1 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * rtl.h (subreg_memory_offset): Declare.
6 * emit-rtl.c (subreg_memory_offset): New function.
7 * expmed.c (store_bit_field_1): Use it.
8 * expr.c (undefined_operand_subword_p): Likewise.
9 * simplify-rtx.c (simplify_subreg): Likewise.
11 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
13 PR rtl-optimization/57448
16 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17 using atomic_load pattern.
18 (expand_atomic_store): Likewise.
20 2017-09-04 Jakub Jelinek <jakub@redhat.com>
23 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
24 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
25 handling. Use replace_call_with_value with NULL instead of
26 gsi_replace, unlink_stmt_vdef and release_defs.
28 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
31 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
33 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
36 * rtl.h (emit_library_call_value_1): Declare.
37 (emit_library_call): Replace declaration with a series of overloads.
38 Remove the parameter count argument.
39 (emit_library_call_value): Likewise.
40 * calls.c (emit_library_call_value_1): Make global. Replace varargs
41 with an "rtx_mode_t *".
42 (emit_library_call_value): Delete.
43 (emit_library_call): Likewise.
44 * asan.c (asan_emit_stack_protection): Update calls accordingly.
45 (asan_emit_allocas_unpoison): Likewise.
46 * builtins.c (expand_builtin_powi): Likewise.
47 (expand_asan_emit_allocas_unpoison): Likewise.
48 * cfgexpand.c (expand_main_function): Likewise.
49 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
50 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
51 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
52 * config/arm/arm.c (arm_trampoline_init): Likewise.
53 (arm_call_tls_get_addr): Likewise.
54 (arm_expand_divmod_libfunc): Likewise.
55 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
56 (smulsi3_highpart): Likewise.
57 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
58 (c6x_expand_compare): Likewise.
59 (c6x_expand_movmem): Likewise.
60 * config/frv/frv.c (frv_trampoline_init): Likewise.
61 * config/i386/i386.c (ix86_trampoline_init): Likewise.
62 (ix86_expand_divmod_libfunc): Likewise.
63 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
64 (ia64_expand_compare): Likewise.
65 (ia64_profile_hook): Likewise.
66 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
67 (nonlocal_goto): Likewise.
68 (restore_stack_nonlocal): Likewise.
69 * config/m32r/m32r.c (block_move_call): Likewise.
70 (m32r_trampoline_init): Likewise.
71 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
72 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
73 (m68k_call_m68k_read_tp): Likewise.
74 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
75 (microblaze_expand_divide): Likewise.
76 * config/mips/mips.h (mips_args): Likewise.
77 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
78 (MIPS_ICACHE_SYNC): Likewise.
79 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
80 (nios2_trampoline_init): Likewise.
81 * config/pa/pa.c (hppa_tls_call): Likewise.
82 (pa_trampoline_init): Likewise.
83 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
84 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
85 (expand_strn_compare): Likewise.
86 (rs6000_generate_compare): Likewise.
87 (rs6000_expand_float128_convert): Likewise.
88 (output_profile_hook): Likewise.
89 (rs6000_trampoline_init): Likewise.
90 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
91 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
92 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
93 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
94 (rs6000_generate_compare): Likewise.
95 (rs6000_expand_float128_convert): Likewise.
96 (output_profile_hook): Likewise.
97 (rs6000_trampoline_init): Likewise.
98 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
99 * config/sh/sh.c (sh_trampoline_init): Likewise.
100 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
101 (sparc_emit_float_lib_cmp): Likewise.
102 (sparc32_initialize_trampoline): Likewise.
103 (sparc64_initialize_trampoline): Likewise.
104 (sparc_profile_hook): Likewise.
105 * config/spu/spu.c (ea_load_store): Likewise.
106 * config/spu/spu.md (floatunssidf2): Likewise.
107 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
108 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
109 * config/visium/visium.c (expand_block_move_4): Likewise.
110 (expand_block_move_2): Likewise.
111 (expand_block_move_1): Likewise.
112 (expand_block_set_4): Likewise.
113 (expand_block_set_2): Likewise.
114 (expand_block_set_1): Likewise.
115 (visium_trampoline_init): Likewise.
116 (visium_profile_hook): Likewise.
117 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
118 (xtensa_setup_frame_addresses): Likewise.
119 (xtensa_trampoline_init): Likewise.
120 * except.c (sjlj_emit_function_enter): Likewise.
121 (sjlj_emit_function_exit): Likewise.
122 * explow.c (allocate_dynamic_stack_space): Likewise.
123 (probe_stack_range): Likewise.
124 * expr.c (convert_mode_scalar): Likewise.
125 * optabs.c (expand_binop): Likewise.
126 (expand_twoval_binop_libfunc): Likewise.
127 (expand_unop): Likewise.
128 (prepare_cmp_insn): Likewise.
129 (prepare_float_lib_cmp): Likewise.
130 (expand_float): Likewise.
131 (expand_fix): Likewise.
132 (expand_fixed_convert): Likewise.
133 (maybe_emit_sync_lock_test_and_set): Likewise.
134 (expand_atomic_compare_and_swap): Likewise.
135 (expand_mem_thread_fence): Likewise.
136 (expand_atomic_fetch_op): Likewise.
138 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
140 * doc/generic.texi (OpenACC): Adjust URL.
141 * doc/invoke.texi (C Dialect Options): Ditto.
143 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
145 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
146 predicate for operand 1. Add (m,<S>) constraint.
147 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
148 Prevent memory operand 1 with register operand 2.
150 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
152 PR rtl-optimization/82024
153 * combine.c (try_combine): If the combination result is a PARALLEL,
154 and we only need to retain the SET in there that would be placed
155 at I2, check that we can place that at I3 instead, before doing so.
157 2017-09-01 Jakub Jelinek <jakub@redhat.com>
160 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
162 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
163 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
166 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
167 Jeff Law <law@redhat.com>
169 * varasm.c (bss_initializer_p): Do not put constants into .bss
170 (categorize_decl_for_section): Handle bss_initializer_p returning
171 false when DECL_INITIAL is NULL.
173 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
176 * config/s390/s390.c (s390_can_inline_p): New function.
178 2017-09-01 Jeff Law <law@redhat.com>
180 PR tree-optimization/82052
181 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
182 Always initialize the returned slot after a hash table miss
185 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
187 * config/s390/s390.md (mem_signal_fence): Remove.
188 * doc/md.texi (mem_signal_fence): Remove.
189 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
191 * target-insns.def (mem_signal_fence): Remove.
193 2017-09-01 Jakub Jelinek <jakub@redhat.com>
196 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
199 * asan.c (create_odr_indicator): Strip name encoding from assembler
200 name before appending it after __odr_asan_.
202 2017-09-01 Martin Liska <mliska@suse.cz>
204 PR tree-optimization/82059
205 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
206 frequency only when an edge is redirected.
208 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
210 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
211 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
212 (arc_conditional_register_usage): Remove ARC600 lp_count
214 (arc_file_start): Emit Tag_ARC_CPU_variation.
215 (arc_can_use_doloop_p): New conditions to use ZOLs.
216 (hwloop_fail): New function.
217 (hwloop_optimize): Likewise.
218 (hwloop_pattern_reg): Likewise.
219 (arc_doloop_hooks): New struct, to be used with reorg_loops.
220 (arc_reorg_loops): New function, calls reorg_loops.
221 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
222 (arc600_corereg_hazard): Remove ZOL checking, case handled by
224 (arc_loop_hazard): Remove function, functionality moved into
226 (arc_hazard): Remove arc_loop_hazard call.
227 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
228 into hwloop_optimize.
229 (arc_label_align): Remove ZOL handling.
230 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
231 * config/arc/arc.md (doloop_begin): Remove pattern.
232 (doloop_begin_i): Likewise.
233 (doloop_end_i): Likewise.
234 (doloop_fallback): Likewise.
235 (doloop_fallback_m): Likewise.
236 (doloop_end): Reimplement expand.
237 (arc_lp): New pattern for LP instruction.
238 (loop_end): New pattern.
239 (loop_fail): Likewise.
240 (decrement_and_branch_until_zero): Likewise.
241 * config/arc/arc.opt (mlpc-width): New option.
242 * doc/invoke.texi (mlpc-width): Document option.
244 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
246 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
247 (arc_ccfsm_advance): Fix checking for delay slots.
248 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
250 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
252 * config/arc/arc.md (movqi_insn): Add stores to save constant long
254 (movhi_insn): Update store instruction constraint which are saving
255 6-bit short immediates.
256 (movsi_insn): Consider also short scaled load operations.
257 (zero_extendhisi2_i): Use Usd constraint instead of T.
258 (extendhisi2_i): Add q constraint.
259 (arc_clzsi2): Add type and length attributes.
260 (arc_ctzsi2): Likewise.
261 * config/arc/constraints.md (Usc): Update constraint, the
262 assembler can parse two relocations for a single instruction.
264 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
266 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
267 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
269 2017-08-31 Olivier Hainque <hainque@adacore.com>
271 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
272 match as powerpc-wrs-vxworks*.
274 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
276 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
277 register constraint for by-element operand.
278 (aarch64_mls_elt_merge<mode>): Likewise.
280 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
282 * config/arc/arc.c (arc_can_follow_jump): Check for short
285 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
287 * config.gcc: Use g.opt for arc.
288 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
289 functionality moved to ...
290 (legitimate_scaled_address_p): New function, ...here.
291 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
292 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
293 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
295 (arc_override_options): Handle G option.
296 (arc_output_pic_addr_const): Correct function definition.
297 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
298 (arc_decl_anon_ns_mem_p): Delete.
299 (arc_in_small_data_p): Overhaul this function to take into
300 consideration the value given via G option.
301 (arc_rewrite_small_data_1): Renamed and corrected old
302 arc_rewrite_small_data function.
303 (arc_rewrite_small_data): New function.
304 (small_data_pattern): Don't use pic_offset_table_rtx.
305 * config/arc/arc.h (CC1_SPEC): Recognize G option.
306 * config/arc/simdext.md (movmisalignv2hi): Use
307 prepare_move_operands function.
309 (movmisalign*): Likewise.
310 * doc/invoke.texi (ARC options): Document -G option.
312 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
314 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
316 * config/arc/arc.c (arc_print_operand): Output scalled address for
317 sdata whenever is possible.
318 (arc_in_small_data_p): Allow sdata for 64bit datum when double
319 load/stores are available.
320 (compact_sda_memory_operand): Check for the alignment required by
321 code density instructions.
322 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
324 * config/arc/constraints.md (Usd): Update constraint.
325 (Us0): New constraint.
326 (Usc): Update constraint.
328 2017-08-31 Richard Biener <rguenther@suse.de>
331 * dwarf2out.c (dwarf2out_early_global_decl): Process each
334 2017-08-31 Tamar Christina <tamar.christina@arm.com>
336 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
337 Resize type_signature.
339 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
340 Alan Hayward <alan.hayward@arm.com>
341 David Sherwood <david.sherwood@arm.com>
343 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
344 subregs whose inner modes can be stored in GPRs.
345 (aarch64_classify_index): Likewise.
347 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
348 Alan Hayward <alan.hayward@arm.com>
349 David Sherwood <david.sherwood@arm.com>
351 * config/aarch64/iterators.md (V_cmp_result): Rename to...
352 (V_INT_EQUIV): ...this.
353 (v_cmp_result): Rename to...
354 (v_int_equiv): ...this.
355 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
356 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
357 (copysign<mode>3): Likewise.
358 (aarch64_simd_bsl<mode>_internal): Likewise.
359 (aarch64_simd_bsl<mode>): Likewise.
360 (vec_cmp<mode><mode>): Likewise.
361 (vcond<mode><mode>): Likewise.
362 (vcond<v_cmp_mixed><mode>): Likewise.
363 (vcondu<mode><v_cmp_mixed>): Likewise.
364 (aarch64_cm<optab><mode>): Likewise.
365 (aarch64_cmtst<mode>): Likewise.
366 (aarch64_fac<optab><mode>): Likewise.
367 (vec_perm_const<mode>): Likewise.
368 (vcond_mask_<mode><v_cmp_result>): Rename to...
369 (vcond_mask_<mode><v_int_equiv>): ...this.
370 (vec_cmp<mode><v_cmp_result>): Rename to...
371 (vec_cmp<mode><v_int_equiv>): ...this.
373 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
374 Alan Hayward <alan.hayward@arm.com>
375 David Sherwood <david.sherwood@arm.com>
377 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
379 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
380 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
381 (UNSPEC_LD4_DREG): New unspecs.
382 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
383 (aarch64_ld2<mode>_dreg_be): Replace with...
384 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
386 (aarch64_ld3<mode>_dreg_le)
387 (aarch64_ld3<mode>_dreg_be): Replace with...
388 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
390 (aarch64_ld4<mode>_dreg_le)
391 (aarch64_ld4<mode>_dreg_be): Replace with...
392 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
395 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
397 PR tree-optimization/81987
398 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
399 insert an initializer in a location not dominated by the stride
402 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
404 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
405 on the entire header of the finally block in the fallthru case.
407 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
409 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
411 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
413 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
414 rs6000_emit_move_from_cr and call renamed function.
415 (rs6000_emit_prologue): Call renamed functions.
416 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
417 movesi_from_cr, remove volatile CRs.
419 2017-08-30 Jon Beniston <jon@beniston.com>
420 Richard Biener <rguenther@suse.de>
422 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
423 of VECTOR_MODE_P check.
424 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
425 element vector types.
427 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
429 * df.h (df_read_modify_subreg_p): Remove in favor of...
430 * rtl.h (read_modify_subreg_p): ...this new function. Take a
431 const_rtx instead of an rtx.
432 * cprop.c (local_cprop_find_used_regs): Update accordingly.
433 * df-problems.c (df_word_lr_mark_ref): Likewise.
434 * ira-lives.c (mark_pseudo_reg_live): Likewise.
435 (mark_pseudo_reg_dead): Likewise.
436 (mark_ref_dead): Likewise.
437 * reginfo.c (init_subregs_of_mode): Likewise.
438 * sched-deps.c (sched_analyze_1): Likewise.
439 * df-scan.c (df_def_record_1): Likewise.
440 (df_uses_record): Likewise.
441 (df_read_modify_subreg_p): Remove in favor of...
442 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
443 const_rtx instead of an rtx.
445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
446 Alan Hayward <alan.hayward@arm.com>
447 David Sherwood <david.sherwood@arm.com>
449 * rtl.h (partial_subreg_p): New function.
450 * caller-save.c (save_call_clobbered_regs): Use it.
451 * calls.c (expand_call): Likewise.
452 * combine.c (combinable_i3pat): Likewise.
453 (simplify_set): Likewise.
454 (make_extraction): Likewise.
455 (make_compound_operation_int): Likewise.
456 (gen_lowpart_or_truncate): Likewise.
457 (force_to_mode): Likewise.
458 (make_field_assignment): Likewise.
459 (reg_truncated_to_mode): Likewise.
460 (record_truncated_value): Likewise.
461 (move_deaths): Likewise.
462 * cse.c (record_jump_cond): Likewise.
463 (cse_insn): Likewise.
464 * cselib.c (cselib_lookup_1): Likewise.
465 * expmed.c (extract_bit_field_using_extv): Likewise.
466 * function.c (assign_parm_setup_reg): Likewise.
467 * ifcvt.c (noce_convert_multiple_sets): Likewise.
468 * ira-build.c (create_insn_allocnos): Likewise.
469 * lra-coalesce.c (merge_pseudos): Likewise.
470 * lra-constraints.c (match_reload): Likewise.
471 (simplify_operand_subreg): Likewise.
472 (curr_insn_transform): Likewise.
473 * lra-lives.c (process_bb_lives): Likewise.
474 * lra.c (new_insn_reg): Likewise.
475 (lra_substitute_pseudo): Likewise.
476 * regcprop.c (mode_change_ok): Likewise.
477 (maybe_mode_change): Likewise.
478 (copyprop_hardreg_forward_1): Likewise.
479 * reload.c (push_reload): Likewise.
480 (find_reloads): Likewise.
481 (find_reloads_subreg_address): Likewise.
482 * reload1.c (alter_reg): Likewise.
483 (eliminate_regs_1): Likewise.
484 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
486 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
488 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
489 back to if statements, including unpack.
491 2017-08-30 Martin Liska <mliska@suse.cz>
494 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
496 (func_checker::compare_asm_inputs_outputs): ... this function.
497 (func_checker::compare_gimple_asm): Use the function to compare
499 * ipa-icf-gimple.h: Rename the function.
501 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
502 Alan Hayward <alan.hayward@arm.com>
503 David Sherwood <david.sherwood@arm.com>
505 * coretypes.h (complex_mode): New type.
506 * gdbhooks.py (build_pretty_printer): Handle it.
507 * machmode.h (complex_mode): New class.
508 (complex_mode::includes_p): New function.
509 (is_complex_int_mode): Likewise.
510 (is_complex_float_mode): Likewise.
511 * genmodes.c (get_mode_class): Handle complex mode classes.
512 * function.c (expand_function_end): Use is_complex_int_mode.
514 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
515 Alan Hayward <alan.hayward@arm.com>
516 David Sherwood <david.sherwood@arm.com>
518 * coretypes.h (scalar_mode_pod): New typedef.
519 * gdbhooks.py (build_pretty_printer): Handle it.
520 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
521 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
522 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
523 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
527 Alan Hayward <alan.hayward@arm.com>
528 David Sherwood <david.sherwood@arm.com>
530 * machmode.h (mode_for_vector): Take a scalar_mode instead
532 * stor-layout.c (mode_for_vector): Likewise.
533 * explow.c (promote_mode): Use as_a <scalar_mode>.
534 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
537 Alan Hayward <alan.hayward@arm.com>
538 David Sherwood <david.sherwood@arm.com>
540 * target.def (preferred_simd_mode): Take a scalar_mode
541 instead of a machine_mode.
542 * targhooks.h (default_preferred_simd_mode): Likewise.
543 * targhooks.c (default_preferred_simd_mode): Likewise.
544 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
545 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
546 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
547 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
548 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
549 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
550 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
551 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
553 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
554 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
555 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
556 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
557 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
558 * doc/tm.texi: Regenerate.
559 * optabs-query.c (can_vec_mask_load_store_p): Return false for
562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
563 Alan Hayward <alan.hayward@arm.com>
564 David Sherwood <david.sherwood@arm.com>
566 * target.def (scalar_mode_supported_p): Take a scalar_mode
567 instead of a machine_mode.
568 * targhooks.h (default_scalar_mode_supported_p): Likewise.
569 * targhooks.c (default_scalar_mode_supported_p): Likewise.
570 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
571 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
572 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
573 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
574 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
575 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
576 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
577 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
578 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
579 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
580 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
581 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
583 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
584 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
585 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
586 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
587 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
589 * doc/tm.texi: Regenerate.
591 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
592 Alan Hayward <alan.hayward@arm.com>
593 David Sherwood <david.sherwood@arm.com>
595 * coretypes.h (opt_scalar_mode): New typedef.
596 * gdbhooks.py (build_pretty_printers): Handle it.
597 * machmode.h (mode_iterator::get_2xwider): Add overload for
599 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
601 * expr.c (convert_mode_scalar): Likewise.
602 * omp-low.c (omp_clause_aligned_alignment): Likewise.
603 * optabs.c (expand_float): Likewise.
604 (expand_fix): Likewise.
605 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
608 Alan Hayward <alan.hayward@arm.com>
609 David Sherwood <david.sherwood@arm.com>
611 * optabs.c (expand_float): Explicitly check for scalars before
612 using a branching expansion.
613 (expand_fix): Likewise.
615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
616 Alan Hayward <alan.hayward@arm.com>
617 David Sherwood <david.sherwood@arm.com>
619 * expr.c (convert_mode): Split scalar handling out into...
620 (convert_mode_scalar): ...this new function. Treat the modes
623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
624 Alan Hayward <alan.hayward@arm.com>
625 David Sherwood <david.sherwood@arm.com>
627 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
629 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
632 Alan Hayward <alan.hayward@arm.com>
633 David Sherwood <david.sherwood@arm.com>
635 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
636 rather than a machine_mode.
637 (fixed_from_string): Likewise.
638 (fixed_convert): Likewise.
639 (fixed_convert_from_int): Likewise.
640 (fixed_convert_from_real): Likewise.
641 (real_convert_from_fixed): Likewise.
642 * fixed-value.c (fixed_from_double_int): Likewise.
643 (fixed_from_string): Likewise.
644 (fixed_convert): Likewise.
645 (fixed_convert_from_int): Likewise.
646 (fixed_convert_from_real): Likewise.
647 (real_convert_from_fixed): Likewise.
648 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
651 Alan Hayward <alan.hayward@arm.com>
652 David Sherwood <david.sherwood@arm.com>
654 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
655 of separate mode class checks. Do not allow vector modes here.
656 (immed_wide_int_const): Use as_a <scalar_mode>.
657 * explow.c (trunc_int_for_mode): Likewise.
658 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
659 (wi::shwi): Likewise.
660 (wi::min_value): Likewise.
661 (wi::max_value): Likewise.
662 * dwarf2out.c (loc_descriptor): Likewise.
663 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
666 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
667 Alan Hayward <alan.hayward@arm.com>
668 David Sherwood <david.sherwood@arm.com>
670 * tree.h (SCALAR_TYPE_MODE): New macro.
671 * expr.c (expand_expr_addr_expr_1): Use it.
672 (expand_expr_real_2): Likewise.
673 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
674 (fold_convert_const_fixed_from_int): Likewise.
675 (fold_convert_const_fixed_from_real): Likewise.
676 (native_encode_fixed): Likewise
677 (native_encode_complex): Likewise
678 (native_encode_vector): Likewise.
679 (native_interpret_fixed): Likewise.
680 (native_interpret_real): Likewise.
681 (native_interpret_complex): Likewise.
682 (native_interpret_vector): Likewise.
683 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
684 (simd_clone_adjust_argument_types): Likewise.
685 (simd_clone_init_simd_arrays): Likewise.
686 (simd_clone_adjust): Likewise.
687 * stor-layout.c (layout_type): Likewise.
688 * tree.c (build_minus_one_cst): Likewise.
689 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
690 * tree-inline.c (estimate_move_cost): Likewise.
691 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
692 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
693 (vectorizable_reduction): Likewise.
694 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
695 (vect_recog_mixed_size_cond_pattern): Likewise.
696 (check_bool_pattern): Likewise.
697 (adjust_bool_pattern): Likewise.
698 (search_type_for_mask_1): Likewise.
699 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
700 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
701 (vectorizable_load): Likewise.
702 (vectorizable_store): Likewise.
703 * ubsan.c (ubsan_encode_value): Likewise.
704 * varasm.c (output_constant): Likewise.
706 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
707 Alan Hayward <alan.hayward@arm.com>
708 David Sherwood <david.sherwood@arm.com>
710 * coretypes.h (scalar_mode): New class.
711 * machmode.h (scalar_mode): Likewise.
712 (scalar_mode::includes_p): New function.
713 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
714 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
715 * genmodes.c (get_mode_class): Handle remaining scalar modes.
716 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
717 * expmed.c (store_bit_field_1): Likewise.
718 (extract_bit_field_1): Likewise.
719 * expr.c (write_complex_part): Likewise.
720 (read_complex_part): Likewise.
721 (emit_move_complex_push): Likewise.
722 (expand_expr_real_2): Likewise.
723 * function.c (assign_parm_setup_reg): Likewise.
724 (assign_parms_unsplit_complex): Likewise.
725 * optabs.c (expand_binop): Likewise.
726 * rtlanal.c (subreg_get_info): Likewise.
727 * simplify-rtx.c (simplify_immed_subreg): Likewise.
728 * varasm.c (output_constant_pool_2): Likewise.
730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
731 Alan Hayward <alan.hayward@arm.com>
732 David Sherwood <david.sherwood@arm.com>
734 * expmed.c (extract_high_half): Use scalar_int_mode and remove
736 (expmed_mult_highpart_optab): Likewise.
737 (expmed_mult_highpart): Likewise.
739 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
740 Alan Hayward <alan.hayward@arm.com>
741 David Sherwood <david.sherwood@arm.com>
743 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
744 instead of a machine_mode.
745 (builtin_memset_read_str): Likewise.
746 * builtins.c (c_readstr): Likewise.
747 (builtin_memcpy_read_str): Likewise.
748 (builtin_strncpy_read_str): Likewise.
749 (builtin_memset_read_str): Likewise.
750 (builtin_memset_gen_str): Likewise.
751 (expand_builtin_signbit): Use scalar_int_mode for local variables.
752 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
753 instead of a machine_mode.
754 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
756 (make_extraction): Likewise.
757 (try_widen_shift_mode): Take and return scalar_int_modes instead
759 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
760 a scalar_int_mode instead of a machine_mode.
761 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
762 (avr_addr_space_pointer_mode): Likewise.
763 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
764 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
765 (msp430_unwind_word_mode): Likewise.
766 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
767 (spu_addr_space_pointer_mode): Likewise.
768 (spu_addr_space_address_mode): Likewise.
769 (spu_libgcc_cmp_return_mode): Likewise.
770 (spu_libgcc_shift_count_mode): Likewise.
771 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
772 (rl78_addr_space_pointer_mode): Likewise.
773 (fl78_unwind_word_mode): Likewise.
774 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
776 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
777 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
778 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
779 (mips_valid_pointer_mode): Likewise.
780 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
781 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
782 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
784 (ft32_addr_space_address_mode): Likewise.
785 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
786 scalar_int_mode instead of a machine_mode.
787 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
789 (m32c_addr_space_address_mode): Likewise.
790 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
791 (rs6000_eh_return_filter_mode): Likewise.
792 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
793 (rs6000_eh_return_filter_mode): Likewise.
794 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
795 (s390_libgcc_shift_count_mode): Likewise.
796 (s390_unwind_word_mode): Likewise.
797 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
799 * target.def (mode_rep_extended): Likewise.
800 (valid_pointer_mode): Likewise.
801 (addr_space.valid_pointer_mode): Likewise.
802 (eh_return_filter_mode): Return a scalar_int_mode rather than
804 (libgcc_cmp_return_mode): Likewise.
805 (libgcc_shift_count_mode): Likewise.
806 (unwind_word_mode): Likewise.
807 (addr_space.pointer_mode): Likewise.
808 (addr_space.address_mode): Likewise.
809 * doc/tm.texi: Regenerate.
810 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
812 (do_jump): Use scalar_int_mode for local variables.
813 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
814 rather than a machine_mode.
815 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
816 (scompare_loc_descriptor_wide): Likewise.
817 (scompare_loc_descriptor_narrow): Likewise.
818 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
820 * except.c (sjlj_emit_dispatch_table): Likewise.
821 (expand_builtin_eh_copy_values): Likewise.
822 * explow.c (convert_memory_address_addr_space_1): Likewise.
823 Take a scalar_int_mode rather than a machine_mode.
824 (convert_memory_address_addr_space): Take a scalar_int_mode rather
826 (memory_address_addr_space): Use scalar_int_mode for local variables.
827 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
828 rather than a machine_mode.
829 * expmed.c (mask_rtx): Likewise.
830 (init_expmed_one_conv): Likewise.
831 (expand_mult_highpart_adjust): Likewise.
832 (extract_high_half): Likewise.
833 (expmed_mult_highpart_optab): Likewise.
834 (expmed_mult_highpart): Likewise.
835 (expand_smod_pow2): Likewise.
836 (expand_sdiv_pow2): Likewise.
837 (emit_store_flag_int): Likewise.
838 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
840 (extract_low_bits): Likewise.
841 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
843 * expr.c (pieces_addr::adjust): Likewise.
844 (can_store_by_pieces): Likewise.
845 (store_by_pieces): Likewise.
846 (clear_by_pieces_1): Likewise.
847 (expand_expr_addr_expr_1): Likewise.
848 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
849 (expand_expr_real_1): Likewise.
850 (try_casesi): Likewise.
851 * final.c (shorten_branches): Likewise.
852 * fold-const.c (fold_convert_const_int_from_fixed): Change the
853 type of "mode" to machine_mode.
854 * internal-fn.c (expand_arith_overflow_result_store): Take a
855 scalar_int_mode rather than a machine_mode.
856 (expand_mul_overflow): Use scalar_int_mode for local variables.
857 * loop-doloop.c (doloop_modify): Likewise.
858 (doloop_optimize): Likewise.
859 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
861 (expand_doubleword_shift_condmove): Likewise.
862 (expand_doubleword_shift): Likewise.
863 (expand_doubleword_clz): Likewise.
864 (expand_doubleword_popcount): Likewise.
865 (expand_doubleword_parity): Likewise.
866 (expand_absneg_bit): Use scalar_int_mode for local variables.
867 (prepare_float_lib_cmp): Likewise.
868 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
869 rather than a machine_mode.
870 (convert_memory_address_addr_space): Likewise.
871 (get_mode_bounds): Likewise.
872 (get_address_mode): Return a scalar_int_mode rather than a
874 * rtlanal.c (get_address_mode): Likewise.
875 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
877 * targhooks.c (default_mode_rep_extended): Likewise.
878 (default_valid_pointer_mode): Likewise.
879 (default_addr_space_valid_pointer_mode): Likewise.
880 (default_eh_return_filter_mode): Return a scalar_int_mode rather
882 (default_libgcc_cmp_return_mode): Likewise.
883 (default_libgcc_shift_count_mode): Likewise.
884 (default_unwind_word_mode): Likewise.
885 (default_addr_space_pointer_mode): Likewise.
886 (default_addr_space_address_mode): Likewise.
887 * targhooks.h (default_eh_return_filter_mode): Likewise.
888 (default_libgcc_cmp_return_mode): Likewise.
889 (default_libgcc_shift_count_mode): Likewise.
890 (default_unwind_word_mode): Likewise.
891 (default_addr_space_pointer_mode): Likewise.
892 (default_addr_space_address_mode): Likewise.
893 (default_mode_rep_extended): Take a scalar_int_mode rather than
895 (default_valid_pointer_mode): Likewise.
896 (default_addr_space_valid_pointer_mode): Likewise.
897 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
899 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
900 rather than a machine_mode.
901 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
903 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
904 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
907 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
908 Alan Hayward <alan.hayward@arm.com>
909 David Sherwood <david.sherwood@arm.com>
911 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
912 the mode argument to scalar_int_mode.
913 (do_jump_by_parts_zero_rtx): Likewise.
914 (do_jump_by_parts_equality_rtx): Likewise.
915 (do_jump_by_parts_greater): Take a mode argument.
916 (do_jump_by_parts_equality): Likewise.
917 (do_jump_1): Update calls accordingly.
919 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
920 Alan Hayward <alan.hayward@arm.com>
921 David Sherwood <david.sherwood@arm.com>
923 * is-a.h (safe_dyn_cast): New function.
924 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
925 (jump_table_for_label): Likewise.
926 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
927 instead of an rtx_insn *.
928 (shorten_branches): Use dyn_cast instead of LABEL_P and
929 JUMP_TABLE_DATA_P. Use jump_table_for_label and
930 rtx_jump_table_data::get_data_mode.
931 (final_scan_insn): Likewise.
933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
934 Alan Hayward <alan.hayward@arm.com>
935 David Sherwood <david.sherwood@arm.com>
937 * combine.c (try_combine): Use is_a <scalar_int_mode> when
938 trying to combine a full-register integer set with a subreg
941 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
942 Alan Hayward <alan.hayward@arm.com>
943 David Sherwood <david.sherwood@arm.com>
945 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
946 that is always either address_mode or pointer_mode.
948 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
949 Alan Hayward <alan.hayward@arm.com>
950 David Sherwood <david.sherwood@arm.com>
952 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
953 when the two are known to be equal.
955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
956 Alan Hayward <alan.hayward@arm.com>
957 David Sherwood <david.sherwood@arm.com>
959 * simplify-rtx.c (simplify_const_unary_operation): Use
960 is_a <scalar_int_mode> instead of checking for a nonzero
961 precision. Forcibly convert op_mode to a scalar_int_mode
962 in that case. More clearly differentiate the operand and
963 result modes and use the former when deciding what the value
964 of a count-bits operation should be. Use is_int_mode instead
965 of checking for a MODE_INT. Remove redundant check for whether
966 this mode has a zero precision.
968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
969 Alan Hayward <alan.hayward@arm.com>
970 David Sherwood <david.sherwood@arm.com>
972 * optabs.c (widen_leading): Change the type of the mode argument
973 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
974 (widen_bswap): Likewise.
975 (expand_parity): Likewise.
976 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
977 (expand_ffs): Likewise.
978 (epand_unop): Check for scalar integer modes before calling the
981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
982 Alan Hayward <alan.hayward@arm.com>
983 David Sherwood <david.sherwood@arm.com>
985 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
987 (expand_expr_real_1): Update call accordingly.
989 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
990 Alan Hayward <alan.hayward@arm.com>
991 David Sherwood <david.sherwood@arm.com>
993 * expmed.c (store_bit_field_using_insv): Add op0_mode and
994 value_mode arguments. Use scalar_int_mode internally.
995 (store_bit_field_1): Rename the new integer mode from imode
996 to op0_mode and use it instead of GET_MODE (op0). Update calls
997 to store_split_bit_field, store_bit_field_using_insv and
998 store_fixed_bit_field.
999 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
1000 Use scalar_int_mode internally. Use a bit count rather than a mode
1001 when calculating the largest bit size for get_best_mode.
1002 Update calls to store_split_bit_field and store_fixed_bit_field_1.
1003 (store_fixed_bit_field_1): Add mode and value_mode arguments.
1004 Remove assertion that OP0 has a scalar integer mode.
1005 (store_split_bit_field): Add op0_mode and value_mode arguments.
1006 Update calls to extract_fixed_bit_field.
1007 (extract_bit_field_using_extv): Add an op0_mode argument.
1008 Use scalar_int_mode internally.
1009 (extract_bit_field_1): Rename the new integer mode from imode to
1010 op0_mode and use it instead of GET_MODE (op0). Update calls to
1011 extract_split_bit_field, extract_bit_field_using_extv and
1012 extract_fixed_bit_field.
1013 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
1014 to extract_split_bit_field and extract_fixed_bit_field_1.
1015 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
1016 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
1018 (extract_split_bit_field): Add an op0_mode argument. Update call
1019 to extract_fixed_bit_field.
1021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1022 Alan Hayward <alan.hayward@arm.com>
1023 David Sherwood <david.sherwood@arm.com>
1025 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
1026 * explow.c (hard_function_value): Likewise.
1027 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
1028 convert_to_mode call outside the loop.
1029 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
1030 for the mode iterator. Require the mode specified by max_pieces
1032 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
1034 (copy_blkmode_to_reg): Likewise.
1035 (set_storage_via_setmem): Likewise.
1036 * optabs.c (prepare_cmp_insn): Likewise.
1037 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1038 * stor-layout.c (finish_bitfield_representative): Likewise.
1040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1041 Alan Hayward <alan.hayward@arm.com>
1042 David Sherwood <david.sherwood@arm.com>
1044 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
1045 * expr.c (convert_move): Use them.
1046 (convert_modes): Likewise.
1047 (store_expr_with_bounds): Likewise.
1049 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1050 Alan Hayward <alan.hayward@arm.com>
1051 David Sherwood <david.sherwood@arm.com>
1053 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
1054 parameter for the mode of "x". Remove the "known_x", "known_mode"
1055 and "known_ret" arguments. Change the type of the mode argument
1057 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
1058 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
1059 (reg_num_sign_bit_copies_for_combine): Likewise.
1060 * rtlanal.c (nonzero_bits1): Likewise.
1061 (num_sign_bit_copies1): Likewise.
1062 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
1063 (reg_num_sign_bit_copies_general): Likewise.
1064 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
1065 (reg_nonzero_bits_general): Likewise.
1067 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1068 Alan Hayward <alan.hayward@arm.com>
1069 David Sherwood <david.sherwood@arm.com>
1071 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
1072 than in subroutines. Return 1 for non-integer modes.
1073 (cached_num_sign_bit_copies): Change the type of the mode parameter
1075 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
1076 classes. Handle CONST_INT_P first and then check whether X also
1077 has a scalar integer mode. Check the same thing for inner registers
1078 of a SUBREG and for values that are being extended or truncated.
1080 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1081 Alan Hayward <alan.hayward@arm.com>
1082 David Sherwood <david.sherwood@arm.com>
1084 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
1085 in subroutines. Return the mode mask for non-integer modes.
1086 (cached_nonzero_bits): Change the type of the mode parameter
1088 (nonzero_bits1): Likewise. Remove early exit for other mode
1089 classes. Handle CONST_INT_P first and then check whether X
1090 also has a scalar integer mode.
