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
24333 instead of DW_AT_data_member_location, DW_AT_bit_offset and
24334 DW_AT_byte_size attributes.
24336 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
24338 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
24339 after forcing to constant memory when the code model is medium.
24341 2017-01-17 Julia Koval <julia.koval@intel.com>
24344 * config/i386/avx512fintrin.h
24345 (_mm512_i32gather_ps): Change __addr type to void const*.
24346 (_mm512_mask_i32gather_ps): Ditto.
24347 (_mm512_i32gather_pd): Ditto.
24348 (_mm512_mask_i32gather_pd): Ditto.
24349 (_mm512_i64gather_ps): Ditto.
24350 (_mm512_mask_i64gather_ps): Ditto.
24351 (_mm512_i64gather_pd): Ditto.
24352 (_mm512_mask_i64gather_pd): Ditto.
24353 (_mm512_i32gather_epi32): Ditto.
24354 (_mm512_mask_i32gather_epi32): Ditto.
24355 (_mm512_i32gather_epi64): Ditto.
24356 (_mm512_mask_i32gather_epi64): Ditto.
24357 (_mm512_i64gather_epi32): Ditto.
24358 (_mm512_mask_i64gather_epi32): Ditto.
24359 (_mm512_i64gather_epi64): Ditto.
24360 (_mm512_mask_i64gather_epi64): Ditto.
24361 (_mm512_i32scatter_ps): Change __addr type to void*.
24362 (_mm512_mask_i32scatter_ps): Ditto.
24363 (_mm512_i32scatter_pd): Ditto.
24364 (_mm512_mask_i32scatter_pd): Ditto.
24365 (_mm512_i64scatter_ps): Ditto.
24366 (_mm512_mask_i64scatter_ps): Ditto.
24367 (_mm512_i64scatter_pd): Ditto.
24368 (_mm512_mask_i64scatter_pd): Ditto.
24369 (_mm512_i32scatter_epi32): Ditto.
24370 (_mm512_mask_i32scatter_epi32): Ditto.
24371 (_mm512_i32scatter_epi64): Ditto.
24372 (_mm512_mask_i32scatter_epi64): Ditto.
24373 (_mm512_i64scatter_epi32): Ditto.
24374 (_mm512_mask_i64scatter_epi32): Ditto.
24375 (_mm512_i64scatter_epi64): Ditto.
24376 (_mm512_mask_i64scatter_epi64): Ditto.
24377 * config/i386/avx512pfintrin.h
24378 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
24379 (_mm512_mask_prefetch_i32gather_ps): Ditto.
24380 (_mm512_mask_prefetch_i64gather_pd): Ditto.
24381 (_mm512_mask_prefetch_i64gather_ps): Ditto.
24382 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
24383 (_mm512_prefetch_i32scatter_ps): Ditto.
24384 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
24385 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
24386 (_mm512_prefetch_i64scatter_pd): Ditto.
24387 (_mm512_prefetch_i64scatter_ps): Ditto.
24388 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
24389 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
24390 * config/i386/avx512vlintrin.h
24391 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
24392 (_mm_mmask_i32gather_ps): Ditto.
24393 (_mm256_mmask_i32gather_pd): Ditto.
24394 (_mm_mmask_i32gather_pd): Ditto.
24395 (_mm256_mmask_i64gather_ps): Ditto.
24396 (_mm_mmask_i64gather_ps): Ditto.
24397 (_mm256_mmask_i64gather_pd): Ditto.
24398 (_mm_mmask_i64gather_pd): Ditto.
24399 (_mm256_mmask_i32gather_epi32): Ditto.
24400 (_mm_mmask_i32gather_epi32): Ditto.
24401 (_mm256_mmask_i32gather_epi64): Ditto.
24402 (_mm_mmask_i32gather_epi64): Ditto.
24403 (_mm256_mmask_i64gather_epi32): Ditto.
24404 (_mm_mmask_i64gather_epi32): Ditto.
24405 (_mm256_mmask_i64gather_epi64): Ditto.
24406 (_mm_mmask_i64gather_epi64): Ditto.
24407 (_mm256_i32scatter_ps): Change __addr type to void*.
24408 (_mm256_mask_i32scatter_ps): Ditto.
24409 (_mm_i32scatter_ps): Ditto.
24410 (_mm_mask_i32scatter_ps): Ditto.
24411 (_mm256_i32scatter_pd): Ditto.
24412 (_mm256_mask_i32scatter_pd): Ditto.
24413 (_mm_i32scatter_pd): Ditto.
24414 (_mm_mask_i32scatter_pd): Ditto.
24415 (_mm256_i64scatter_ps): Ditto.
24416 (_mm256_mask_i64scatter_ps): Ditto.
24417 (_mm_i64scatter_ps): Ditto.
24418 (_mm_mask_i64scatter_ps): Ditto.
24419 (_mm256_i64scatter_pd): Ditto.
24420 (_mm256_mask_i64scatter_pd): Ditto.
24421 (_mm_i64scatter_pd): Ditto.
24422 (_mm_mask_i64scatter_pd): Ditto.
24423 (_mm256_i32scatter_epi32): Ditto.
24424 (_mm256_mask_i32scatter_epi32): Ditto.
24425 (_mm_i32scatter_epi32): Ditto.
24426 (_mm_mask_i32scatter_epi32): Ditto.
24427 (_mm256_i32scatter_epi64): Ditto.
24428 (_mm256_mask_i32scatter_epi64): Ditto.
24429 (_mm_i32scatter_epi64): Ditto.
24430 (_mm_mask_i32scatter_epi64): Ditto.
24431 (_mm256_i64scatter_epi32): Ditto.
24432 (_mm256_mask_i64scatter_epi32): Ditto.
24433 (_mm_i64scatter_epi32): Ditto.
24434 (_mm_mask_i64scatter_epi32): Ditto.
24435 (_mm256_i64scatter_epi64): Ditto.
24436 (_mm256_mask_i64scatter_epi64): Ditto.
24437 (_mm_i64scatter_epi64): Ditto.
24438 (_mm_mask_i64scatter_epi64): Ditto.
24439 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
24440 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
24441 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
24442 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
24443 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
24444 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
24445 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
24446 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
24447 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
24448 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
24449 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
24450 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
24451 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
24452 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
24453 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
24454 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
24455 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
24456 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
24457 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
24458 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
24459 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
24460 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
24461 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
24462 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
24463 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
24464 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
24465 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
24466 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
24467 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
24468 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
24469 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
24470 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
24471 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
24472 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
24473 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
24474 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
24475 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
24476 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
24477 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
24478 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
24479 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
24480 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
24481 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
24482 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
24483 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
24484 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
24485 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
24486 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
24487 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
24488 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
24489 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
24490 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
24491 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
24492 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
24493 definitions accordingly.