1092 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1093 Alan Hayward <alan.hayward@arm.com>
1094 David Sherwood <david.sherwood@arm.com>
1096 * expr.c (widest_int_mode_for_size): Make the comment match the code.
1097 Return a scalar_int_mode and assert that the size is greater than
1099 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
1100 (op_by_pieces_d::op_by_pieces_d): Likewise.
1101 (op_by_pieces_d::run): Likewise.
1102 (can_store_by_pieces): Likewise.
1104 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1105 Alan Hayward <alan.hayward@arm.com>
1106 David Sherwood <david.sherwood@arm.com>
1108 * combine.c (extract_left_shift): Add a mode argument and update
1110 (make_compound_operation_int): Change the type of the mode parameter
1111 to scalar_int_mode and update the call to extract_left_shift.
1113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1114 Alan Hayward <alan.hayward@arm.com>
1115 David Sherwood <david.sherwood@arm.com>
1117 * combine.c (simplify_and_const_int): Change the type of the mode
1118 parameter to scalar_int_mode.
1119 (simplify_and_const_int_1): Likewise. Update recursive call.
1121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1122 Alan Hayward <alan.hayward@arm.com>
1123 David Sherwood <david.sherwood@arm.com>
1125 * combine.c (simplify_compare_const): Check that the mode is a
1126 scalar_int_mode (rather than VOIDmode) before testing its
1128 (simplify_comparison): Move COMPARISON_P handling out of the
1129 loop and restrict the latter part of the loop to scalar_int_modes.
1130 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
1131 and when considering SUBREG_REGs. Use is_int_mode instead of
1132 checking GET_MODE_CLASS against MODE_INT.
1134 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1135 Alan Hayward <alan.hayward@arm.com>
1136 David Sherwood <david.sherwood@arm.com>
1138 * combine.c (try_widen_shift_mode): Move check for equal modes to...
1139 (simplify_shift_const_1): ...here. Use scalar_int_mode for
1140 shift_unit_mode and for modes involved in scalar shifts.
1142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1143 Alan Hayward <alan.hayward@arm.com>
1144 David Sherwood <david.sherwood@arm.com>
1146 * combine.c (force_int_to_mode): New function, split out from...
1147 (force_to_mode): ...here. Keep xmode up-to-date and use it
1148 instead of GET_MODE (x).
1150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1151 Alan Hayward <alan.hayward@arm.com>
1152 David Sherwood <david.sherwood@arm.com>
1154 * optabs-query.h (extraction_insn::struct_mode): Change type to
1155 opt_scalar_int_mode and update comment.
1156 (extraction_insn::field_mode): Change type to scalar_int_mode.
1157 (extraction_insn::pos_mode): Likewise.
1158 * combine.c (make_extraction): Update accordingly.
1159 * optabs-query.c (get_traditional_extraction_insn): Likewise.
1160 (get_optab_extraction_insn): Likewise.
1161 * recog.c (simplify_while_replacing): Likewise.
1162 * expmed.c (narrow_bit_field_mem): Change the type of the mode
1163 parameter to opt_scalar_int_mode.
1165 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1166 Alan Hayward <alan.hayward@arm.com>
1167 David Sherwood <david.sherwood@arm.com>
1169 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
1170 to a scalar_int_mode instead of a machine_mode.
1171 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
1172 (get_best_mode): Return a boolean and use a pointer argument to store
1173 the selected mode. Replace the limit mode parameter with a bit limit.
1174 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
1175 for the values returned by bit_field_mode_iterator::next_mode.
1176 (store_bit_field): Update call to get_best_mode.
1177 (store_fixed_bit_field): Likewise.
1178 (extract_fixed_bit_field): Likewise.
1179 * expr.c (optimize_bitfield_assignment_op): Likewise.
1180 * fold-const.c (optimize_bit_field_compare): Likewise.
1181 (fold_truth_andor_1): Likewise.
1182 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
1183 Update for new type of m_mode.
1184 (get_best_mode): As above.
1186 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1187 Alan Hayward <alan.hayward@arm.com>
1188 David Sherwood <david.sherwood@arm.com>
1190 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
1191 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
1192 (store_bit_field): Check is_a <scalar_int_mode> before calling
1193 strict_volatile_bitfield_p.
1194 (extract_bit_field): Likewise.
1196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1197 Alan Hayward <alan.hayward@arm.com>
1198 David Sherwood <david.sherwood@arm.com>
1200 * target.def (cstore_mode): Return a scalar_int_mode.
1201 * doc/tm.texi: Regenerate.
1202 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
1203 * targhooks.h (default_cstore_mode): Likewise.
1204 * targhooks.c (default_cstore_mode): Likewise, using a forced
1206 * expmed.c (emit_cstore): Expect the target of the cstore to be
1209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1210 Alan Hayward <alan.hayward@arm.com>
1211 David Sherwood <david.sherwood@arm.com>
1213 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
1215 (niter_desc): Likewise mode.
1216 (iv_analyze): Add a mode parameter.
1218 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
1219 and change its type to scalar_int_mode.
1220 * loop-iv.c: Update commentary at head of file.
1221 (iv_constant): Pass the mode paraeter before the rtx it describes
1222 and change its type to scalar_int_mode. Remove VOIDmode handling.
1223 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
1224 (iv_extend): Likewise.
1225 (shorten_into_mode): Likewise.
1226 (iv_add): Use scalar_int_mode.
1227 (iv_mult): Likewise.
1228 (iv_shift): Likewise.
1229 (canonicalize_iv_subregs): Likewise.
1230 (get_biv_step_1): Pass the outer_mode parameter before the rtx
1231 it describes and change its mode to scalar_int_mode. Also change
1232 the type of the returned inner_mode to scalar_int_mode.
1233 (get_biv_step): Likewise, turning outer_mode from a pointer
1234 into a direct parameter. Update call to get_biv_step_1.
1235 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
1236 iv_constant and get_biv_step.
1237 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
1238 and change its type to scalar_int_mode. Don't initialise iv->mode
1239 to VOIDmode and remove later checks for its still being VOIDmode.
1240 Update calls to iv_analyze_op and iv_analyze_expr. Check
1241 is_a <scalar_int_mode> when changing the mode under consideration.
1242 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
1243 Update call to iv_analyze_expr.
1244 (iv_analyze_op): Add a mode parameter. Reject subregs whose
1245 inner register is not also a scalar_int_mode. Update call to
1247 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
1248 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
1249 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
1250 separate mode class checks. Update calls to iv_analyze. Remove
1251 fix-up of VOIDmodes after iv_analyze_biv.
1252 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
1253 don't have a scalar_int_mode. Update call to biv_p.
1255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1256 Alan Hayward <alan.hayward@arm.com>
1257 David Sherwood <david.sherwood@arm.com>
1259 * cfgexpand.c (convert_debug_memory_address): Use
1260 as_a <scalar_int_mode>.
1261 * combine.c (expand_compound_operation): Likewise.
1262 (make_extraction): Likewise.
1263 (change_zero_ext): Likewise.
1264 (simplify_comparison): Likewise.
1265 * cse.c (cse_insn): Likewise.
1266 * dwarf2out.c (minmax_loc_descriptor): Likewise.
1267 (mem_loc_descriptor): Likewise.
1268 (loc_descriptor): Likewise.
1269 * expmed.c (init_expmed_one_mode): Likewise.
1270 (synth_mult): Likewise.
1271 (emit_store_flag_1): Likewise.
1272 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
1273 of a comparison with size.
1274 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
1275 (reduce_to_bit_field_precision): Likewise.
1276 * function.c (expand_function_end): Likewise.
1277 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
1278 * loop-doloop.c (doloop_modify): Likewise.
1279 * optabs.c (expand_binop): Likewise.
1280 (expand_unop): Likewise.
1281 (expand_copysign_absneg): Likewise.
1282 (prepare_cmp_insn): Likewise.
1283 (maybe_legitimize_operand): Likewise.
1284 * recog.c (const_scalar_int_operand): Likewise.
1285 * rtlanal.c (get_address_mode): Likewise.
1286 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1287 (simplify_cond_clz_ctz): Likewise.
1288 * tree-nested.c (get_nl_goto_field): Likewise.
1289 * tree.c (build_vector_type_for_mode): Likewise.
1290 * var-tracking.c (use_narrower_mode): Likewise.
1292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1293 Alan Hayward <alan.hayward@arm.com>
1294 David Sherwood <david.sherwood@arm.com>
1296 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
1297 * builtins.c (expand_builtin_signbit): Use it.
1298 * cfgexpand.c (expand_debug_expr): Likewise.
1299 * dojump.c (do_jump): Likewise.
1300 (do_compare_and_jump): Likewise.
1301 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
1302 * expmed.c (make_tree): Likewise.
1303 * expr.c (expand_expr_real_2): Likewise.
1304 (expand_expr_real_1): Likewise.
1305 (try_casesi): Likewise.
1306 * fold-const-call.c (fold_const_call_ss): Likewise.
1307 * fold-const.c (unextend): Likewise.
1308 (extract_muldiv_1): Likewise.
1309 (fold_single_bit_test): Likewise.
1310 (native_encode_int): Likewise.
1311 (native_encode_string): Likewise.
1312 (native_interpret_int): Likewise.
1313 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
1314 * internal-fn.c (expand_addsub_overflow): Likewise.
1315 (expand_neg_overflow): Likewise.
1316 (expand_mul_overflow): Likewise.
1317 (expand_arith_overflow): Likewise.
1318 * match.pd: Likewise.
1319 * stor-layout.c (layout_type): Likewise.
1320 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
1321 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
1322 * tree-ssanames.c (get_range_info): Likewise.
1323 * tree-switch-conversion.c (array_value_type) Likewise.
1324 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
1325 (vect_recog_divmod_pattern): Likewise.
1326 (vect_recog_mixed_size_cond_pattern): Likewise.
1327 * tree-vrp.c (extract_range_basic): Likewise.
1328 (simplify_float_conversion_using_ranges): Likewise.
1329 * tree.c (int_fits_type_p): Likewise.
1330 * ubsan.c (instrument_bool_enum_load): Likewise.
1331 * varasm.c (mergeable_string_section): Likewise.
1332 (narrowing_initializer_constant_valid_p): Likewise.
1333 (output_constant): Likewise.
1335 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1336 Alan Hayward <alan.hayward@arm.com>
1337 David Sherwood <david.sherwood@arm.com>
1339 * machmode.h (NARROWEST_INT_MODE): New macro.
1340 * expr.c (alignment_for_piecewise_move): Use it instead of
1341 GET_CLASS_NARROWEST_MODE (MODE_INT).
1342 (push_block): Likewise.
1343 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1345 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1347 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1348 Alan Hayward <alan.hayward@arm.com>
1349 David Sherwood <david.sherwood@arm.com>
1351 * postreload.c (move2add_valid_value_p): Change the type of the
1352 mode parameter to scalar_int_mode.
1353 (move2add_use_add2_insn): Add a mode parameter and use it instead
1355 (move2add_use_add3_insn): Likewise.
1356 (reload_cse_move2add): Update accordingly.
1358 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1359 Alan Hayward <alan.hayward@arm.com>
1360 David Sherwood <david.sherwood@arm.com>
1362 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
1364 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
1365 * optabs.c (expand_unop): Likewise.
1367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1368 Alan Hayward <alan.hayward@arm.com>
1369 David Sherwood <david.sherwood@arm.com>
1371 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
1372 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
1373 (popcount_loc_descriptor): Likewise.
1374 (bswap_loc_descriptor): Likewise.
1375 (rotate_loc_descriptor): Likewise.
1376 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
1377 calling the functions above.
1379 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1380 Alan Hayward <alan.hayward@arm.com>
1381 David Sherwood <david.sherwood@arm.com>
1383 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
1385 (try_combine): Likewise.
1386 (simplify_if_then_else): Likewise.
1387 * cse.c (cse_insn): Likewise.
1388 * dwarf2out.c (mem_loc_descriptor): Likewise.
1389 * emit-rtl.c (gen_lowpart_common): Likewise.
1390 * simplify-rtx.c (simplify_truncation): Likewise.
1391 (simplify_binary_operation_1): Likewise.
1392 (simplify_const_relational_operation): Likewise.
1393 (simplify_ternary_operation): Likewise.
1394 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
1396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1397 Alan Hayward <alan.hayward@arm.com>
1398 David Sherwood <david.sherwood@arm.com>
1400 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
1401 * reload.c (push_reload): Likewise.
1402 (find_reloads): Likewise.
1404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1405 Alan Hayward <alan.hayward@arm.com>
1406 David Sherwood <david.sherwood@arm.com>
1408 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
1409 (make_compound_operation_int): Likewise.
1410 (change_zero_ext): Likewise.
1411 * expr.c (convert_move): Likewise.
1412 (convert_modes): Likewise.
1413 * fwprop.c (forward_propagate_subreg): Likewise.
1414 * loop-iv.c (get_biv_step_1): Likewise.
1415 * optabs.c (widen_operand): Likewise.
1416 * postreload.c (move2add_valid_value_p): Likewise.
1417 * recog.c (simplify_while_replacing): Likewise.
1418 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1419 (simplify_binary_operation_1): Likewise. Remove redundant
1420 mode equality check.
1422 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1423 Alan Hayward <alan.hayward@arm.com>
1424 David Sherwood <david.sherwood@arm.com>
1426 * combine.c (combine_simplify_rtx): Add checks for
1427 is_a <scalar_int_mode>.
1428 (simplify_if_then_else): Likewise.
1429 (make_field_assignment): Likewise.
1430 (simplify_comparison): Likewise.
1431 * ifcvt.c (noce_try_bitop): Likewise.
1432 * loop-invariant.c (canonicalize_address_mult): Likewise.
1433 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1436 Alan Hayward <alan.hayward@arm.com>
1437 David Sherwood <david.sherwood@arm.com>
1439 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
1440 is_a <scalar_int_mode> instead of != BLKmode.
1442 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1443 Alan Hayward <alan.hayward@arm.com>
1444 David Sherwood <david.sherwood@arm.com>
1446 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
1447 instead of != VOIDmode.
1448 * combine.c (if_then_else_cond): Likewise.
1449 (change_zero_ext): Likewise.
1450 * dwarf2out.c (mem_loc_descriptor): Likewise.
1451 (loc_descriptor): Likewise.
1452 * rtlanal.c (canonicalize_condition): Likewise.
1453 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
1455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1456 Alan Hayward <alan.hayward@arm.com>
1457 David Sherwood <david.sherwood@arm.com>
1459 * simplify-rtx.c (simplify_binary_operation_1): Use
1460 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
1462 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1463 Alan Hayward <alan.hayward@arm.com>
1464 David Sherwood <david.sherwood@arm.com>
1466 * wide-int.h (int_traits<unsigned char>) New class.
1467 (int_traits<unsigned short>) Likewise.
1468 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
1469 Use GET_MODE_UNIT_PRECISION and remove redundant test for
1471 * combine.c (set_nonzero_bits_and_sign_copies): Use
1472 is_a <scalar_int_mode>.
1473 (find_split_point): Likewise.
1474 (combine_simplify_rtx): Likewise.
1475 (simplify_logical): Likewise.
1476 (expand_compound_operation): Likewise.
1477 (expand_field_assignment): Likewise.
1478 (make_compound_operation): Likewise.
1479 (extended_count): Likewise.
1480 (change_zero_ext): Likewise.
1481 (simplify_comparison): Likewise.
1482 * dwarf2out.c (scompare_loc_descriptor): Likewise.
1483 (ucompare_loc_descriptor): Likewise.
1484 (minmax_loc_descriptor): Likewise.
1485 (mem_loc_descriptor): Likewise.
1486 (loc_descriptor): Likewise.
1487 * expmed.c (init_expmed_one_mode): Likewise.
1488 * lra-constraints.c (lra_constraint_offset): Likewise.
1489 * optabs.c (prepare_libcall_arg): Likewise.
1490 * postreload.c (move2add_note_store): Likewise.
1491 * reload.c (operands_match_p): Likewise.
1492 * rtl.h (load_extend_op): Likewise.
1493 * rtlhooks.c (gen_lowpart_general): Likewise.
1494 * simplify-rtx.c (simplify_truncation): Likewise.
1495 (simplify_unary_operation_1): Likewise.
1496 (simplify_binary_operation_1): Likewise.
1497 (simplify_const_binary_operation): Likewise.
1498 (simplify_const_relational_operation): Likewise.
1499 (simplify_subreg): Likewise.
1500 * stor-layout.c (bitwise_mode_for_mode): Likewise.
1501 * var-tracking.c (adjust_mems): Likewise.
1502 (prepare_call_arguments): Likewise.
1504 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1505 Alan Hayward <alan.hayward@arm.com>
1506 David Sherwood <david.sherwood@arm.com>
1508 * machmode.h (is_int_mode): New fuction.
1509 * combine.c (find_split_point): Use it.
1510 (combine_simplify_rtx): Likewise.
1511 (simplify_if_then_else): Likewise.
1512 (simplify_set): Likewise.
1513 (simplify_shift_const_1): Likewise.
1514 (simplify_comparison): Likewise.
1515 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
1516 * cse.c (notreg_cost): Likewise.
1517 (cse_insn): Likewise.
1518 * cselib.c (cselib_lookup_1): Likewise.
1519 * dojump.c (do_jump_1): Likewise.
1520 (do_compare_rtx_and_jump): Likewise.
1521 * dse.c (get_call_args): Likewise.
1522 * dwarf2out.c (rtl_for_decl_init): Likewise.
1523 (native_encode_initializer): Likewise.
1524 * expmed.c (emit_store_flag_1): Likewise.
1525 (emit_store_flag): Likewise.
1526 * expr.c (convert_modes): Likewise.
1527 (store_field): Likewise.
1528 (expand_expr_real_1): Likewise.
1529 * fold-const.c (fold_read_from_constant_string): Likewise.
1530 * gimple-ssa-sprintf.c (get_format_string): Likewise.
1531 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
1532 * optabs.c (expand_binop): Likewise.
1533 (expand_unop): Likewise.
1534 (expand_abs_nojump): Likewise.
1535 (expand_one_cmpl_abs_nojump): Likewise.
1536 * simplify-rtx.c (mode_signbit_p): Likewise.
1537 (val_signbit_p): Likewise.
1538 (val_signbit_known_set_p): Likewise.
1539 (val_signbit_known_clear_p): Likewise.
1540 (simplify_relational_operation_1): Likewise.
1541 * tree.c (vector_type_mode): Likewise.
1543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1544 Alan Hayward <alan.hayward@arm.com>
1545 David Sherwood <david.sherwood@arm.com>
1547 * machmode.h (smallest_mode_for_size): Fix formatting.
1548 (smallest_int_mode_for_size): New function.
1549 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
1550 instead of smallest_mode_for_size.
1551 * combine.c (make_extraction): Likewise.
1552 * config/arc/arc.c (arc_expand_movmem): Likewise.
1553 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
1554 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
1555 * config/s390/s390.c (s390_expand_insv): Likewise.
1556 * config/sparc/sparc.c (assign_int_registers): Likewise.
1557 * config/spu/spu.c (spu_function_value): Likewise.
1558 (spu_function_arg): Likewise.
1559 * coverage.c (get_gcov_type): Likewise.
1560 (get_gcov_unsigned_t): Likewise.
1561 * dse.c (find_shift_sequence): Likewise.
1562 * expmed.c (store_bit_field_1): Likewise.
1563 * expr.c (convert_move): Likewise.
1564 (store_field): Likewise.
1565 * internal-fn.c (expand_arith_overflow): Likewise.
1566 * optabs-query.c (get_best_extraction_insn): Likewise.
1567 * optabs.c (expand_twoval_binop_libfunc): Likewise.
1568 * stor-layout.c (layout_type): Likewise.
1569 (initialize_sizetypes): Likewise.
1570 * targhooks.c (default_get_mask_mode): Likewise.
1571 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
1573 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1574 Alan Hayward <alan.hayward@arm.com>
1575 David Sherwood <david.sherwood@arm.com>
1577 * machmode.h (opt_mode::else_blk): New function.
1578 (int_mode_for_mode): Declare.
1579 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
1580 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
1582 * cfgexpand.c (expand_debug_expr): Likewise.
1583 * combine.c (gen_lowpart_or_truncate): Likewise.
1584 (gen_lowpart_for_combine): Likewise.
1585 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
1586 * config/avr/avr.c (avr_to_int_mode): Likewise.
1587 (avr_out_plus_1): Likewise.
1588 (avr_out_plus): Likewise.
1589 (avr_out_round): Likewise.
1590 * config/i386/i386.c (ix86_split_to_parts): Likewise.
1591 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
1592 (s390_expand_vcond): Likewise.
1593 * config/spu/spu.c (spu_split_immediate): Likewise.
1594 (spu_expand_mov): Likewise.
1595 * dse.c (get_stored_val): Likewise.
1596 * expmed.c (store_bit_field_1): Likewise.
1597 (convert_extracted_bit_field): Use int_mode_for_mode instead of
1599 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
1600 (extract_low_bits): Likewise.
1601 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
1602 handling rather than repeating the check.
1603 (emit_group_store): Likewise.
1604 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
1605 * optabs.c (expand_absneg_bit): Likewise.
1606 (expand_copysign_absneg): Likewise.
1607 (expand_copysign_bit): Likewise.
1608 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
1609 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
1610 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
1611 * var-tracking.c (prepare_call_arguments): Likewise.
1612 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
1613 int_mode_for_mode instead of mode_for_size.
1614 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
1616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1617 Alan Hayward <alan.hayward@arm.com>
1618 David Sherwood <david.sherwood@arm.com>
1620 * machmode.h (int_mode_for_size): New function.
1621 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
1622 instead of mode_for_size.
1623 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
1625 * combine.c (expand_field_assignment): Use int_mode_for_size
1626 instead of mode_for_size.
1627 (make_extraction): Likewise.
1628 (simplify_shift_const_1): Likewise.
1629 (simplify_comparison): Likewise.
1630 * dojump.c (do_jump): Likewise.
1631 * dwarf2out.c (mem_loc_descriptor): Likewise.
1632 * emit-rtl.c (init_derived_machine_modes): Likewise.
1633 * expmed.c (flip_storage_order): Likewise.
1634 (convert_extracted_bit_field): Likewise.
1635 * expr.c (copy_blkmode_from_reg): Likewise.
1636 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
1637 * internal-fn.c (expand_mul_overflow): Likewise.
1638 * lower-subreg.c (simple_move): Likewise.
1639 * optabs-libfuncs.c (init_optabs): Likewise.
1640 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1641 * tree.c (vector_type_mode): Likewise.
1642 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
1643 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
1644 * tree-vect-generic.c (expand_vector_parallel): Likewise.
1645 * tree-vect-stmts.c (vectorizable_load): Likewise.
1646 (vectorizable_store): Likewise.
1648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1649 Alan Hayward <alan.hayward@arm.com>
1650 David Sherwood <david.sherwood@arm.com>
1652 * coretypes.h (pod_mode): New type.
1653 (scalar_int_mode_pod): New typedef.
1654 * machmode.h (pod_mode): New class.
1655 (int_n_data_t::m): Change type to scalar_int_mode_pod.
1656 * genmodes.c (emit_mode_int_n): Update accordingly.
1657 * lower-subreg.h (target_lower_subreg): Change type to
1658 scalar_int_mode_pod.
1659 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
1660 scalar_int_mode_pod.
1662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1663 Alan Hayward <alan.hayward@arm.com>
1664 David Sherwood <david.sherwood@arm.com>
1666 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
1667 machine_mode to scalar_int_mode.
1668 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
1669 (rs6000_option_override_internal): Remove cast to int.
1670 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
1671 machine_mode to scalar_int_mode.
1672 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
1673 (rs6000_option_override_internal): Remove cast to int.
1674 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
1675 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
1677 * config/s390/s390.c (s390_expand_builtin): Likewise.
1678 * coretypes.h (scalar_int_mode): New type.
1679 (opt_scalar_int_mode): New typedef.
1680 * machmode.h (scalar_int_mode): New class.
1681 (scalar_int_mode::includes_p): New function.
1682 (byte_mode): Change type to scalar_int_mode.
1683 (word_mode): Likewise.
1684 (ptr_mode): Likewise.
1685 * emit-rtl.c (byte_mode): Likewise.
1686 (word_mode): Likewise.
1687 (ptr_mode): Likewise.
1688 (init_derived_machine_modes): Update accordingly.
1689 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
1690 and MODE_PARTIAL_INT.
1691 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
1692 opt_scalar_int_mode.
1694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1695 Alan Hayward <alan.hayward@arm.com>
1696 David Sherwood <david.sherwood@arm.com>
1698 * target.def (libgcc_floating_mode_supported_p): Take a
1700 * doc/tm.texi: Regenerate.
1701 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
1703 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
1704 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
1707 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1708 Alan Hayward <alan.hayward@arm.com>
1709 David Sherwood <david.sherwood@arm.com>
1711 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
1712 * doc/tm.texi: Regenerate.
1713 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
1714 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
1715 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
1716 * targhooks.h (default_floatn_mode): Likewise.
1717 * targhooks.c (default_floatn_mode): Likewise.
1718 * tree.c (build_common_tree_nodes): Update accordingly.
1720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1721 Alan Hayward <alan.hayward@arm.com>
1722 David Sherwood <david.sherwood@arm.com>
1724 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
1725 (mode_iterator::iterate_p): Likewise.
1726 (mode_iterator::get_wider): Likewise.
1727 * expr.c (init_expr_target): Use opt_scalar_float_mode.
1729 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1730 Alan Hayward <alan.hayward@arm.com>
1731 David Sherwood <david.sherwood@arm.com>
1733 * coretypes.h (opt_scalar_float_mode): New typedef.
1734 * machmode.h (float_mode_for_size): New function.
1735 * emit-rtl.c (double_mode): Delete.
1736 (init_emit_once): Use float_mode_for_size.
1737 * stor-layout.c (layout_type): Likewise.
1738 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
1740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1741 Alan Hayward <alan.hayward@arm.com>
1742 David Sherwood <david.sherwood@arm.com>
1744 * output.h (assemble_real): Take a scalar_float_mode.
1745 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
1746 * config/arm/arm.md (consttable_4): Likewise.
1747 (consttable_8): Likewise.
1748 (consttable_16): Likewise.
1749 * config/mips/mips.md (consttable_float): Likewise.
1750 * config/s390/s390.c (s390_output_pool_entry): Likewise.
1751 * varasm.c (assemble_real): Take a scalar_float_mode.
1752 (output_constant_pool_2): Update accordingly.
1753 (output_constant): Likewise.
1755 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1756 Alan Hayward <alan.hayward@arm.com>
1757 David Sherwood <david.sherwood@arm.com>
1759 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
1760 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
1761 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
1762 (native_encode_real): Likewise.
1763 (native_interpret_real): Likewise.
1764 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
1765 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1768 Alan Hayward <alan.hayward@arm.com>
1769 David Sherwood <david.sherwood@arm.com>
1771 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
1772 <scalar_float_mode>. Simplify.
1773 (gen_extend_conv_libfunc): Likewise.
1775 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1776 Alan Hayward <alan.hayward@arm.com>
1777 David Sherwood <david.sherwood@arm.com>
1779 * coretypes.h (scalar_float_mode): New type.
1780 * machmode.h (mode_traits::from_int): Use machine_mode if
1781 USE_ENUM_MODES is defined.
1782 (is_a): New function.
1784 (dyn_cast): Likewise.
1785 (scalar_float_mode): New class.
1786 (scalar_float_mode::includes_p): New function.
1787 (is_float_mode): Likewise.
1788 * gdbhooks.py (MachineModePrinter): New class.
1789 (build_pretty_printer): Use it for scalar_float_mode.
1790 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
1791 (format_helper::format_helper): Turn into a template.
1792 * genmodes.c (get_mode_class): New function.
1793 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
1794 or machine_mode if none.
1795 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
1796 as_a <scalar_float_mode>.
1797 * dwarf2out.c (mem_loc_descriptor): Likewise.
1798 (insert_float): Likewise.
1799 (add_const_value_attribute): Likewise.
1800 * simplify-rtx.c (simplify_immed_subreg): Likewise.
1801 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
1802 (expand_unop): Update accordingly.
1803 (expand_abs_nojump): Likewise.
1804 (expand_copysign_absneg): Take a scalar_float_mode.
1805 (expand_copysign_bit): Likewise.
1806 (expand_copysign): Update accordingly.
1808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1809 Alan Hayward <alan.hayward@arm.com>
1810 David Sherwood <david.sherwood@arm.com>
1812 * coretypes.h (opt_mode): New class.
1813 * machmode.h (opt_mode): Likewise.
1814 (opt_mode::else_void): New function.
1815 (opt_mode::require): Likewise.
1816 (opt_mode::exists): Likewise.
1817 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
1818 (GET_MODE_2XWIDER_MODE): Likewise.
1819 (mode_iterator::get_wider): Update accordingly.
1820 (mode_iterator::get_2xwider): Likewise.
1821 (mode_iterator::get_known_wider): Likewise, turning into a template.
1822 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
1823 forcing a wider mode to exist.
1824 * config/cr16/cr16.h (LONG_REG_P): Likewise.
1825 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1826 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
1827 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
1828 * lower-subreg.c (init_lower_subreg): Likewise.
1829 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
1830 on the final iteration.
1831 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
1832 a wider mode exists before asking for a move pattern.
1833 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
1834 forcing a wider mode to exist.
1835 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
1836 returning false if no such mode exists.
1837 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
1838 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
1839 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
1840 Avoid checking for a MODE_INT if we already know the mode is not a
1842 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
1843 forcing a wider mode to exist.
1844 (expmed_mult_highpart_optab): Likewise.
1845 (expmed_mult_highpart): Likewise.
1846 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
1848 * lto-streamer-in.c (lto_input_mode_table): Likewise.
1849 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
1850 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1851 * internal-fn.c (expand_mul_overflow): Update use of
1852 GET_MODE_2XWIDER_MODE.
1853 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1854 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
1855 GET_MODE_WIDER_MODE.
1856 (convert_plusminus_to_widen): Likewise.
1857 * tree-switch-conversion.c (array_value_type): Likewise.
1858 * var-tracking.c (emit_note_insn_var_location): Likewise.
1859 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1860 Return false inside rather than outside the loop if no wider mode
1862 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
1863 and GET_MODE_2XWIDER_MODE
1864 (can_compare_p): Use else_void.
1865 * gdbhooks.py (OptMachineModePrinter): New class.
1866 (build_pretty_printer): Use it for opt_mode.
1868 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1869 Alan Hayward <alan.hayward@arm.com>
1870 David Sherwood <david.sherwood@arm.com>
1872 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
1873 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
1875 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1876 Alan Hayward <alan.hayward@arm.com>
1877 David Sherwood <david.sherwood@arm.com>
1879 * machmode.h (mode_traits): New structure.
1880 (get_narrowest_mode): New function.
1881 (mode_iterator::start): Likewise.
1882 (mode_iterator::iterate_p): Likewise.
1883 (mode_iterator::get_wider): Likewise.
1884 (mode_iterator::get_known_wider): Likewise.
1885 (mode_iterator::get_2xwider): Likewise.
1886 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
1887 (FOR_EACH_MODE): Likewise.
1888 (FOR_EACH_MODE_FROM): Likewise.
1889 (FOR_EACH_MODE_UNTIL): Likewise.
1890 (FOR_EACH_WIDER_MODE): Likewise.
1891 (FOR_EACH_2XWIDER_MODE): Likewise.
1892 * builtins.c (expand_builtin_strlen): Use new mode iterators.
1893 * combine.c (simplify_comparison): Likewise
1894 * config/i386/i386.c (type_natural_mode): Likewise.
1895 * cse.c (cse_insn): Likewise.
1896 * dse.c (find_shift_sequence): Likewise.
1897 * emit-rtl.c (init_derived_machine_modes): Likewise.
1898 (init_emit_once): Likewise.
1899 * explow.c (hard_function_value): Likewise.
1900 * expmed.c (extract_fixed_bit_field_1): Likewise.
1901 (extract_bit_field_1): Likewise.
1902 (expand_divmod): Likewise.
1903 (emit_store_flag_1): Likewise.
1904 * expr.c (init_expr_target): Likewise.
1905 (convert_move): Likewise.
1906 (alignment_for_piecewise_move): Likewise.
1907 (widest_int_mode_for_size): Likewise.
1908 (emit_block_move_via_movmem): Likewise.
1909 (copy_blkmode_to_reg): Likewise.
1910 (set_storage_via_setmem): Likewise.
1911 (compress_float_constant): Likewise.
1912 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1913 * optabs-query.c (get_best_extraction_insn): Likewise.
1914 * optabs.c (expand_binop): Likewise.
1915 (expand_twoval_unop): Likewise.
1916 (expand_twoval_binop): Likewise.
1917 (widen_leading): Likewise.
1918 (widen_bswap): Likewise.
1919 (expand_parity): Likewise.
1920 (expand_unop): Likewise.
1921 (prepare_cmp_insn): Likewise.
1922 (prepare_float_lib_cmp): Likewise.
1923 (expand_float): Likewise.
1924 (expand_fix): Likewise.
1925 (expand_sfix_optab): Likewise.
1926 * postreload.c (move2add_use_add2_insn): Likewise.
1927 * reg-stack.c (reg_to_stack): Likewise.
1928 * reginfo.c (choose_hard_reg_mode): Likewise.
1929 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1930 * stor-layout.c (mode_for_size): Likewise.
1931 (smallest_mode_for_size): Likewise.
1932 (mode_for_vector): Likewise.
1933 (finish_bitfield_representative): Likewise.
1934 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
1935 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
1936 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1937 * var-tracking.c (prepare_call_arguments): Likewise.
1939 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1940 Alan Hayward <alan.hayward@arm.com>
1941 David Sherwood <david.sherwood@arm.com>
1943 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
1944 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
1945 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
1946 * machmode.h (mode_size): Move earlier in file.
1947 (mode_precision): Likewise.
1948 (mode_inner): Likewise.
1949 (mode_nunits): Likewise.
1950 (mode_unit_size): Likewise.
1951 (unit_unit_precision): Likewise.
1952 (mode_wider): Likewise.
1953 (mode_2xwider): Likewise.
1954 (machine_mode): New class.
1955 (mode_to_bytes): New function.
1956 (mode_to_bits): Likewise.
1957 (mode_to_precision): Likewise.
1958 (mode_to_inner): Likewise.
1959 (mode_to_unit_size): Likewise.
1960 (mode_to_unit_precision): Likewise.
1961 (mode_to_nunits): Likewise.
1962 (GET_MODE_SIZE): Use mode_to_bytes.
1963 (GET_MODE_BITSIZE): Use mode_to_bits.
1964 (GET_MODE_PRECISION): Use mode_to_precision.
1965 (GET_MODE_INNER): Use mode_to_inner.
1966 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
1967 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
1968 (GET_MODE_NUNITS): Use mode_to_nunits.
1969 * system.h (ALWAYS_INLINE): New macro.
1970 * config/powerpcspe/powerpcspe-c.c
1971 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
1972 int for arg1_mode and arg2_mode.
1974 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1975 Alan Hayward <alan.hayward@arm.com>
1976 David Sherwood <david.sherwood@arm.com>
1978 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
1979 Prefix mode names with E_ in case statements.
1980 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1981 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
1982 (aarch64_split_simd_move): Likewise.
1983 (aarch64_gen_storewb_pair): Likewise.
1984 (aarch64_gen_loadwb_pair): Likewise.
1985 (aarch64_gen_store_pair): Likewise.
1986 (aarch64_gen_load_pair): Likewise.
1987 (aarch64_get_condition_code_1): Likewise.
1988 (aarch64_constant_pool_reload_icode): Likewise.
1989 (get_rsqrte_type): Likewise.
1990 (get_rsqrts_type): Likewise.
1991 (get_recpe_type): Likewise.
1992 (get_recps_type): Likewise.
1993 (aarch64_gimplify_va_arg_expr): Likewise.
1994 (aarch64_simd_container_mode): Likewise.
1995 (aarch64_emit_load_exclusive): Likewise.
1996 (aarch64_emit_store_exclusive): Likewise.
1997 (aarch64_expand_compare_and_swap): Likewise.
1998 (aarch64_gen_atomic_cas): Likewise.
1999 (aarch64_emit_bic): Likewise.
2000 (aarch64_emit_atomic_swap): Likewise.
2001 (aarch64_emit_atomic_load_op): Likewise.
2002 (aarch64_evpc_trn): Likewise.