24495 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
24496 Kuan-Lin Chen <kuanlinchentw@gmail.com>
24499 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
24502 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
24505 * ira-conflicts.c (ira_build_conflicts): Update total conflict
24506 hard regs for inner regno.
24508 2017-01-17 Martin Liska <mliska@suse.cz>
24511 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
24512 assumption and add comment.
24514 2017-01-17 Nathan Sidwell <nathan@acm.org>
24516 * ipa-visibility.c (localize_node): New function, broken out of ...
24517 (function_and_variable_visibility): ... here. Call it.
24519 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
24521 PR middle-end/77445
24522 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
24523 correctly set frequency of oudgoing edge.
24524 (duplicate_thread_path): Fix profile updating.
24526 2017-01-17 Jakub Jelinek <jakub@redhat.com>
24529 * configure.ac: Add GCC_BASE_VER.
24530 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
24531 version from BASE-VER file.
24532 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
24533 (gcc.o): Depend on $(BASEVER).
24534 * common.opt (dumpfullversion): New option.
24535 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
24536 * doc/invoke.texi: Document -dumpfullversion.
24537 * doc/install.texi: Document --with-gcc-major-version-only.
24538 * configure: Regenerated.
24540 2017-01-17 Richard Biener <rguenther@suse.de>
24542 PR tree-optimization/71433
24543 * tree-vrp.c (register_new_assert_for): Merge same asserts
24544 on all incoming edges.
24545 (process_assert_insertions_for): Handle insertions at the
24548 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
24550 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
24551 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
24553 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
24556 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
24559 2017-01-17 Alan Modra <amodra@gmail.com>
24562 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
24563 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
24564 symbolic stack limit when pic.
24566 2017-01-16 Martin Sebor <msebor@redhat.com>
24568 PR tree-optimization/78608
24569 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
24571 2017-01-16 Jeff Law <law@redhat.com>
24574 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
24575 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
24576 for several include directories that may be relative to sysroot.
24577 * config/i386/x-mingw32 (gplus_includedir): Define.
24578 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
24579 (native_system_includedir): Likewise.
24580 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
24581 override if TARGET_SYSTEM_ROOT is defined.
24582 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
24584 PR tree-optimization/79090
24585 PR tree-optimization/33562
24586 PR tree-optimization/61912
24587 PR tree-optimization/77485
24588 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
24589 and computed trims into the dump file.
24591 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
24593 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
24595 2017-01-16 Jakub Jelinek <jakub@redhat.com>
24598 * gimplify.c (gimplify_init_constructor): If want_value and
24599 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
24603 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
24604 sequence. Formatting fixes.
24605 (doloop_optimize): Formatting fixes.
24608 * gcc.c (debug_level_greater_than_spec_func): New function.
24609 (static_spec_functions): Add debug-level-gt spec function.
24610 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
24612 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24613 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
24614 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
24615 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
24616 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
24617 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
24619 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
24621 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
24622 QImode fixups to general and mask registers only.
24624 2017-01-16 Carl Love <cel@us.ibm.com>
24626 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
24627 for built-in functions
24628 vector signed char vec_nabs (vector signed char)
24629 vector signed short vec_nabs (vector signed short)
24630 vector signed int vec_nabs (vector signed int)
24631 vector signed long long vec_nabs (vector signed long long)
24632 vector float vec_nabs (vector float)
24633 vector double vec_nabs (vector double)
24634 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
24636 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
24637 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
24638 * doc/extend.texi: Update the documentation file for the new built-in
24641 2017-01-16 Martin Sebor <msebor@redhat.com>
24643 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
24646 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24648 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
24649 UNSPEC_VSX__XXSPLTD to require special splat handling.
24651 2017-01-16 David Malcolm <dmalcolm@redhat.com>
24654 * system.h: Poison strndup.
24656 2017-01-16 Alan Modra <amodra@gmail.com>
24659 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
24662 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
24664 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
24666 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
24668 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
24669 call recog here. Assert that INSN_CODE (insn) is non-negative.
24671 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24674 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
24676 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
24677 in the currently scheduled RTL fragment.
24679 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24681 PR rtl-optimization/78751
24682 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
24685 2017-01-14 Jeff Law <law@redhat.com>
24687 PR tree-optimization/79090
24688 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
24689 variable length stores.
24690 (compute_trims): Delete dead assignment to *trim_tail.
24691 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
24694 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
24696 PR rtl-optimization/78626
24697 PR rtl-optimization/78727
24698 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
24699 of a block, and split such blocks after everything else is finished.
24701 2017-01-14 Alan Modra <amodra@gmail.com>
24704 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
24705 target legitimate_combined_insn.
24706 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
24707 (rs6000_legitimate_combined_insn): New function.
24708 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
24710 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
24711 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
24712 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
24714 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
24716 * doc/frontends.texi (G++ and GCC): Remove references to Java.
24718 2017-01-13 Jeff Law <law@redhat.com>
24720 PR tree-optimization/33562
24721 PR tree-optimization/61912
24722 PR tree-optimization/77485
24723 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
24725 (delete_dead_assignment): Likewise.
24726 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
24727 statement to delete_dead_call and delete_dead_assignment.
24729 2017-01-13 David Malcolm <dmalcolm@redhat.com>
24732 * substring-locations.c (format_warning_va): Strengthen case 1 so
24733 that both endpoints of the substring must be within the format
24734 range for just the substring to be printed.
24736 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
24738 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
24739 * config/i386/i386.c (ix86_target_string): Add missing options
24740 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
24741 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
24742 flags_other and ix86_target_other to flags2_other. Display unknown
24744 (ix86_valid_target_attribute_inner_p): Add missing options and
24745 reorder options by implied ISAs, as in ix86_target_string.
24747 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24749 * hash-table.h (hash_table::too_empty_p): New function.
24750 (hash_table::expand): Use it.
24751 (hash_table::traverse): Likewise.
24752 (hash_table::empty_slot): Use sizeof (value_type) instead of
24753 sizeof (PTR) to convert bytes to elements. Shrink the table
24754 if the current size is excessive for the current number of
24757 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24759 * ira-costs.c (record_reg_classes): Break from the inner loop
24760 early once alt_fail is known to be true. Update outer loop
24761 handling accordingly.
24763 2017-01-13 Jeff Law <law@redhat.com>
24765 * tree-ssa-dse.c (decrement_count): New function.
24766 (increment_start_addr, maybe_trim_memstar_call): Likewise.
24767 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
24768 when we know the partially dead statement is a mem* function.
24770 PR tree-optimization/61912
24771 PR tree-optimization/77485
24772 * tree-ssa-dse.c: Include expr.h.
24773 (maybe_trim_constructor_store): New function.
24774 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
24776 PR tree-optimization/33562
24777 PR tree-optimization/61912
24778 PR tree-optimization/77485
24779 * doc/invoke.texi: Document new dse-max-object-size param.
24780 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
24781 * tree-ssa-dse.c: Include params.h.
24782 (dse_store_status): New enum.
24783 (initialize_ao_ref_for_dse): New, partially extracted from
24785 (valid_ao_ref_for_dse, normalize_ref): New.
24786 (setup_live_bytes_from_ref, compute_trims): Likewise.
24787 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
24788 (maybe_trim_partially_dead_store): Likewise.
24789 (maybe_trim_complex_store): Likewise.
24790 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
24791 Track what bytes live from the original store. Return tri-state
24792 for dead, partially dead or live.
24793 (dse_dom_walker): Add constructor, destructor and new private members.
24794 (delete_dead_call, delete_dead_assignment): New extracted from
24796 (dse_optimize_stmt): Make a member of dse_dom_walker.
24797 Use initialize_ao_ref_for_dse.
24799 PR tree-optimization/33562
24800 PR tree-optimization/61912
24801 PR tree-optimization/77485
24802 * sbitmap.h (bitmap_count_bits): Prototype.
24803 (bitmap_clear_range, bitmap_set_range): Likewise.
24804 * sbitmap.c (bitmap_clear_range): New function.
24805 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
24807 2017-01-13 Martin Liska <mliska@suse.cz>
24810 * function.c (set_cfun): Add new argument force.
24811 * function.h (set_cfun): Likewise.
24812 * ipa-inline-transform.c (inline_call): Use the function when
24813 strict alising from is dropped for function we inline to.
24815 2017-01-13 Richard Biener <rguenther@suse.de>
24817 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
24818 for dumping GIMPLE INTEGER_CSTs.
24820 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24822 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
24823 to 201112L since C++17.
24825 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24828 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
24829 if -fsanitize=kernel-address is present.
24831 2017-01-13 Richard Biener <rguenther@suse.de>
24833 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
24834 as _Literal ( type ) number in case usual suffixes do not
24835 preserve all information.
24837 2017-01-13 Richard Biener <rguenther@suse.de>
24839 PR tree-optimization/77283
24840 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
24841 and ssa-iterators.h.
24842 (is_feasible_trace): Implement a cost model based on joiner
24845 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24848 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
24849 char or short to __float128/_Float128 directly.
24851 2017-01-12 Martin Sebor <msebor@redhat.com>
24853 to -Wformat-overflow.
24854 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
24855 (min_bytes_remaining): Same.
24856 (get_string_length): Same.
24857 (format_string): Same.
24858 (format_directive): Same.
24860 (pass_sprintf_length::handle_gimple_call): Same.
24862 2017-01-12 Jakub Jelinek <jakub@redhat.com>
24864 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
24865 info.nowrite calls with no lhs that can't throw. Return bool
24866 whether gsi_remove has been called or not.
24867 (pass_sprintf_length::handle_gimple_call): Return bool whether
24868 try_substitute_return_value called gsi_remove. Formatting fix.
24869 (pass_sprintf_length::execute): Don't use gsi_remove if
24870 handle_gimple_call returned true.
24873 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
24874 be removed due to side-effects, don't remove following barrier nor
24875 turn the successor edge into fallthru edge.
24877 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24880 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
24881 element-reversing loads and stores as not swappable.
24883 2017-01-12 Nathan Sidwell <nathan@acm.org>
24884 Nicolai Stange <nicstange@gmail.com>
24886 * combine.c (try_combine): Don't ignore result of overlap checking
24887 loop. Combine overlap & asm check into single loop.
24889 2017-01-12 Richard Biener <rguenther@suse.de>
24891 * tree-pretty-print.c (dump_generic_node): Provide -gimple
24892 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
24894 2017-01-12 Richard Biener <rguenther@suse.de>
24896 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
24897 and TS_TARGET_OPTION directly derive from TS_BASE.
24898 * tree-core.h (tree_optimization_option): Derive from tree_base.
24899 (tree_target_option): Likewise.
24901 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
24903 * config/i386/i386.c (memory_address_length): Increase len
24904 only when rip_relative_addr_p returns false.
24906 2017-01-11 Julia Koval <julia.koval@intel.com>
24908 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
24909 (OPTION_MASK_ISA_SGX_SET): New.
24910 (ix86_handle_option): Handle OPT_msgx.
24911 * config.gcc: Added sgxintrin.h.
24912 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
24913 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
24914 * config/i386/i386.c (ix86_target_string): Add -msgx.
24916 (ix86_option_override_internal): Handle new options.
24917 (ix86_valid_target_attribute_inner_p): Add sgx.
24918 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
24919 * config/i386/i386.opt: Add msgx.
24920 * config/i386/sgxintrin.h: New file.
24921 * config/i386/x86intrin.h: Add sgxintrin.h.
24923 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24926 * fold-const.c (maybe_nonzero_address): Return 1 for function
24928 (tree_single_nonzero_warnv_p): Don't handle function local objects
24932 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
24935 2017-01-11 David Malcolm <dmalcolm@redhat.com>
24938 * opts.c: Include "spellcheck.h"
24939 (struct string_fragment): New struct.
24940 (struct edit_distance_traits<const string_fragment &>): New
24942 (get_closest_sanitizer_option): New function.
24943 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
24945 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24947 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
24949 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
24950 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
24951 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
24952 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
24953 for initial die_offset if dwarf_split_debug_info.
24954 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
24955 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
24957 (output_skeleton_debug_sections): Formatting fix. Use
24958 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
24959 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
24961 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
24963 * config/arm/cortex-a53.md: Add bypasses for
24964 cortex_a53_r2f_cvt.
24965 (cortex_a53_r2f): Only use for transfers.
24966 (cortex_a53_f2r): Likewise.
24967 (cortex_a53_r2f_cvt): Add reservation for conversions.
24968 (cortex_a53_f2r_cvt): Likewise.
24970 2017-01-11 Tamar Christina <tamar.christina@arm.com>
24972 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
24973 to all inlined functions, change static to extern.
24975 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
24978 * config/arm/arm.c (legitimize_pic_address): Handle reference to
24980 (arm_assemble_integer): Likewise.
24982 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
24984 * config.gcc: Use new awk script to check CPU, FPU and architecture
24985 parameters for --with-... options.
24986 * config/arm/parsecpu.awk: New file
24987 * config/arm/arm-cpus.in: New file.
24988 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
24990 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
24992 * config/arm/t-arm: Update dependency rules.
24993 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
24994 of processing .def files.
24995 * config/arm/genopt.sh: Deleted.
24996 * config/arm/gentune.sh: Deleted.
24997 * config/arm/arm-cores.def: Deleted.
24998 * config/arm/arm-arches.def: Deleted.
24999 * config/arm/arm-fpus.def: Deleted.
25000 * config/arm/arm-tune.md: Regenerated.
25001 * config/arm/arm-tables.opt: Regenerated.