2003 (aarch64_evpc_uzp): Likewise.
2004 (aarch64_evpc_zip): Likewise.
2005 (aarch64_evpc_ext): Likewise.
2006 (aarch64_evpc_rev): Likewise.
2007 (aarch64_evpc_dup): Likewise.
2008 (aarch64_gen_ccmp_first): Likewise.
2009 (aarch64_gen_ccmp_next): Likewise.
2010 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
2011 (alpha_emit_xfloating_libcall): Likewise.
2012 (emit_insxl): Likewise.
2013 (alpha_arg_type): Likewise.
2014 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
2015 (arc_preferred_simd_mode): Likewise.
2016 (arc_secondary_reload): Likewise.
2017 (get_arc_condition_code): Likewise.
2018 (arc_print_operand): Likewise.
2019 (arc_legitimate_constant_p): Likewise.
2020 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2021 * config/arc/arc.md (casesi_load): Likewise.
2022 (casesi_compact_jump): Likewise.
2023 * config/arc/predicates.md (proper_comparison_operator): Likewise.
2024 (cc_use_register): Likewise.
2025 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2026 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2027 (arm_init_iwmmxt_builtins): Likewise.
2028 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
2029 (neon_expand_vector_init): Likewise.
2030 (arm_attr_length_move_neon): Likewise.
2031 (maybe_get_arm_condition_code): Likewise.
2032 (arm_emit_vector_const): Likewise.
2033 (arm_preferred_simd_mode): Likewise.
2034 (arm_output_iwmmxt_tinsr): Likewise.
2035 (thumb1_output_casesi): Likewise.
2036 (thumb2_output_casesi): Likewise.
2037 (arm_emit_load_exclusive): Likewise.
2038 (arm_emit_store_exclusive): Likewise.
2039 (arm_expand_compare_and_swap): Likewise.
2040 (arm_evpc_neon_vuzp): Likewise.
2041 (arm_evpc_neon_vzip): Likewise.
2042 (arm_evpc_neon_vrev): Likewise.
2043 (arm_evpc_neon_vtrn): Likewise.
2044 (arm_evpc_neon_vext): Likewise.
2045 (arm_validize_comparison): Likewise.
2046 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
2047 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
2048 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
2049 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
2050 (c6x_preferred_simd_mode): Likewise.
2051 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
2052 (epiphany_rtx_costs): Likewise.
2053 * config/epiphany/predicates.md (proper_comparison_operator):
2055 * config/frv/frv.c (condexec_memory_operand): Likewise.
2056 (frv_emit_move): Likewise.
2057 (output_move_single): Likewise.
2058 (output_condmove_single): Likewise.
2059 (frv_hard_regno_mode_ok): Likewise.
2060 (frv_matching_accg_mode): Likewise.
2061 * config/h8300/h8300.c (split_adds_subs): Likewise.
2062 (h8300_rtx_costs): Likewise.
2063 (h8300_print_operand): Likewise.
2064 (compute_mov_length): Likewise.
2065 (output_logical_op): Likewise.
2066 (compute_logical_op_length): Likewise.
2067 (compute_logical_op_cc): Likewise.
2068 (h8300_shift_needs_scratch_p): Likewise.
2069 (output_a_shift): Likewise.
2070 (compute_a_shift_length): Likewise.
2071 (compute_a_shift_cc): Likewise.
2072 (expand_a_rotate): Likewise.
2073 (output_a_rotate): Likewise.
2074 * config/i386/i386.c (classify_argument): Likewise.
2075 (function_arg_advance_32): Likewise.
2076 (function_arg_32): Likewise.
2077 (function_arg_64): Likewise.
2078 (function_value_64): Likewise.
2079 (ix86_gimplify_va_arg): Likewise.
2080 (ix86_legitimate_constant_p): Likewise.
2081 (put_condition_code): Likewise.
2082 (split_double_mode): Likewise.
2083 (ix86_avx256_split_vector_move_misalign): Likewise.
2084 (ix86_expand_vector_logical_operator): Likewise.
2085 (ix86_split_idivmod): Likewise.
2086 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
2087 (ix86_build_const_vector): Likewise.
2088 (ix86_build_signbit_mask): Likewise.
2089 (ix86_match_ccmode): Likewise.
2090 (ix86_cc_modes_compatible): Likewise.
2091 (ix86_expand_branch): Likewise.
2092 (ix86_expand_sse_cmp): Likewise.
2093 (ix86_expand_sse_movcc): Likewise.
2094 (ix86_expand_int_sse_cmp): Likewise.
2095 (ix86_expand_vec_perm_vpermi2): Likewise.
2096 (ix86_expand_vec_perm): Likewise.
2097 (ix86_expand_sse_unpack): Likewise.
2098 (ix86_expand_int_addcc): Likewise.
2099 (ix86_split_to_parts): Likewise.
2100 (ix86_vectorize_builtin_gather): Likewise.
2101 (ix86_vectorize_builtin_scatter): Likewise.
2102 (avx_vpermilp_parallel): Likewise.
2103 (inline_memory_move_cost): Likewise.
2104 (ix86_tieable_integer_mode_p): Likewise.
2105 (x86_maybe_negate_const_int): Likewise.
2106 (ix86_expand_vector_init_duplicate): Likewise.
2107 (ix86_expand_vector_init_one_nonzero): Likewise.
2108 (ix86_expand_vector_init_one_var): Likewise.
2109 (ix86_expand_vector_init_concat): Likewise.
2110 (ix86_expand_vector_init_interleave): Likewise.
2111 (ix86_expand_vector_init_general): Likewise.
2112 (ix86_expand_vector_set): Likewise.
2113 (ix86_expand_vector_extract): Likewise.
2114 (emit_reduc_half): Likewise.
2115 (ix86_emit_i387_round): Likewise.
2116 (ix86_mangle_type): Likewise.
2117 (ix86_expand_round_sse4): Likewise.
2118 (expand_vec_perm_blend): Likewise.
2119 (canonicalize_vector_int_perm): Likewise.
2120 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
2121 (expand_vec_perm_1): Likewise.
2122 (expand_vec_perm_interleave3): Likewise.
2123 (expand_vec_perm_even_odd_pack): Likewise.
2124 (expand_vec_perm_even_odd_1): Likewise.
2125 (expand_vec_perm_broadcast_1): Likewise.
2126 (ix86_vectorize_vec_perm_const_ok): Likewise.
2127 (ix86_expand_vecop_qihi): Likewise.
2128 (ix86_expand_mul_widen_hilo): Likewise.
2129 (ix86_expand_sse2_abs): Likewise.
2130 (ix86_expand_pextr): Likewise.
2131 (ix86_expand_pinsr): Likewise.
2132 (ix86_preferred_simd_mode): Likewise.
2133 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
2134 * config/i386/sse.md (*andnot<mode>3): Likewise.
2135 (<mask_codefor><code><mode>3<mask_name>): Likewise.
2136 (*<code><mode>3): Likewise.
2137 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
2138 (ia64_expand_atomic_op): Likewise.
2139 (ia64_arg_type): Likewise.
2140 (ia64_mode_to_int): Likewise.
2141 (ia64_scalar_mode_supported_p): Likewise.
2142 (ia64_vector_mode_supported_p): Likewise.
2143 (expand_vec_perm_broadcast): Likewise.
2144 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
2145 (iq2000_function_arg_advance): Likewise.
2146 (iq2000_function_arg): Likewise.
2147 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
2148 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
2149 (m68k_libcall_value): Likewise.
2150 (m68k_function_value): Likewise.
2151 (sched_attr_op_type): Likewise.
2152 * config/mcore/mcore.c (mcore_output_move): Likewise.
2153 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
2155 (microblaze_function_arg): Likewise.
2156 * config/mips/mips.c (mips16_build_call_stub): Likewise.
2157 (mips_print_operand): Likewise.
2158 (mips_mode_ok_for_mov_fmt_p): Likewise.
2159 (mips_vector_mode_supported_p): Likewise.
2160 (mips_preferred_simd_mode): Likewise.
2161 (mips_expand_vpc_loongson_even_odd): Likewise.
2162 (mips_expand_vec_unpack): Likewise.
2163 (mips_expand_vi_broadcast): Likewise.
2164 (mips_expand_vector_init): Likewise.
2165 (mips_expand_vec_reduc): Likewise.
2166 (mips_expand_msa_cmp): Likewise.
2167 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
2168 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
2169 (cc_flags_for_mode): Likewise.
2170 * config/msp430/msp430.c (msp430_print_operand): Likewise.
2171 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
2172 (nds32_output_casesi_pc_relative): Likewise.
2173 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2174 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
2175 (nvptx_gen_unpack): Likewise.
2176 (nvptx_gen_pack): Likewise.
2177 (nvptx_gen_shuffle): Likewise.
2178 (nvptx_gen_wcast): Likewise.
2179 (nvptx_preferred_simd_mode): Likewise.
2180 * config/pa/pa.c (pa_secondary_reload): Likewise.
2181 * config/pa/predicates.md (base14_operand): Likewise.
2182 * config/powerpcspe/powerpcspe-c.c
2183 (altivec_resolve_overloaded_builtin): Likewise.
2184 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
2186 (rs6000_preferred_simd_mode): Likewise.
2187 (output_vec_const_move): Likewise.
2188 (rs6000_expand_vector_extract): Likewise.
2189 (rs6000_split_vec_extract_var): Likewise.
2190 (reg_offset_addressing_ok_p): Likewise.
2191 (rs6000_legitimate_offset_address_p): Likewise.
2192 (rs6000_legitimize_address): Likewise.
2193 (rs6000_emit_set_const): Likewise.
2194 (rs6000_const_vec): Likewise.
2195 (rs6000_emit_move): Likewise.
2196 (spe_build_register_parallel): Likewise.
2197 (rs6000_darwin64_record_arg_recurse): Likewise.
2198 (swap_selector_for_mode): Likewise.
2199 (spe_init_builtins): Likewise.
2200 (paired_init_builtins): Likewise.
2201 (altivec_init_builtins): Likewise.
2202 (do_load_for_compare): Likewise.
2203 (rs6000_generate_compare): Likewise.
2204 (rs6000_expand_float128_convert): Likewise.
2205 (emit_load_locked): Likewise.
2206 (emit_store_conditional): Likewise.
2207 (rs6000_output_function_epilogue): Likewise.
2208 (rs6000_handle_altivec_attribute): Likewise.
2209 (rs6000_function_value): Likewise.
2210 (emit_fusion_gpr_load): Likewise.
2211 (emit_fusion_p9_load): Likewise.
2212 (emit_fusion_p9_store): Likewise.
2213 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
2214 (fusion_gpr_mem_load): Likewise.
2215 (fusion_addis_mem_combo_load): Likewise.
2216 (fusion_addis_mem_combo_store): Likewise.
2217 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
2218 (fusion_gpr_mem_load): Likewise.
2219 (fusion_addis_mem_combo_load): Likewise.
2220 (fusion_addis_mem_combo_store): Likewise.
2221 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2223 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
2224 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
2225 (rs6000_preferred_simd_mode): Likewise.
2226 (output_vec_const_move): Likewise.
2227 (rs6000_expand_vector_extract): Likewise.
2228 (rs6000_split_vec_extract_var): Likewise.
2229 (reg_offset_addressing_ok_p): Likewise.
2230 (rs6000_legitimate_offset_address_p): Likewise.
2231 (rs6000_legitimize_address): Likewise.
2232 (rs6000_emit_set_const): Likewise.
2233 (rs6000_const_vec): Likewise.
2234 (rs6000_emit_move): Likewise.
2235 (rs6000_darwin64_record_arg_recurse): Likewise.
2236 (swap_selector_for_mode): Likewise.
2237 (paired_init_builtins): Likewise.
2238 (altivec_init_builtins): Likewise.
2239 (rs6000_expand_float128_convert): Likewise.
2240 (emit_load_locked): Likewise.
2241 (emit_store_conditional): Likewise.
2242 (rs6000_output_function_epilogue): Likewise.
2243 (rs6000_handle_altivec_attribute): Likewise.
2244 (rs6000_function_value): Likewise.
2245 (emit_fusion_gpr_load): Likewise.
2246 (emit_fusion_p9_load): Likewise.
2247 (emit_fusion_p9_store): Likewise.
2248 * config/rx/rx.c (rx_gen_move_template): Likewise.
2249 (flags_from_mode): Likewise.
2250 * config/s390/predicates.md (s390_alc_comparison): Likewise.
2251 (s390_slb_comparison): Likewise.
2252 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
2253 (s390_vector_mode_supported_p): Likewise.
2254 (s390_cc_modes_compatible): Likewise.
2255 (s390_match_ccmode_set): Likewise.
2256 (s390_canonicalize_comparison): Likewise.
2257 (s390_emit_compare_and_swap): Likewise.
2258 (s390_branch_condition_mask): Likewise.
2259 (s390_rtx_costs): Likewise.
2260 (s390_secondary_reload): Likewise.
2261 (__SECONDARY_RELOAD_CASE): Likewise.
2262 (s390_expand_cs): Likewise.
2263 (s390_preferred_simd_mode): Likewise.
2264 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
2265 * config/sh/sh.c (sh_print_operand): Likewise.
2266 (dump_table): Likewise.
2267 (sh_secondary_reload): Likewise.
2268 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2269 * config/sh/sh.md (casesi_worker_1): Likewise.
2270 (casesi_worker_2): Likewise.
2271 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
2272 (fcc_comparison_operator): Likewise.
2273 * config/sparc/sparc.c (sparc_expand_move): Likewise.
2274 (emit_soft_tfmode_cvt): Likewise.
2275 (sparc_preferred_simd_mode): Likewise.
2276 (output_cbranch): Likewise.
2277 (sparc_print_operand): Likewise.
2278 (sparc_expand_vec_perm_bmask): Likewise.
2279 (vector_init_bshuffle): Likewise.
2280 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
2281 (spu_vector_mode_supported_p): Likewise.
2282 (spu_expand_insv): Likewise.
2283 (spu_emit_branch_or_set): Likewise.
2284 (spu_handle_vector_attribute): Likewise.
2285 (spu_builtin_splats): Likewise.
2286 (spu_builtin_extract): Likewise.
2287 (spu_builtin_promote): Likewise.
2288 (spu_expand_sign_extend): Likewise.
2289 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
2290 (tilegx_simd_int): Likewise.
2291 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
2292 (tilepro_simd_int): Likewise.
2293 * config/v850/v850.c (const_double_split): Likewise.
2294 (v850_print_operand): Likewise.
2295 (ep_memory_offset): Likewise.
2296 * config/vax/vax.c (vax_rtx_costs): Likewise.
2297 (vax_output_int_move): Likewise.
2298 (vax_output_int_add): Likewise.
2299 (vax_output_int_subtract): Likewise.
2300 * config/visium/predicates.md (visium_branch_operator): Likewise.
2301 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
2302 (visium_print_operand_address): Likewise.
2303 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2304 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
2305 (xtensa_expand_conditional_branch): Likewise.
2306 (xtensa_copy_incoming_a7): Likewise.
2307 (xtensa_output_literal): Likewise.
2308 * dfp.c (decimal_real_maxval): Likewise.
2309 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
2311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2312 Alan Hayward <alan.hayward@arm.com>
2313 David Sherwood <david.sherwood@arm.com>
2315 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
2316 (mode_nunits_inline): Likewise.
2317 (mode_inner_inline): Likewise.
2318 (mode_unit_size_inline): Likewise.
2319 (mode_unit_precision_inline): Likewise.
2320 (emit_insn_modes_h): Likewise. Also emit a #define of the
2322 (emit_mode_wider): Add an E_ prefix to mode names.
2323 (emit_mode_complex): Likewise.
2324 (emit_mode_inner): Likewise.
2325 (emit_mode_adjustments): Likewise.
2326 (emit_mode_int_n): Likewise.
2327 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
2328 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
2329 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
2330 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
2331 (CRC32_BUILTIN, ENTRY): Likewise.
2332 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
2333 (aarch64_pop_regs): Likewise.
2334 (aarch64_process_components): Likewise.
2335 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
2336 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
2337 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
2338 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
2339 * config/arm/arm.c (arm_init_libfuncs): Likewise.
2340 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
2342 * config/i386/i386-builtin.def (pcmpestr): Likewise.
2343 (pcmpistr): Likewise.
2344 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
2345 * config/mmix/mmix.c (mmix_output_condition): Likewise.
2346 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2348 * config/rl78/rl78.c (mduc_regs): Likewise.
2349 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2350 (htm_expand_builtin): Likewise.
2351 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
2352 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
2353 * config/xtensa/xtensa.c (print_operand): Likewise.
2354 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
2355 (NUM_MODE_VECTOR_INT): Likewise.
2356 * genoutput.c (null_operand): Likewise.
2357 (output_operand_data): Likewise.
2358 * genrecog.c (print_parameter_value): Likewise.
2359 * lra.c (debug_operand_data): Likewise.
2361 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2363 * dwarf2out.c (scompare_loc_descriptor_wide)
2364 (scompare_loc_descriptor_narrow): New functions, split out from...
2365 (scompare_loc_descriptor): ...here.
2366 * expmed.c (emit_store_flag_int): New function, split out from...
2367 (emit_store_flag): ...here.
2369 2017-08-30 Richard Biener <rguenther@suse.de>
2371 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
2372 (dwarf2out_early_finish): Move setting of AT_pubnames from
2373 early debug output to early finish.
2375 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
2377 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
2378 and -mdata-region to the assembler.
2380 2017-08-30 Richard Biener <rguenther@suse.de>
2382 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
2384 (gen_subprogram_die): Add DW_AT_object_pointer only early.
2385 (dwarf2out_early_global_decl): Only generate a DIE for the
2386 abstract origin if it doesn't already exist or is a declaration DIE.
2387 (resolve_addr): Do not add the linkage name twice when
2388 generating a stub DIE for the DW_TAG_GNU_call_site target.
2390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2392 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2393 Use machine_mode rather than int for arg1_mode.
2395 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
2398 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
2399 that the second argument of the built-in functions to unpack
2400 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
2401 switch statement instead a lot of if statements.
2402 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
2403 Allow 64-bit values to be in Altivec registers as well as
2404 traditional floating point registers.
2405 (pack<mode>, FMOVE128_VSX iterator): Likewise.
2407 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
2409 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
2410 MAX_REGS_PER_ADDRESS == 1.
2412 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
2414 * config/i386/i386.opt (flag_fentry): Do not init to -1.
2415 * config/i386/i386.c (ix86_option_override_internal): Simplify
2416 setting of opts->x_flag_entry.
2418 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2419 Jakub Jelinek <jakub@redhat.com>
2420 Richard Biener <rguenther@suse.de>
2422 PR tree-optimization/81503
2423 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
2424 folded constant fits in the target type; reorder tests for clarity.
2426 2017-08-29 Martin Liska <mliska@suse.cz>
2428 * passes.def: Include pass_lower_switch.
2429 * stmt.c (dump_case_nodes): Remove and move to
2430 tree-switch-conversion.
2431 (case_values_threshold): Likewise.
2432 (expand_switch_as_decision_tree_p): Likewise.
2433 (emit_case_decision_tree): Likewise.
2434 (expand_case): Likewise.
2435 (balance_case_nodes): Likewise.
2436 (node_has_low_bound): Likewise.
2437 (node_has_high_bound): Likewise.
2438 (node_is_bounded): Likewise.
2439 (emit_case_nodes): Likewise.
2440 (struct simple_case_node): New struct.
2441 (add_case_node): Remove.
2442 (emit_case_dispatch_table): Use vector instead of case_list.
2443 (reset_out_edges_aux): Remove.
2444 (compute_cases_per_edge): Likewise.
2445 (expand_case): Build list of simple_case_node.
2446 (expand_sjlj_dispatch_table): Use it.
2447 * tree-switch-conversion.c (struct case_node): Moved from
2448 stmt.c and adjusted.
2449 (emit_case_nodes): Likewise.
2450 (node_has_low_bound): Likewise.
2451 (node_has_high_bound): Likewise.
2452 (node_is_bounded): Likewise.
2453 (case_values_threshold): Likewise.
2454 (reset_out_edges_aux): Likewise.
2455 (compute_cases_per_edge): Likewise.
2456 (add_case_node): Likewise.
2457 (dump_case_nodes): Likewise.
2458 (balance_case_nodes): Likewise.
2459 (expand_switch_as_decision_tree_p): Likewise.
2460 (emit_jump): Likewise.
2461 (emit_case_decision_tree): Likewise.
2462 (try_switch_expansion): Likewise.
2463 (do_jump_if_equal): Likewise.
2464 (emit_cmp_and_jump_insns): Likewise.
2465 (fix_phi_operands_for_edge): New function.
2466 (record_phi_operand_mapping): Likewise.
2467 (class pass_lower_switch): New pass.
2468 (pass_lower_switch::execute): New function.
2469 (make_pass_lower_switch): Likewise.
2470 (conditional_probability):
2471 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
2472 * tree-pass.h: Add make_pass_lower_switch.
2474 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
2477 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
2480 2017-08-29 Richard Biener <rguenther@suse.de>
2482 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
2483 we do not add a DW_AT_inline attribute twice.
2484 (gen_subprogram_die): Remove code setting DW_AT_inline on
2485 DECL_ABSTRACT_P nodes.
2487 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
2489 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
2490 calls to internal functions.
2491 (gimplify_modify_expr): Likewise.
2492 * tree-call-cdce.c (use_internal_fn): Likewise.
2493 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
2494 (convert_to_divmod): Set the nothrow flag.
2495 * tree-if-conv.c (predicate_mem_writes): Likewise.
2496 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
2497 (vectorizable_call): Likewise.
2498 (vectorizable_store): Likewise.
2499 (vectorizable_load): Likewise.
2500 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
2501 (vect_recog_mask_conversion_pattern): Likewise.
2503 2017-08-29 Martin Liska <mliska@suse.cz>
2506 * gcc.c (driver_handle_option): Add new argument.
2507 * opts-common.c (handle_option): Pass
2508 target_option_override_hook.
2509 * opts-global.c (lang_handle_option): Add new option.
2510 (set_default_handlers): Add new argument.
2511 (decode_options): Likewise.
2512 * opts.c (target_handle_option): Likewise.
2513 (common_handle_option): Call target_option_override_hook.
2514 * opts.h (struct cl_option_handler_func): Add hook for
2515 target option override.
2516 (struct cl_option_handlers): Likewise.
2517 (set_default_handlers): Add new argument.
2518 (decode_options): Likewise.
2519 (common_handle_option): Likewise.
2520 (target_handle_option): Likewise.
2521 * toplev.c (toplev::main): Pass targetm.target_option.override
2524 2017-08-29 Richard Biener <rguenther@suse.de>
2525 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2527 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
2528 life to the active subtree.
2530 2017-08-28 Jeff Law <law@redhat.com>
2532 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
2533 derive_equivalences.
2534 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
2536 (edge_info::derive_equivalences): New private member function
2538 * tree-ssa-dom.c (class edge_info): Changed from a struct
2539 to a class. Add ctor/dtor, methods and data members.
2540 (edge_info::edge_info): Renamed from allocate_edge_info.
2541 Initialize additional members.
2542 (edge_info::~edge_info): New.
2543 (free_dom_edge_info): Delete the edge info.
2544 (record_edge_info): Use new class & associated member functions.
2545 Tighten forms for testing for edge equivalences.
2546 (record_temporary_equivalences): Iterate over the simple
2547 equivalences rather than assuming there's only one per edge.
2548 (cprop_into_successor_phis): Iterate over the simple
2549 equivalences rather than assuming there's only one per edge.
2550 (optimize_stmt): Use operand_equal_p rather than pointer
2551 equality for mini-DSE code.
2553 2017-08-28 Nathan Sidwell <nathan@acm.org>
2555 * gcc.c (execute): Fold SIGPIPE handling into switch
2556 statement. Adjust internal error message.
2558 2017-08-28 Richard Biener <rguenther@suse.de>
2561 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
2562 Do nothing for removed DIEs.
2564 2017-08-28 Richard Biener <rguenther@suse.de>
2566 PR tree-optimization/81977
2567 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
2570 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
2573 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
2574 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
2575 using targetm.gen_mem_thread_fence.
2577 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
2580 * config/i386/i386.md (*<btsc><mode>): Change operand 2
2581 predicate to register_operand. Reorder operands.
2582 (*btr<mode>): Ditto.
2583 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
2584 (*btr<mode>_mask): Ditto.
2586 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
2588 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
2589 * config/rs6000/xmmintrin.h: New file.
2590 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
2592 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2595 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
2596 parameter and_insn and return it.
2597 (recombine_lvx_pattern): Insert a copy to ensure availability of
2598 the base register of the copied masking operation at the point of
2599 the instruction replacement.
2600 (recombine_stvx_pattern): Likewise.
2602 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
2604 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
2605 undocumented switches.
2606 (-mpower9-dform-vector): Likewise.
2607 (-mpower9-dform): Likewise.
2608 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
2609 comments to delete references to -mpower9-dform* switches.
2610 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
2611 Delete reference to -mpower9-dform* switches, test for
2612 -mpower9-vector instead.
2613 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
2614 (OTHER_P9_VECTOR_MASKS): Likewise.
2615 (POWERPC_MASKS): Likewise.
2616 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
2617 tests against -mpower9-dform* to -mpower9-vector. Delete code
2618 that checked for -mpower9-dform* consistancy with other options.
2619 Add test for -mpower9-misc to enable other power9 switches.
2620 (rs6000_init_hard_regno_mode_ok): Likewise.
2621 (rs6000_option_override_internal): Likewise.
2622 (rs6000_emit_prologue): Likewise.
2623 (rs6000_emit_epilogue): Likewise.
2624 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
2625 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
2626 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
2628 (emit_fusion_p9_store): Likewise.
2629 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
2630 resetting these macros if the assembler does not support ISA 3.0
2632 (TARGET_P9_DFORM_VECTOR): Likewise.
2633 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
2634 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
2636 2017-08-25 Alan Modra <amodra@gmail.com>
2638 PR rtl-optimization/81747
2639 * cse.c (cse_extended_basic_block): Don't attempt to record
2640 equivalences for degenerate conditional branches that branch
2641 to their fall-through.
2643 2017-08-24 Martin Sebor <msebor@redhat.com>
2646 * gimple-fold.c (size_must_be_zero_p): New function.
2647 (gimple_fold_builtin_memory_op): Call it.
2649 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
2651 * config/rs6000/mm_malloc.h: New file.
2653 2017-08-24 Bin Cheng <bin.cheng@arm.com>
2655 PR tree-optimization/81913
2656 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
2657 analysis when either IVs in condition can wrap.
2659 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
2661 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
2662 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
2664 2017-08-24 Richard Biener <rguenther@suse.de>
2667 * targhooks.c (default_target_can_inline_p): Properly
2668 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
2669 is present and always compare.
2670 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
2671 infer -mfpmath=sse from TARGET_SSE_P.
2672 (ix86_can_inline_p): Properly use target_option_default_node when
2673 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
2675 2017-08-24 Richard Biener <rguenther@suse.de>
2678 * dwarf2out.c (output_die): Handle flag_generate_offload like
2680 (output_comp_unit): Likewise.
2681 (gen_array_type_die): Likewise.
2682 (dwarf2out_early_finish): Likewise.
2683 (note_variable_value_in_expr): Likewise.
2684 (dwarf2out_finish): Likewise. Adjust assert.
2685 * cgraphunit.c (symbol_table::compile): Move setting of
2686 flag_generate_offload earlier ...
2687 (symbol_table::finalize_compilation_unit): ... here, before
2688 early debug finalization.
2690 2017-08-24 Richard Biener <rguenther@suse.de>
2692 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
2693 and ipa-fnsummary.h.
2694 (ix86_can_inline_p): When ix86_fpmath flags do not match
2695 check whether the callee uses FP math at all.
2697 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
2700 * tree-ssanames.c (get_nonzero_bits): Use element_precision
2701 instead of TYPE_PRECISION.
2703 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
2704 Alan Hayward <alan.hayward@arm.com>
2705 David Sherwood <david.sherwood@arm.com>
2707 * combine.c (make_extraction): Use subreg_offset_from_lsb.
2709 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
2711 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
2713 (ix86_frame::stack_realign_allocate): New field.
2714 (struct machine_frame_state): Modify comments.
2715 (machine_frame_state::sp_realigned_fp_end): New field.
2716 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
2718 (sp_valid_at): Add assertion to assure no attempt to access invalid
2719 offset of a realigned stack.
2720 (fp_valid_at): Likewise.
2721 (choose_baseaddr): Modify comments.
2722 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
2723 ix86_expand_prologue.
2724 (ix86_expand_prologue): Modify stack realignment and allocation.
2725 (ix86_expand_epilogue): Modify comments.
2726 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
2727 avx2_runtime, avx512f, and avx512f_runtime.
2729 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
2731 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
2732 (mstackrealign): Do not init to -1.
2733 * config/i386/i386.c (ix86_option_override_internal):
2734 Check opts_set, not opts when setting default value of
2735 opts->x_ix86_force_align_arg_pointer.
2737 2017-08-23 Richard Biener <rguenther@suse.de>
2739 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
2740 lang_hooks.decl_printable_name.
2741 * print-rtl-function.c (print_rtx_function): Likewise.
2742 * tree-pretty-print.c (dump_function_header): Likewise.
2744 2017-08-23 Richard Biener <rguenther@suse.de>
2747 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
2748 -g0 at compile-time.
2750 2017-08-23 Tamar Christina <tamar.christina@arm.com>
2753 * doc/sourcebuild.texi (Other hardware attributes):
2756 2017-08-23 Tamar Christina <tamar.christina@arm.com>
2759 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
2760 Fix single-use check.
2762 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
2764 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
2766 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
2768 * doc/install.texi: Modify to add more details on running selected
2771 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
2773 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
2774 is combined with -mabi=ms.
2775 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
2778 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2780 PR tree-optimization/81488
2781 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
2782 and cached_basis fields.
2783 (MAX_SPREAD): New constant.
2784 (alloc_cand_and_find_basis): Initialize new fields.
2785 (clear_visited): New function.
2786 (create_phi_basis_1): Rename from create_phi_basis, set visited
2787 and cached_basis fields.
2788 (create_phi_basis): New wrapper function.
2789 (phi_add_costs_1): Rename from phi_add_costs, add spread
2790 parameter, set visited field, short-circuit when limits reached.
2791 (phi_add_costs): New wrapper function.
2792 (record_phi_increments_1): Rename from record_phi_increments, set
2794 (record_phi_increments): New wrapper function.
2795 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
2796 (phi_incr_cost): New wrapper function.
2797 (all_phi_incrs_profitable_1): Rename from
2798 all_phi_incrs_profitable, set visited field.
2799 (all_phi_incrs_profitable): New wrapper function.
2801 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
2802 Alan Hayward <alan.hayward@arm.com>
2803 David Sherwood <david.sherwood@arm.com>
2805 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
2806 that takes the outer and inner modes.
2807 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
2808 comparison as the canonical test for a paradoxical subreg.
2809 * combine.c (simplify_set): Use paradoxical_subreg_p.
2810 (make_extraction): Likewise.
2811 (force_to_mode): Likewise.
2812 (rtx_equal_for_field_assignment_p): Likewise.
2813 (gen_lowpart_for_combine): Likewise.
2814 (simplify_comparison): Likewise.
2815 * cse.c (equiv_constant): Likewise.
2816 * expmed.c (store_bit_field_1): Likewise.
2817 * final.c (alter_subreg): Likewise.
2818 * fwprop.c (propagate_rtx): Likewise.
2819 (forward_propagate_subreg): Likewise.
2820 * ira-conflicts.c (ira_build_conflicts): Likewise.
2821 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
2822 * lra-constraints.c (curr_insn_transform): Likewise.
2823 (split_reg): Likewise.
2824 * lra-eliminations.c (move_plus_up): Likewise.
2825 (lra_eliminate_regs_1): Likewise.
2826 * recog.c (general_operand): Likewise.
2827 * ree.c (combine_reaching_defs): Likewise.
2828 * reload.c (push_reload): Likewise.
2829 (find_reloads): Likewise.
2830 * reload1.c (elimination_effects): Likewise.
2831 (compute_reload_subreg_offset): Likewise.
2832 (choose_reload_regs): Likewise.
2833 * rtlanal.c (subreg_lsb_1): Likewise.
2834 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2835 (simplify_subreg): Likewise.
2836 * var-tracking.c (track_loc_p): Likewise.
2837 * emit-rtl.c (byte_lowpart_offset): Likewise.
2838 (paradoxical_subreg_p): Delete out-of-line definition.
2840 2017-08-22 Jeff Law <law@redhat.com>
2842 PR tree-optimization/81741
2843 PR tree-optimization/71947
2844 * tree-ssa-dom.c: Include tree-inline.h.
2845 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
2846 equivalences if one is more expensive to compute than the other.
2847 * tree-ssa-scopedtables.h (class const_or_copies): Make
2848 record_const_or_copy_raw method private.
2849 (class avail_exprs_stack): New method simplify_binary_operation.
2850 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
2851 avail_exprs_stack::simplify_binary_operation as needed.
2852 (avail_exprs_stack::simplify_binary_operation): New function.
2854 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
2856 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
2857 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
2858 (DOT_SYMBOLS): Likewise.
2859 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
2860 (RELOCATABLE_NEEDS_FIXUP): Likewise.
2861 (RS6000_ABI_NAME): Likewise.
2862 (TARGET_CMODEL): Likewise.
2863 (TOC_SECTION_ASM_OP): Likewise.
2864 (SET_CMODEL): New macro.
2865 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
2867 2017-08-22 Richard Biener <rguenther@suse.de>
2869 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
2870 to allow for free-lang-data replacements similar to verify_type_variant.
2872 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
2873 Alan Hayward <alan.hayward@arm.com>
2874 David Sherwood <david.sherwood@arm.com>
2876 * config/aarch64/aarch64.md (casesi): Use DImode rather than
2877 VOIDmode for the LABEL_REF.
2879 2017-08-22 Richard Biener <rguenther@suse.de>
2881 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
2883 2017-08-22 Richard Biener <rguenther@suse.de>
2885 * common.opt (feliminate-dwarf2-dups): Ignore.
2886 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
2887 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
2888 same_die_p_wrap, compute_section_prefix,
2889 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
2890 (comdat_symbol_id, comdat_symbol_number): Likewise.
2891 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
2893 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
2894 (output_die): Mark unreachable path unreachable.
2895 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
2896 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
2897 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
2898 (dwarf2out_early_finish): Likewise.
2900 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
2902 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
2904 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
2907 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
2908 not VAR_P. Filter attribute warnings with OPT_Wattributes.
2909 (avr_attribute_table) <io, io_low, address>: Initialize
2910 .decl_required with true.
2912 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
2914 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
2915 undocumented debugging options.
2916 (-mvsx-scalar-double): Likewise.
2917 (-mallow-df-permute): Likewise.
2918 (-mvectorize-builtins): Likewise.
2919 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2920 (rs6000_builtin_vectorized_function): Likewise.
2921 (rs6000_builtin_md_vectorized_function): Likewise.
2922 (rs6000_opt_vars): Likewise.
2924 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
2927 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
2928 (*btrq_imm): Rename from *btrq.
2929 (*btcq_imm): Rename from *btcq.
2930 (btsc): New code attribute.
2931 (*<btsc><mode>): New insn pattern.
2932 (*btr<mode>): Ditto.
2933 (*<btsc><mode>_mask): New insn_and_split pattern.
2934 (*btr<mode>_mask): Ditto.
2936 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2937 Alan Hayward <alan.hayward@arm.com>
2938 David Sherwood <david.sherwood@arm.com>
2940 * function.c (pad_below): Simplify padding calculation.
2942 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2943 Alan Hayward <alan.hayward@arm.com>
2944 David Sherwood <david.sherwood@arm.com>
2946 * target.def (function_prologue): Remove frame size argument.
2947 (function_epilogue): Likewise.
2948 * doc/tm.texi: Regenerate.
2949 * final.c (final_start_function): Update call to function_prologue.
2950 (final_end_function): Update call to function_epilogue.
2951 (default_function_pro_epilogue): Remove frame size argument.
2952 * output.h (default_function_pro_epilogue): Likewise.
2953 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
2954 (arm_output_function_prologue): Likewise.
2955 * config/frv/frv.c (frv_function_prologue): Likewise.
2956 (frv_function_epilogue): Likewise.
2957 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
2958 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
2959 (ia64_output_function_epilogue): Likewise.
2960 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
2961 (m32r_output_function_epilogue): Likewise.