25002 * config/arm/arm-cpu.h: New generated file.
25003 * config/arm/arm-cpu-data.h: New generated file.
25004 * config/arm/arm-cpu-cdata.h: New generated file.
25006 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
25009 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
25011 (input_varpool_node): Unpack dynamically_initialized bit.
25013 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
25015 PR rtl-optimization/79032
25016 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
25017 the alignment of the adjusted memory reference against that of MODE,
25018 instead of the alignment of the original memory reference.
25020 2017-01-11 Martin Jambor <mjambor@suse.cz>
25022 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
25024 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
25025 decorated functions.
25027 2017-01-11 Richard Biener <rguenther@suse.de>
25029 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
25030 set range/nonnull info for PHI results. Do not set it on
25031 stmts marked for removal.
25033 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
25035 * expr.c (store_field): In the bitfield case, fetch the return value
25036 from the registers before applying a single big-endian adjustment.
25037 Always do a final load for a BLKmode value not larger than a word.
25039 2017-01-10 David Malcolm <dmalcolm@redhat.com>
25042 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25043 that we correctly handle column numbers greater than
25044 LINE_MAP_MAX_COLUMN_NUMBER.
25046 2017-01-10 Martin Sebor <msebor@redhat.com>
25048 PR middle-end/78245
25049 * gimple-ssa-sprintf.c (get_destination_size): Call
25050 {init,fini}object_sizes.
25051 * tree-object-size.c (addr_object_size): Adjust.
25052 (pass_through_call): Adjust.
25053 (pass_object_sizes::execute): Adjust.
25054 * tree-object-size.h (fini_object_sizes): Declare.
25056 2017-01-10 Martin Sebor <msebor@redhat.com>
25058 PR tree-optimization/78775
25059 * builtins.c (get_size_range): Move...
25060 * calls.c: ...to here.
25061 (alloc_max_size): Accept zero argument.
25062 (operand_signed_p): Remove.
25063 (maybe_warn_alloc_args_overflow): Call get_size_range.
25064 * calls.h (get_size_range): Declare.
25066 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
25068 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
25069 from TI's devices.csv file as of September 2016.
25070 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
25072 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
25074 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
25075 * doc/invoke.texi: Likewise.
25076 * doc/md.texi: Likewise.
25077 * doc/objc.texi: Likewise.
25079 2017-01-10 Joshua Conner <joshconner@google.com>
25081 * config/arm/fuchsia-elf.h: New file.
25082 * config/fuchsia.h: New file.
25083 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
25084 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
25086 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
25088 2016-01-10 Richard Biener <rguenther@suse.de>
25090 PR tree-optimization/79034
25091 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25092 Propagate out degenerate PHIs in the joiner.
25094 2017-01-10 Martin Liska <mliska@suse.cz>
25096 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
25097 (sort_congruence_classes_by_decl_uid): Likewise.
25098 (sort_congruence_class_groups_by_decl_uid): Likewise.
25099 (sem_item_optimizer::merge_classes): Sort class, groups in these
25100 classes and members in the groups by DECL_UID of declarations.
25101 This would make merge operations stable.
25103 2017-01-10 Martin Liska <mliska@suse.cz>
25105 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
25106 usage of m_classes_vec.
25107 (sem_item_optimizer::~sem_item_optimizer): Likewise.
25108 (sem_item_optimizer::get_group_by_hash): Likewise.
25109 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
25110 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
25111 (sem_item_optimizer::verify_classes): Likewise.
25112 (sem_item_optimizer::process_cong_reduction): Likewise.
25113 (sem_item_optimizer::dump_cong_classes): Likewise.
25114 (sem_item_optimizer::merge_classes): Likewise.
25115 * ipa-icf.h (congruence_class_hash): Rename from
25116 congruence_class_group_hash. Remove declaration of m_classes_vec.
25118 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
25120 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
25121 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
25122 * config.gcc: Add avx512vpopcntdqintrin.h.
25123 * config/i386/avx512vpopcntdqintrin.h: New.
25124 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
25125 * config/i386/i386-builtin-types.def: Add new types.
25126 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
25127 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
25128 __builtin_ia32_vpopcountq_v8di_mask): New.
25129 * config/i386/i386-c.c (ix86_target_macros_internal): Define
25130 __AVX512VPOPCNTDQ__.
25131 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
25132 (PTA_AVX512VPOPCNTDQ): Define.
25133 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
25134 TARGET_AVX512VPOPCNTDQ_P): Define.
25135 * config/i386/i386.opt: Add mavx512vpopcntdq.
25136 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
25137 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
25139 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25141 PR middle-end/77484
25142 * predict.def (PRED_CALL): Set to 67.
25144 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
25146 * expr.c (store_field): In the bitfield case, if the value comes from
25147 a function call and is of an aggregate type returned in registers, do
25148 not modify the field mode; extract the value in all cases if the mode
25149 is BLKmode and the size is not larger than a word.
25151 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
25154 * config/i386/cpuid.h: Fix undefined behavior.
25156 2017-01-04 Jeff Law <law@redhat.com>
25158 PR tree-optimization/79007
25159 PR tree-optimization/67955
25160 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
25161 conservative for pt.null when flag_non_call_exceptions is on.
25163 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25165 PR translation/79019
25166 PR translation/79020
25167 * params.def (PARAM_INLINE_MIN_SPEEDUP,
25168 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
25169 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
25171 * config/avr/avr.opt (maccumulate-args): Likewise.
25172 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
25173 * common.opt (freport-bug): Likewise.
25174 * cif-code.def (CIF_FINAL_ERROR): Likewise.
25175 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
25176 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
25177 translatable string.
25178 * config/i386/i386.c (function_value_32): Likewise.
25179 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
25180 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
25182 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
25183 * common/config/msp430/msp430-common.c (msp430_handle_option):
25185 * symtab.c (symtab_node::verify_base): Likewise.
25186 * opts.c (set_debug_level): Likewise.
25187 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
25188 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
25189 missing whitespace to translatable strings.
25190 * config/avr/avr.md (bswapsi2): Fix typo in comment.
25191 * config/sh/superh.h: Likewise.
25192 * config/i386/xopintrin.h: Likewise.
25193 * config/i386/znver1.md: Likewise.
25194 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
25195 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
25196 * double-int.h (struct double_int): Likewise.
25197 * double-int.c (div_and_round_double): Likewise.
25198 * wide-int.cc: Likewise.
25199 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
25200 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
25201 * cfgcleanup.c (crossjumps_occured): Renamed to ...
25202 (crossjumps_occurred): ... this.
25203 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
25206 PR tree-optimization/78899
25207 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
25208 returning bool return struct loop *, NULL for failure and the new
25210 (versionable_outer_loop_p): Don't version outer loop if it has
25211 dont_vectorized bit set.
25212 (tree_if_conversion): When versioning outer loop, ensure
25213 tree_if_conversion is performed also on the inner loop of the
25214 non-vectorizable outer loop copy.
25215 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
25216 LOOP_VECTORIZED in inner loop of the scalar outer loop and
25217 prevent vectorization of it.
25218 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
25219 the outer loop vectorization of the non-scalar version is attempted
25220 before vectorization of the inner loop in scalar version. If
25221 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
25222 vectorization of its inner loop.
25223 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
25224 has 2 inner loops, rename also on edges from bb whose single pred
25225 is outer_loop->header. Fix typo in function comment.
25227 2017-01-09 Martin Sebor <msebor@redhat.com>
25230 * asan.c (asan_emit_stack_protection): Increase local buffer size
25231 to avoid snprintf truncation warning.
25233 2017-01-09 Andrew Pinski <apinski@cavium.com>
25235 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
25236 to reference thunderx2t99 for the tuning structure
25237 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
25239 (thunderx2t99_extra_costs): This.
25240 * config/aarch64/aarch64-tune.md: Regenerate.
25241 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
25242 (vulcan_addrcost_table): This.
25243 (vulcan_regmove_cost): Rename to ...
25244 (thunderx2t99_regmove_cost): This.
25245 (vulcan_vector_cost): Rename to ...
25246 (thunderx2t99_vector_cost): this.
25247 (vulcan_branch_cost): Rename to ...
25248 (thunderx2t99_branch_cost): This.
25249 (vulcan_tunings): Rename to ...
25250 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
25251 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
25253 2017-01-09 Martin Jambor <mjambor@suse.cz>
25257 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
25258 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
25259 (propagate_vr_accross_jump_function): Use the above function for all
25260 value range computations for pass-through jump functions and type
25261 converasion from explicit value range values.
25262 (ipcp_propagate_stage): Do not attempt to deduce types of formal
25263 parameters from TYPE_ARG_TYPES.
25264 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
25265 (ipa_write_node_info): Stream type of the actual argument.
25266 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
25268 2017-01-09 Martin Liska <mliska@suse.cz>
25271 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
25272 (lookup_compiler): Do not show error message with have_E.
25274 2017-01-09 Jakub Jelinek <jakub@redhat.com>
25276 PR tree-optimization/78938
25277 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
25278 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
25279 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
25280 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
25283 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25285 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
25288 2017-01-09 Richard Biener <rguenther@suse.de>
25290 PR tree-optimization/78997
25291 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
25292 name condition properly.
25294 2017-01-09 Richard Biener <rguenther@suse.de>
25297 * dwarf2out.c (is_cxx): New overload with context.
25298 (is_naming_typedef_decl): Use it.
25300 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25302 * invoke.texi (Option Summary): Correct spacing in option lists
25303 and add line breaks to fix over-long lines.
25305 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25307 PR middle-end/17660
25309 * extend.texi (Common Variable Attributes): Add xref to GCC
25310 Internals manual to explain mode attribute keywords.
25312 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
25315 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
25316 and Preprocessor Options.
25317 (Options for Linking): Document -pthread here....
25318 (RS/6000 and PowerPC Options): ...not here.
25319 (Solaris 2 Options): ...or here.
25320 * doc/cppopts.texi: Document -pthread.
25322 2017-01-08 Martin Sebor <msebor@redhat.com>
25324 PR middle-end/77708
25325 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
25326 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
25327 New member functions.
25328 (format_directive): Used them.
25330 (pass_sprintf_length::handle_gimple_call): Same.
25331 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
25332 to avoid truncation for any argument.
25333 (extract_affine_mul): Same.
25334 * tree.c (get_file_function_name): Same.
25336 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25338 PR middle-end/77484
25339 * predict.def (PRED_INDIR_CALL): Set to 86.
25341 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25343 PR preprocessor/54124
25344 * doc/cppopts.texi: Reformat -d subtable to list the full name
25345 of the options. Add cross-reference to the docs for the general
25346 compiler -d options.
25347 * doc/invoke.texi (Developer Options): Add cross-reference to the
25348 preprocessor-specific -d option documentation.
25350 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25352 PR preprocessor/13498
25353 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
25354 redudant material, and reflect new command-line options.
25355 (System Headers): Likewise.
25357 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
25359 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
25360 -isystem, and -idirafter. Copy-edit.
25361 * doc/cppopts.texi: Copy-edit. Remove contradiction about
25362 default for -ftrack-macro-expansion. Delete obsolete and
25363 badly-formatted implementation details about -fdebug-cpp output.
25364 * doc/cppwarnopts.texi: Copy-edit.
25366 2017-01-07 David Malcolm <dmalcolm@redhat.com>
25369 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
25370 that the transition from a max line width >= 1<<10 to narrower
25371 lines works correctly.
25373 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
25375 * doc/options.texi (PerFunction): New.
25376 * opt-functions.awk (switch_flags): Map both Optimization and
25377 PerFunction to CL_OPTIMIZATION.
25378 * opth-gen.awk: Test for PerFunction flag along with
25380 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
25381 it only when the latter is present. Skip those that don't in
25382 the hash function generator.
25383 * common.opt (fvar-tracking): Mark as PerFunction instead of
25385 (fvar-tracking-assignments): Likewise.
25386 (fvar-tracking-assignments-toggle): Likewise.
25387 (fvar-tracking-uninit): Likewise.
25389 2017-01-07 Jakub Jelinek <jakub@redhat.com>
25391 PR translation/79018
25392 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
25395 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
25398 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
25399 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
25400 TARGET_LONG_JUMP_TABLE_OFFSETS.
25401 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
25402 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
25403 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
25404 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
25405 * config/m68k/m68k.md (tablejump expander): Likewise.
25406 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
25407 TARGET_LONG_JUMP_TABLE_OFFSETS.
25408 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
25409 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
25411 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
25412 David Holsgrove <david.holsgrove@xilinx.com>
25414 * common/config/microblaze/microblaze-common.c
25415 (TARGET_EXCEPT_UNWIND_INFO): Remove.
25416 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
25418 * config/microblaze/microblaze.c (microblaze_must_save_register)
25419 (microblaze_expand_epilogue, microblaze_return_addr): Handle
25421 (microblaze_eh_return): New function.
25422 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
25423 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
25424 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
25425 * config/microblaze/microblaze.md (eh_return): New pattern.
25427 2017-01-06 Jakub Jelinek <jakub@redhat.com>
25429 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
25430 GCC_DIAGNOSTIC_STRINGIFY): Define.
25432 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
25434 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25436 * config/arm/arm.md (<mcrr>): New.
25438 * config/arm/arm.c (arm_arch5te): New.
25439 (arm_option_override): Set arm_arch5te.
25440 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
25442 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
25443 (arm_mcrr_qualifiers): ... this. New.
25444 (MRRC_QUALIFIERS): Define to...