2962 * config/microblaze/microblaze.c (microblaze_function_prologue)
2963 (microblaze_function_epilogue): Likewise.
2964 * config/mips/mips.c (mips_output_function_prologue): Likewise.
2965 (mips_output_function_epilogue): Likewise.
2966 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
2967 (mmix_target_asm_function_epilogue): Likewise.
2968 * config/msp430/msp430.c (msp430_start_function): Likewise.
2969 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
2970 (nds32_asm_function_epilogue): Likewise.
2971 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
2972 * config/pa/pa.c (pa_output_function_prologue): Likewise.
2973 (pa_output_function_epilogue): Likewise.
2974 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
2975 (rs6000_output_function_epilogue): Likewise.
2976 * config/rl78/rl78.c (rl78_start_function): Likewise.
2977 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
2978 (rs6000_output_function_epilogue): Likewise.
2979 * config/rx/rx.c (rx_output_function_prologue): Likewise.
2980 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
2981 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
2982 (sparc_asm_function_epilogue): Likewise.
2984 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2986 * tree.h (type_has_mode_precision_p): New function.
2987 * convert.c (convert_to_integer_1): Use it.
2988 * expr.c (expand_expr_real_2): Likewise.
2989 (expand_expr_real_1): Likewise.
2990 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
2991 * match.pd: Likewise.
2992 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
2993 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
2994 * tree-tailcall.c (process_assignment): Likewise.
2995 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2996 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
2997 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
2998 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2999 (vectorizable_assignment): Likewise.
3000 (vectorizable_shift): Likewise.
3001 (vectorizable_operation): Likewise.
3002 * tree-vrp.c (register_edge_assert_for_2): Likewise.
3004 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
3006 * match.pd: Add pow (C, x) simplification.
3008 2017-08-21 Richard Biener <rguenther@suse.de>
3010 PR tree-optimization/81900
3011 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
3012 for blocks with abnormal predecessors.
3013 (compute_antic): Do not set visited flag prematurely.
3015 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
3018 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
3020 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3022 * stor-layout.h (vector_type_mode): Move to...
3023 * tree.h (vector_type_mode): ...here.
3024 * stor-layout.c (vector_type_mode): Move to...
3025 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
3027 2017-08-21 Richard Biener <rguenther@suse.de>
3029 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
3030 register_external_die hooks.
3031 (debug_false_tree_charstarstar_uhwistar): Declare.
3032 (debug_nothing_tree_charstar_uhwi): Likewise.
3033 * debug.c (do_nothing_debug_hooks): Adjust.
3034 (debug_false_tree_charstarstar_uhwistar): New do nothing.
3035 (debug_nothing_tree_charstar_uhwi): Likewise.
3036 * dbxout.c (dbx_debug_hooks): Adjust.
3037 (xcoff_debug_hooks): Likewise.
3038 * sdbout.c (sdb_debug_hooks): Likewise.
3039 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3040 * dwarf2out.c (macinfo_label_base): New global.
3041 (dwarf2out_register_external_die): New function for the
3042 register_external_die hook.
3043 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
3044 (dwarf2_debug_hooks): Use them.
3045 (dwarf2_lineno_debug_hooks): Adjust.
3046 (struct die_struct): Add with_offset flag.
3047 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
3048 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
3049 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
3050 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
3051 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
3052 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
3053 defining section names for the early LTO debug variants.
3054 (reset_indirect_string): New helper.
3055 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
3056 (print_dw_val): Add support for offsetted symbol references.
3057 (get_ultimate_context): Split out from is_cxx.
3058 (is_cxx): Use get_ultimate_context.
3059 (is_fortran): Add decl overload.
3060 (compute_comp_unit_symbol): Split out worker from
3061 compute_section_prefix.
3062 (compute_section_prefix): Call compute_comp_unit_symbol and
3063 set comdat_type_p here.
3064 (output_die): Skip DIE symbol output for the LTO added one.
3065 Handle DIE symbol references with offset.
3066 (output_comp_unit): Guard section name mangling properly.
3067 For LTO debug sections emit a symbol at the section beginning
3068 which we use to refer to its DIEs.
3069 (add_abstract_origin_attribute): For DIEs registered via
3070 dwarf2out_register_external_die directly refer to the early
3071 DIE rather than indirectly through the shadow one we created.
3072 Remove obsolete call to dwarf2out_abstract_function for
3073 non-function/block origins.
3074 (gen_array_type_die): When generating early LTO debug do
3075 not emit DW_AT_string_length.
3076 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
3077 late when in LTO. As suggested place a gcc_unreachable for
3078 the DECL_ABSTRACT_P case.
3079 (gen_subprogram_die): Avoid another specification DIE
3080 for early built declarations/definitions for the late LTO case.
3081 (gen_variable_die): Add type references for late duplicated VLA dies
3083 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
3084 we have the abstract instance already.
3085 (process_scope_var): Adjust decl DIE contexts in LTO which
3086 first puts them in limbo.
3087 (gen_decl_die): Do not generate type DIEs late apart from
3088 types for VLAs or for decls we do not yet have a DIE. Do not
3089 call dwarf2out_abstract_function late.
3090 (dwarf2out_early_global_decl): Make sure to create DIEs
3091 for abstract instances of a decl first.
3092 (dwarf2out_late_global_decl): Adjust comment.
3093 (output_macinfo_op): With multiple macro sections use
3094 macinfo_label_base to distinguish labels.
3095 (output_macinfo): Likewise. Update macinfo_label_base.
3096 Pass in the line info label.
3097 (note_variable_value_in_expr): When generating LTO resolve
3098 all variable values here by generating DIEs as needed.
3099 (init_sections_and_labels): Add early LTO debug flag parameter
3100 and generate different sections and names if set. Add generation
3101 counter for the labels so we can have multiple of them.
3102 (reset_dies): Helper to allow DIEs to be output multiple times.
3103 (dwarf2out_finish): When outputting DIEs to the fat part of an
3104 LTO object first reset DIEs.
3105 (dwarf2out_early_finish): Output early DIEs when generating LTO.
3106 (modified_type_die): Check for decl_ultimate_origin being self
3108 (gen_type_die_with_usage): Likewise.
3109 (gen_typedef_die): Allow decl_ultimate_origin being self.
3110 (set_decl_abstract_flags): Remove.
3111 (set_block_abstract_flags): Likewise.
3112 (dwarf2out_abstract_function): Treat the early generated DIEs
3113 as the abstract copy and only add DW_AT_inline and
3114 DW_AT_artificial here and call set_decl_origin_self.
3115 If the DIE has an abstract origin don't do anything.
3116 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
3117 if we have none yet (Go fails to build one, PR78628).
3118 (variably_modified_type_p): Prevent endless recursion for Ada
3119 cyclic pointer types.
3120 * lto-streamer-in.c: Include debug.h.
3121 (dref_queue): New global.
3122 (lto_read_tree_1): Stream in DIE references.
3123 (lto_input_tree): Register DIE references.
3124 (input_function): Stream DECL_DEBUG_ARGS.
3125 * lto-streamer-out.c: Include debug.h.
3126 (lto_write_tree_1): Output DIE references.
3127 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
3128 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
3129 (output_function): Stream DECL_DEBUG_ARGS.
3130 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
3131 Stream DECL_ABSTRACT_ORIGIN.
3132 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
3133 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
3134 DECL_CONTEXT for file-scope decls.
3135 * lto-streamer.h (struct dref_entry): Declare.
3136 (dref_queue): Likewise.
3137 * cfgexpand.c (pass_expand::execute): Do not call the
3138 outlining_inline_function hook here.
3139 * lto-wrapper.c (debug_obj): New global.
3140 (tool_cleanup): Unlink it if required.
3141 (debug_objcopy): New function.
3142 (run_gcc): Handle early debug sections in the IL files by
3143 extracting them to separate files, partially linkin them and
3144 feeding the result back as result to the linker.
3145 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
3146 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
3147 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
3148 sections into a separate segment.
3149 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
3151 (darwin_asm_dwarf_section): Likewise.
3152 (darwin_asm_output_dwarf_offset): Likewise.
3153 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
3155 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3156 Alan Hayward <alan.hayward@arm.com>
3157 David Sherwood <david.sherwood@arm.com>
3159 * read-md.h (md_reader::record_potential_iterator_use): Replace
3160 pointer argument with an rtx and an index.
3161 * read-rtl.c (iterator_group::apply_iterator): Likewise.
3162 (apply_mode_iterator): Likewise.
3163 (apply_code_iterator): Likewise.
3164 (apply_int_iterator): Likewise.
3165 (apply_subst_iterator): Likewise.
3166 (record_iterator_use): Likewise.
3167 (record_attribute_use): Likewise.
3168 (md_reader::record_potential_iterator_use): Likewise. Update calls
3169 to record_iterator_use and apply_iterator.
3170 (iterator_use): Replace ptr with x and index.
3171 (attribute_use): Likewise.
3172 (apply_attribute_uses): Update calls to apply_iterator.
3173 (apply_iterators): Likewise. Update initialization of iterator_use.
3174 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
3175 and record_potential_iterator_use.
3176 (rtx_reader::read_rtx_operand): Likewise.
3178 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3179 Alan Hayward <alan.hayward@arm.com>
3180 David Sherwood <david.sherwood@arm.com>
3182 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
3185 2017-08-21 Richard Biener <rguenther@suse.de>
3188 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
3189 at struct end conservatively when comparing common bases.
3191 2017-08-21 Richard Biener <rguenther@suse.de>
3193 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
3194 (mem_ref_in_stmt): Remove.
3195 (determine_max_movement): Use ref index to get at the reference.
3196 (invariantness_dom_walker::before_dom_children): Deal with
3197 lim data already initialized.
3198 (gather_mem_refs_stmt): Initialize lim data and record ref index.
3200 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
3202 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
3203 (TARGET_ISA_ROUND): Ditto.
3204 (TARGET_ROUND): Ditto.
3205 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
3206 * config/i386/i386.md: Ditto.
3207 * config/i386/sse.md: Ditto.
3208 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
3209 with OPTION_MASK_ISA_SSE4_1.
3211 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
3214 * doc/extend.texi (x86 Built-in Functions): Correct the name of
3215 __builtin_ia32_lzcnt_u16.
3217 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
3220 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
3221 (rs6000_set_current_function): Rewrite function to use it.
3223 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
3226 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
3227 and TYPE_WARN_IF_NOT_ALIGN.
3228 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
3229 (handle_warn_if_not_align): New.
3230 (place_union_field): Call handle_warn_if_not_align.
3231 (place_field): Call handle_warn_if_not_align.
3232 Copy TYPE_WARN_IF_NOT_ALIGN.
3233 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
3234 (layout_type): Likewise.
3235 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
3237 (tree_decl_common): Add warn_if_not_align.
3238 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
3239 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
3240 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
3241 (DECL_WARN_IF_NOT_ALIGN): Likewise.
3242 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
3243 * doc/extend.texi: Document warn_if_not_aligned attribute.
3244 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
3246 2017-08-17 Martin Liska <mliska@suse.cz>
3249 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
3250 (get_data_dependence): Use it as pointer type.
3251 (distribute_loop): Likewise.
3253 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3255 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
3256 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
3257 (p8_vmrgow_<mode>_direct): New define_insn.
3258 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
3259 handle endianness for vmrgew and vmrgow permute patterns.
3261 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
3263 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
3264 * config/rs6000/rs6000-cpus.def: Remove comment.
3265 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
3266 (POWERPC_MASKS): Likewise.
3267 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
3268 use of TARGET_VSX_TIMODE.
3269 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
3270 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
3271 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
3272 (rs6000_option_override_internal): Remove dead code.
3273 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
3274 (rs6000_legitimize_reload_address): Likewise.
3275 (rs6000_legitimate_address_p): Likewise.
3276 (rs6000_opt_masks): Delete "vsx-timode".
3277 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
3278 from function comment.
3279 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
3280 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
3281 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
3283 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
3284 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
3285 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
3288 (define_peephole2): Likewise.
3290 2017-08-17 Martin Sebor <msebor@redhat.com>
3293 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
3294 past the end of an array.
3295 (test_pp_format): Add test cases.
3297 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
3299 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
3300 missing ECF_NOTHROW flags.
3302 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
3305 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
3306 operands residing in integer registers.
3307 (*vsx_le_perm_load_<mode>): Likewise.
3308 (*vsx_le_perm_store_<mode>): Likewise.
3309 (define_peephole2): Add peepholes to optimize the above.
3311 2017-08-17 Marek Polacek <polacek@redhat.com>
3314 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
3315 to mimic what shorten_compare did. Change the return type to bool.
3316 (fold_cond_expr_with_comparison): Update call to
3317 operand_equal_for_comparison_p.
3318 (fold_ternary_loc): Likewise.
3320 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
3322 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
3324 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
3326 2017-08-17 Richard Biener <rguenther@suse.de>
3328 * tree-ssa-structalias.c (solve_graph): When propagating
3329 to successors update the graphs succ edges and avoid duplicate work.
3331 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
3334 * config/i386/i386.c (ix86_option_override_internal): Save target
3335 specific options after ix86_stack_protector_guard_reg was changed.
3337 2017-08-17 Richard Biener <rguenther@suse.de>
3339 PR tree-optimization/81827
3340 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
3341 (new_var_info): Initialize it conservatively.
3342 (get_call_vi): Mark register vars.
3343 (new_scalar_tmp_constraint_exp): Likewise.
3344 (handle_rhs_call): Likewise.
3345 (handle_const_call): Likewise.
3346 (create_function_info_for): Likewise.
3347 (solve_constraints): Sort varinfos to separate register from
3348 non-register vars to pack points-to solution bitmaps during
3351 2017-08-17 Marek Polacek <polacek@redhat.com>
3353 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
3355 2017-08-17 Richard Biener <rguenther@suse.de>
3357 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
3358 to true when overflow is undefined and we saturated the result.
3360 2017-08-17 Alan Modra <amodra@gmail.com>
3363 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
3364 Don't use store multiple if only one reg needs saving.
3365 (interesting_frame_related_regno): New function.
3366 (rs6000_frame_related): Don't emit frame info for regs that
3368 (rs6000_emit_epilogue): Likewise.
3370 2017-08-16 Nathan Sidwell <nathan@acm.org>
3372 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
3373 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
3374 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
3375 * tree.c (free_lang_data_in_type): Use else-if chain. Always
3376 clear TYPE_LANG_1. Remove obsolete member-function stripping.
3377 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
3378 (verify_type): Adjust for TYPE_BINFO move.
3379 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
3380 process TYPE_BINFO directly.
3381 (hash_tree): Likewise.
3382 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
3384 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
3387 2017-08-16 David Malcolm <dmalcolm@redhat.com>
3389 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
3391 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
3394 * config/i386/i386.md (*anddi_1_btr): Change predicates of
3395 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
3396 Add ix86_binary_operator_ok to insn constraint.
3397 (*iordi_1_bts): Ditto.
3398 (*xordi_1_btc): Ditto.
3399 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
3400 Update corresponding peephole2 pattern.
3404 2017-08-16 Bin Cheng <bin.cheng@arm.com>
3406 PR tree-optimization/81832
3407 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
3408 copy loop header which has IFN_LOOP_DIST_ALIAS call.
3410 2017-08-16 Marek Polacek <polacek@redhat.com>
3413 * fold-const.c (fold_indirect_ref_1): Restore original behavior
3414 regarding size_zero_node.
3416 2017-08-16 Martin Liska <mliska@suse.cz>
3419 * config.gcc: Respect previously set extra_objs in case
3422 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
3424 PR tree-optimization/81835
3425 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
3426 the phi SSA_NAME. Check that the condition in a COND_EXPR does
3427 not depend on the phi.
3429 2017-08-16 Alan Modra <amodra@gmail.com>
3431 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
3434 2017-08-16 Alan Modra <amodra@gmail.com>
3436 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
3437 (save_reg_p): ..into this. Update all callers.
3438 (first_reg_to_save): Simplify.
3440 2017-08-16 Alan Modra <amodra@gmail.com>
3442 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
3445 2017-08-15 Joseph Myers <joseph@codesourcery.com>
3449 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
3450 constant count if that count is less than 32.
3452 2017-08-15 Nathan Sidwell <nathan@acm.org>
3454 * gcc.c (execute): Emit friendlier message if inferior is killed
3455 by an external cause.
3457 2017-08-15 Richard Biener <rguenther@suse.de>
3459 PR tree-optimization/81790
3460 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
3461 CONSTRUCTORs from simplifying and VN.
3463 2017-08-14 Martin Sebor <msebor@redhat.com>
3465 * builtin-attrs.def: Add comments.
3467 2017-08-14 Martin Sebor <msebor@redhat.com>
3470 * doc/extend.texi (attribute nonstring): Document new attribute.
3472 2017-08-14 Martin Sebor <msebor@redhat.com>
3475 * tree-diagnostic.c (default_tree_printer): Handle %G.
3476 * gimple-pretty-print.h (percent_G_format): Declare new function.
3477 * gimple-pretty-print.c (percent_G_format): Define.
3478 * tree-pretty-print.c (percent_K_format): Add argument.
3480 2017-08-14 Martin Sebor <msebor@redhat.com>
3482 PR translation/79998
3483 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
3484 Remove a stray space.
3486 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
3489 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
3490 (*iordi_1_bts): Ditto.
3491 (*xordi_1_btc): Ditto.
3493 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3496 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
3497 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3499 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
3500 quoted strings, and make more translator-friendly.
3501 (darwin_rs6000_override_options): Likewise.
3502 (rs6000_option_override_internal): Likewise.
3503 (rs6000_return_in_memory): Fix overlong line.
3504 (init_cmulative_args): Use quoted strings, and make more
3505 translator-friendly.
3506 (rs6000_pass_by_reference): Fix overlong line.
3507 (def_builtin): Use quoted strings.
3508 (altivec_expand_predicate_builtin): Use quoted strings, and make
3509 more translator-friendly.
3510 (htm_expand_builtin): Use quoted strings.
3511 (cpu_expand_builtin): Use quoted strings, and make more
3512 translator-friendly.
3513 (altivec_expand_builtin): Likewise.
3514 (paired_expand_predicate_builtin): Likewise.
3515 (rs6000_invalid_builtin): Likewise.
3516 (builtin_function_type): Use quoted strings.
3517 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
3518 more translator-friendly.
3519 (rs6000_trampoline_init): Likewise.
3520 (rs6000_handle_altivec_attribute): Likewise.
3521 (rs6000_inner_target_options): Use quoted strings.
3522 (rs6000_disable_incompatible_switches): Likewise.
3523 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
3524 strings, and make more translator-friendly.
3525 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
3527 2017-08-14 Bin Cheng <bin.cheng@arm.com>
3529 PR tree-optimization/81799
3530 * tree-loop-distribution.c (version_loop_by_alias_check): Force
3531 cond_expr to simple gimple operand.
3533 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
3536 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
3538 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
3542 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
3543 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
3544 TARGET_GASISR_PROLOGUES.
3545 * config/avr/avr.c (avr_option_override): Same.
3546 (avr_pass_pre_proep::execute): Same.
3548 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
3551 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
3552 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
3554 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
3556 * config/i386/i386.md (*load_tp_<mode>): Redefine as
3557 define_insn_and_split. Split to a memory load from 0 in
3558 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
3559 using PTR mode iterator.
3560 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
3561 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
3562 (*add_tp_<mode>): Redefine as define_insn_and_split.
3563 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
3564 address space. Merge with *add_tp_x32 using PTR mode iterator.
3565 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
3566 Split to an add with a memory load from 0 in
3567 DEFAULT_TLS_SEG_REG address space.
3569 2017-08-12 Andrew Pinski <apinski@cavium.com>
3571 * config/aarch64/aarch64-option-extensions.def (rdma):
3572 Fix feature string to what Linux prints out in /proc/cpuinfo.
3574 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
3577 * dwarf2out.c (modified_type_die): For C typedef types that have
3578 an ultimate origin, process the ultimate origin instead of the
3580 (gen_typedef_die): Assert that input DECLs have no ultimate
3582 (gen_type_die_with_usage): For typedef variants that have an
3583 ultimate origin, just call gen_decl_die on the original DECL.
3584 (process_scope_var): Avoid creating DIEs for local typedefs and
3585 concrete static variables.
3587 2017-08-12 Alan Modra <amodra@gmail.com>
3591 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
3592 match gnu-user.h startfile.
3593 (ENDFILE_LINUX_SPEC): Similarly.
3595 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
3598 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
3600 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
3602 2017-08-11 Tamar Christina <tamar.christina@arm.com>
3603 * config/aarch64/aarch64.md (mov<mode>): Change.
3604 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
3605 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
3606 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
3608 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
3610 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
3611 for storage order barriers.
3613 2017-08-11 Martin Liska <mliska@suse.cz>
3616 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
3617 variables of void type.
3619 2017-08-11 Martin Liska <mliska@suse.cz>
3621 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
3622 TARGET_SUPPORTS_ALIASES.
3623 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
3624 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
3625 (optimize_weakref): Likewise.
3626 * symtab.c (symtab_node::noninterposable_alias): Likewise.
3627 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
3628 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
3630 2017-08-11 Martin Liska <mliska@suse.cz>
3633 * config/i386/i386.c (make_resolver_func): Do complete
3634 refactoring of the function.
3636 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
3639 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
3640 * config/i386/i386.c (ix86_stack_protect_guard): Use
3641 ix86_stack_protect_guard_symbol_str to generate varible declaration.
3642 * doc/invoke.texi (x86 Options): Document
3643 -mstack-protector-guard-symbol= option.
3645 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
3647 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
3648 * config/i386/i386.c (ix86_split_stack_guard): New function.
3649 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
3650 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
3651 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
3652 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
3653 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
3654 (split_stack_space_check): Call ix86_split_stack_guard.
3656 2017-08-10 Martin Sebor <msebor@redhat.com>
3658 * print-tree.c (print_node): Print location using the established
3660 Replace spaces with colons.
3661 (debug_raw, debug): Ditto.
3663 2017-08-10 Martin Sebor <msebor@redhat.com>
3666 * pretty-print.c (pp_format): Correct the handling of %s precision.
3668 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
3671 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
3673 (ix86_finalize_stack_frame_flags): This. Also clear
3674 frame_pointer_needed if -fno-omit-frame-pointer is used without
3676 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
3677 with ix86_finalize_stack_frame_flags.
3678 (ix86_expand_epilogue): Likewise.
3679 (ix86_expand_split_stack_prologue): Likewise.
3680 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
3682 2017-08-10 Martin Liska <mliska@suse.cz>
3685 * c-attribs.c (handle_target_attribute):
3686 Report warning for an empty string argument of target attribute.
3688 2017-08-09 Jakub Jelinek <jakub@redhat.com>
3691 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
3693 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
3694 or DECL_NONLOCAL labels.
3695 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
3696 or DECL_NONLOCAL labels here.
3698 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
3700 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
3701 to indicate when early gimple folding has been disabled.
3702 (rs6000_gimple_fold_builtin): Add debug content.
3703 (rs6000_invalid_builtin): Fix whitespace.
3704 (rs6000_expand_builtin): Fix whitespace.
3705 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
3707 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
3710 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
3711 SAVE_MULTIPLE if not all the registers that saves, should be saved.
3713 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
3715 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
3716 (qdf24xx): Likewise.
3717 * config/aarch64/aarch64.md: Include falkor.md.
3718 * config/aarch64/falkor.md: New.
3720 2017-08-09 Marek Polacek <polacek@redhat.com>
3723 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
3724 * diagnostic.c (emit_diagnostic): Add a comment.
3725 (emit_diagnostic_valist): New function.
3727 2017-08-09 Marek Polacek <polacek@redhat.com>
3730 * input.c (make_location): New overload.
3731 * input.h (make_location): Declare.
3733 2017-08-08 Alan Modra <amodra@gmail.com>
3734 H.J. Lu <hongjiu.lu@intel.com>
3737 * gcc.c (NO_PIE_SPEC): Delete.
3738 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
3740 (LINK_PIE_SPEC): ..to here.
3741 (LINK_COMMAND_SPEC): Support -no-pie.
3742 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
3743 chain of crtbegin*.o selection, update for PIE_SPEC changes and
3745 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
3746 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
3747 (ENDFILE_CRTEND_SPEC): Similarly.
3749 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
3752 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
3753 (mstack-protector-guard-offset=): Ditto.
3754 * config/i386/i386.c (ix86_option_override): Handle
3755 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
3757 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
3758 ix86_stack_protect_guard_offset variables.
3759 (TARGET_STACK_PROTECT_GUARD): Always define.
3760 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
3761 and -mstack-protector-guard-offset= options.
3763 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3765 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
3766 boundary case for the last candidate.
3768 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3770 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
3771 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
3773 2017-08-08 Tamar Christina <tamar.christina@arm.com>
3776 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
3777 * config/aarch64/aarch64-builtins.c
3778 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
3779 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
3780 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
3782 2017-08-08 Tamar Christina <tamar.christina@arm.com>
3783 Andrew Pinski <pinskia@gmail.com>
3786 * internal-fn.def (XORSIGN): New.
3787 * optabs.def (xorsign_optab): New.
3788 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
3789 (convert_expand_mult_copysign): New.
3790 (pass_optimize_widening_mul::execute): Call
3791 convert_expand_mult_copysign.
3793 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3795 PR tree-optimization/81354
3796 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
3797 Insert on edges rather than explicitly creating landing pads.
3798 (analyze_candidates_and_replace): Commit edge inserts.
3800 2017-08-08 Richard Biener <rguenther@suse.de>
3803 * tree-ssa-loop-niter.c: Include tree-dfa.h.
3804 (expand_simple_operations): Also look through ADDR_EXPRs with
3805 MEM_REF bases treating them as POINTER_PLUS_EXPR.
3807 2017-08-08 Richard Biener <rguenther@suse.de>
3809 PR tree-optimization/81723
3810 * tree-vect-slp.c (struct bst_traits): New hash traits.
3811 (bst_fail): New global.
3812 (vect_build_slp_tree_2): New worker, split out from ...
3813 (vect_build_slp_tree): ... this now wrapping it with using
3814 bst_fail set to cache SLP tree build fails. Properly handle
3816 (vect_analyze_slp_instance): Allocate and free bst_fail.
3818 2017-08-08 Martin Liska <mliska@suse.cz>
3821 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
3822 LABEL_DECLs that can be from a different function.
3824 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3826 PR tree-optimization/81744
3827 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
3828 loop's number of iterations.
3830 2017-08-08 Martin Liska <mliska@suse.cz>
3832 * asan.c: Include header files.
3833 * attribs.c (build_decl_attribute_variant): New function moved
3835 (build_type_attribute_qual_variant): Likewise.
3836 (cmp_attrib_identifiers): Likewise.
3837 (simple_cst_list_equal): Likewise.
3838 (omp_declare_simd_clauses_equal): Likewise.
3839 (attribute_value_equal): Likewise.
3840 (comp_type_attributes): Likewise.
3841 (build_type_attribute_variant): Likewise.
3842 (lookup_ident_attribute): Likewise.
3843 (remove_attribute): Likewise.
3844 (merge_attributes): Likewise.
3845 (merge_type_attributes): Likewise.
3846 (merge_decl_attributes): Likewise.
3847 (merge_dllimport_decl_attributes): Likewise.
3848 (handle_dll_attribute): Likewise.
3849 (attribute_list_equal): Likewise.
3850 (attribute_list_contained): Likewise.
3851 * attribs.h (lookup_attribute): New function moved from tree.[ch].
3852 (lookup_attribute_by_prefix): Likewise.
3853 * bb-reorder.c: Include header files.
3854 * builtins.c: Likewise.
3855 * calls.c: Likewise.
3856 * cfgexpand.c: Likewise.
3857 * cgraph.c: Likewise.
3858 * cgraphunit.c: Likewise.
3859 * convert.c: Likewise.
3860 * dwarf2out.c: Likewise.
3861 * final.c: Likewise.
3862 * fold-const.c: Likewise.
3863 * function.c: Likewise.
3864 * gimple-expr.c: Likewise.
3865 * gimple-fold.c: Likewise.
3866 * gimple-pretty-print.c: Likewise.
3867 * gimple.c: Likewise.
3868 * gimplify.c: Likewise.
3869 * hsa-common.c: Likewise.
3870 * hsa-gen.c: Likewise.
3871 * internal-fn.c: Likewise.
3872 * ipa-chkp.c: Likewise.
3873 * ipa-cp.c: Likewise.
3874 * ipa-devirt.c: Likewise.
3875 * ipa-fnsummary.c: Likewise.
3876 * ipa-inline.c: Likewise.
3877 * ipa-visibility.c: Likewise.
3879 * lto-cgraph.c: Likewise.
3880 * omp-expand.c: Likewise.
3881 * omp-general.c: Likewise.
3882 * omp-low.c: Likewise.
3883 * omp-offload.c: Likewise.
3884 * omp-simd-clone.c: Likewise.
3885 * opts-global.c: Likewise.
3886 * passes.c: Likewise.
3887 * predict.c: Likewise.
3888 * sancov.c: Likewise.
3889 * sanopt.c: Likewise.
3890 * symtab.c: Likewise.
3891 * toplev.c: Likewise.
3892 * trans-mem.c: Likewise.
3893 * tree-chkp.c: Likewise.
3894 * tree-eh.c: Likewise.
3895 * tree-into-ssa.c: Likewise.
3896 * tree-object-size.c: Likewise.
3897 * tree-parloops.c: Likewise.
3898 * tree-profile.c: Likewise.
3899 * tree-ssa-ccp.c: Likewise.
3900 * tree-ssa-live.c: Likewise.
3901 * tree-ssa-loop.c: Likewise.
3902 * tree-ssa-sccvn.c: Likewise.
3903 * tree-ssa-structalias.c: Likewise.
3904 * tree-ssa.c: Likewise.
3905 * tree-streamer-in.c: Likewise.
3906 * tree-vectorizer.c: Likewise.
3907 * tree-vrp.c: Likewise.
3909 * ubsan.c: Likewise.
3910 * varasm.c: Likewise.
3911 * varpool.c: Likewise.
3912 * tree.c: Remove functions moved to attribs.[ch].
3914 * config/aarch64/aarch64.c: Add attrs.h header file.
3915 * config/alpha/alpha.c: Likewise.
3916 * config/arc/arc.c: Likewise.
3917 * config/arm/arm.c: Likewise.
3918 * config/avr/avr.c: Likewise.
3919 * config/bfin/bfin.c: Likewise.
3920 * config/c6x/c6x.c: Likewise.
3921 * config/cr16/cr16.c: Likewise.
3922 * config/cris/cris.c: Likewise.
3923 * config/darwin.c: Likewise.
3924 * config/epiphany/epiphany.c: Likewise.
3925 * config/fr30/fr30.c: Likewise.
3926 * config/frv/frv.c: Likewise.
3927 * config/ft32/ft32.c: Likewise.
3928 * config/h8300/h8300.c: Likewise.
3929 * config/i386/winnt.c: Likewise.
3930 * config/ia64/ia64.c: Likewise.
3931 * config/iq2000/iq2000.c: Likewise.
3932 * config/lm32/lm32.c: Likewise.
3933 * config/m32c/m32c.c: Likewise.
3934 * config/m32r/m32r.c: Likewise.
3935 * config/m68k/m68k.c: Likewise.
3936 * config/mcore/mcore.c: Likewise.
3937 * config/microblaze/microblaze.c: Likewise.
3938 * config/mips/mips.c: Likewise.
3939 * config/mmix/mmix.c: Likewise.
3940 * config/mn10300/mn10300.c: Likewise.
3941 * config/moxie/moxie.c: Likewise.
3942 * config/msp430/msp430.c: Likewise.
3943 * config/nds32/nds32-isr.c: Likewise.
3944 * config/nds32/nds32.c: Likewise.
3945 * config/nios2/nios2.c: Likewise.
3946 * config/nvptx/nvptx.c: Likewise.
3947 * config/pa/pa.c: Likewise.
3948 * config/pdp11/pdp11.c: Likewise.
3949 * config/powerpcspe/powerpcspe.c: Likewise.
3950 * config/riscv/riscv.c: Likewise.
3951 * config/rl78/rl78.c: Likewise.
3952 * config/rx/rx.c: Likewise.
3953 * config/s390/s390.c: Likewise.
3954 * config/sh/sh.c: Likewise.
3955 * config/sol2.c: Likewise.
3956 * config/sparc/sparc.c: Likewise.
3957 * config/spu/spu.c: Likewise.
3958 * config/stormy16/stormy16.c: Likewise.
3959 * config/tilegx/tilegx.c: Likewise.
3960 * config/tilepro/tilepro.c: Likewise.
3961 * config/v850/v850.c: Likewise.
3962 * config/vax/vax.c: Likewise.
3963 * config/visium/visium.c: Likewise.
3964 * config/xtensa/xtensa.c: Likewise.
3966 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3969 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
3970 constraints since the -mupper-regs-* switches have been
3972 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
3973 into a vector from a double word element that was extracted from
3974 another vector, and eliminate extra XXPERMDI instructions.
3975 (vsx_concat_<mode>_2): Likewise.
3976 (vsx_concat_<mode>_3): Likewise.
3977 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
3978 concat to allow optimizing inserts from previous extracts.
3980 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
3982 * config/i386/i386.c (ix86_stack_protect_guard): Generate
3983 memory reference to a SSP offset in TLS address space.
3984 (ix86_print_operand) <case '@'>: Remove.
3985 (ix86_print_operand_punct_valid_p): Remove '@' code.
3986 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
3988 (stack_tls_protect_set_<mode>): Remove.
3989 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
3990 (stack_tls_protect_test_<mode>): Remove.
3991 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
3993 2017-08-07 Olivier Hainque <hainque@adacore.com>
3996 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
3998 2017-08-07 Douglas Rupp <rupp@adacore.com>
4000 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
4001 variable was referenced as multidir in command.
4003 2017-08-07 Jakub Jelinek <jakub@redhat.com>
4006 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
4009 2017-08-07 Martin Liska <mliska@suse.cz>
4011 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
4012 * config/rl78/rl78.c: Add include of attribs.h.
4013 * config/sh/sh.c: Likewise.
4014 * config/v850/v850.c: Likewise.
4016 2017-08-07 Tom de Vries <tom@codesourcery.com>
4019 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
4021 2017-08-07 Martin Liska <mliska@suse.cz>
4023 * config/mips/mips.c: Include attribs.h.
4025 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
4028 * doc/invoke.texi: Document change in behvaior for -Ofast for
4031 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
4033 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
4035 (aarch64_pop_regs): Likewise.
4036 (aarch64_gen_load_pair): Likewise.
4037 (aarch64_save_callee_saves): Likewise.
4038 (aarch64_restore_callee_saves): Likewise.
4040 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4042 * config/i386/i386.c: Revert the last change.
4044 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4047 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
4049 (ix86_finalize_stack_frame_flags): This. Also clear
4050 frame_pointer_needed if -fno-omit-frame-pointer is used without
4052 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
4053 with ix86_finalize_stack_frame_flags.
4054 (ix86_expand_epilogue): Likewise.
4055 (ix86_expand_split_stack_prologue): Likewise.
4057 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4060 * config/i386/i386.c (get_builtin_code_for_version): Set priority
4061 to P_AES for Westmere.
4063 2017-08-07 Jonathan Yong <10walls@gmail.com>
4065 * config/i386/mingw.opt (fset-stack-executable): Removed.
4066 * config/i386/cygming.opt (fset-stack-executable): Moved
4068 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
4070 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
4072 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
4074 2017-08-07 Marek Polacek <polacek@redhat.com>
4077 * fold-const.c (fold_indirect_ref_1): Check type_domain.
4079 2017-08-07 Martin Liska <mliska@suse.cz>
4081 * attribs.h (canonicalize_attr_name): New function.
4082 (cmp_attribs): Move from c-format.c and adjusted.
4083 (is_attribute_p): Moved from tree.h.
4084 * tree-inline.c: Add new includes.
4085 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
4086 (private_is_attribute_p): Remove.
4087 (private_lookup_attribute): Likewise.
4088 (private_lookup_attribute_by_prefix): Simplify.
4089 (remove_attribute): Use is_attribute_p.
4090 * tree.h: Remove removed declarations.
4092 2017-08-07 Jakub Jelinek <jakub@redhat.com>
4095 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
4096 instead of computing it in the function. Formatting fix.
4097 (expand_case): Don't rely on default_edge being the first edge,
4098 clear it if removing it, pass default_edge to
4099 emit_case_dispatch_table.
4100 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
4103 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
4105 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
4106 insn in the function, emit NOP after the insn.
4108 2017-08-06 Tom de Vries <tom@codesourcery.com>
4110 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
4113 2017-08-06 Tom de Vries <tom@codesourcery.com>
4115 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
4118 2017-08-04 Yury Gribov <tetra2005@gmail.com>
4120 PR tree-optimization/57371
4121 * match.pd: New pattern.
4123 2017-08-04 Marek Polacek <polacek@redhat.com>
4126 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
4127 perform the computation in offset_int.
4129 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
4131 PR tree-optimization/81136
4132 * tree-vectorizer.h: Include tree-hash-traits.h.
4133 (vec_base_alignments): New typedef.
4134 (vec_info): Add a base_alignments field.
4135 (vect_record_base_alignments): Declare.
4136 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
4138 (DR_IS_CONDITIONAL_IN_STMT): New macro.
4139 (create_data_ref): Add an is_conditional_in_stmt argument.
4140 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
4141 the is_conditional_in_stmt field.
4142 (data_ref_loc): Add an is_conditional_in_stmt field.
4143 (get_references_in_stmt): Set the is_conditional_in_stmt field.
4144 (find_data_references_in_stmt): Update call to create_data_ref.
4145 (graphite_find_data_references_in_stmt): Likewise.
4146 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
4147 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
4148 (vect_record_base_alignment): New function.
4149 (vect_record_base_alignments): Likewise.
4150 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
4151 for nested statements even if we fail to compute a misalignment.
4152 Use pooled base alignments for unconditional references.
4153 (vect_find_same_alignment_drs): Compare base addresses instead
4155 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
4156 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
4158 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
4160 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
4161 Add an explicit name for the enum. Use auto_vec for slp_instances
4163 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
4165 (_bb_vec_info): Add a constructor and destructor.
4166 (vinfo_for_stmt): Return NULL for uids of -1 as well.
4167 (destroy_loop_vec_info): Delete.
4168 (vect_destroy_datarefs): Likewise.
4169 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
4170 (vec_info::vec_info): New function.
4171 (vec_info::~vec_info): Likewise.
4172 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
4173 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
4174 destroy_loop_vec_info.
4175 * tree-vect-loop.c (new_loop_vec_info): Replace with...
4176 (_loop_vec_info::_loop_vec_info): ...this.
4177 (destroy_loop_vec_info): Replace with...
4178 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
4179 the stmt_vec_infos. Leave handling of vec_info information to its
4180 destructor. Remove explicit vector releases.
4181 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
4182 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
4183 * tree-vect-slp.c (new_bb_vec_info): Replace with...
4184 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
4185 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
4186 (destroy_bb_vec_info): Replace with...
4187 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
4188 information to its destructor.
4189 (vect_slp_analyze_bb_1): Use new and delete instead of
4190 new_bb_vec_info and destroy_bb_vec_info.
4191 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
4194 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
4196 * tree-data-ref.h (subscript): Add access_fn field.
4197 (data_dependence_relation): Add could_be_independent_p.
4198 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
4199 (same_access_functions): Move to tree-data-ref.c.
4200 * tree-data-ref.c (ref_contains_union_access_p): New function.
4201 (access_fn_component_p): Likewise.
4202 (access_fn_components_comparable_p): Likewise.
4203 (dr_analyze_indices): Add a reference to access_fn_component_p.
4204 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
4206 (constant_access_functions): Likewise.
4207 (add_other_self_distances): Likewise.
4208 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
4209 (initialize_data_dependence_relation): Use XCNEW and remove
4210 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
4211 of access functions that have the same type. Allow the
4212 subsequence to end with different bases in some circumstances.
4213 Record the chosen access functions in SUB_ACCESS_FN.
4214 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
4215 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
4216 (subscript_dependence_tester_1): Likewise dra and drb.
4217 (build_classic_dist_vector): Update calls accordingly.
4218 (subscript_dependence_tester): Likewise.
4219 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
4220 DDR_COULD_BE_INDEPENDENT_P.
4221 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
4222 comp_alias_ddrs instead of may_alias_ddrs.
4223 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
4225 (vect_analyze_data_ref_dependence): Use it if
4226 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
4227 distance vectors if that fails.
4228 (dependence_distance_ge_vf): New function.
4229 (vect_prune_runtime_alias_test_list): Use it. Don't clear
4230 LOOP_VINFO_MAY_ALIAS_DDRS.
4232 2017-08-04 Richard Biener <rguenther@suse.de>
4235 * fold-const.c (fold_binary_loc): Properly restrict
4236 minus_var0 && minus_var1 case when associating undefined overflow
4239 2017-08-04 Tom de Vries <tom@codesourcery.com>
4241 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
4243 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4245 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4246 Don't start diagnostic messages with a capital letter.
4247 * config/rs6000/rs6000.c (rs6000_option_override_internal):
4249 (rs6000_invalid_builtin): Likewise.
4250 (rs6000_trampoline_init): Likewise.
4252 2017-08-03 Jakub Jelinek <jakub@redhat.com>
4255 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
4256 after setting changeable df flags.
4258 2017-08-03 Richard Biener <rguenther@suse.de>
4260 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
4261 up if the use is in USE - X.
4263 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
4265 * toplev.c (dumpfile.h): New include.
4266 (internal_error_reentered): New static function. Use it...
4267 (internal_error_function): ...here to handle reentered internal_error.
4269 2017-08-03 Richard Biener <rguenther@suse.de>
4272 * fold-const.c (split_tree): Add minus_var and minus_con
4273 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
4274 here but always use minus_*.
4275 (associate_trees): Assert we never associate with MINUS_EXPR
4276 and NULL first operand. Do not recurse for PLUS_EXPR operands
4277 when associating as MINUS_EXPR either.
4278 (fold_binary_loc): Track minus_var and minus_con.
4280 2017-08-03 Tom de Vries <tom@codesourcery.com>
4283 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
4284 ACCEL_COMPILER, apply finish_options on
4285 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
4287 2017-08-03 Tom de Vries <tom@codesourcery.com>
4290 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
4291 function_entry_patch_area_size > 0.
4293 2017-08-03 Jakub Jelinek <jakub@redhat.com>
4296 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
4297 instead of 10??LU, perform unit multiplication in wide_int,
4298 don't change alloc_object_size_limit if the limit is larger
4301 PR tree-optimization/81655
4302 PR tree-optimization/81588
4303 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
4304 the case when ranges[i].low and high are 1 for unsigned type with
4308 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
4309 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
4311 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4313 * tree-vrp.h: Add include guard.
4315 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
4318 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
4319 (ud2): New insn pattern.
4320 * config/i386/i386.c (ix86_expand_epilogue):
4321 For naked functions, generate ud2 instead of trap insn.
4323 2017-08-02 Marek Polacek <polacek@redhat.com>
4326 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
4328 2017-08-02 Tom de Vries <tom@codesourcery.com>
4329 Cesar Philippidis <cesar@codesourcery.com>
4331 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
4332 Add missing edge probabilities.
4334 2017-08-02 Tamar Christina <tamar.christina@arm.com>
4336 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
4339 2017-08-02 Jakub Jelinek <jakub@redhat.com>
4342 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
4343 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
4344 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
4345 prologue_seq sequences - if any.
4347 2017-08-02 Richard Biener <rguenther@suse.de>
4349 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
4350 via vectors if supported, integer extracts via punning if supported
4351 or otherwise vector extracts.
4353 2017-08-02 Richard Biener <rguenther@suse.de>
4355 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
4357 (bitmap_insert_into_set): ... this.
4359 2017-08-02 Richard Biener <rguenther@suse.de>
4361 PR tree-optimization/81633
4363 2015-08-17 Alan Hayward <alan.hayward@arm.com>
4365 PR tree-optimization/71752
4366 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
4368 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
4370 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
4371 (machine_function::call_ms2sysv_pad_out): Remove field.
4372 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
4373 (ix86_compute_frame_layout): Likewise.
4375 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
4378 * config/i386/i386.c (ix86_set_func_type): Disallow naked
4379 attribute with interrupt attribute.
4381 2017-08-01 Andrew Pinski <apinski@cavium.com>
4383 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
4384 BIT_INSERT_EXPR's operand 1
4385 to see if the types precision matches.
4387 2017-08-01 Martin Liska <mliska@suse.cz>
4390 * builtins.c (expand_builtin_memcpy_args): Remove.
4391 (expand_builtin_memcpy): Call newly added function
4392 expand_builtin_memory_copy_args.
4393 (expand_builtin_memcpy_with_bounds): Likewise.
4394 (expand_builtin_mempcpy): Remove last argument.
4395 (expand_builtin_mempcpy_with_bounds): Likewise.
4396 (expand_builtin_memory_copy_args): New function created from
4397 expand_builtin_mempcpy_args with small modifications.
4398 (expand_builtin_mempcpy_args): Remove.
4399 (expand_builtin_stpcpy): Remove unused argument.
4400 (expand_builtin): Likewise.
4401 (expand_builtin_with_bounds): Likewise.
4403 2017-08-01 Martin Liska <mliska@suse.cz>
4406 Make mempcpy more optimal (PR middle-end/70140).
4408 2017-08-01 Jakub Jelinek <jakub@redhat.com>
4411 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
4412 __builtin_vec_cmpne verify both arguments are compatible vectors
4413 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
4414 verify arg1_type is a pointer or array type. For __builtin_vec_st,
4415 move computation of aligned to after checking the argument types.
4419 * config/rs6000/vsx.md (vextract_fp_from_shorth,
4420 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
4423 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
4424 Doug Rupp <rupp@adacore.com>
4425 Olivier Hainque <hainque@adacore.com>
4427 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
4428 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
4430 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
4431 for TARGET_OS_CPP_BUILTIN.
4432 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
4433 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
4435 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
4436 passing required abi options to the assembler for EABI configurations.
4437 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
4438 of .text.hot and .text.unlikely sections for kernel modules when
4439 using ARM style exceptions.
4440 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
4441 options. Add EXTRA_CC1_SPEC.
4442 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
4444 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
4446 (ARM_TARGET2_DWARF_FORMAT): Define.
4447 * config/arm/t-vxworks: Adjust multilib control to removal of the
4448 Diab command line options.
4450 2017-08-01 Martin Liska <mliska@suse.cz>
4452 PR gcov-profile/81561
4453 * gcov.c (unblock): Make unblocking safe as we need to preserve
4454 index correspondence of blocks and block_lists.
4456 2017-08-01 Richard Biener <rguenther@suse.de>
4458 PR tree-optimization/81181
4459 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4460 (compute_antic): ... end of iteration here.
4462 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
4464 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
4465 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
4466 (ftree-slp-vectorize): Likewise.
4467 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
4468 can no longer be set independent of flag_tree_loop_vectorize.
4469 * omp-general.c (emp_max_vf): Likewise.
4470 * opts.c (enable_fdo_optimizations): Remove references to
4471 flag_tree_vectorize, these are now implicit.
4472 (common_handle_option): Remove handling for OPT_ftree_vectorize,
4473 and leave it for the options machinery.
4475 2017-08-01 Martin Liska <mliska@suse.cz>
4478 * builtins.c (expand_builtin_memcpy_args): Remove.
4479 (expand_builtin_memcpy): Call newly added function
4480 expand_builtin_memory_copy_args.
4481 (expand_builtin_memcpy_with_bounds): Likewise.
4482 (expand_builtin_mempcpy): Remove last argument.
4483 (expand_builtin_mempcpy_with_bounds): Likewise.
4484 (expand_builtin_memory_copy_args): New function created from
4485 expand_builtin_mempcpy_args with small modifications.
4486 (expand_builtin_mempcpy_args): Remove.
4487 (expand_builtin_stpcpy): Remove unused argument.
4488 (expand_builtin): Likewise.
4489 (expand_builtin_with_bounds): Likewise.
4491 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
4494 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
4495 print "ds:" only for immediates in generic address space.
4497 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
4500 * config/i386/i386.c (ix86_funciton_naked): New prototype.
4501 (ix86_function_ok_for_sibcall): Return false for naked functions.
4503 2017-08-01 Richard Biener <rguenther@suse.de>
4505 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
4506 (compute_antic): Seed worklist with exit block predecessors.
4507 * cfganal.c (dfs_find_deadend): For a cycle return the source
4508 of the edge closing it.
4510 2017-08-01 Tamar Christina <tamar.christina@arm.com>
4512 * config/aarch64/aarch64.c
4513 (aarch64_can_const_movi_rtx_p): Move 0 check.
4515 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4517 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
4518 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
4520 * match.pd: Ditto in address comparison pattern.
4522 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4524 PR tree-optimization/81627
4525 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
4526 closed ssa form for store-store chain.
4528 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4530 PR tree-optimization/81620
4531 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
4532 for store-store chain.
4534 2017-08-01 Jakub Jelinek <jakub@redhat.com>
4536 PR tree-optimization/81588
4537 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
4538 ranges[i].in_p, invert comparison code ccode. For >/>=,
4539 swap rhs1 and rhs2 and comparison code unconditionally,
4540 for </<= don't do that. Don't swap rhs1/rhs2 again if
4541 ranges[i].in_p, instead invert comparison code ccode if
4542 opcode or oe->rank is BIT_IOR_EXPR.
4545 * optabs.def (vec_extract_optab, vec_init_optab): Change from
4546 a direct optab to conversion optab.
4547 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
4548 with GET_MODE_INNER as last argument instead of optab_handler.
4549 * expmed.c (extract_bit_field_1): Likewise. Use vector from
4550 vector extraction if possible and optab is available.
4551 * expr.c (store_constructor): Use convert_optab_handler instead
4552 of optab_handler. Use vector initialization from smaller
4553 vectors if possible and optab is available.
4554 * tree-vect-stmts.c (vectorizable_load): Likewise.
4555 * doc/md.texi (vec_extract, vec_init): Document that the optabs
4557 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
4558 of vec_init from half-sized vectors with the same element mode.
4559 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
4560 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
4561 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
4562 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
4563 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
4564 after mode in gen_vec_extract* calls.
4565 (vec_extract<mode>): Renamed to ...
4566 (vec_extract<mode><ssescalarmodelower>): ... this.
4567 (vec_extract<mode><ssehalfvecmodelower>): New expander.
4568 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
4569 element mode after mode in gen_vec_init* calls.
4570 (VEC_INIT_HALF_MODE): New mode iterator.
4571 (vec_init<mode>): Renamed to ...
4572 (vec_init<mode><ssescalarmodelower>): ... this.
4573 (vec_init<mode><ssehalfvecmodelower>): New expander.
4574 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
4575 (vec_extractv2sfsf): ... this.
4576 (vec_initv2sf): Renamed to ...
4577 (vec_initv2sfsf): ... this.
4578 (vec_extractv2si): Renamed to ...
4579 (vec_extractv2sisi): ... this.
4580 (vec_initv2si): Renamed to ...
4581 (vec_initv2sisi): ... this.
4582 (vec_extractv4hi): Renamed to ...
4583 (vec_extractv4hihi): ... this.
4584 (vec_initv4hi): Renamed to ...
4585 (vec_initv4hihi): ... this.
4586 (vec_extractv8qi): Renamed to ...
4587 (vec_extractv8qiqi): ... this.
4588 (vec_initv8qi): Renamed to ...
4589 (vec_initv8qiqi): ... this.
4590 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
4591 (vec_init<mode>): Renamed to ...
4592 (vec_init<mode><VEC_base_l>): ... this.
4593 (vec_extract<mode>): Renamed to ...
4594 (vec_extract<mode><VEC_base_l>): ... this.
4595 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
4596 (vec_initv2sfsf): ... this.
4597 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
4598 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
4599 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
4600 element mode after mode in gen_vec_init* calls.
4601 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
4602 (vec_init<mode><Vel>): ... this.
4603 (vec_extract<mode>): Renamed to ...
4604 (vec_extract<mode><Vel>): ... this.
4605 * config/aarch64/iterators.md (Vel): New mode attribute.
4606 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
4607 Add element mode after mode in gen_vec_extract* calls.
4608 * config/s390/vector.md (non_vec_l): New mode attribute.
4609 (vec_extract<mode>): Renamed to ...
4610 (vec_extract<mode><non_vec_l>): ... this.
4611 (vec_init<mode>): Renamed to ...
4612 (vec_init<mode><non_vec_l>): ... this.
4613 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
4614 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
4616 * config/arm/iterators.md (V_elem_l): New mode attribute.
4617 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
4618 (vec_extract<mode><V_elem_l>): ... this.
4619 (vec_extractv2di): Renamed to ...
4620 (vec_extractv2didi): ... this.
4621 (vec_init<mode>): Renamed to ...
4622 (vec_init<mode><V_elem_l>): ... this.
4623 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
4624 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
4625 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
4626 Add element mode after gen_vec_extract* calls.
4627 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
4628 (vec_init<mode><unitmode>): ... this.
4629 (vec_extract<mode>): Renamed to ...
4630 (vec_extract<mode><unitmode>): ... this.
4631 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
4632 (vec_init<mode><unitmode>): ... this.
4633 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
4634 (vec_initv2sfsf): ... this.
4635 (vec_extractv2sf): Renamed to ...
4636 (vec_extractv2sfsf): ... this.
4637 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
4638 Add element mode after gen_vec_extract* calls.
4639 * config/mips/mips.md (unitmode): New mode iterator.
4640 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
4641 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
4642 * config/spu/spu.md (inner_l): New mode attribute.
4643 (vec_init<mode>): Renamed to ...
4644 (vec_init<mode><inner_l>): ... this.
4645 (vec_extract<mode>): Renamed to ...
4646 (vec_extract<mode><inner_l>): ... this.
4647 * config/sparc/sparc.md (veltmode): New mode iterator.
4648 (vec_init<VMALL:mode>): Renamed to ...
4649 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
4650 * config/ia64/vect.md (vec_initv2si): Renamed to ...
4651 (vec_initv2sisi): ... this.
4652 (vec_initv2sf): Renamed to ...
4653 (vec_initv2sfsf): ... this.
4654 (vec_extractv2sf): Renamed to ...
4655 (vec_extractv2sfsf): ... this.
4656 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
4657 (vec_init<mode>): Renamed to ...
4658 (vec_init<mode><VEC_base_l>): ... this.
4659 (vec_extract<mode>): Renamed to ...
4660 (vec_extract<mode><VEC_base_l>): ... this.
4661 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
4662 (vec_initv2sfsf): ... this.
4663 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
4664 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
4665 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
4666 gen_vec_init* calls.
4668 2017-08-01 Richard Biener <rguenther@suse.de>
4670 PR tree-optimization/81297
4671 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
4672 TREE_OVERFLOW from INTEGER_CSTs.
4674 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
4676 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
4678 2017-07-31 Carl Love <cel@us.ibm.com>
4680 * config/rs6000/rs6000-c: Add support for built-in functions
4681 vector signed char vec_xl_be (signed long long, signed char *);
4682 vector unsigned char vec_xl_be (signed long long, unsigned char *);
4683 vector signed int vec_xl_be (signed long long, signed int *);
4684 vector unsigned int vec_xl_be (signed long long, unsigned int *);
4685 vector signed long long vec_xl_be (signed long long, signed long long *);
4686 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
4687 vector signed short vec_xl_be (signed long long, signed short *);
4688 vector unsigned short vec_xl_be (signed long long, unsigned short *);
4689 vector double vec_xl_be (signed long long, double *);
4690 vector float vec_xl_be (signed long long, float *);
4691 * config/rs6000/altivec.h (vec_xl_be): Add #define.
4692 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
4693 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
4695 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
4696 (altivec_expand_builtin): Add switch statement to call
4697 altivec_expand_xl_be for each builtin.
4698 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
4699 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
4700 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
4701 __builtin_vsx_le_be_v16qi.
4702 * doc/extend.texi: Update the built-in documentation file for the
4703 new built-in functions.
4705 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
4708 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
4710 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
4712 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4714 * config.gcc: Add z14.
4715 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
4716 CPU model numbers for z13s and z14.
4717 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
4719 * config/s390/s390-opts.h (enum processor_type): Rename
4720 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
4721 * config/s390/s390.c (processor_table): Add field for CPU name to
4722 be passed to Binutils.
4723 (s390_asm_output_machine_for_arch): Use the new field in
4724 processor_table for Binutils.
4725 (s390_expand_builtin): Replace arch12 with z14.
4726 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
4727 (s390_get_sched_attrmask): Likewise.
4728 (s390_get_unit_mask): Likewise.
4729 * config/s390/s390.opt: Add z14 to processor_type enum.
4731 2017-07-31 Martin Jambor <mjambor@suse.cz>
4734 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
4735 regardless of optimization level.
4737 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
4738 Martin Liska <mliska@suse.cz>
4740 * predict.def: Remove old comment and adjust probability.
4741 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
4744 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
4747 * config/i386/i386.c (ix86_function_naked): New function.
4748 (ix86_can_use_return_insn_p): Return false for naked functions.
4749 (ix86_expand_prologue): Skip prologue for naked functions.
4750 (ix86_expand_epilogue): Skip epilogue for naked functions
4751 and emit trap instruction.
4752 (ix86_warn_func_return): New function.
4753 (ix86_attribute_table): Add "naked" attribute specification.
4754 (TARGET_WARN_FUNC_RETURN): Define.
4755 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
4757 2017-07-31 Martin Liska <mliska@suse.cz>
4759 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
4760 (dump_gimple_bb_header): Always dump BB info.
4761 (pp_cfg_jump): Do not append info about BB when dumping a jump.
4763 2017-07-31 Martin Liska <mliska@suse.cz>
4766 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
4767 also with current_function_decl non-null equality.
4769 2017-07-31 Jakub Jelinek <jakub@redhat.com>
4772 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
4773 change type to the element type, instead add eltype variable and
4774 use it where we are interested in the element type.
4776 PR tree-optimization/81603
4777 * ipa-polymorphic-call.c
4778 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
4779 offset arithmetic in offset_int, bail out if the resulting bit offset
4780 doesn't fit into shwi.
4782 2017-07-31 Martin Liska <mliska@suse.cz>
4784 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
4785 (gimplify_save_expr): Fix comment.
4787 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
4790 * config/i386/i386.c (ix86_function_arg): Update arguments for
4792 (ix86_compute_frame_layout): Set the initial stack offset to
4793 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
4794 INCOMING_FRAME_SP_OFFSET.
4795 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
4796 stack before exception handler returns.
4797 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
4798 the 'ERROR_CODE' for exception handler.
4800 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
4802 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
4803 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
4804 (ASM_OUTPUT_REG_POP): Ditto.
4805 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
4806 instead of asm_fprintf to output pure string.
4808 2017-07-29 Jakub Jelinek <jakub@redhat.com>
4810 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
4811 to imported_module_or_decl hook.
4812 (debug_nothing_tree_tree_tree_bool): Remove.
4813 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
4814 * debug.c (do_nothing_debug_hooks): Use
4815 debug_nothing_tree_tree_tree_bool_bool instead of
4816 debug_nothing_tree_tree_tree_bool.
4817 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4818 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
4819 * sdbout.c (sdb_debug_hooks): Likewise.
4820 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
4821 (gen_namespace_die): Add DW_AT_export_symbols attribute if
4823 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
4824 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
4825 attribute, don't add anything.
4827 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4829 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
4830 (fold_build2_stat_loc): Likewise.
4831 (fold_build3_stat_loc): Likewise.
4832 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
4833 (fold_build1_loc): Remove macro.
4834 (fold_build2_loc): Likewise.
4835 (fold_build3_loc): Likewise.
4837 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4839 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
4840 (gimple_build_debug_bind_source_stat): Likewise.
4841 * gimple.h (gimple_build_debug_bind): Remove macro.
4842 (gimple_build_debug_bind_source): Likewise.
4844 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4846 * bitmap.c (bitmap_alloc): Adjust.
4847 (bitmap_gc_alloc): Likewise.
4848 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
4850 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4852 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
4853 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
4854 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
4855 (bitmap_gc_alloc_stat): Likewise.
4856 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
4858 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4860 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
4861 * rtl.h (shallow_copy_rtx): Remove macro.
4863 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4865 * emit-rtl.c (gen_raw_REG): Adjust.
4866 * gengenrtl.c (gendef): Likewise.
4867 * rtl.c (rtx_alloc_stat): Remove _stat from name.
4868 * rtl.h (rtx_alloc): Remove macro.
4870 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4872 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
4873 (build_tree_list_stat): Likewise.
4874 * tree.h (build_tree_list): Remove macro.
4875 (build_tree_list_vec): Likewise.
4877 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4879 * tree.c (make_vector_stat): Remove _stat from name.
4880 (build_vector_stat): Likewise.
4881 * tree.h (make_vector_stat): Remove macro.
4882 (build_vector_stat): Likewise.
4884 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4886 * tree.h (build_var_debug_value): Remove prototype.
4888 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4890 * tree.c (tree_cons_stat): Remove _stat from name.
4891 * tree.h (tree_cons): Remove macro.
4893 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4895 * tree.c (build_vl_exp_stat): Remove _stat from name.
4896 * tree.h (build_vl_exp): Remove macro.
4898 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4900 * tree.c (build_decl_stat): Remove _stat from name.
4901 * tree.h (build_decl): Remove macro.
4903 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4905 * gimple.c (gimple_build_with_ops_stat): Adjust.
4906 (gimple_alloc_stat): Remove _stat from name.
4907 * gimple.h (gimple_alloc): Remove macro.
4909 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4911 * tree.c (make_tree_vec_stat): Remove _stat from name.
4912 (grow_tree_vec_stat): Likewise.
4913 * tree.h (make_tree_vec_stat): Adjust prototype.
4914 (grow_tree_vec_stat): Likewise.
4915 (make_tree_vec): Remove macro.
4916 (grow_tree_vec): Likewise.
4918 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4920 * fold-const.c (fold_build1_stat_loc): Adjust.
4921 (fold_build2_stat_loc): Likewise.
4922 (fold_build3_stat_loc): Likewise.
4923 * tree.c (build0_stat): Remove _stat from name.
4924 (build1_stat): Likewise.
4925 (build2_stat): Likewise.
4926 (build3_stat): Likewise.
4927 (build4_stat): Likewise.
4928 (build5_stat): Likewise.
4929 * tree.h (build1_loc): Remove macro, and rename _stat function
4931 (build2_loc): Likewise.
4932 (build3_loc): Likewise.
4933 (build4_loc): Likewise.
4934 (build5_loc): Likewise.
4936 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4938 * tree.c (make_int_cst_stat): Remove _stat from name.
4939 * tree.h (make_int_cst_stat): Adjust prototype.
4940 (make_int_cst): Remove macro.
4942 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4944 * tree.c (make_tre_binfo_stat): Remove _stat from name.
4945 * tree.h (make_tree_binfo_stat): Adjust prototype.
4946 (make_tree_binfo): Remove.
4948 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4950 * tree.c (copy_node_stat): Rename to copy_node.
4951 (build_distinct_type_copy): Adjust.
4952 * tree.h (copy_node_stat): Adjust prototype.
4953 (copy_node): Remove macro.
4955 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4957 * tree.c (make_node_stat): rename to make_node.
4958 (build_tree_list_stat): Adjust.
4959 (build0_stat): Likewise.
4960 (build2_stat): Likewise.
4961 (build3_stat): Likewise.
4962 (build4_stat): Likewise.
4963 (build5_stat): Likewise.
4964 (build_decl_stat): Likewise.
4965 * tree.h (make_node_stat): Adjust prototype.
4966 (make_node): remove macro.
4968 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
4970 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
4971 (PPC_FEATURE2_SCV): Likewise.
4972 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
4974 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4976 * config/aarch64/aarch64.c
4977 (aarch64_internal_mov_immediate): Add new special pattern.
4978 * config/aarch64/aarch64.md (*movdi_aarch64):
4979 Add reg/32bit const mov case.
4981 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4982 Richard Sandiford <richard.sandiford@linaro.org>
4984 * config/aarch64/aarch64.md (mov<mode>): Generalize.
4985 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4986 Add integer and movi cases.
4987 (movi-split-hf-df-sf split, fp16): New.
4988 (enabled): Added TARGET_FP_F16INST.
4989 * config/aarch64/iterators.md (GPF_HF): New.
4990 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
4992 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4994 * config/aarch64/aarch64.c
4995 (aarch64_simd_container_mode): Add prototype.
4996 (aarch64_expand_mov_immediate): Add HI support.
4997 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
4998 (aarch64_can_const_movi_rtx_p): New.
4999 (aarch64_preferred_reload_class):
5000 Remove restrictions of using FP registers for certain SIMD operations.
5001 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
5002 (aarch64_valid_floating_const): Add integer move validation.
5003 (aarch64_simd_imm_scalar_p): Remove.
5004 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
5005 (aarch64_legitimate_constant_p): Expand list of supported cases.
5006 * config/aarch64/aarch64-protos.h
5007 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
5008 (aarch64_reinterpret_float_as_int): New.
5009 (aarch64_simd_imm_scalar_p): Remove.
5010 * config/aarch64/constraints.md (Uvi): New.
5011 (Dd): Split into Ds and new Dd.
5012 * config/aarch64/aarch64.md (*movsi_aarch64):
5014 (*movdi_aarch64): Add SIMD mov case.
5016 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5018 * tree-predcom.c: (struct chain): Handle store-store chain in which
5019 stores for elimination only store loop invariant values.
5020 (execute_pred_commoning_chain): Ditto.
5021 (prepare_initializers_chain_store_elim): Ditto.
5022 (prepare_finalizers): Ditto.
5023 (is_inv_store_elimination_chain): New function.
5024 (initialize_root_vars_store_elim_1): New function.
5026 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5028 * tree-predcom.c: Revise general description of the pass.
5029 (enum chain_type): New enum type for store elimination.
5030 (struct chain): New field supporting store elimination.
5031 (struct component): Ditto.
5032 (dump_chain): Dump store-stores chain.
5033 (release_chain): Release resources.
5034 (split_data_refs_to_components): Compute and create component
5035 contains only stores for elimination.
5036 (get_chain_last_ref_at): New function.
5037 (make_invariant_chain): Initialization.
5038 (make_rooted_chain): Specify chain type in parameter and record it.
5039 (add_looparound_copies): Skip for store-stores chain.
5040 (determine_roots_comp): Compute type of chain and pass it to
5042 (initialize_root_vars_store_elim_2): New function.
5043 (finalize_eliminated_stores): New function.
5044 (remove_stmt): Handle store for elimination.
5045 (execute_pred_commoning_chain): Execute predictive commoning on
5047 (determine_unroll_factor): Skip unroll for store-stores chain.
5048 (prepare_initializers_chain_store_elim): New function.
5049 (prepare_initializers_chain): Hanlde store-store chain.
5050 (prepare_finalizers_chain, prepare_finalizers): New function.
5051 (tree_predictive_commoning_loop): Return integer value indicating
5052 if loop is unrolled or lcssa form is corrupted.
5053 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
5055 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5057 * tree-predcom.c (initialize_root): Delete.
5058 (execute_pred_commoning_chain): Initialize root vars and replace
5059 reference of non-combined chain directly, rather than call above
5062 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5064 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
5065 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
5067 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5069 * tree-predcom.c (struct chain): New field init_seq.
5070 (release_chain): Release init_seq.
5071 (prepare_initializers_chain): Record intialization stmts in above
5073 (insert_init_seqs): New function.
5074 (tree_predictive_commoning_loop): Call insert_init_seqs.
5076 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5078 * tree-predcom.c (determine_roots_comp): Skip trivial components.
5080 2017-07-28 Richard Biener <rguenther@suse.de>
5082 * match.pd: Remove superfluous :c.
5083 * genmatch.c (simplify::id): Add member.
5084 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
5086 (current_id): New global.
5087 (dt_node::parent): Move from ...
5088 (dt_operand::parent): ... here. Add for_id member.
5089 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
5090 (decision_tree::find_node): Relax order requirement when
5091 merging DT_TRUE nodes to ones inbetween the current simplify
5092 and the one we try to merge with. Add diagnostic whenever
5093 we need to enforce pattern order by not merging.
5094 (decision_tree::insert): Set current_id.
5095 (decision_tree::print_node): Dump parent node and for_id.
5096 (parser::last_id): Add member.
5097 (parser::push_simplify): Assign unique id.
5098 (parser::parser): Initialize last_id.
5100 2017-07-28 Martin Liska <mliska@suse.cz>
5103 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
5104 gimple_build_debug_bind.
5106 2017-07-28 Richard Biener <rguenther@suse.de>
5108 PR tree-optimization/81502
5109 * match.pd: Add pattern combining BIT_INSERT_EXPR with
5111 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
5113 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
5114 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
5115 for BIT_FIELD_REF args.
5116 * fold-const.c (make_bit_field_ref): Likewise.
5117 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5119 2017-07-28 Jakub Jelinek <jakub@redhat.com>
5122 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
5123 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
5124 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
5125 Or it into SANITIZER_UNDEFINED.
5126 * ubsan.c: Include gimple-fold.h and varasm.h.
5127 (ubsan_expand_ptr_ifn): New function.
5128 (instrument_pointer_overflow): New function.
5129 (maybe_instrument_pointer_overflow): New function.
5130 (instrument_object_size): Formatting fix.
5131 (pass_ubsan::execute): Call instrument_pointer_overflow
5132 and maybe_instrument_pointer_overflow.
5133 * internal-fn.c (expand_UBSAN_PTR): New function.
5134 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
5135 * sanitizer.def (__ubsan_handle_pointer_overflow,
5136 __ubsan_handle_pointer_overflow_abort): New builtins.
5137 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
5138 * internal-fn.def (UBSAN_PTR): New internal function.
5139 * opts.c (sanitizer_opts): Add pointer-overflow.
5140 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
5141 * fold-const.c (build_range_check): Compute pointer range check in
5142 integral type if pointer arithmetics would be needed. Formatting
5145 2017-07-28 Martin Liska <mliska@suse.cz>
5148 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
5149 parameters that are of a variable-length.
5151 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
5153 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
5155 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
5156 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
5157 (CRT_CALL_STATIC_FUNCTION): Likewise.
5158 (ASM_DEFAULT_SPEC): New define.
5159 (ASM_SPEC32): Likewise.
5160 (ASM_SPEC64): Likewise.
5161 (ASM_SPEC_COMMON): Likewise.
5162 (ASM_SPEC): Likewise.
5163 (INVALID_64BIT): Likewise.
5164 (LINK_OS_DEFAULT_SPEC): Likewise.
5165 (LINK_OS_SPEC32): Likewise.
5166 (LINK_OS_SPEC64): Likewise.
5167 (POWERPC_LINUX): Likewise.
5168 (PTRDIFF_TYPE): Likewise.
5169 (RESTORE_FP_PREFIX): Likewise.
5170 (RESTORE_FP_SUFFIX): Likewise.
5171 (SAVE_FP_PREFIX): Likewise.
5172 (SAVE_FP_SUFFIX): Likewise.
5173 (SIZE_TYPE): Likewise.
5174 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
5175 (TARGET_64BIT): Likewise.
5176 (TARGET_64BIT): Likewise.
5177 (TARGET_AIX): Likewise.
5178 (WCHAR_TYPE_SIZE): Likewise.
5179 (WCHAR_TYPE): Undefine.
5180 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
5181 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
5182 (CPP_OS_RTEMS_SPEC): Delete.
5183 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
5184 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
5186 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
5188 2017-07-28 Jakub Jelinek <jakub@redhat.com>
5190 PR tree-optimization/81578
5191 * tree-parloops.c (build_new_reduction): Bail out if
5192 reduction_code isn't one of the standard OpenMP reductions.
5193 Move the details printing after that decision.
5195 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
5197 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
5198 related to reload_in_progress.
5199 (splat_input_operand): Likewise.
5200 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
5202 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
5204 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
5205 (TARGET_INSTANTIATE_DECLS): Likewise.
5206 (legitimate_indexed_address_p): Delete reload_in_progress code.
5207 (rs6000_debug_legitimate_address_p): Likewise.
5208 (rs6000_eliminate_indexed_memrefs): Likewise.
5209 (rs6000_emit_le_vsx_store): Likewise.
5210 (rs6000_emit_move_si_sf_subreg): Likewise.
5211 (rs6000_emit_move): Likewise.
5212 (register_to_reg_type): Likewise.
5213 (rs6000_pre_atomic_barrier): Likewise.
5214 (rs6000_machopic_legitimize_pic_address): Likewise.
5215 (rs6000_allocate_stack_temp): Likewise.
5216 (rs6000_address_for_fpconvert): Likewise.
5217 (rs6000_address_for_altivec): Likewise.
5218 (rs6000_secondary_memory_needed_rtx): Delete function.
5219 (rs6000_check_sdmode): Likewise.