25445 (arm_mrrc_qualifiers): ... this. New.
25446 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
25448 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
25449 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
25450 (MRRCI, mrrc, MRRC): New.
25451 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
25452 VUNSPEC_MRRC2): New.
25454 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25456 * config/arm/arm.md (<mcr>): New.
25458 * config/arm/arm.c (arm_coproc_builtin_available): Add
25459 support for mcr, mrc, mcr2 and mrc2.
25460 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
25461 (arm_mcr_qualifiers): ... this. New.
25462 (MRC_QUALIFIERS): Define to ...
25463 (arm_mrc_qualifiers): ... this. New.
25464 (MCR_QUALIFIERS): Define to ...
25465 (arm_mcr_qualifiers): ... this. New.
25466 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
25468 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
25469 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
25470 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
25471 VUNSPEC_MRC2): New.
25473 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25475 * config/arm/arm.md (*ldc): New.
25479 * config/arm/arm.c (arm_coproc_builtin_available): Add
25480 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
25481 (arm_coproc_ldc_stc_legitimate_address): New.
25482 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
25483 'qualifier_const_pointer'.
25484 (LDC_QUALIFIERS): Define to...
25485 (arm_ldc_qualifiers): ... this. New.
25486 (STC_QUALIFIERS): Define to...
25487 (arm_stc_qualifiers): ... this. New.
25488 * config/arm/arm-protos.h
25489 (arm_coproc_ldc_stc_legitimate_address): New.
25490 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
25491 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
25492 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
25493 stc2, stcl, stc2l): New.
25494 * config/arm/constraints.md (Uz): New.
25495 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
25496 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
25497 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
25498 VUNSPEC_STC2L): New.
25500 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25502 * config/arm/arm.md (<cdp>): New.
25503 * config/arm/arm.c (neon_const_bounds): Rename this ...
25504 (arm_const_bounds): ... this.
25505 (arm_coproc_builtin_available): New.
25506 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
25507 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
25508 (CDP_QUALIFIERS): Define to...
25509 (arm_cdp_qualifiers): ... this. New.
25511 (arm_expand_builtin_args): Add case for 6 arguments.
25512 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
25513 (arm_const_bounds): ... this.
25514 (arm_coproc_builtin_available): New.
25515 * config/arm/arm_acle.h (__arm_cdp): New.
25517 * config/arm/arm_acle_builtins.def (cdp): New.
25519 * config/arm/iterators.md (CDPI,CDP,cdp): New.
25520 * config/arm/neon.md: Rename all 'neon_const_bounds' to
25521 'arm_const_bounds'.
25522 * config/arm/types.md (coproc): New.
25523 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
25524 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
25525 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
25526 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
25528 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25530 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
25531 (UBINOP_QUALIFIERS): New.
25533 (acle_builtin_data): New. Change comment.
25534 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
25535 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
25536 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
25537 arm_acle_builtins.def.
25538 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
25539 (arm_init_acle_builtins): New.
25540 (CRC32_BUILTIN): Remove.
25541 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
25542 crc32cb, crc32ch and crc32cw.
25543 (arm_init_crc32_builtins): Remove.
25544 (arm_init_builtins): Use arm_init_acle_builtins rather
25545 than arm_init_crc32_builtins.
25546 (arm_expand_acle_builtin): New.
25547 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
25548 * config/arm/arm_acle_builtins.def: New.
25550 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
25552 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
25553 (arm_builtin_datum): ... this.
25554 (arm_init_neon_builtin): Rename to ...
25555 (arm_init_builtin): ... this. Add a new parameters PREFIX
25556 and USE_SIG_IN_NAME.
25557 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
25558 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
25559 'arm_builtin_datum'.
25560 (arm_init_vfp_builtins): Likewise.
25561 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
25562 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
25563 (arm_expand_neon_args): Rename to ...
25564 (arm_expand_builtin_args): ... this. Rename builtin_arg
25565 enum values and differentiate between ARG_BUILTIN_MEMORY
25566 and ARG_BUILTIN_NEON_MEMORY.
25567 (arm_expand_neon_builtin_1): Rename to ...
25568 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
25569 values, arm_expand_builtin_args and add bool parameter NEON.
25570 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
25571 (arm_expand_vfp_builtin): Likewise.
25572 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
25574 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25576 PR middle-end/77484
25577 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
25578 * predict.c (tree_estimate_probability_bb): Reverse direction of
25579 polymorphic call predictor.
25581 2017-01-06 David Malcolm <dmalcolm@redhat.com>
25583 * passes.c (execute_one_pass): Split out pass-skipping logic into...
25584 (determine_pass_name_match): ...this new function and...
25585 (should_skip_pass_p): ...this new function.
25587 2017-01-06 Nathan Sidwell <nathan@acm.org>
25589 * ipa-visibility.c (function_and_variable_visibility): Reformat
25590 comments and long lines. Remove extrneous if.
25591 * symtab.c (symtab_node::make_decl_local): Fix code format.
25592 (symtab_node::set_section_for_node): Fix comment typo.
25594 2017-01-06 Martin Liska <mliska@suse.cz>
25597 * lra-constraints.c: Rename invariant to lra_invariant.
25598 * predict.c (set_even_probabilities): Initialize e to NULL.
25600 2017-01-05 Martin Sebor <msebor@redhat.com>
25602 PR tree-optimization/78910
25603 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
25604 (format_integer): Correct off-by-one error in the handling
25605 of precision with negative numbers in signed conversions..
25607 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
25609 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
25611 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25613 PR tree-optimization/71016
25614 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
25615 factor_out_conditional_conversion. Formatting fix.
25616 (factor_out_conditional_conversion): Add cond_stmt argument.
25617 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
25618 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
25621 2017-01-05 David Malcolm <dmalcolm@redhat.com>
25623 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
25624 read-rtl-function.o, and selftest-rtl.o.
25625 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
25626 (selftest::aarch64_test_loading_full_dump): New function.
25627 (selftest::aarch64_run_selftests): New function.
25628 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
25629 selftest::aarch64_run_selftests.
25630 * config/i386/i386.c
25631 (selftest::ix86_test_loading_dump_fragment_1): New function.
25632 (selftest::ix86_test_loading_call_insn): New function.
25633 (selftest::ix86_test_loading_full_dump): New function.
25634 (selftest::ix86_test_loading_unspec): New function.
25635 (selftest::ix86_run_selftests): Call the new functions.
25636 * emit-rtl.c (maybe_set_max_label_num): New function.
25637 * emit-rtl.h (maybe_set_max_label_num): New decl.
25638 * function.c (instantiate_decls): Guard call to
25639 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
25640 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
25642 * gensupport.c (gen_reader::gen_reader): Pass "false"
25643 for new "compact" param of rtx_reader.
25644 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
25645 rather than an empty string for NULL strings.