5220 (rs6000_alloc_sdmode_stack_slot): Likewise.
5221 (rs6000_instantiate_decls): Likewise.
5222 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
5223 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
5224 Delete reload_in_progress.
5225 (*vec_reload_and_plus_<mptrsize>): Likewise.
5226 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
5227 (vsx_div_v2di): Likewise.
5228 (vsx_udiv_v2di): Likewise.
5230 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
5232 * config/rs6000/rs6000.opt (mlra): Replace with stub.
5233 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
5234 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
5235 (rs6000_debug_reg_global): Delete print of LRA status.
5236 (rs6000_option_override_internal): Delete dead LRA related code.
5237 (rs6000_lra_p): Delete function.
5238 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
5240 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
5242 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
5243 * config/riscv/rtems.h: New file.
5245 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5246 Sudakshina Das <sudi.das@arm.com>
5248 * config/aarch64/aarch64.md
5249 (define_split for and<mode>3nr_compare): Move
5250 non aarch64_logical_operand to a register.
5251 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
5252 register immediate operand to a register.
5253 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
5255 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
5258 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
5260 2017-07-27 Richard Biener <rguenther@suse.de>
5262 PR tree-optimization/81573
5263 PR tree-optimization/81494
5264 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
5265 multi defuse cycle case.
5267 2017-07-27 Richard Biener <rguenther@suse.de>
5269 PR tree-optimization/81571
5270 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
5273 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
5275 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
5276 earlier and only if MASK_FPU is set. Adjust formatting.
5278 2017-07-27 Martin Liska <mliska@suse.cz>
5280 * opt-functions.awk: Add validation of value of Init.
5281 * optc-gen.awk: Pass new argument.
5283 2017-07-27 Martin Liska <mliska@suse.cz>
5285 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
5286 Fix wrong condition.
5288 2017-07-27 Martin Liska <mliska@suse.cz>
5290 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
5291 BBs and edges seen by autoFDO.
5293 2017-07-27 Richard Biener <rguenther@suse.de>
5295 PR tree-optimization/81502
5296 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
5297 with incompatible but same sized type.
5298 (execute_update_addresses_taken): Likewise.
5300 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
5302 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
5303 flag_tree_loop_vectorize rather than flag_tree_vectorize.
5305 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5308 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
5309 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
5310 Change s_operand to memory_operand.
5312 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
5314 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
5315 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
5316 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
5317 Emit instructions rather than returning an expression. Handle TFmode
5318 and KFmode by casting to TImode.
5319 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
5320 (rs6000_emit_le_vsx_store): Likewise.
5321 * config/rs6000/vsx.md (VSX_TI): New iterator.
5322 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
5323 (*vsx_le_undo_permute_<mode>): Likewise.
5324 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
5325 emit the split sequence.
5326 (*vsx_le_perm_store_<mode>): Likewise.
5328 2017-07-27 Jakub Jelinek <jakub@redhat.com>
5330 PR tree-optimization/81555
5331 PR tree-optimization/81556
5332 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
5333 if true, force CHANGED for the recursive invocation.
5334 (reassociate_bb): Remember original length of ops array, pass
5335 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
5337 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
5338 attributes for noipa attribute. For naked attribute use
5339 lookup_attribute first before lookup_attribute_spec.
5340 * final.c (rest_of_handle_final): Disable IPA RA for functions with
5342 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
5343 for functions with noipa attribute.
5344 (cgraph_externally_visible_p): Return true for functions with noipa
5346 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
5347 for functions with noipa attribute.
5348 * doc/extend.texi: Document noipa function attribute.
5349 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
5350 also for functions with noipa attribute.
5351 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
5353 2017-07-26 Andrew Pinski <apinski@cavium.com>
5355 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
5356 vec_unalign_load_cost and vec_unalign_store_cost.
5358 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5360 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
5361 -mvsx-small-integer option.
5362 (ISA_3_0_MASKS_IEEE): Likewise.
5363 (OTHER_VSX_VECTOR_MASKS): Likewise.
5364 (POWERPC_MASKS): Likewise.
5365 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
5366 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
5367 code, only testing for DImode being allowed in non-VSX floating
5369 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
5370 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
5372 (rs6000_option_override_internal): Delete -mvsx-small-integer.
5373 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
5374 TARGET_P8_VECTOR test.
5375 (rs6000_secondary_reload_simple_move): Likewise.
5376 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
5377 since TARGET_P9_VECTOR was already tested.
5378 (rs6000_opt_masks): Remove -mvsx-small-integer.
5379 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
5380 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
5382 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
5383 test for TARGET_VEXTRACTUB was used, and that uses
5385 (p9 extract splitter): Likewise.
5386 (vsx_extract_<mode>_di_p9): Likewise.
5387 (vsx_extract_<mode>_store_p9): Likewise.
5388 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
5389 for TARGET_P9_VECTOR was used. Delete code that is now dead with
5390 the elimination of TARGET_VSX_SMALL_INTEGER.
5391 (vsx_extract_<mode>_p8): Likewise.
5392 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
5393 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
5394 (vsx_set_<mode>_p9): Likewise.
5395 (vsx_set_v4sf_p9): Likewise.
5396 (vsx_set_v4sf_p9_zero): Likewise.
5397 (vsx_insert_extract_v4sf_p9): Likewise.
5398 (vsx_insert_extract_v4sf_p9_2): Likewise.
5399 * config/rs6000/rs6000.md (sign extend splitter): Change
5400 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
5401 (floatsi<mode>2_lfiwax_mem): Likewise.
5402 (floatunssi<mode>2_lfiwzx_mem): Likewise.
5403 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
5404 since a test for TARGET_P9_VECTOR was used.
5405 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
5406 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
5407 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
5408 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
5409 TARGET_P8_VECTOR test.
5410 (fix_trunc<mode>si2_stfiwx): Likewise.
5411 (fix_trunc<mode>si2_internal): Likewise.
5412 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
5413 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
5415 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
5416 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
5417 TARGET_P8_VECTOR test.
5418 (fixuns_trunc<mode>si2_stfiwx): Likewise.
5419 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
5420 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
5422 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
5423 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
5424 since a test for TARGET_P9_VECTOR was used.
5425 (splitter for loading small constants): Likewise.
5427 2017-07-26 Andrew Pinski <apinski@cavium.com>
5429 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
5432 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
5435 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
5436 (fp_valid_at): Likewise.
5438 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
5440 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
5441 (qdf24xx_addrcost_table): Likewise.
5442 (cortexa57_tunings): Update to use generic_branch_cost.
5443 (cortexa72_tunings): Likewise.
5444 (cortexa73_tunings): Likewise.
5445 (qdf24xx_tunings): Likewise.
5447 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
5449 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
5450 (thunderx2t99_branch_cost): Likewise.
5451 (cortexa35_tunings): Update to use generic_branch_cost.
5452 (cortexa53_tunings): Likewise.
5453 (cortexa57_tunings): Likewise.
5454 (cortexa72_tunings): Likewise.
5455 (cortexa73_tunings): Likewise.
5456 (thunderx2t99_tunings): Likewise.
5458 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
5460 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
5461 (sparc_option_override): Honour MASK_FSMULD.
5462 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
5463 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
5464 * config/sparc/sparc.opt (mfsmuld): New option.
5465 * doc/invoke.texi (mfsmuld): Document option.
5467 2017-07-26 Marek Polacek <polacek@redhat.com>
5470 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
5472 2017-07-26 Richard Biener <rguenther@suse.de>
5474 * gimple-match-head.c (do_valueize): Return OP if valueize
5476 (get_def): New helper to get at the def stmt of a SSA name
5478 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
5479 do_valueize to get at the def stmt.
5480 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
5482 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
5485 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
5487 2017-07-26 Martin Liska <mliska@suse.cz>
5490 * function.c (expand_function_start): Make expansion of
5491 nonlocal_goto_save_area after parm_birth_insn.
5493 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
5495 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
5496 from all CPU target flags enable members.
5498 2017-07-26 Richard Biener <rguenther@suse.de>
5500 * genmatch.c (dt_simplify::gen): Make iterator vars const.
5501 (decision_tree::gen): Make 'type' const.
5502 (write_predicate): Likewise.
5504 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
5506 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
5507 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
5508 (rs6000_option_override_internal): Likewise.
5509 (rs6000_expand_vector_set): Likewise.
5510 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
5511 (TARGET_UPPER_REGS_SF): Likewise.
5512 (TARGET_UPPER_REGS_DI): Likewise.
5513 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
5514 (TARGET_DIRECT_MOVE_64BIT): Likewise.
5515 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
5516 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
5517 (Splitters for DI constants in Altivec registers): Likewise.
5518 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
5519 (vsx_set_v4sf_p9): Likewise.
5520 (vsx_set_v4sf_p9_zero): Likewise.
5521 (vsx_insert_extract_v4sf_p9): Likewise.
5522 (vsx_insert_extract_v4sf_p9_2): Likewise.
5524 2017-07-25 Carl Love <cel@us.ibm.com>
5526 * doc/extend.texi: Update the built-in documentation file for the
5527 existing built-in functions
5528 vector signed char vec_cnttz (vector signed char);
5529 vector unsigned char vec_cnttz (vector unsigned char);
5530 vector signed short vec_cnttz (vector signed short);
5531 vector unsigned short vec_cnttz (vector unsigned short);
5532 vector signed int vec_cnttz (vector signed int);
5533 vector unsigned int vec_cnttz (vector unsigned int);
5534 vector signed long long vec_cnttz (vector signed long long);
5535 vector unsigned long long vec_cnttz (vector unsigned long long);
5537 2017-07-25 Andrew Pinski <apinski@cavium.com>
5539 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
5540 accesses where the use is for the first operand of a BIT_INSERT.
5542 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
5545 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
5546 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
5548 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
5550 * config/i386/gstabs.h: Delete.
5551 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
5553 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
5555 * config/i386/i386.c (ix86_decompose_address): Do not check for
5556 register RTX when looking at index_reg or base_reg.
5557 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
5559 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
5561 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
5562 to update EH info here.
5564 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5566 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
5568 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5570 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
5572 2017-07-25 Torsten Duwe <duwe@suse.de>
5574 * common.opt: Introduce -fpatchable-function-entry
5575 command line option, and its variables function_entry_patch_area_size
5576 and function_entry_patch_area_start.
5577 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
5578 including a two-value parser.
5579 * target.def (print_patchable_function_entry): New target hook.
5580 * targhooks.h (default_print_patchable_function_entry): New function.
5581 * targhooks.c (default_print_patchable_function_entry): Likewise.
5582 * toplev.c (process_options): Switch off IPA-RA if
5583 patchable function entries are being generated.
5584 * varasm.c (assemble_start_function): Look at the
5585 patchable-function-entry command line switch and current
5586 function attributes and maybe generate NOP instructions by
5587 calling the print_patchable_function_entry hook.
5588 * doc/extend.texi: Document patchable_function_entry attribute.
5589 * doc/invoke.texi: Document -fpatchable_function_entry
5590 command line option.
5591 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
5593 * doc/tm.texi: Re-generate.
5595 2017-07-25 Jakub Jelinek <jakub@redhat.com>
5598 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
5599 TARGET_AVX512DQ rather than TARGET_AVX512BW.
5601 2017-07-25 Tamar Christina <tamar.christina@arm.com>
5603 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
5605 2017-07-25 Richard Biener <rguenther@suse.de>
5607 PR tree-optimization/81455
5608 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
5609 not walk in cycles when looking for guards.
5611 2017-07-25 Richard Biener <rguenther@suse.de>
5613 PR tree-optimization/81529
5614 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
5615 when optimizing backedge uses.
5617 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
5619 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
5621 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
5622 to dl_section_ref. On AIX, append an expression to subtract
5623 the size of the section length to dl_section_ref.
5625 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
5627 * configure.ac: If any of the config.* scripts fail, exit 1.
5628 * configure: Regenerate.
5630 2017-07-25 Richard Biener <rguenther@suse.de>
5633 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
5634 of immediate uses, be more verbose on errors.
5636 2017-07-25 Richard Biener <rguenther@suse.de>
5638 PR tree-optimization/81510
5639 * tree-vect-loop.c (vect_is_simple_reduction): When the
5640 reduction stmt is not inside the loop bail out.
5642 2017-07-25 Richard Biener <rguenther@suse.de>
5644 PR tree-optimization/81303
5645 * tree-vect-loop-manip.c (vect_loop_versioning): Build
5646 profitability check against LOOP_VINFO_NITERSM1.
5648 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5650 * domwalk.c (cmp_bb_postorder): Simplify.
5651 (sort_bbs_postorder): New function. Use it...
5652 (dom_walker::walk): ...here to optimize common cases.
5654 2017-07-25 Martin Liska <mliska@suse.cz>
5657 * ipa-visibility.c (function_and_variable_visibility): Make the
5658 redirection just on target that supports aliasing.
5659 Fix GNU coding style.
5661 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
5664 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
5666 * config/arm/rtems.h: Likewise.
5667 * config/bfin/rtems.h: Likewise.
5668 * config/i386/rtemself.h: Likewise.
5669 * config/lm32/rtems.h: Likewise.
5670 * config/m32c/rtems.h: Likewise.
5671 * config/m68k/rtemself.h: Likewise.
5672 * config/microblaze/rtems.h: Likewise.
5673 * config/mips/rtems.h: Likewise.
5674 * config/moxie/rtems.h: Likewise.
5675 * config/nios2/rtems.h: Likewise.
5676 * config/powerpcspe/rtems.h: Likewise.
5677 * config/rs6000/rtems.h: Likewise.
5678 * config/rtems.h: Likewise.
5679 * config/sh/rtems.h: Likewise.
5680 * config/sh/rtemself.h: Likewise.
5681 * config/sparc/rtemself.h: Likewise.
5683 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
5686 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
5687 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
5688 * tree-ssa-structalias.c (alias_get_name): Same.
5690 2017-07-25 Bin Cheng <bin.cheng@arm.com>
5693 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
5694 instructions if no du chain is found.
5696 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
5698 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
5700 2017-07-25 Richard Biener <rguenther@suse.de>
5703 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
5706 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
5708 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
5710 (ISA_2_7_MASKS_SERVER): Likewise.
5711 (ISA_3_0_MASKS_IEEE): Likewise.
5712 (OTHER_P8_VECTOR_MASKS): Likewise.
5713 (OTHER_VSX_VECTOR_MASKS): Likewise.
5714 (POWERPC_MASKS): Likewise.
5715 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
5716 duplicate list of options.
5717 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
5718 explicit -mupper-regs options.
5719 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
5720 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
5721 alias for -mupper-regs-df.
5722 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
5723 (rs6000_init_hard_regno_mode_ok): Likewise.
5724 (rs6000_option_override_internal): Likewise.
5725 (rs6000_opt_masks): Likewise.
5726 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
5727 options in terms of whether -mvsx or -mpower8-vector was used.
5728 (TARGET_UPPER_REGS_DI): Likewise.
5729 (TARGET_UPPER_REGS_SF): Likewise.
5730 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
5731 -mupper-regs-* options.
5733 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
5735 * passes.c (emergency_dump_function): Print some empty lines and a
5736 header before the RTL dump.
5738 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
5740 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
5742 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
5745 * config/aarch64/aarch64.c (aarch64_classify_symbol):
5746 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
5748 2017-07-24 Carl Love <cel@us.ibm.com>
5750 * config/rs6000/rs6000-c.c: Add support for built-in functions
5751 vector float vec_extract_fp32_from_shorth (vector unsigned short);
5752 vector float vec_extract_fp32_from_shortl (vector unsigned short);
5753 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
5754 vec_extract_fp_from_shortl): Add defines for the two builtins.
5755 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
5756 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
5758 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
5759 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
5760 * doc/extend.texi: Update the built-in documentation file for the
5761 new built-in function.
5763 2017-07-24 Jakub Jelinek <jakub@redhat.com>
5766 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
5768 * doc/generic.texi: Likewise.
5769 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
5770 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
5772 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
5774 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
5775 (aarch64_mls_elt_merge<mode>): Likewise.
5777 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
5779 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
5780 having __cxa_atexit.
5782 2017-07-23 Michael Collison <michael.collison@arm.com>
5784 * config/arm/arm.c (arm_option_override): Deprecate
5785 use of -mstructure-size-boundary.
5786 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
5787 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
5789 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5792 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
5793 Reduce cost estimate for direct moves.
5795 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
5798 * config/i386/i386.c (ix86_option_override_internal): Disable
5799 BMI, BMI2 and TBM instructions for -m16.
5801 2017-07-21 Carl Love <cel@us.ibm.com>
5803 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5804 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5805 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5806 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5807 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5808 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5809 VMULOSW): New enum "unspec" values.
5810 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5811 altivec_vmulosw): New patterns.
5812 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5813 VMULOSW): Add definitions.
5815 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
5817 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
5818 (qdf24xx): Likewise.
5819 * config/aarch64/aarch64-options-extensions.def (rdma); New.
5820 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
5821 (AARCH64_FL_V8_1): Renumber.
5822 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
5823 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
5824 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
5825 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
5826 rdma to feature modifiers list.
5828 2017-07-21 Yury Gribov <tetra2005@gmail.com>
5831 * ipa-visibility (function_and_variable_visibility): Convert
5832 recursive PLT call to direct call if appropriate.
5834 2017-07-21 Andrew Pinski <apinski@cavium.com>
5836 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
5837 operand 1 to see if the types precision matches.
5838 * fold-const.c (operand_equal_p): Likewise.
5840 2017-07-21 Richard Biener <rguenther@suse.de>
5842 PR tree-optimization/81303
5843 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
5844 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
5845 (vect_peeling_hash_get_lowest_cost): Adjust.
5846 (vect_enhance_data_refs_alignment): Likewise. Use
5847 vect_get_peeling_costs_all_drs to compute the penalty for no
5848 peeling to match up costs.
5850 2017-07-21 Richard Biener <rguenther@suse.de>
5852 PR tree-optimization/81500
5853 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
5854 we didn't identify a reduction path.
5856 2017-07-21 Tom de Vries <tom@codesourcery.com>
5857 Cesar Philippidis <cesar@codesourcery.com>
5859 PR gcov-profile/81442
5860 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
5863 2017-07-21 Tom de Vries <tom@codesourcery.com>
5866 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
5868 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
5869 nvptx_override_options_after_change.
5871 2017-07-21 Ulrich Drepper <drepper@redhat.com>
5873 * dwarf2out.c (output_file_names): Avoid double testing for
5876 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
5878 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
5880 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5882 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
5884 (try_crossjump_to_edge): Do not punt on partitioned functions.
5886 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5888 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
5889 Put all BBs reachable only via paths crossing cold region to cold
5891 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
5893 2016-07-21 Richard Biener <rguenther@suse.de>
5895 PR tree-optimization/81303
5896 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
5897 into account prologue and epilogue iterations when raising
5898 min_profitable_iters to sth at least covering one vector iteration.
5900 2017-07-21 Tamar Christina <tamar.christina@arm.com>
5902 * config/arm/arm.c (arm_test_cpu_arch_dat):
5905 2017-07-20 Nathan Sidwell <nathan@acm.org>
5907 Remove TYPE_METHODS.
5908 * tree.h (TYPE_METHODS): Delete.
5909 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
5910 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
5911 (dbxout_type_methods): Scan TYPE_FIELDS.
5912 (dbxout_type): Don't check TYPE_METHODS here.
5913 * function.c (use_register_for_decl): Always ignore register for
5914 class types when not optimizing.
5915 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
5916 * tree.c (free_lang_data_in_type): Stitch out member functions and
5917 templates from TYPE_FIELDS.
5918 (build_distinct_type_copy, verify_type_variant,
5919 verify_type): Member fns are on TYPE_FIELDS.
5920 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
5921 * tree-pretty-print.c (dump_generic_node): Likewise.
5923 2017-07-20 Jakub Jelinek <jakub@redhat.com>
5926 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
5927 V2TImode and V4TImode.
5928 (ix86_expand_vector_extract): Likewise.
5929 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
5930 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
5931 (ssescalarmode): Handle V4TImode and V2TImode.
5932 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
5933 (*vec_extractv2ti, *vec_extractv4ti): New insns.
5934 (VEXTRACTI128_MODE): New mode iterator.
5935 (splitter for *vec_extractv?ti first element): New.
5936 (VEC_INIT_MODE): New mode iterator.
5937 (vec_init<mode>): Consolidate 3 expanders into one using
5938 VEC_INIT_MODE mode iterator.
5940 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
5942 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
5943 non_spilled_static_chain_regno_p.
5945 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
5947 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
5949 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
5951 * bb-reorder.c (connect_traces): Allow copying of blocks within
5954 2017-07-20 Richard Biener <rguenther@suse.de>
5956 * gimple.h (gimple_phi_result): Add gphi * overload.
5957 (gimple_phi_result_ptr): Likewise.
5958 (gimple_phi_arg): Likewise. Adjust index assert to only
5959 allow actual argument accesses rather than all slots available
5961 (gimple_phi_arg_def): Add gphi * overload.
5962 * tree-phinodes.c (make_phi_node): Initialize only actual
5964 (resize_phi_node): Clear memory not covered by old node,
5965 do not initialize excess argument slots.
5966 (reserve_phi_args_for_new_edge): Initialize new argument slot
5969 2017-07-20 Bin Cheng <bin.cheng@arm.com>
5971 PR tree-optimization/81388
5973 2016-07-21 Bin Cheng <bin.cheng@arm.com>
5975 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
5976 by removing computation of may_be_zero.
5978 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5979 Tom de Vries <tom@codesourcery.com>
5982 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
5983 when gimple level profile disagrees with what RTL expander did.
5985 2017-07-20 Richard Biener <rguenther@suse.de>
5987 PR tree-optimization/61171
5988 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
5989 (vect_analyze_stmt): Add slp instance parameter.
5990 (vectorizable_reduction): Likewise.
5991 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
5992 (vect_is_simple_reduction): Deal with chains not detected
5993 as SLP reduction chain, specifically not properly associated
5994 chains containing a mix of plus/minus.
5995 (get_reduction_op): Remove.
5996 (get_initial_defs_for_reduction): Simplify, pass in whether
5997 this is a reduction chain, pass in the SLP node for the PHIs.
5998 (vect_create_epilog_for_reduction): Get the SLP instance as
6000 (vectorizable_reduction): Get the SLP instance as arg.
6001 During analysis remember the SLP node with the PHIs in the
6002 instance. Simplify getting at the vectorized reduction PHIs.
6003 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
6004 through SLP instance.
6005 (vect_slp_analyze_operations): Likewise.
6006 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
6007 (vect_transform_stmt): Likewise.
6009 2017-07-20 Tom de Vries <tom@codesourcery.com>
6011 PR tree-optimization/81489
6012 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
6013 read of phi arg location to before loop that modifies phi.
6015 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
6017 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
6020 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6023 * except.c (execute): Fix ordering issue.
6025 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
6027 PR rtl-optimization/81423
6028 * combine.c (make_compound_operation_int): Don't try to optimize
6029 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
6031 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
6033 PR rtl-optimization/81423
6034 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
6035 with a constant that is -1 in the truncated to mode.
6037 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6039 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
6040 (determine_unlikely_bbs): ... here.
6041 * predict.h (propagate_unlikely_bbs_forward): Declare.
6042 * cfgexpand.c (pass_expand::execute): Use it.
6043 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
6045 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
6046 propagate_unlikely_bbs_forward.
6048 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6051 * except.c (maybe_add_nop_after_section_switch): New function.
6054 2017-07-19 Tom de Vries <tom@codesourcery.com>
6056 * gimple.h (gimple_phi_set_arg): Make assert more strict.
6058 2017-07-19 Tom de Vries <tom@codesourcery.com>
6060 * gimple.h (gimple_phi_arg): Make assert more strict.
6062 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
6064 * config.gcc (powerpc*-*-*): Add mmintrin.h.
6065 * config/rs6000/mmintrin.h: New file.
6066 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
6068 2017-07-19 Jakub Jelinek <jakub@redhat.com>
6070 PR tree-optimization/81346
6071 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
6073 2017-07-19 Tom de Vries <tom@codesourcery.com>
6075 * config/nvptx/nvptx.md (VECIM): Add V2DI.
6077 2017-07-19 Tom de Vries <tom@codesourcery.com>
6079 * config/nvptx/nvptx-modes.def: Add V2DImode.
6080 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
6081 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
6082 (nvptx_output_mov_insn): Handle lack of mov.b128.
6083 (nvptx_print_operand): Handle 'H' and 'L' codes.
6084 (nvptx_vector_mode_supported): Allow V2DImode.
6085 (nvptx_preferred_simd_mode): New function.
6086 (nvptx_data_alignment): New function.
6087 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
6088 nvptx_preferred_simd_mode.
6089 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
6091 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
6093 2017-07-19 Tom de Vries <tom@codesourcery.com>
6095 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
6096 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
6097 (nvptx_vector_mode_supported): New function. Allow V2SImode.
6098 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
6099 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
6100 (mov<VECIM>_insn): New define_insn.
6101 (define_expand "mov<VECIM>): New define_expand.
6103 2017-07-19 Tom de Vries <tom@codesourcery.com>
6105 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
6107 2017-07-19 Jakub Jelinek <jakub@redhat.com>
6109 PR tree-optimization/81346
6110 * fold-const.h (fold_div_compare, range_check_type): Declare.
6111 * fold-const.c (range_check_type): New function.
6112 (build_range_check): Use range_check_type.
6113 (fold_div_compare): No longer static, rewritten into
6114 a match.pd helper function.
6115 (fold_comparison): Don't call fold_div_compare here.
6116 * match.pd (X / C1 op C2): New optimization using fold_div_compare
6119 2017-07-19 Nathan Sidwell <nathan@acm.org>
6121 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
6122 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
6123 * tree.c (find_decls_types_r, verify_type): Use
6124 TYPE_{MIN,MAX}_VALUE_RAW.
6125 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
6126 (hash_tree): Likewise.
6127 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
6129 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
6132 2017-07-18 Tom de Vries <tom@codesourcery.com>
6135 * omp-expand.c (expand_omp_for_static_chunk): Handle
6136 equal-argument loop exit phi.
6138 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
6141 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
6142 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
6143 operand 2 predicate.
6144 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
6145 operand 2 predicate.
6146 (ror,rol -> rorx splitters): Use const_int_operand as
6147 operand 2 predicate.
6149 2017-06-18 Richard Biener <rguenther@suse.de>
6151 PR tree-optimization/81410
6152 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
6153 the gap in the ! slp_perm SLP case after each group.
6155 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
6158 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
6161 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
6164 * predict.c (set_even_probabilities): Cleanup; do not affect
6165 probabilities that are already known.
6166 (combine_predictions_for_bb): Call even when count is set.
6168 2017-07-18 Nathan Sidwell <nathan@acm.org>
6170 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
6173 2017-07-18 Bin Cheng <bin.cheng@arm.com>
6176 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
6177 optimization for loop niter analysis.
6179 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
6182 * config/avr/avr.c (avr_optimize_casesi): Don't use
6183 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
6185 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
6187 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
6188 body_cost_vec from _vect_peel_extended_info.
6189 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
6190 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
6193 2017-07-18 Bin Cheng <bin.cheng@arm.com>
6195 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
6197 2017-07-18 Richard Biener <rguenther@suse.de>
6199 PR tree-optimization/80620
6200 PR tree-optimization/81403
6201 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
6202 info when re-using a VN table entry.
6204 2017-07-18 Richard Biener <rguenther@suse.de>
6206 PR tree-optimization/81418
6207 * tree-vect-loop.c (vectorizable_reduction): Properly compute
6208 vectype_in. Verify that with lane-reducing reduction operations
6209 we have a single def-use cycle.
6211 2017-07-17 Carl Love <cel@us.ibm.com>
6213 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
6215 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6216 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6217 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
6218 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
6219 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
6220 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
6221 VMULOSW): New enum "unspec" values.
6222 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
6223 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
6224 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
6225 altivec_vmulosw): New patterns.
6226 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
6227 VMULOSW): Add definitions.
6228 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
6230 * config/alpha/alpha.c: Include predict.h.
6232 2017-07-17 Yury Gribov <tetra2005@gmail.com>
6234 * tree-vrp.c (compare_assert_loc): Fix comparison function
6235 to return predictable results.
6237 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6239 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
6242 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
6243 * doc/invoke.texi (mexpand-adddi): Update text.
6245 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6247 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
6248 that also clobbers the CC register. The old expand code is moved
6250 (*arc_clzsi2): ... here.
6251 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
6252 the CC register. The old expand code is moved to ...
6253 (arc_ctzsi2): ... here.
6255 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6257 * config/arc/arc.opt (mindexed-loads): Use initial value
6258 TARGET_INDEXED_LOADS_DEFAULT.
6259 (mauto-modify-reg): Use initial value
6260 TARGET_AUTO_MODIFY_REG_DEFAULT.
6261 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
6262 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
6263 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
6264 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
6266 2017-07-17 Martin Liska <mliska@suse.cz>
6269 * opts.c (finish_options): Do not allow -fgnu-tm
6270 w/ -fsanitize={kernel-,}address. Say sorry.
6272 2017-07-17 Bin Cheng <bin.cheng@arm.com>
6275 * tree-loop-distribution.c (classify_partition): Only assert on
6276 numer of iterations.
6277 (merge_dep_scc_partitions): Delete prameter. Update function call.
6278 (distribute_loop): Remove code handling loop with unknown niters.
6279 (pass_loop_distribution::execute): Skip loop with unknown niters.
6281 2017-07-17 Bin Cheng <bin.cheng@arm.com>
6284 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
6285 function sort_partitions_by_post_order.
6287 2017-07-17 Bin Cheng <bin.cheng@arm.com>
6289 PR tree-optimization/81374
6290 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
6291 the max index of basic blocks, rather than number of basic blocks.
6293 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6295 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
6297 (arc_legitimate_pic_operand_p): Likewise.
6298 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
6300 (arc_needs_pcl_p): Likewise.
6301 (arc_legitimate_pc_offset_p): Likewise.
6302 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
6303 function is also used in constrains.md.
6304 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
6305 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
6306 PLUS. Only return true/false in known cases, otherwise assert.
6307 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
6308 is already called in arc_legitimate_constant_p.
6309 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
6311 (LEGITIMATE_PIC_OPERAND_P): Use
6312 arc_raw_symbolic_reference_mentioned_p function.
6313 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
6318 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6319 Andrew Burgess <andrew.burgess@embecosm.com>
6321 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
6322 (arc_return_address_register): New function.
6323 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
6324 (arc_handle_fndecl_attribute): Add naked attribute.
6325 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
6326 (TARGET_WARN_FUNC_RETURN): Likewise.
6327 (arc_allocate_stack_slots_for_args): New function.
6328 (arc_warn_func_return): Likewise.
6329 (machine_function): Change type fn_type.
6330 (arc_compute_function_type): Consider new naked function type,
6331 change function return type.
6332 (arc_must_save_register): Adapt to handle new
6333 arc_compute_function_type's return type.
6334 (arc_expand_prologue): Likewise.
6335 (arc_expand_epilogue): Likewise.
6336 (arc_return_address_regs): Delete.
6337 (arc_return_address_register): New function.
6338 (arc_epilogue_uses): Use above function.
6339 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
6340 (arc_function_type): Change encoding, add naked type.
6341 (ARC_INTERRUPT_P): Change to handle the new encoding.
6342 (ARC_FAST_INTERRUPT_P): Likewise.
6343 (ARC_NORMAL_P): Define.
6344 (ARC_NAKED_P): Likewise.
6345 (arc_compute_function_type): Delete prototype.
6346 * config/arc/arc.md (in_ret_delay_slot): Use
6347 arc_return_address_register function.
6348 (simple_return): Likewise.
6349 (p_return_i): Likewise.
6351 2017-07-17 Jakub Jelinek <jakub@redhat.com>
6353 PR tree-optimization/81428
6354 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
6355 can't be built for those types.
6357 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
6359 Remove stuff dead since r239246.
6361 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
6362 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
6363 (avr_inform_devices): Remove dead stuff.
6365 2017-07-17 Tamar Christina <tamar.christina@arm.com>
6367 * config/arm/arm_neon.h: Fix softp typo.
6369 2017-07-17 Jakub Jelinek <jakub@redhat.com>
6371 PR tree-optimization/81365
6372 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
6373 aggregate moves onto bb predecessor edges, make sure there are no
6374 loads that could alias the lhs in between the start of bb and the
6377 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
6380 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
6381 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
6382 [LSHIFTRT, outer_code = TRUNCATE]: Same.
6384 2017-07-17 Jakub Jelinek <jakub@redhat.com>
6386 PR tree-optimization/81396
6387 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
6388 (init_symbolic_number): Initialize it to 1.
6389 (perform_symbolic_merge): Add n_ops from both operands into the new
6391 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
6392 without base_addr as useless if they need more than one operation.
6393 (bswap_replace): Handle !bswap case for NULL base_addr.
6395 2017-07-17 Tom de Vries <tom@codesourcery.com>
6398 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
6401 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
6403 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
6404 conditional builtin define __FIX_LEON3FT_B2BST.
6406 2017-07-17 Daniel Cederman <cederman@gaisler.com>
6408 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
6409 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
6412 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
6414 PR rtl-optimization/81424
6415 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
6416 to remove potential trapping from operands if -fnon-call-exceptions.
6418 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6420 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
6421 profile_proability for scalling.
6422 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
6424 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6426 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
6428 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6430 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
6431 fixpoint arithmetics.
6433 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6435 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
6436 fixpoint arithmetics.
6438 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6440 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
6441 fixpoint arithmetics.
6443 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6445 * profile-count.h (profile_probability::from_reg_br_prob_note,
6446 profile_probability::to_reg_br_prob_note): New functions.
6447 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
6448 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
6449 * predict.c (probability_reliable_p): Update.
6450 (edge_probability_reliable_p): Update.
6451 (br_prob_note_reliable_p): Update.
6452 (invert_br_probabilities): Update.
6453 (add_reg_br_prob_note): New function.
6454 (combine_predictions_for_insn): Update.
6455 * asan.c (asan_clear_shadow): Update.
6456 * cfgbuild.c (compute_outgoing_frequencies): Update.
6457 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6458 (update_br_prob_note): Update.
6459 (rtl_verify_edges): Update.
6460 (purge_dead_edges): Update.
6461 (fixup_reorder_chain): Update.
6462 * emit-rtl.c (try_split): Update.
6463 * ifcvt.c (cond_exec_process_insns): Update.
6464 (cond_exec_process_if_block): Update.
6465 (dead_or_predicable): Update.
6466 * internal-fn.c (expand_addsub_overflow): Update.
6467 (expand_neg_overflow): Update.
6468 (expand_mul_overflow): Update.
6469 * loop-doloop.c (doloop_modify): Update.
6470 * loop-unroll.c (compare_and_jump_seq): Update.
6471 * optabs.c (emit_cmp_and_jump_insn_1): Update.
6472 * predict.h: Update.
6473 * reorg.c (mostly_true_jump): Update.
6475 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
6476 * config/alpha/alpha.c (emit_unlikely_jump): Update.
6477 * config/arc/arc.c: (emit_unlikely_jump): Update.
6478 * config/arm/arm.c: (emit_unlikely_jump): Update.
6479 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
6480 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
6481 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
6482 (ix86_print_operand): Update.
6483 (ix86_split_fp_branch): Update.
6484 (predict_jump): Update.
6485 * config/ia64/ia64.c (ia64_print_operand): Update.
6486 * config/mmix/mmix.c (mmix_print_operand): Update.
6487 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
6488 (rs6000_expand_split_stack_prologue): Update.
6489 * config/rs6000/rs6000.c: Update.
6490 * config/s390/s390.c (s390_expand_vec_strlen): Update.
6491 (s390_expand_vec_movstr): Update.
6492 (s390_expand_cs_tdsi): Update.
6493 (s390_expand_split_stack_prologue): Update.
6494 * config/sh/sh.c (sh_print_operand): Update.
6495 (expand_cbranchsi4): Update.
6496 (expand_cbranchdi4): Update.
6497 * config/sparc/sparc.c (output_v9branch): Update.
6498 * config/spu/spu.c (get_branch_target): Update.
6499 (ea_load_store_inline): Update.
6500 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
6501 * config/tilepro/tilepro.c: Update.
6503 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
6505 * gimplify.c (mostly_copy_tree_r): Revert latest change.
6506 (gimplify_save_expr): Likewise.
6508 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6510 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
6512 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6514 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
6516 * timevar.def (TV_IPA_FNSUMMARY): Define.
6518 2017-07-16 Daniel Cederman <cederman@gaisler.com>
6520 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
6521 to back store errata sensitive sequence from being generated.
6522 (sqrtdf2_fix): Likewise.
6524 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6526 * tree-ssa-threadupdate.c (compute_path_counts,
6527 update_joiner_offpath_counts): Use profile_probability.
6529 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
6532 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6534 * config/arm/arm-c.c (arm_cpu_builtins): Define
6535 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
6537 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
6539 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6540 array entries to represent __ieee128 versions of the
6541 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
6542 scalar_extract_sig, and scalar_insert_exp built-in functions.
6543 (altivec_resolve_overloaded_builtin): Add special case handling
6544 for the __builtin_scalar_insert_exp function, as represented by
6545 the P9V_BUILTIN_VEC_VSIEDP constant.
6546 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
6547 exponent support for __ieee128 argument.
6548 (VSESQP): Add scalar extract signature support for __ieee128
6550 (VSTDCNQP): Add scalar test negative support for __ieee128
6552 (VSIEQP): Add scalar insert exponent support for __int128 argument
6553 with __ieee128 result.
6554 (VSIEQPF): Add scalar insert exponent support for __ieee128
6555 argument with __ieee128 result.
6556 (VSTDCQP): Add scalar test data class support for __ieee128
6558 (VSTDCNQP): Add overload support for scalar test negative with
6560 (VSTDCQP): Add overload support for scalar test data class
6562 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
6564 (UNSPEC_VSX_SIEXPQP): New constant.
6565 (xsxexpqp): New insn for VSX scalar extract exponent quad
6567 (xsxsigqp): New insn for VSX scalar extract significand quad
6569 (xsiexpqpf): New insn for VSX scalar insert exponent quad
6570 precision with floating point argument.
6571 (xststdcqp): New expand for VSX scalar test data class quad
6573 (xststdcnegqp): New expand for VSX scalar test negative quad
6575 (xststdcqp): New insn to match expansions for VSX scalar test data
6576 class quad precision and VSX scalar test negative quad precision.
6577 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6578 special case operand checking to enforce that second operand of
6579 VSX scalar test data class with quad precision argument is a 7-bit
6581 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
6582 prototypes and descriptions of __ieee128 versions of
6583 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
6584 scalar_test_data_class, and scalar_test_neg built-in functions.
6586 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6588 PR tree-optimization/81162
6589 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
6590 replace a negate with an add.
6592 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
6594 * doc/invoke.texi (arm/-mcpu): Document +crypto.
6596 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6598 * config/arm/arm-c.c (arm_cpu_builtins): Define
6599 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
6601 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6603 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
6604 (armv8-r): Set ARM Cortex-R52 as default CPU.
6605 * config/arm/arm-tables.opt: Regenerate.
6606 * config/arm/arm-tune.md: Regenerate.
6607 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
6609 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
6610 extension for -mcpu=cortex-r52.
6612 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6614 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
6615 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
6616 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
6617 (fp-armv8): Define it as FP_ARMv8 only.
6618 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
6619 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
6621 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
6622 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
6623 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
6624 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
6625 than TARGET_FPU_ARMV8.
6626 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
6627 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
6628 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
6630 * config/arm/neon.md (neon_vrint): Likewise.
6631 (neon_vcvt): Likewise.
6632 (neon_<fmaxmin_op><mode>): Likewise.
6633 (<fmaxmin><mode>3): Likewise.
6634 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
6635 * config/arm/predicates.md (arm_cond_move_operator): Check against
6636 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
6638 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
6640 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
6643 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6645 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
6646 loop in comment with memset.
6648 2017-07-14 Martin Liska <mliska@suse.cz>
6650 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
6651 * dwarf2out.c (is_java): Remove the function.
6652 (output_pubname): Remove usage of the function.
6653 (lower_bound_default): Remove usage of DW_LANG_Java.
6654 (gen_compile_unit_die): Likewise.
6655 * gcc.c: Remove compiler defaults for .java and .zip files.
6656 * gimple-expr.c (remove_suffix): Change as there's no longer
6657 extension than 4-letter one.
6658 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
6659 (gimplify_save_expr): Likewise.
6660 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
6661 as it's possible even for other languages than Java.
6662 * langhooks.h (struct lang_hooks): Remove Java from a comment.
6663 * lto-opts.c (lto_write_options): Remove reference to Java.
6664 * opts.c (strip_off_ending): Update file extension handling.
6665 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
6666 * tree-eh.c (lower_resx): Likewise.
6667 * tree.c (free_lang_data_in_type): Remove dead code.
6668 (find_decls_types_r): Likewise.
6669 (build_common_builtin_nodes): Remove Java from a comment.
6670 (verify_type): Remove dead code.
6671 * varasm.c (assemble_external): Remove Java from a comment.
6673 2017-07-14 Martin Liska <mliska@suse.cz>
6675 * opts.c (finish_options): Add quotes.
6676 (common_handle_option): Likewise.
6678 2017-07-14 Martin Liska <mliska@suse.cz>
6680 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
6681 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
6683 * dwarf2out.c (lower_bound_default): Do not handle
6685 (gen_compile_unit_die): Likewise.
6686 * gcc.c: Remove default extension binding for GNU Pascal.
6687 * stmt.c: Remove Pascal language from a comment.
6688 * xcoffout.c: Likewise.
6690 2017-07-13 David Malcolm <dmalcolm@redhat.com>
6693 * diagnostic-show-locus.c (fixit_cmp): New function.
6694 (layout::layout): Sort m_fixit_hints.
6695 (column_range::column_range): Assert that the values are valid.
6696 (struct char_span): New struct.
6697 (correction::overwrite): New method.
6698 (struct source_line): New struct.
6699 (line_corrections::add_hint): Add assertions. Reimplement memcpy
6700 calls in terms of classes source_line and char_span, and
6701 correction::overwrite.
6702 (selftest::test_overlapped_fixit_printing_2): New function.
6703 (selftest::diagnostic_show_locus_c_tests): Call it.
6705 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6707 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
6708 early if there is no lhs.
6710 2017-07-13 Martin Liska <mliska@suse.cz>
6712 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
6713 (gen_reference_type_die): Likewise.
6714 * stor-layout.c: Remove Pascal-related comment.
6716 2017-07-13 Martin Liska <mliska@suse.cz>
6718 * opts.c (finish_options): Add quotes to error messages.
6719 (parse_sanitizer_options): Likewise.
6721 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
6723 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
6725 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
6727 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
6729 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
6731 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
6733 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
6735 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6738 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
6739 provides the hardware capability bits, define the macro
6740 __BUILTIN_CPU_SUPPORTS__.
6741 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
6742 if GLIBC does not provide the hardware capability bits. Add a
6743 gcc_unreachable call if the built-in cpu function is neither
6744 __builtin_cpu_is nor __builtin_cpu_supports.
6745 (rs6000_get_function_versions_dispatcher): Change the warning
6746 that an old GLIBC is used which does not export the capability
6747 bits to be an error.
6748 * doc/extend.texi (target_clones attribute): Document the
6749 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
6750 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
6751 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
6752 the macros defined by GCC if the newer GLIBC is available.
6754 2017-07-12 Jeff Law <law@redhat.com>
6756 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
6757 remaining includes slightly.
6758 * config/riscv/riscv-builtins.c: Include profile-count.h.
6760 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
6763 * config/avr/avr.c (avr_set_current_function): In diagnostic
6764 messages: Quote keywords and (parts of) identifiers.
6765 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
6768 2017-07-12 Carl Love <cel@us.ibm.com>
6770 * config/rs6000/rs6000-c.c: Add support for built-in functions
6771 vector bool char vec_revb (vector bool char);
6772 vector bool short vec_revb (vector short char);
6773 vector bool int vec_revb (vector bool int);
6774 vector bool long long vec_revb (vector bool long long);
6775 * doc/extend.texi: Update the built-in documentation file for the
6776 new built-in functions.
6778 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6780 * config/s390/s390.md: Remove movcc splitter.
6782 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6784 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
6785 load/store on condition.
6787 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
6790 * config/avr/avr.c (avr_encode_section_info)
6791 [progmem && !TREE_READONLY]: Error if progmem object needs
6794 2017-07-11 Michael Collison <michael.collison@arm.com>
6796 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
6799 2017-07-11 Carl Love <cel@us.ibm.com>
6801 * config/rs6000/rs6000-c.c: Add support for builtins
6802 vector unsigned int vec_parity_lsbb (vector signed int);
6803 vector unsigned int vec_parity_lsbb (vector unsigned int);
6804 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
6805 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
6806 vector unsigned long long vec_parity_lsbb (vector signed long long);
6807 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
6808 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
6809 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
6810 * doc/extend.texi: Update the built-in documentation file for the
6811 new built-in functions.
6813 2017-07-11 David Malcolm <dmalcolm@redhat.com>
6815 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
6816 (layout::m_primary_loc): New field.
6817 (layout::layout): Initialize new field. Move location filtering
6818 logic from here to...
6819 (layout::maybe_add_location_range): ...this new method. Add
6820 support for filtering to just the lines already specified by other
6822 (layout::will_show_line_p): New method.
6823 (gcc_rich_location::add_location_if_nearby): New method.
6824 (selftest::test_add_location_if_nearby): New test function.
6825 (selftest::diagnostic_show_locus_c_tests): Call it.
6826 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
6829 2017-07-11 Tom de Vries <tom@codesourcery.com>
6831 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
6832 (bb_first_real_insn): New function.
6833 (nvptx_single): Add extra initialization of broadcasted condition
6836 2017-07-11 Nathan Sidwell <nathan@acm.org>
6838 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
6840 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
6842 * doc/extend.texi (AVR Function Attributes): Remove weblink to
6843 Binutils doc as TEXI will mess them up.
6844 * doc/invoke.texi (AVR Options): Same here.
6846 2017-07-11 Daniel Cederman <cederman@gaisler.com>
6848 * config/sparc/sparc.opt (mfix-ut700): New option.
6849 (mfix-gr712rc): Likewise.
6850 (sparc_fix_b2bst): New variable.
6851 * doc/invoke.texi (SPARC options): Document them.
6852 (ARM options): Fix warnings.
6853 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
6854 instructions to prevent sequences that can trigger the store-store
6855 errata for certain LEON3FT processors.
6856 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
6857 (sparc_option_override): Set sparc_fix_b2bst appropriately.
6858 * config/sparc/sparc.md (fix_b2bst): New attribute.
6859 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
6861 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
6864 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
6866 (*rsqrtsf2_sse): Ditto.
6868 (div<mode>3): Macroize insn from divdf3 and divsf3
6869 using MODEF mode iterator.
6871 2017-07-10 Martin Sebor <msebor@redhat.com>
6873 PR tree-optimization/80397
6874 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
6875 instead of testing for equality to INTEGER_TYPE.
6877 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
6879 * config.gcc: Remove uclibc from arc target spec.
6881 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
6883 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
6885 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6888 * lto-wrapper.c (remove_option): New function.
6889 (merge_and_complain): Merge PIC/PIE options more realistically.
6891 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
6893 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
6897 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
6898 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
6899 * config.in: Regenerate.
6900 * configure: Regenerate.
6901 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
6902 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
6903 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
6904 (TARGET_GASISR_PROLOGUES): ...target mask.
6905 * common/config/avr/avr-common.c
6906 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
6907 Set -mgas-isr-prologues.
6908 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
6909 INSERT_PASS_BEFORE for it.
6910 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
6911 * config/avr/avr.c (avr_option_override)
6912 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
6913 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
6914 (avr_attribute_table) <no_gccisr>: Add new function attribute.
6915 (avr_set_current_function) <is_no_gccisr>: Init machine field.
6916 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
6918 (make_avr_pass_pre_proep): New function.
6919 (emit_push_sfr) <treg>: Add argument to function and use it
6921 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
6922 and set machine->gasisr.yes.
6923 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
6924 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
6925 __gcc_isr.n_pushed to .L__stack_usage.
6926 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
6927 (avr_asm_final_postscan_insn): ...this new static function.
6928 * config/avr/avr.h (machine_function)
6929 <is_no_gccisr, use_L__stack_usage>: New fields.
6930 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
6931 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
6932 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
6933 (gasisr, *gasisr): New expander and insn.
6934 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
6935 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
6936 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
6938 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
6940 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
6943 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
6945 Move jump-tables out of .text again.
6948 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
6949 (ASM_OUTPUT_ADDR_VEC): New function.
6950 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
6951 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
6952 INSN_ADDRESSes as asm comment.
6953 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
6954 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
6955 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
6956 * config/avr/avr.md (*tablejump): Adjust comment.
6957 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
6958 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
6960 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
6961 (avr_output_addr_vec): New proto.
6962 (avr_log_t) <insn_addresses>: New field.
6964 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
6967 * config/i386/i386.c (ix86_function_arg_advance): Set
6968 outgoing_args_on_stack to true if there are outgoing arguments
6970 (ix86_function_arg): Likewise.
6971 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
6972 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
6973 * config/i386/i386.h (machine_function): Add
6974 outgoing_args_on_stack.
6976 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
6978 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
6980 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
6982 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
6984 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
6985 (REAL_H): Remove $(MACHMODE_H).
6986 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
6988 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
6989 $(MACHMODE_H) and double-int.h.
6990 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
6992 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
6995 2017-07-07 Andrew Pinski <apinski@cavium.com>
6997 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
6998 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
7000 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7002 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
7003 Add warning if GCC was not configured to link against a GLIBC that
7004 exports the hardware capability bits.
7005 (make_resolver_func): Make resolver function private and not a
7006 COMDAT function. Create the name with clone_function_name instead
7007 of make_unique_name.
7010 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
7011 correct operand in doing the split.
7013 2017-07-07 Carl Love <cel@us.ibm.com>
7015 * config/rs6000/rs6000-c: Add support for built-in function
7016 vector unsigned short vec_pack_to_short_fp32 (vector float,
7018 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
7019 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
7020 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
7021 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
7022 (convert_4f32_8i16): Add define_expand.
7023 * doc/extend.texi: Update the built-in documentation file for the
7024 new built-in function.
7026 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7028 * config/sparc/m8.md: New file.
7029 * config/sparc/sparc.md: Include m8.md.
7031 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7033 * config/sparc/sparc.opt: New option -mvis4b.
7034 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
7035 (sparc_option_override): Handle VIS4B.
7036 (enum sparc_builtins): Define
7037 SPARC_BUILTIN_DICTUNPACK{8,16,32},
7038 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
7039 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
7040 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
7041 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
7042 (check_constant_argument): New function.
7043 (sparc_vis_init_builtins): Define builtins
7044 __builtin_vis_dictunpack{8,16,32},
7045 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
7046 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
7047 __builtin_vis_fpcmpde{8,16,32}shl and
7048 __builtin_vis_fpcmpur{8,16,32}shl.
7049 (sparc_expand_builtin): Check that the constant operands to
7050 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
7051 constant and in range.
7052 * config/sparc/sparc-c.c (sparc_target_macros): Handle
7054 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
7055 (SPARC_IMM5_P): Likewise.
7056 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
7057 (enabled): Handle vis4b.
7058 (UNSPEC_DICTUNPACK): New unspec.
7059 (UNSPEC_FPCMPSHL): Likewise.
7060 (UNSPEC_FPUCMPSHL): Likewise.
7061 (UNSPEC_FPCMPDESHL): Likewise.
7062 (UNSPEC_FPCMPURSHL): Likewise.
7063 (cpu_feature): New CPU feature `vis4b'.
7064 (dictunpack{8,16,32}): New insns.
7065 (FPCSMODE): New mode iterator.
7066 (fpcscond): New code iterator.
7067 (fpcsucond): Likewise.
7068 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
7069 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
7070 (fpcmpde{8,16,32}{si,di}shl): Likewise.
7071 (fpcmpur{8,16,32}{si,di}shl): Likewise.
7072 * config/sparc/constraints.md: Define constraints `q' for unsigned
7073 2-bit integer constants and `t' for unsigned 5-bit integer
7075 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
7077 (imm5_operand_dictunpack16): Likewise.
7078 (imm5_operand_dictunpack32): Likewise.
7079 (imm2_operand): Likewise.
7080 * doc/invoke.texi (SPARC Options): Document -mvis4b.
7081 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
7082 ditunpack* and fpcmp*shl builtins.
7084 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7086 * config.gcc: Handle m8 in --with-{cpu,tune} options.
7087 * config.in: Add HAVE_AS_SPARC6 define.
7088 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
7090 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
7092 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
7093 (CPP_CPU_SPEC): Handle m8.
7094 (ASM_CPU_SPEC): Likewise.
7095 * config/sparc/sparc-opts.h (enum processor_type): Add
7097 * config/sparc/sparc.c (m8_costs): New struct.
7098 (sparc_option_override): Handle TARGET_CPU_m8.
7099 (sparc32_initialize_trampoline): Likewise.
7100 (sparc64_initialize_trampoline): Likewise.
7101 (sparc_issue_rate): Likewise.
7102 (sparc_register_move_cost): Likewise.
7103 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
7104 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
7105 (ASM_CPU64_DEFAULT_SPEC): Likewise.
7106 (CPP_CPU_SPEC): Handle M8.
7107 (ASM_CPU_SPEC): Likewise.
7108 (AS_M8_FLAG): Define.
7109 * config/sparc/sparc.md: Add m8 to the cpu attribute.
7110 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
7111 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
7113 * configure: Regenerate.
7114 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
7117 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7119 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
7121 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
7122 ("*movdi_insn_sp32"): Do not set v3pipe.
7123 ("*movsi_insn"): Likewise.
7124 ("*movdi_insn_sp64"): Likewise.
7125 ("*movsf_insn"): Likewise.
7126 ("*movdf_insn_sp32"): Likewise.
7127 ("*movdf_insn_sp64"): Likewise.
7128 ("*zero_extendsidi2_insn_sp64"): Likewise.
7129 ("*sign_extendsidi2_insn"): Likewise.
7130 ("*mov<VM32:mode>_insn"): Likewise.
7131 ("*mov<VM64:mode>_insn_sp64"): Likewise.
7132 ("*mov<VM64:mode>_insn_sp32"): Likewise.
7133 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
7134 ("<vlop:code><VL:mode>3"): Likewise.
7135 ("*not_<vlop:code><VL:mode>3"): Likewise.
7136 ("*nand<VL:mode>_vis"): Likewise.
7137 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
7138 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
7139 ("one_cmpl<VL:mode>2"): Likewise.
7140 ("faligndata<VM64:mode>_vis"): Likewise.
7141 ("alignaddrsi_vis"): Likewise.
7142 ("alignaddrdi_vis"): Likweise.
7143 ("alignaddrlsi_vis"): Likewise.
7144 ("alignaddrldi_vis"): Likewise.
7145 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
7146 ("bmaskdi_vis"): Likewise.
7147 ("bmasksi_vis"): Likewise.
7148 ("bshuffle<VM64:mode>_vis"): Likewise.
7149 ("cmask8<P:mode>_vis"): Likewise.
7150 ("cmask16<P:mode>_vis"): Likewise.
7151 ("cmask32<P:mode>_vis"): Likewise.
7152 ("pdistn<P:mode>_vis"): Likewise.
7153 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
7155 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7157 * config/sparc/sparc.md ("subtype"): New insn attribute.
7158 ("*wrgsr_sp64"): Set insn subtype.
7159 ("*rdgsr_sp64"): Likewise.
7160 ("alignaddrsi_vis"): Likewise.
7161 ("alignaddrdi_vis"): Likewise.
7162 ("alignaddrlsi_vis"): Likewise.
7163 ("alignaddrldi_vis"): Likewise.
7164 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
7165 ("fexpand_vis"): Likewise.
7166 ("fpmerge_vis"): Likewise.
7167 ("faligndata<VM64:mode>_vis"): Likewise.
7168 ("bshuffle<VM64:mode>_vis"): Likewise.
7169 ("cmask8<P:mode>_vis"): Likewise.
7170 ("cmask16<P:mode>_vis"): Likewise.
7171 ("cmask32<P:mode>_vis"): Likewise.
7172 ("fchksm16_vis"): Likewise.
7173 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
7174 ("fmean16_vis"): Likewise.
7175 ("fp<plusminus_insn>64_vis"): Likewise.
7176 ("<plusminus_insn>v8qi3"): Likewise.
7177 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
7178 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
7179 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
7180 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
7181 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
7182 ("*movqi_insn"): Likewise.
7183 ("*movhi_insn"): Likewise.
7184 ("*movsi_insn"): Likewise.
7185 ("movsi_pic_gotdata_op"): Likewise.
7186 ("*movdi_insn_sp32"): Likewise.
7187 ("*movdi_insn_sp64"): Likewise.
7188 ("movdi_pic_gotdata_op"): Likewise.
7189 ("*movsf_insn"): Likewise.
7190 ("*movdf_insn_sp32"): Likewise.
7191 ("*movdf_insn_sp64"): Likewise.
7192 ("*zero_extendhisi2_insn"): Likewise.
7193 ("*zero_extendqihi2_insn"): Likewise.
7194 ("*zero_extendqisi2_insn"): Likewise.
7195 ("*zero_extendqidi2_insn"): Likewise.
7196 ("*zero_extendhidi2_insn"): Likewise.
7197 ("*zero_extendsidi2_insn_sp64"): Likewise.
7198 ("ldfsr"): Likewise.
7199 ("prefetch_64"): Likewise.
7200 ("prefetch_32"): Likewise.
7201 ("tie_ld32"): Likewise.
7202 ("tie_ld64"): Likewise.
7203 ("*tldo_ldub_sp32"): Likewise.
7204 ("*tldo_ldub1_sp32"): Likewise.
7205 ("*tldo_ldub2_sp32"): Likewise.
7206 ("*tldo_ldub_sp64"): Likewise.
7207 ("*tldo_ldub1_sp64"): Likewise.
7208 ("*tldo_ldub2_sp64"): Likewise.
7209 ("*tldo_ldub3_sp64"): Likewise.
7210 ("*tldo_lduh_sp32"): Likewise.
7211 ("*tldo_lduh1_sp32"): Likewise.
7212 ("*tldo_lduh_sp64"): Likewise.
7213 ("*tldo_lduh1_sp64"): Likewise.
7214 ("*tldo_lduh2_sp64"): Likewise.
7215 ("*tldo_lduw_sp32"): Likewise.
7216 ("*tldo_lduw_sp64"): Likewise.
7217 ("*tldo_lduw1_sp64"): Likewise.
7218 ("*tldo_ldx_sp64"): Likewise.
7219 ("*mov<VM32:mode>_insn"): Likewise.
7220 ("*mov<VM64:mode>_insn_sp64"): Likewise.
7221 ("*mov<VM64:mode>_insn_sp32"): Likewise.
7223 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7225 * config/sparc/sparc.md ("type"): New insn type viscmp.
7226 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
7228 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
7229 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
7230 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
7231 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
7233 ("n7_vis_logical_11cycle"): Likewise.
7234 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
7235 * config/sparc/niagara2.md ("niag3_vis": Likewise.
7236 * config/sparc/niagara.md ("niag_vis"): Likewise.
7237 * config/sparc/ultra3.md ("us3_fga"): Likewise.
7238 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
7240 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7242 * config/sparc/sparc.md: New instruction type `bmask'.
7243 (bmaskdi_vis): Use the `bmask' type.
7244 (bmasksi_vis): Likewise.
7245 * config/sparc/ultra3.md (us3_array): Likewise.
7246 * config/sparc/niagara7.md (n7_array): Likewise.
7247 * config/sparc/niagara4.md (n4_array): Likewise.
7248 * config/sparc/niagara2.md (niag2_vis): Likewise.
7249 (niag3_vis): Likewise.
7250 * config/sparc/niagara.md (niag_vis): Likewise.
7252 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7254 * ipa-comdats.c: Remove optimize check from gate.
7255 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
7256 for functions not optimized.
7257 (ipa_fn_summary_read): Skip optimize check.
7258 (ipa_fn_summary_write): Likewise.
7259 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
7261 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
7263 (can_inline_edge_p): Check flag_pcc_struct_return for match.
7264 (check_callers): Give up on caller which is not optimized.
7265 (inline_small_functions): Likewise.
7266 (ipa_inline): Do not give up when not optimizing.
7267 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
7268 away unoptimizes cdtors.
7269 (whole_program_function_and_variable_visibility): Do
7270 ipa_discover_readonly_nonaddressable_vars in LTO mode.
7271 * ipa.c (process_references): Do not check optimize.
7272 (symbol_table::remove_unreachable_nodes): Update optimize check.
7273 (set_writeonly_bit): Update optimize check.
7274 (pass_ipa_cdtor_merge::gate): Do not check optimize.
7275 (pass_ipa_single_use::gate): Remove.
7277 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7279 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
7280 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
7281 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
7282 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
7283 permute_load, permute_store, adjust_extract, adjust_splat,
7284 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
7285 replace_swap_with_copy, dump_swap_insn_table,
7286 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
7287 recombine_lvx_pattern, recombine_stvx_pattern,
7288 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
7289 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
7290 to file rs6000-p8swap.c.
7291 * config/rs6000/rs6000-p8swap.c: New file.
7292 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
7293 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
7294 and rs6000*-*-* targets.
7296 2017-07-06 David Malcolm <dmalcolm@redhat.com>
7298 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
7300 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7302 * lto-wrapper.c (merge_and_complain): Do not merge
7303 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
7304 fsigned_zeros, ftrapping_math, fwrapv.
7305 (append_compiler_options): Do not track these options.
7306 (append_linker_options): Likewie
7308 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7310 * cgraphunit.c (cgraph_node::finalize_function): When
7311 !flag_toplevel_reorde set no_reorder flag.
7312 (varpool_node::finalize_decl): Likewise.
7313 (symbol_table::compile): Drop no toplevel reorder path.
7315 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7317 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
7318 edges; zero probability is not better than uninitialized.
7320 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
7322 * asan.h (asan_sanitize_allocas_p): Declare.
7323 * asan.c (asan_sanitize_allocas_p): New function.
7324 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
7325 (handle_builtin_alloca): Likewise.
7326 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
7327 if !asan_sanitize_allocas_p.
7328 * params.def (asan-instrument-allocas): Add new option.
7329 * params.h (ASAN_PROTECT_ALLOCAS): Define.
7330 * opts.c (common_handle_option): Disable allocas sanitization for
7333 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
7335 * asan.c: Include gimple-fold.h.
7336 (get_last_alloca_addr): New function.
7337 (handle_builtin_stackrestore): Likewise.
7338 (handle_builtin_alloca): Likewise.
7339 (asan_emit_allocas_unpoison): Likewise.
7340 (get_mem_refs_of_builtin_call): Add new parameter, remove const
7341 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
7342 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
7343 (instrument_builtin_call): Pass gimple iterator to
7344 get_mem_refs_of_builtin_call.
7345 (last_alloca_addr): New global.
7346 * asan.h (asan_emit_allocas_unpoison): Declare.
7347 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
7348 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
7349 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
7350 if function calls alloca.
7351 * gimple-fold.c (replace_call_with_value): Remove static keyword.
7352 * gimple-fold.h (replace_call_with_value): Declare.
7353 * internal-fn.c: Include asan.h.
7354 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
7355 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
7357 2017-07-06 David Malcolm <dmalcolm@redhat.com>
7359 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
7360 (C_SELFTEST_FLAGS): New.
7361 (CPP_SELFTEST_FLAGS): New.
7362 (SELFTEST_DEPS): New, from deps of s-selftest.
7363 (C_SELFTEST_DEPS): New, from deps of s-selftest.
7364 (CPP_SELFTEST_DEPS): New.
7365 (selftest): Add dependency on s-selftest-c++.
7366 (s-selftest): Rename to...
7367 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
7368 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
7369 than SELFTEST_FLAGS.
7370 (selftest-gdb): Rename to...
7371 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
7373 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
7374 (selftest-valgrind): Rename to...
7375 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
7377 (selftest-valgrind): Reintroduce as an alias for
7378 selftest-c-valgrind.
7379 (s-selftest-c++): New.
7380 (selftest-c++-gdb): New.
7381 (selftest-c++-valgrind): New.
7383 2017-07-06 Olivier Hainque <hainque@adacore.com>
7385 * gcc.c (process_command): When deciding if undefined variables
7386 should be ignored when processing specs, accept "gcc -v" as well.
7388 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7390 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
7391 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
7393 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
7395 * config/arm/arm-cpus.in (armv8-r): Add new entry.
7396 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
7397 * config/arm/arm-tables.opt: Regenerate.
7398 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
7400 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
7402 2017-07-06 Carl Love <cel@us.ibm.com>
7404 * ChangeLog: Clean up from mid air collision
7406 2017-07-06 Carl Love <cel@us.ibm.com>
7408 * config/rs6000/rs6000-c.c: Add support for built-in functions
7409 vector signed int vec_subc (vector signed int, vector signed int);
7410 vector signed __int128 vec_subc (vector signed __int128,
7411 vector signed __int128);
7412 vector unsigned __int128 vec_subc (vector unsigned __int128,
7413 vector unsigned __int128);
7414 vector signed int vec_sube (vector signed int, vector signed int,
7416 vector unsigned int vec_sube (vector unsigned int,
7417 vector unsigned int,
7418 vector unsigned int);
7419 vector signed __int128 vec_sube (vector signed __int128,
7420 vector signed __int128,
7421 vector signed__int128);
7422 vector unsigned __int128 vec_sube (vector unsigned __int128,
7423 vector unsigned __int128,
7424 vector unsigned __int128);
7425 vector signed int vec_subec (vector signed int, vector signed int,
7427 vector unsigned int vec_subec (vector unsigned int,
7428 vector unsigned int,
7429 vector unsigned int);
7430 vector signed __int128 vec_subec (vector signed __int128,
7431 vector signed __int128,
7432 vector signed__int128);
7433 vector unsigned __int128 vec_subec (vector unsigned __int128,
7434 vector unsigned __int128,
7435 vector unsigned __int128);
7436 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
7437 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
7438 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
7439 BU_ALTIVEC_OVERLOAD_X definitions.
7440 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
7441 * doc/extend.texi: Update the built-in documentation file for the new
7444 2017-07-06 David Malcolm <dmalcolm@redhat.com>
7447 * diagnostic-show-locus.c (layout::layout): Use start and finish
7448 spelling location for the start and finish of each range.
7449 * genmatch.c (linemap_client_expand_location_to_spelling_point):
7450 Add unused aspect param.
7451 * input.c (expand_location_1): Add "aspect" param, and use it
7452 to access the correct part of the location.
7453 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
7455 (expand_location_to_spelling_point): Likewise.
7456 (linemap_client_expand_location_to_spelling_point): Add "aspect"
7457 param, and pass it to expand_location_1.
7459 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
7461 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
7462 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
7463 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
7464 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
7465 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
7466 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
7467 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
7468 _mm_maskz_getmant_ss): New intrinsics.
7469 (__builtin_ia32_getexpss128_mask): Changed to ...
7470 __builtin_ia32_getexpss128_round ... this.
7471 (__builtin_ia32_getexpsd128_mask): Changed to ...
7472 __builtin_ia32_getexpsd128_round ... this.
7473 * config/i386/i386-builtin-types.def
7474 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
7475 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
7476 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
7477 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
7478 __builtin_ia32_getmantss_mask_round): New builtins.
7479 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
7480 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
7481 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
7482 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
7483 * config/i386/sse.md
7484 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
7485 avx512f_sgetexp<mode><mask_scalar_name>
7486 <round_saeonly_scalar_name> ... this.
7487 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
7488 %0, %1, %2<round_saeonly_op3>}): Changed to ...
7489 vgetexp<ssescalarmodesuffix>
7490 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7491 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
7492 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
7493 avx512f_vgetmant<mode><mask_scalar_name>
7494 <round_saeonly_scalar_name> ... this.
7495 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
7496 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
7497 vgetmant<ssescalarmodesuffix>
7498 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
7499 %0<mask_scalar_operand4>, %1, %2
7500 <round_saeonly_scalar_mask_op4>, %3} ... this.
7501 * config/i386/subst.md (mask_scalar_operand4,
7502 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
7503 round_saeonly_scalar_nimm_predicate): New subst attributes.
7505 2017-07-06 Julia Koval <julia.koval@intel.com>
7507 * config/i386/i386.c (ix86_erase_embedded_rounding):
7508 Remove code for old rounding pattern.
7510 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
7512 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
7514 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
7516 * doc/sourcebuild.texi (Test Directives, Variants of
7517 dg-require-support): Add documentation for dg-require-stack-check.
7519 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
7521 * config/i386/subst.md (mask_scalar, round_scalar,
7522 round_saeonly_scalar): New meta-templates.
7523 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
7524 round_scalar_mask_operand3, round_scalar_mask_op3,
7525 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
7526 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
7527 round_saeonly_scalar_constraint,
7528 round_saeonly_scalar_prefix): New subst attribute.
7529 * config/i386/sse.md
7530 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
7531 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
7532 <round_scalar_name> ... this.
7533 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
7534 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
7535 <round_scalar_name> ... this.
7536 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
7537 <sse>_vm<code><mode>3<mask_scalar_name>
7538 <round_saeonly_scalar_name> ... this.
7539 (v<plusminus_mnemonic><ssescalarmodesuffix>
7540 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
7541 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
7542 v<plusminus_mnemonic><ssescalarmodesuffix>
7543 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7544 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
7545 (v<multdiv_mnemonic><ssescalarmodesuffix>
7546 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
7547 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
7548 v<multdiv_mnemonic><ssescalarmodesuffix>
7549 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7550 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
7551 (v<maxmin_float><ssescalarmodesuffix>
7552 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
7553 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
7554 v<maxmin_float><ssescalarmodesuffix>
7555 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7556 %0<mask_scalar_operand3>, %1, %<iptr>2
7557 <round_saeonly_scalar_mask_op3>} ... this.
7559 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
7561 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
7562 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
7564 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
7565 Alan Hayward <alan.hayward@arm.com>
7566 David Sherwood <david.sherwood@arm.com>
7568 * combine.c (simplify_if_then_else): Remove "enum" before
7570 * compare-elim.c (can_eliminate_compare): Likewise.
7571 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
7573 (aarch64_lookup_simd_builtin_type): Likewise.
7574 (aarch64_simd_builtin_type): Likewise.
7575 (aarch64_init_simd_builtin_types): Likewise.
7576 (aarch64_simd_expand_args): Likewise.
7577 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
7579 (aarch64_reverse_mask): Likewise.
7580 (aarch64_simd_emit_reg_reg_move): Likewise.
7581 (aarch64_gen_adjusted_ldpstp): Likewise.
7582 (aarch64_ccmp_mode_to_code): Likewise.
7583 (aarch64_operands_ok_for_ldpstp): Likewise.
7584 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7585 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
7587 (aarch64_min_divisions_for_recip_mul): Likewise.
7588 (aarch64_reassociation_width): Likewise.
7589 (aarch64_get_condition_code_1): Likewise.
7590 (aarch64_simd_emit_reg_reg_move): Likewise.
7591 (aarch64_simd_attr_length_rglist): Likewise.
7592 (aarch64_reverse_mask): Likewise.
7593 (aarch64_operands_ok_for_ldpstp): Likewise.
7594 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7595 (aarch64_gen_adjusted_ldpstp): Likewise.
7596 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
7598 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
7599 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
7600 (arm_lookup_simd_builtin_type): Likewise.
7601 (arm_simd_builtin_type): Likewise.
7602 (arm_init_simd_builtin_types): Likewise.
7603 (arm_expand_builtin_args): Likewise.
7604 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
7605 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
7606 (ft32_setup_incoming_varargs): Likewise.
7607 (ft32_function_arg): Likewise.
7608 (ft32_function_arg_advance): Likewise.
7609 (ft32_pass_by_reference): Likewise.
7610 (ft32_arg_partial_bytes): Likewise.
7611 (ft32_valid_pointer_mode): Likewise.
7612 (ft32_addr_space_pointer_mode): Likewise.
7613 (ft32_addr_space_legitimate_address_p): Likewise.
7614 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
7616 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
7617 (ix86_emit_outlined_ms2sysv_restore): Likewise.
7618 (iamcu_alignment): Likewise.
7619 (canonicalize_vector_int_perm): Likewise.
7620 (ix86_noce_conversion_profitable_p): Likewise.
7621 (ix86_mpx_bound_mode): Likewise.
7622 (ix86_operands_ok_for_move_multiple): Likewise.
7623 * config/microblaze/microblaze-protos.h
7624 (microblaze_expand_conditional_branch_reg): Likewise.
7625 * config/microblaze/microblaze.c
7626 (microblaze_expand_conditional_branch_reg): Likewise.