25646 * read-md.c: Potentially include config.h rather than bconfig.h.
25647 Wrap include of errors.h with #ifdef GENERATOR_FILE.
25648 (have_error): New global, copied from errors.c.
25649 (md_reader::read_name): Rename to...
25650 (md_reader::read_name_1): ...this, adding "out_loc" param,
25651 and converting "missing name or number" to returning false, rather
25653 (md_reader::read_name): Reimplement in terms of read_name_1.
25654 (md_reader::read_name_or_nil): New function.
25655 (md_reader::read_string): Handle "(nil)" by returning NULL.
25656 (md_reader::md_reader): Add new param "compact".
25657 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
25658 (md_reader::read_file): New method.
25659 * read-md.h (md_reader::md_reader): Add new param "compact".
25660 (md_reader::read_file): New method.
25661 (md_reader::is_compact): New accessor.
25662 (md_reader::read_name): Convert return type from void to file_location.
25663 (md_reader::read_name_or_nil): New decl.
25664 (md_reader::read_name_1): New decl.
25665 (md_reader::m_compact): New field.
25666 (noop_reader::noop_reader): Pass "false" for new "compact" param
25668 (rtx_reader::rtx_reader): Add new "compact" param.
25669 (rtx_reader::read_rtx_operand): Make virtual and convert return
25670 type from void to rtx.
25671 (rtx_reader::read_until): New decl.
25672 (rtx_reader::handle_any_trailing_information): New virtual function.
25673 (rtx_reader::postprocess): New virtual function.
25674 (rtx_reader::finalize_string): New virtual function.
25675 (rtx_reader::m_in_call_function_usage): New field.
25676 (rtx_reader::m_reuse_rtx_by_id): New field.
25677 * read-rtl-function.c: New file.
25678 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
25679 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
25680 (selftest::verify_three_block_rtl_cfg): New decl.
25681 * read-rtl-function.h: New file.
25682 * read-rtl.c: Potentially include config.h rather than bconfig.h.
25683 For host, include function.h, memmodel.h, and emit-rtl.h.
25684 (one_time_initialization): New function.
25685 (struct compact_insn_name): New struct.
25686 (compact_insn_names): New array.
25687 (find_code): Handle insn codes in compact dumps.
25688 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
25689 (bind_subst_iter_and_attr): Likewise.
25690 (add_condition_to_string): Likewise.
25691 (add_condition_to_rtx): Likewise.
25692 (apply_attribute_uses): Likewise.
25693 (add_current_iterators): Likewise.
25694 (apply_iterators): Likewise.
25695 (initialize_iterators): Guard usage of apply_subst_iterator with
25696 #ifdef GENERATOR_FILE.
25697 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
25698 (md_reader::read_mapping): Likewise.
25699 (add_define_attr_for_define_subst): Likewise.
25700 (add_define_subst_attr): Likewise.
25701 (read_subst_mapping): Likewise.
25702 (check_code_iterator): Likewise.
25703 (rtx_reader::read_rtx): Likewise. Move one-time initialization
25705 (one_time_initialization): New function.
25706 (rtx_reader::read_until): New method.
25707 (read_flags): New function.
25708 (parse_reg_note_name): New function.
25709 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
25710 Handle reuse_rtx ids.
25711 Wrap iterator lookup within #ifdef GENERATOR_FILE.
25712 Add parsing support for RTL dumps, mirroring the special-cases in
25713 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
25714 values, and calling handle_any_trailing_information.
25715 (rtx_reader::read_rtx_operand): Convert return type from void
25716 to rtx, returning return_rtx. Handle case 'e'. Call
25717 finalize_string on XSTR and XTMPL fields.
25718 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
25719 "(nil)" values were omitted. Call the postprocess vfunc on the
25721 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
25722 class ctor. Initialize m_in_call_function_usage. Call
25723 one_time_initialization.
25724 * rtl-tests.c (selftest::test_uncond_jump): Call
25725 set_new_first_and_last_insn.
25726 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
25727 * selftest-rtl.c: New file.
25728 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
25729 (selftest::get_insn_by_uid): New decl.
25730 * selftest-run-tests.c (selftest::run_tests): Call
25731 read_rtl_function_c_tests.
25732 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
25733 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
25736 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
25738 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
25739 operands in a special way. Assert that pos+len <= mode precision.
25741 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25743 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
25744 3 argument Alias with unlimited for the negative form.
25745 (fno-vect-cost-model): Removed.
25747 2017-01-05 Martin Liska <mliska@suse.cz>
25749 * hsa-gen.c (gen_hsa_divmod): New function.
25750 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
25752 2017-01-05 Martin Liska <mliska@suse.cz>
25755 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
25758 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25760 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
25761 small constant length operands.
25763 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25765 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
25766 between loop iterations.
25768 2017-01-05 Martin Liska <mliska@suse.cz>
25771 * gimplify.c (gimplify_decl_expr): Compare to
25772 asan_poisoned_variables instread of checking flags.
25773 (gimplify_target_expr): Likewise.
25774 (gimplify_expr): Likewise.
25775 (gimplify_function_tree): Conditionally initialize
25776 asan_poisoned_variables.
25778 2017-01-04 Jeff Law <law@redhat.com>
25780 PR tree-optimizatin/78812
25781 * rtl.h (contains_mem_rtx_p): Prototype.
25782 * ifcvt.c (containts_mem_rtx_p): Move from here to...
25783 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
25784 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
25785 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
25786 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
25788 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25790 * input.c (assert_char_at_range): Default-initialize actual_range.
25792 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25794 * df-scan.c (df_ref_create_structure): Make regno unsigned,
25795 to match the caller.
25797 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25799 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
25800 insns after final jump in test to emit dummy move.
25802 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25804 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
25805 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
25807 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25809 * multiple_target.c (create_dispatcher_calls): Init e_next.
25810 * tree-ssa-loop-split.c (split_loop): Init border.
25811 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
25814 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
25819 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
25820 (altivec_register_operand): Do not return true if the operand
25821 contains a SUBREG mixing SImode and SFmode.
25822 (vsx_register_operand): Likewise.
25823 (vsx_reg_sfsubreg_ok): New predicate.
25824 (vfloat_operand): Do not return true if the operand contains a
25825 SUBREG mixing SImode and SFmode.
25826 (vint_operand): Likewise.
25827 (vlogical_operand): Likewise.
25828 (gpc_reg_operand): Likewise.
25829 (int_reg_operand): Likewise.
25830 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
25831 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
25832 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
25834 (rs6000_emit_move_si_sf_subreg): New helper function.
25835 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
25836 fixup SUBREGs involving SImode and SFmode.
25837 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
25838 numbers for the new peephole2 optimization.
25839 (peephole2 for SFmode unions): New peephole2 to optimize cases in
25840 the GLIBC math library that do AND/IOR/XOR operations on single
25841 precision floating point.
25842 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
25843 target macros to say whether we need to avoid SUBREGs mixing
25845 (TARGET_ALLOW_SF_SUBREG): Likewise.
25846 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
25847 (UNSPEC_SI_FROM_SF): Likewise.
25848 (iorxor): Change spacing.
25849 (and_ior_xor): New iterator for AND, IOR, and XOR.
25850 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
25851 (movdi_from_sf_zero_ext): Likewise.
25852 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
25853 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
25854 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
25855 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
25856 (fms<mode>4): Likewise.
25857 (fnma<mode>4): Likewise.
25858 (fnms<mode>4): Likewise.
25859 (nfma<mode>4): Likewise.
25860 (nfms<mode>4): Likewise.
25862 2017-01-04 Marek Polacek <polacek@redhat.com>
25865 * doc/invoke.texi: Document -Wpointer-compare.
25867 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25869 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
25872 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
25873 descriptions for -gdwarf-5 and emit them as uleb128 instead of
25876 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
25879 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
25880 documentation of the powerpc_popcntb_ok attribute.
25881 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25882 code to issue warning messages if a requested CPU configuration is
25883 not supported by the binary (assembler and loader) toolchain.
25884 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
25885 made to define a built-in function that has been disabled.
25886 (paired_init_builtins): Add assertion to prevent ICE if attempt is
25887 made to define a built-in function that has been disabled.
25888 (altivec_init_builtins): Add comment explaining why definition
25889 of the DST built-in functions is not preceded by an assertion
25890 check. Add assertions to prevent ICE if attempts are made to
25891 define an altivec predicate or an abs* built-in function that has
25893 (htm_init_builtins): Add comment explaining why definition of the
25894 htm built-in functions is not preceded by an assertion check.
25896 2017-01-04 Jeff Law <law@redhat.com>
25898 PR tree-optimizatin/67955
25899 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
25900 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
25901 the points-to solution does not include pt_null. Use DECL_PT_UID
25904 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
25906 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
25907 Use gen_int_mode instead of gen_lopwart for const_int operands.
25909 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25911 PR tree-optimization/71563
25912 * match.pd: Simplify X << Y into X if Y is known to be 0 or
25913 out of range value - has low bits known to be zero.
25915 2017-01-04 Alan Modra <amodra@gmail.com>
25917 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
25918 * configure: Regenerate.
25919 * config.in: Regenerate.
25921 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25924 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
25925 a substring of the message, but strcmp with the whole message. Ifdef
25926 ENABLE_NLS, translate the message first using dgettext.
25928 2017-01-03 Jeff Law <law@redhat.com>
25930 PR tree-optimizatin/78856
25931 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
25932 (mark_threaded_blocks): Remove code to truncate thread paths that
25933 cross multiple loop headers. Instead invalidate the cached loop
25934 iteration information and handle case of a thread path walking
25935 into an irreducible region.
25937 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
25940 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
25941 assertions. Add support for doing the signbit if the IEEE 128-bit
25942 floating point value is in a GPR.
25943 * config/rs6000/rs6000.md (Fsignbit): Delete.
25944 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
25945 Update the length attribute if the value is in a GPR.
25946 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
25947 the sign or zero extension instruction, since the value is always 0/1.
25948 (signbit<mode>2_dm2): Delete using <Fsignbit>.
25951 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
25952 extracting SImode to a GPR register so that we can generate a
25953 store, limit the vector to be in a traditional Altivec register
25954 for the vextuwrx instruction.
25956 2017-01-03 Ian Lance Taylor <iant@google.com>
25958 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
25960 2017-01-03 Martin Sebor <msebor@redhat.com>
25962 PR tree-optimization/78696
25963 * gimple-ssa-sprintf.c (format_floating): Correct handling of
25964 precision. Use MPFR for %f for greater fidelity. Correct handling
25966 (pass_sprintf_length::compute_format_length): Set width and precision
25967 specified by asrerisk to void_node for vararg functions.
25968 (try_substitute_return_value): Adjust dump output.
25970 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
25972 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
25974 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
25976 * doc/invoke.texi (SPARC options): Document -mlra as the default.
25977 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
25978 -mlra/-mno-lra was passed to the compiler.
25980 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
25982 PR rtl-optimization/65618
25983 * emit-rtl.c (try_split): Move initialization of "before" and
25984 "after" to just before the call to emit_insn_after_setloc.
25986 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
25988 * doc/md.texi (Standard Names): Remove reference to Java frontend.
25990 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
25992 * dwarf2out.c (gen_enumeration_type_die): When
25993 -gno-strict-dwarf, add a DW_AT_encoding attribute.
25995 2017-01-03 Jakub Jelinek <jakub@redhat.com>
25997 PR tree-optimization/78965
25998 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
25999 Change first argument from const call_info & to call_info &. For %n
26000 set info.nowrite to false.
26002 PR middle-end/78901
26003 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
26004 possibly throwing calls.
26006 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
26007 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
26008 and fns handling, rather than in a separate case SSA_NAME.
26010 2017-01-02 Jeff Law <law@redhat.com>
26012 * config/darwin-driver.c (darwin_driver_init): Const-correctness
26013 fixes for first_period and second_period variables.
26015 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
26018 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
26019 (*insvqi_1): New insn pattern.
26020 (*insvqi_1_mem_rex64): Ditto.
26021 (*insvqi_2): Ditto.
26022 (*insvqi_3): Rename from *insvqi.
26024 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
26026 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
26028 * doc/cfg.texi (Edges): Remove reference to Java.
26029 (Maintaining the CFG): Ditto.
26031 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26033 PR middle-end/77674
26034 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
26035 transparent aliases.
26037 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
26039 PR middle-end/77484
26040 * predict.def (PRED_CALL): Update hitrate.
26041 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
26042 * predict.c (tree_estimate_probability_bb): Split CALL predictor
26043 into direct/indirect/polymorphic variants.
26045 2017-01-01 Jakub Jelinek <jakub@redhat.com>
26047 Update copyright years.
26049 * gcc.c (process_command): Update copyright notice dates.
26050 * gcov-dump.c (print_version): Ditto.
26051 * gcov.c (print_version): Ditto.
26052 * gcov-tool.c (print_version): Ditto.
26053 * gengtype.c (create_file): Ditto.
26054 * doc/cpp.texi: Bump @copying's copyright year.
26055 * doc/cppinternals.texi: Ditto.
26056 * doc/gcc.texi: Ditto.
26057 * doc/gccint.texi: Ditto.
26058 * doc/gcov.texi: Ditto.
26059 * doc/install.texi: Ditto.
26060 * doc/invoke.texi: Ditto.
26062 Copyright (C) 2017 Free Software Foundation, Inc.
26064 Copying and distribution of this file, with or without modification,
26065 are permitted in any medium without royalty provided the copyright
26066 notice and this notice are preserved.