7627 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
7629 (rs6000_reassociation_width): Likewise.
7630 (rs6000_invalid_binary_op): Likewise.
7631 (fusion_p9_p): Likewise.
7632 (emit_fusion_p9_load): Likewise.
7633 (emit_fusion_p9_store): Likewise.
7634 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
7636 (riscv_hard_regno_mode_ok_p): Likewise.
7637 (riscv_address_insns): Likewise.
7638 (riscv_split_symbol): Likewise.
7639 (riscv_legitimize_move): Likewise.
7640 (riscv_function_value): Likewise.
7641 (riscv_hard_regno_nregs): Likewise.
7642 (riscv_expand_builtin): Likewise.
7643 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
7644 (riscv_build_integer): Likewise.
7645 (riscv_split_integer): Likewise.
7646 (riscv_legitimate_constant_p): Likewise.
7647 (riscv_cannot_force_const_mem): Likewise.
7648 (riscv_regno_mode_ok_for_base_p): Likewise.
7649 (riscv_valid_base_register_p): Likewise.
7650 (riscv_valid_offset_p): Likewise.
7651 (riscv_valid_lo_sum_p): Likewise.
7652 (riscv_classify_address): Likewise.
7653 (riscv_legitimate_address_p): Likewise.
7654 (riscv_address_insns): Likewise.
7655 (riscv_load_store_insns): Likewise.
7656 (riscv_force_binary): Likewise.
7657 (riscv_split_symbol): Likewise.
7658 (riscv_force_address): Likewise.
7659 (riscv_legitimize_address): Likewise.
7660 (riscv_move_integer): Likewise.
7661 (riscv_legitimize_const_move): Likewise.
7662 (riscv_legitimize_move): Likewise.
7663 (riscv_address_cost): Likewise.
7664 (riscv_subword): Likewise.
7665 (riscv_output_move): Likewise.
7666 (riscv_canonicalize_int_order_test): Likewise.
7667 (riscv_emit_int_order_test): Likewise.
7668 (riscv_function_arg_boundary): Likewise.
7669 (riscv_pass_mode_in_fpr_p): Likewise.
7670 (riscv_pass_fpr_single): Likewise.
7671 (riscv_pass_fpr_pair): Likewise.
7672 (riscv_get_arg_info): Likewise.
7673 (riscv_function_arg): Likewise.
7674 (riscv_function_arg_advance): Likewise.
7675 (riscv_arg_partial_bytes): Likewise.
7676 (riscv_function_value): Likewise.
7677 (riscv_pass_by_reference): Likewise.
7678 (riscv_setup_incoming_varargs): Likewise.
7679 (riscv_print_operand): Likewise.
7680 (riscv_elf_select_rtx_section): Likewise.
7681 (riscv_save_restore_reg): Likewise.
7682 (riscv_for_each_saved_reg): Likewise.
7683 (riscv_register_move_cost): Likewise.
7684 (riscv_hard_regno_mode_ok_p): Likewise.
7685 (riscv_hard_regno_nregs): Likewise.
7686 (riscv_class_max_nregs): Likewise.
7687 (riscv_memory_move_cost): Likewise.
7688 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
7689 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
7690 (rl78_addr_space_address_mode): Likewise.
7691 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7693 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7694 (rs6000_reassociation_width): Likewise.
7695 (rs6000_invalid_binary_op): Likewise.
7696 (fusion_p9_p): Likewise.
7697 (emit_fusion_p9_load): Likewise.
7698 (emit_fusion_p9_store): Likewise.
7699 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
7700 (ok_for_simple_move_operands): Likewise.
7701 (ok_for_simple_move_strict_operands): Likewise.
7702 (ok_for_simple_arith_logic_operands): Likewise.
7703 (visium_legitimize_reload_address): Likewise.
7704 (visium_select_cc_mode): Likewise.
7705 (output_cbranch): Likewise.
7706 (visium_split_double_move): Likewise.
7707 (visium_expand_copysign): Likewise.
7708 (visium_expand_int_cstore): Likewise.
7709 (visium_expand_fp_cstore): Likewise.
7710 * config/visium/visium.c (visium_pass_by_reference): Likewise.
7711 (visium_function_arg): Likewise.
7712 (visium_function_arg_advance): Likewise.
7713 (visium_libcall_value): Likewise.
7714 (visium_setup_incoming_varargs): Likewise.
7715 (visium_legitimate_constant_p): Likewise.
7716 (visium_legitimate_address_p): Likewise.
7717 (visium_legitimize_address): Likewise.
7718 (visium_secondary_reload): Likewise.
7719 (visium_register_move_cost): Likewise.
7720 (visium_memory_move_cost): Likewise.
7721 (prepare_move_operands): Likewise.
7722 (ok_for_simple_move_operands): Likewise.
7723 (ok_for_simple_move_strict_operands): Likewise.
7724 (ok_for_simple_arith_logic_operands): Likewise.
7725 (visium_function_value_1): Likewise.
7726 (rtx_ok_for_offset_p): Likewise.
7727 (visium_legitimize_reload_address): Likewise.
7728 (visium_split_double_move): Likewise.
7729 (visium_expand_copysign): Likewise.
7730 (visium_expand_int_cstore): Likewise.
7731 (visium_expand_fp_cstore): Likewise.
7732 (visium_split_cstore): Likewise.
7733 (visium_select_cc_mode): Likewise.
7734 (visium_split_cbranch): Likewise.
7735 (output_cbranch): Likewise.
7736 (visium_print_operand_address): Likewise.
7737 * expmed.c (flip_storage_order): Likewise.
7738 * expmed.h (emit_cstore): Likewise.
7739 (flip_storage_order): Likewise.
7740 * genrecog.c (validate_pattern): Likewise.
7741 * hsa-gen.c (gen_hsa_addr): Likewise.
7742 * internal-fn.c (expand_arith_overflow): Likewise.
7743 * ira-color.c (allocno_copy_cost_saving): Likewise.
7744 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7745 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
7746 (process_invariant_for_inheritance): Likewise.
7747 * lra-eliminations.c (move_plus_up): Likewise.
7748 * omp-low.c (lower_oacc_reductions): Likewise.
7749 * simplify-rtx.c (simplify_subreg): Likewise.
7750 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
7751 (TARGET_CHKP_BOUND_MODE): Likewise..
7752 * targhooks.c (default_chkp_bound_mode): Likewise.
7753 (default_setup_incoming_vararg_bounds): Likewise.
7754 * targhooks.h (default_chkp_bound_mode): Likewise.
7755 (default_setup_incoming_vararg_bounds): Likewise.
7756 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
7757 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
7758 (have_whole_vector_shift): Likewise.
7759 * tree-vect-stmts.c (vectorizable_load): Likewise.
7760 * doc/tm.texi: Regenerate.
7762 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
7764 Graceful degrade if Binutils PR21472 is not available.
7767 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
7768 .rodata in flash test fails.
7769 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
7770 * confgure: Regenerate.
7771 * config.in: Regenerate.
7772 * config/avr/avr.c (avr_asm_named_section)
7773 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
7774 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
7775 (avr_asm_init_sections): Same.
7777 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7779 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
7780 (fma<VH:mode>4_intrinsic): Likewise.
7781 (*fmsub<VCVTF:mode>4): Likewise.
7782 (*fmsub<VH:mode>4_intrinsic): Likewise.
7784 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
7787 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
7788 Don't depend on "optimize > 0".
7789 (out_movhi_r_mr, out_movqi_mr_r): Same.
7790 (out_movhi_mr_r, out_movqi_r_mr): Same.
7791 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
7792 io_address_operand on "optimize > 0".
7794 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7796 * tree-loop-distribution.c: Add general explanantion on the pass.
7797 (generate_loops_for_partition): Mark distributed loop.
7798 (pg_add_dependence_edges): New parameter. Handle alias data
7799 dependence specially and record it in the parameter if asked.
7800 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
7801 (init_partition_graph_vertices, add_partition_graph_edge): New.
7802 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
7803 (free_partition_graph_vdata, build_partition_graph): New.
7804 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
7805 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
7806 (data_ref_segment_size, latch_dominated_by_data_ref): New.
7807 (compute_alias_check_pairs, version_loop_by_alias_check): New.
7808 (version_for_distribution_p, finalize_partitions): New.
7809 (distribute_loop): Handle alias data dependence specially. Factor
7810 out loop fusion code as functions and call these functions.
7812 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7814 * tree-loop-distribution.c (classify_partition): New parameter and
7815 better handle reduction statement.
7816 (rdg_build_partitions): Revise comment.
7817 (distribute_loop): Compute statements in all partitions and pass it
7818 to classify_partition.
7820 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7822 * tree-loop-distribution.c (enum partition_type): New.
7823 (struct partition): New field type.
7824 (partition_merge_into): Add parameter. Update partition type.
7825 (data_dep_in_cycle_p, update_type_for_merge): New functions.
7826 (build_rdg_partition_for_vertex): Compute partition type.
7827 (rdg_build_partitions): Dump partition type.
7828 (distribute_loop): Update calls to partition_merge_into.
7830 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7832 * tree-loop-distribution.c (struct ddr_hasher): New.
7833 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
7835 (classify_partition): Call get_data_dependence.
7836 (pg_add_dependence_edges): Ditto.
7837 (distribute_loop): Release data dependence hash table.
7839 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7841 * tree-loop-distribution.c (ref_base_address): Delete.
7842 (similar_memory_accesses): Rename ...
7843 (share_memory_accesses): ... to this. Check if partitions access
7844 the same memory reference.
7845 (distribute_loop): Call share_memory_accesses.
7847 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7849 * tree-loop-distribution.c (struct partition): New field recording
7851 (partition_alloc, partition_free): Init and release data refs.
7852 (partition_merge_into): Merge data refs.
7853 (build_rdg_partition_for_vertex): Collect data refs for partition.
7854 (pg_add_dependence_edges): Change parameters from vector to bitmap.
7856 (distribute_loop): Remve data refs from vertice data of partition
7859 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7861 * tree-loop-distribution.c (params.h): Include header file.
7862 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
7863 (datarefs_vec): New global var.
7864 (create_rdg_vertices): Use datarefs_vec directly.
7865 (free_rdg): Don't free data references.
7866 (build_rdg): Update use. Don't free data references.
7867 (distribute_loop): Compute global variable for data references.
7868 Bail out if there are too many data references.
7870 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7872 * tree-loop-distribution.c (loop_nest): New global var.
7873 (build_rdg): Use loop directly, rather than loop nest.
7874 (pg_add_dependence_edges): Remove loop nest parameter. Use global
7876 (distribute_loop): Compute global variable loop nest. Update use.
7878 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7880 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
7881 (partition_merge_into): New parameter. Dump reason for fusion.
7882 (distribute_loop): Update use of partition_merge_into.
7884 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7886 * tree-loop-distribution.c (bb_top_order_index): New.
7887 (bb_top_order_index_size, bb_top_order_cmp): New.
7888 (stmts_from_loop): Use topological order.
7889 (pass_loop_distribution::execute): Compute and release topological
7890 order for basic blocks.
7892 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7894 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
7897 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7899 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
7900 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
7901 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
7902 * internal-fn.def (LOOP_DIST_ALIAS): New.
7903 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
7904 (fold_loop_internal_call): ... this.
7905 (vect_loop_dist_alias_call): New function.
7906 (set_uid_loop_bbs): Call fold_loop_internal_call.
7907 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
7910 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7913 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
7914 Require dead FLAGS_REG at the beginning of a peephole.
7916 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7919 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
7920 arguments in the call to __builtin_ia32_sbb_u32.
7921 (_subborrow_u64): Swap _X and _Y arguments in the call to
7922 __builtin_ia32_sbb_u64.
7924 2017-07-04 Jakub Jelinek <jakub@redhat.com>
7927 * tree-vrp.c (compare_assert_loc): Turn into a function template
7928 with stable template parameter. Only test if a->e is NULL,
7929 !a->e == !b->e has been verified already. Use e == NULL or
7930 e != NULL instead of e or ! e tests. If stable is true, don't use
7931 iterative_hash_expr, on the other side allow a or b or both NULL
7932 and sort the NULLs last.
7933 (process_assert_insertions): Sort using compare_assert_loc<false>
7934 instead of compare_assert_loc, later sort using
7935 compare_assert_loc<true> before calling process_assert_insertions_for
7936 in a loop. Use break instead of continue once seen NULL pointer.
7938 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
7940 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7941 Cortex-R7 and Cortex-R8 processors.
7943 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7945 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
7946 uninitialized while src is not.
7948 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
7950 * common/config/arm/arm-common.c: Adjust include path for
7952 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
7953 (arm-cpu.h): Create in build directory. Adjust dependency rules.
7954 (arm-cpu-data.h): Likewise.
7955 (arm-cpu-cdata.h): Likewise.
7956 * config/arm/arm-cpu.h: Delete.
7957 * config/arm/arm-cpu-cdata.h: Delete.
7958 * config/arm/arm-cpu-data.h: Delete.
7960 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
7962 * config/arm/arm-cpus.in (cortex-a55): New.
7963 (cortex-a75): Likewise.
7964 (cortex-a75.cortex-a55): Likewise.
7965 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
7967 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
7968 * config/arm/arm-cpu-cdata.h: Regenerate.
7969 * config/arm/arm-cpu-data.h: Regenerate.
7970 * config/arm/arm-cpu.h: Regenerate.
7971 * config/arm/arm-tables.opt: Regenerate.
7972 * config/arm/arm-tune.md: Regenerate.
7974 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7976 * haifa-sched.c (sched_create_recovery_edges): Update profile.
7978 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7980 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
7983 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
7985 PR tree-optimization/81292
7986 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
7987 full_string_p, also call adjust_related_strinfos if the adjustment
7988 is simple, otherwise invalidate related strinfos.
7990 2017-07-04 Martin Liska <mliska@suse.cz>
7993 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
7994 newly created variable as DECL_IGNORED_P.
7996 2017-07-04 Martin Liska <mliska@suse.cz>
7999 * ipa-inline.c (inline_small_functions):
8000 Use xstrdup_for_dump.
8002 2017-07-04 Tom de Vries <tom@codesourcery.com>
8004 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
8006 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
8009 * config/darwin.c (darwin_function_switched_text_sections):
8012 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
8014 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
8016 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
8018 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
8020 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8022 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
8023 min_profitable_iters, and th as inclusive lower bounds.
8024 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
8025 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
8026 for min_profitable_iters and min_profitable_estimate.
8027 (vect_transform_loop): Treat th as an inclusive lower bound.
8028 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
8030 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
8033 * config/darwin.c (darwin_function_switched_text_sections):
8034 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
8035 in two pieces, and suppress the use of buf.
8037 2017-07-03 Nathan Sidwell <nathan@acm.org>
8039 * hash-table.h (hash_table_mod1): Fix indentation.
8041 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8044 * predict.c (force_edge_cold): Be more careful about propagation
8046 * profile-count.h (profile_probability::guessed,
8047 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
8049 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
8051 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
8053 * doc/invoke.texi (rcpc architecture extension): Document it.
8055 2017-07-03 Richard Biener <rguenther@suse.de>
8057 PR tree-optimization/60510
8058 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
8059 the scalar reduction PHI and use it.
8060 (vectorizable_reduction): Properly guard the single_defuse_cycle
8061 path for non-SLP reduction chains where we cannot use it.
8062 Rework reduc_def/index and vector type deduction. Rework
8063 vector operand gathering during reduction op code-gen.
8064 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
8065 chains dissolve the chain and leave it to non-SLP reduction
8068 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8070 * tree-data-ref.h (dr_alignment): Declare.
8071 * tree-data-ref.c (dr_alignment): New function.
8072 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
8073 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
8075 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
8077 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8079 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
8080 and base_misalignment fields.
8081 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
8082 * tree-data-ref.c: Include builtins.h.
8083 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
8084 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
8085 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
8086 * tree-vect-data-refs.c: Include tree-cfg.h.
8087 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
8088 fields instead of calculating an alignment here.
8089 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
8090 innermost_loop_behavior fields.
8092 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8094 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
8096 (DR_STEP_ALIGNMENT): New macro.
8097 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
8098 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
8099 (create_data_ref): Print it.
8100 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
8101 to tell whether the step preserves vector (mis)alignment.
8102 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
8103 Move the check for an integer step and generalise to all INTEGER_CST.
8104 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
8105 Print the outer step alignment.
8107 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8109 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
8110 with offset_alignment.
8111 (DR_ALIGNED_TO): Delete.
8112 (DR_OFFSET_ALIGNMENT): New macro.
8113 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
8114 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
8115 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
8116 (create_data_ref): Likewise.
8117 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
8118 (vect_analyze_data_refs): Likewise.
8119 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
8120 creating dummy innermost behavior.
8122 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8124 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
8125 with a "innermost_loop_behavior *" and refeence tree.
8126 * tree-data-ref.c (dr_analyze_innermost): Likewise.
8127 (create_data_ref): Update call accordingly.
8128 * tree-predcom.c (find_looparound_phi): Likewise.
8130 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8132 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
8133 fields with dr_wrt_vec_loop.
8134 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
8135 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
8136 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
8137 (vect_dr_behavior): New function.
8138 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
8139 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
8140 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
8141 track whether the step preserves the misalignment.
8142 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
8143 Use vect_dr_behavior.
8144 (vect_setup_realignment): Update call accordingly.
8145 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
8146 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
8147 call to vect_create_addr_base_for_vector_ref.
8148 (vect_create_cond_for_align_checks): Likewise.
8149 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
8150 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
8151 (vect_recog_mask_conversion_pattern): Likewise.
8152 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
8153 (new_stmt_vec_info): Remove redundant zeroing.
8155 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
8157 * common/config/arm/arm-common.c (arm_be8_option): New function.
8158 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
8159 (ISA_ARMv6): Add isa_bit_be8.
8160 * config/arm/arm.h (arm_be8_option): Add prototype.
8161 (BE8_SPEC_FUNCTION): New define.
8162 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
8163 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
8164 (mlittle-endian): Similarly.
8165 (mbe8, mbe32): New options.
8166 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
8167 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
8169 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8171 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
8173 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8175 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
8176 (cleanup_tree_cfg_bb): Use it.
8177 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
8179 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
8181 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8184 * loop-doloop.c (add_test): Update profile.
8186 2017-07-03 Martin Liska <mliska@suse.cz>
8189 * sanopt.c (rewrite_usage_of_param): New function.
8190 (sanitize_rewrite_addressable_params): Likewise.
8191 (pass_sanopt::execute): Call rewrite_usage_of_param.
8193 2017-07-03 Richard Biener <rguenther@suse.de>
8195 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
8196 back to using VIEW_CONVERT_EXPR.
8198 2017-07-03 Martin Liska <mliska@suse.cz>
8201 * doc/extend.texi: Document when a resolver function is
8202 generated for target_clones.
8204 2017-07-03 Martin Liska <mliska@suse.cz>
8206 * asan.c (asan_emit_stack_protection): Unpoison just red zones
8207 and shadow memory of auto variables which are subject of
8208 use-after-scope sanitization.
8209 (asan_expand_mark_ifn): Add do set only when is_poison.
8211 2016-07-03 Richard Biener <rguenther@suse.de>
8213 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
8215 (vect_force_simple_reduction): Record reduction def -> phi mapping.
8216 (vectorizable_reduction): Perform reduction PHI creation when
8217 visiting a reduction PHI and adjust and simplify code generation
8218 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
8219 (vect_transform_loop): Visit reduction PHIs.
8220 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
8221 defs into the SLP tree.
8222 (vect_build_slp_tree): Reduction defs terminate the recursion.
8223 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
8225 (vect_get_vec_defs_for_stmt_copy): Export.
8226 (vect_get_vec_defs): Likewise.
8227 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
8229 (vect_get_vec_defs_for_stmt_copy): Declare.
8230 (vect_get_vec_defs): Likewise.
8232 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8234 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
8235 parameter with a "loop" parameter and use it instead of the
8236 loop containing DR_STMT. Don't check simple_iv when doing
8237 BB analysis. Describe the two analysis modes in the comment.
8239 2017-07-03 Tom de Vries <tom@codesourcery.com>
8241 PR tree-optimization/69468
8242 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
8243 (find_same_succ_bb): Handle ignore_edge_flags.
8245 2017-07-03 Tom de Vries <tom@codesourcery.com>
8247 PR tree-optimization/81192
8248 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
8250 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
8252 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
8254 2017-07-03 Tom de Vries <tom@codesourcery.com>
8256 PR tree-optimization/81192
8257 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
8258 BB_SAME_SUCC (bb) == NULL.
8260 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8262 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
8265 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8267 * dumpfile.c: Include profile-count.h
8268 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
8270 (insert_cond_bb): Update profile.
8271 * tree-cfg.h (insert_cond_bb): Update prototype.
8272 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
8273 * tree-dump.c: Do not include tree-cfg.
8275 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8277 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
8279 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8281 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
8284 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8286 * tree-complex.c (expand_complex_div_wide): update profile.
8288 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8289 Alan Hayward <alan.hayward@arm.com>
8290 David Sherwood <david.sherwood@arm.com>
8292 * Makefile.in (MACHMODE_H): Remove insn-modes.h
8293 (CORETYPES_H): New define.
8294 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
8295 (insn-modes-inline.h, s-modes-inline-h): New rules.
8296 (generated_files): Add insn-modes-inline.h.
8297 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
8298 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
8299 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
8300 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
8301 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
8302 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
8303 (build/gencodes.o, build/genconditions.o): Likewise.
8304 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
8305 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
8306 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
8307 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
8308 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
8309 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
8310 * coretypes.h: Include everything up to real.h for generators.
8311 Include insn-modes.h first. Include wide-int-print.h after
8312 wide-int.h. Include insn-modes-inline.h and then machmode.h.
8313 * machmode.h: Don't include insn-modes.h here.
8314 * function-tests.c: Remove includes of signop.h, machmode.h,
8315 double-int.h and wide-int.h.
8317 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
8319 * optc-save-gen.awk: Likewise.
8320 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
8321 * godump.c: Remove include of wide-int-print.h.
8322 * pretty-print.h: Likewise.
8323 * wide-int-print.cc: Likewise.
8324 * wide-int.cc: Likewise.
8325 * hash-map-tests.c: Remove include of signop.h.
8326 * hash-set-tests.c: Likewise.
8327 * rtl-tests.c: Likewise.
8328 * mkconfig.sh: Remove include of machmode.h.
8329 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
8331 (emit_insn_modes_inline_h): ...this new function. Emit the code
8332 into an insn-modes-inline.h header file, adding appropriate
8333 include guards and end comments.
8334 (emit_insn_modes_c_header): Remove include of machmode.h.
8335 (emit_min_insn_modes_c_header): Include coretypes.h rather than
8337 (main): Handle -i flag and call emit_insn_modes_inline_h when
8340 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8342 * tree-ssa-strlen.c (strinfo): Rename the length field to
8343 nonzero_chars. Add a full_string_p field.
8344 (compare_nonzero_chars, zero_length_string_p): New functions.
8345 (get_addr_stridx): Add an offset_out parameter.
8346 Use compare_nonzero_chars.
8347 (get_stridx): Update accordingly. Use compare_nonzero_chars.
8348 (new_strinfo): Update after above changes to strinfo.
8349 (set_endptr_and_length): Set full_string_p.
8350 (get_string_length): Update after above changes to strinfo.
8351 (unshare_strinfo): Update call to new_strinfo.
8352 (maybe_invalidate): Likewise.
8353 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
8354 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
8355 as a uhwi instead of an shwi. Update after above changes to
8356 strinfo and new_strinfo.
8357 (zero_length_string): Assert that chainsi contains full strings.
8358 Use zero_length_string_p. Update call to new_strinfo.
8359 (adjust_related_strinfos): Update after above changes to strinfo.
8360 Copy full_string_p from origsi.
8361 (adjust_last_stmt): Use zero_length_string_p.
8362 (handle_builtin_strlen): Update after above changes to strinfo and
8363 new_strinfo. Install the lhs as the string length if the previous
8364 entry didn't describe a full string.
8365 (handle_builtin_strchr): Update after above changes to strinfo
8367 (handle_builtin_strcpy): Likewise.
8368 (handle_builtin_strcat): Likewise.
8369 (handle_builtin_malloc): Likewise.
8370 (handle_pointer_plus): Likewise.
8371 (handle_builtin_memcpy): Likewise. Track nonzero characters
8372 that aren't necessarily followed by a nul terminator.
8373 (handle_char_store): Likewise.
8375 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8377 PR tree-optimization/80769
8378 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
8379 for malloc and calloc. Document the new invariant that all related
8380 strinfos have delayed lengths or none do.
8381 (verify_related_strinfos): Move earlier in file.
8382 (set_endptr_and_length): New function, split out from...
8383 (get_string_length): ...here. Also set the lengths of related
8385 (zero_length_string): Assert that chainsi has known (rather than
8387 (adjust_related_strinfos): Likewise.
8389 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8391 PR tree-optimization/81136
8392 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
8393 assert that two references with the same misalignment have the same
8394 compile-time misalignment if those compile-time misalignments
8397 2017-07-01 Andi Kleen <ak@linux.intel.com>
8399 * print-tree.c (print_node): Print all attributes.
8401 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8403 * cfg.c (scale_bbs_frequencies): New function.
8404 * cfg.h (scale_bbs_frequencies): Declare it.
8405 * cfgloopanal.c (single_likely_exit): Cleanup.
8406 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
8408 (scale_loop_profile): Likewise.
8409 (loop_version): Likewise.
8410 (create_empty_loop_on_edge): Update.
8411 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
8412 scale_loop_frequencies, scale_loop_profile, loopify,
8413 loop_version): Update prototypes.
8414 * modulo-sched.c (sms_schedule): Update.
8415 * predict.c (unlikely_executed_edge_p): Also check probability.
8416 (probably_never_executed_edge_p): Fix typo.
8417 * tree-if-conv.c (version_loop_for_if_conversion): Update.
8418 * tree-parloops.c (gen_parallel_loop): Update.
8419 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
8420 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8421 * tree-ssa-loop-split.c (split_loop): Update.
8422 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
8423 * tree-vect-loop-manip.c (vect_do_peeling): Update.
8424 (vect_loop_versioning): Update.
8425 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8427 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8429 * trans-mem.c (split_bb_make_tm_edge): Update profile.
8431 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8433 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
8434 to keep profile consistent.
8436 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8438 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
8439 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
8440 * profile-count.h (max_safe_multiplier): Make unsigned.
8441 (profile_count::guessed_zero): New.
8443 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8445 * bb-reorder.c (fix_up_crossing_landing_pad,
8446 fix_crossing_conditional_branches): Use make_single_succ_edge
8447 to keep profile consistent.
8449 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8451 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
8454 2017-07-01 Jakub Jelinek <jakub@redhat.com>
8457 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
8458 the right scopes, make sure cond_jump isn't preserved between multiple
8459 iterations. Search for fallthru edge whenever there are 3+ edges and
8460 use find_fallthru_edge for it.
8462 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8464 Patch by Alexander Monakov <amonakov@ispras.ru>
8465 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
8466 probabilities consistently.
8468 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8470 * pa.c (pa_expand_compare_and_swap_loop): Update call of
8471 emit_cmp_and_jump_insns.
8473 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8476 * tree-inline.c (expand_call_inline): Combine profile statuses.
8478 2017-06-30 Andrew Pinski <apinski@cavium.com>
8480 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
8481 fold_stmt returned true.
8483 2017-06-30 Nathan Sidwell <nathan@acm.org>
8485 * ggc.h (empty_string): Delete.
8486 * cfgexpand.c (expand_asm_stmt): Use plain "".
8487 * optabs.c (expand_asm_memory_barrier): Likewise.
8488 * stringpool.c (empty_string): Delete.
8489 (digit_vector, digit_string): Delete.
8490 (ggc_alloc_string): Use plain "", don't optimize single digit
8491 strings. Use ggc_alloc_atomic.
8493 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
8495 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
8496 comparison set and one other set, use the cost of the non-comparison
8499 2017-06-30 Nathan Sidwell <nathan@acm.org>
8501 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
8504 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
8506 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
8507 loops. Remove now unneeded calls to gimple_switch_set_label() that
8508 just set removed labels to NULL_TREE.
8510 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
8512 * tree-ssanames.c (set_range_info_raw): Abstract from ...
8513 (set_range_info): ...here. Only call set_range_info_raw if domain
8515 (set_nonzero_bits): Call set_range_info_raw.
8516 * tree-ssanames.h (set_range_info_raw): New.
8518 2017-06-30 Jakub Jelinek <jakub@redhat.com>
8521 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
8522 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
8523 of nonimmediate_operand and <store_mask_constraint> instead of m
8524 for the input operand. For V8FI iterator, always split if input
8525 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
8526 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
8527 <store_mask_predicate> instead of register_operand and
8528 <store_mask_constraint> instead of v for the input operand. Make
8529 sure both operands aren't MEMs for if not <mask_applied>.
8531 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
8533 * lto-wrapper.c (copy_file) Close both file descriptors before
8536 2017-06-30 Martin Liska <mliska@suse.cz>
8539 * multiple_target.c (create_dispatcher_calls): Make ifunc
8540 also for function that don't have calls or are not referenced.
8542 2017-06-30 Richard Biener <rguenther@suse.de>
8544 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
8545 analyze the first scalar stmt. Move vector type computation
8546 for the BB case here from ...
8547 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
8548 live operation processing in the SLP case properly.
8550 2017-06-30 Richard Biener <rguenther@suse.de>
8552 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
8554 2017-06-30 Martin Liska <mliska@suse.cz>
8557 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
8558 before BUILT_IN_UNWIND_RESUME when ASAN is used.
8560 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
8562 * doc/invoke.texi (AArch64): Add missing options and remove redundant
8565 2017-06-30 Richard Biener <rguenther@suse.de>
8567 PR tree-optimization/81249
8568 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
8569 condition reduction result to original scalar type.
8571 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8573 * profile-count.h (enum profile_quality): Fix typos and whitespace
8576 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8578 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
8579 type for branch probabilities.
8581 2017-06-29 Julian Brown <julian@codesourcery.com>
8582 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8584 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
8585 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
8586 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
8587 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
8589 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8591 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
8592 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
8593 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
8594 CC usage from generic code to here.
8595 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
8596 CC usage into the target macros.
8598 2017-06-29 Maya Rashish <coypu@sdf.org>
8600 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
8603 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8605 * arm/arm-builtins.c: Include profile-count.h
8606 * except.c (sjlj_emit_function_enter): Use
8607 profile_probability::unlikely.
8609 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8611 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
8612 and tocrel_offset be pointer args rather than implicitly using
8614 (legitimate_constant_pool_address_p, rs6000_emit_move,
8615 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
8616 tocrel_offset and use in toc_relative_expr_p call.
8617 (print_operand, print_operand_address): Use static tocrel_base_oac
8618 and tocrel_offset_oac.
8619 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
8622 2017-06-29 Maya Rashish <coypu@sdf.org>
8624 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
8626 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
8628 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
8629 objects, take into account only the alignment of 'op0' and 'mode1' if
8632 2017-06-29 Steve Ellcey <sellcey@cavium.com>
8634 * ccmp.c (ccmp_tree_comparison_p): New function.
8635 (ccmp_candidate_p): Update to use above function.
8636 (get_compare_parts): New function.
8637 (expand_ccmp_next): Update to use new functions.
8638 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
8640 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
8641 take mode as argument.
8642 * ccmp.h (expand_ccmp_expr): Add mode as argument.
8643 * expr.c (expand_expr_real_1): Pass mode as argument.
8645 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
8647 * combine.c (combine_instructions): Print insns to dump_file, together
8650 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8652 * asan.c (asan_emit_stack_protection): Update.
8653 (create_cond_insert_point): Update.
8654 * auto-profile.c (afdo_propagate_circuit): Update.
8655 * basic-block.h (struct edge_def): Turn probability to
8656 profile_probability.
8657 (EDGE_FREQUENCY): Update.
8658 * bb-reorder.c (find_traces_1_round): Update.
8659 (better_edge_p): Update.
8660 (sanitize_hot_paths): Update.
8661 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
8662 (make_single_succ_edge): Update.
8663 (check_bb_profile): Update.
8664 (dump_edge_info): Update.
8665 (update_bb_profile_for_threading): Update.
8666 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
8667 probabilitycount to 0.
8668 * cfgbuild.c (compute_outgoing_frequencies): Update.
8669 * cfgcleanup.c (try_forward_edges): Update.
8670 (outgoing_edges_match): Update.
8671 (try_crossjump_to_edge): Update.
8672 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
8673 (expand_gimple_tailcall): Update.
8674 (construct_init_block): Use make_single_succ_edge.
8675 (construct_exit_block): Use make_single_succ_edge.
8676 * cfghooks.c (verify_flow_info): Update.
8677 (redirect_edge_succ_nodup): Update.
8678 (split_edge): Update.
8679 (account_profile_record): Update.
8680 * cfgloopanal.c (single_likely_exit): Update.
8681 * cfgloopmanip.c (scale_loop_profile): Update.
8682 (set_zero_probability): Remove.
8683 (duplicate_loop_to_header_edge): Update.
8684 * cfgloopmanip.h (loop_version): Update prototype.
8685 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8686 (force_nonfallthru_and_redirect): Update.
8687 (update_br_prob_note): Update.
8688 (rtl_verify_edges): Update.
8689 (purge_dead_edges): Update.
8690 (rtl_lv_add_condition_to_bb): Update.
8691 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
8692 * cgraphunit.c (init_lowered_empty_function): Update.
8693 (cgraph_node::expand_thunk): Update.
8694 * cilk-common.c: Include profile-count.h
8695 * dojump.c (inv): Remove.
8696 (jumpifnot): Update.
8697 (jumpifnot_1): Update.
8698 (do_jump_1): Update.
8700 (do_jump_by_parts_greater_rtx): Update.
8701 (do_compare_rtx_and_jump): Update.
8702 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
8703 do_jump_1. do_compare_rtx_and_jump): Update prototype.
8704 * dwarf2cfi.c: Include profile-count.h
8705 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
8706 (sjlj_emit_dispatch_table): Likewise.
8707 * explow.c: Include profile-count.h
8708 * expmed.c (emit_store_flag_force): Update.
8709 (do_cmp_and_jump): Update.
8710 * expr.c (compare_by_pieces_d::generate): Update.
8711 (compare_by_pieces_d::finish_mode): Update.
8712 (emit_block_move_via_loop): Update.
8713 (store_expr_with_bounds): Update.
8714 (store_constructor): Update.
8715 (expand_expr_real_2): Update.
8716 (expand_expr_real_1): Update.
8717 * expr.h (try_casesi, try_tablejump): Update prototypes.
8718 * gimple-pretty-print.c (dump_probability): Update.
8719 (dump_profile): New.
8720 (dump_gimple_label): Update.
8721 (dump_gimple_bb_header): Update.
8722 * graph.c (draw_cfg_node_succ_edges): Update.
8723 * hsa-gen.c (convert_switch_statements): Update.
8724 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
8725 (find_if_case_1): Update.
8726 (find_if_case_2): Update.
8727 * internal-fn.c (expand_arith_overflow_result_store): Update.
8728 (expand_addsub_overflow): Update.
8729 (expand_neg_overflow): Update.
8730 (expand_mul_overflow): Update.
8731 (expand_vector_ubsan_overflow): Update.
8732 * ipa-cp.c (good_cloning_opportunity_p): Update.
8733 * ipa-split.c (split_function): Use make_single_succ_edge.
8734 * ipa-utils.c (ipa_merge_profiles): Update.
8735 * loop-doloop.c (add_test): Update.
8736 (doloop_modify): Update.
8737 * loop-unroll.c (compare_and_jump_seq): Update.
8738 (unroll_loop_runtime_iterations): Update.
8739 * lra-constraints.c (lra_inheritance): Update.
8740 * lto-streamer-in.c (input_cfg): Update.
8741 * lto-streamer-out.c (output_cfg): Update.
8742 * mcf.c (adjust_cfg_counts): Update.
8743 * modulo-sched.c (sms_schedule): Update.
8744 * omp-expand.c (expand_omp_for_init_counts): Update.
8745 (extract_omp_for_update_vars): Update.
8746 (expand_omp_ordered_sink): Update.
8747 (expand_omp_for_ordered_loops): Update.
8748 (expand_omp_for_generic): Update.
8749 (expand_omp_for_static_nochunk): Update.
8750 (expand_omp_for_static_chunk): Update.
8751 (expand_cilk_for): Update.
8752 (expand_omp_simd): Update.
8753 (expand_omp_taskloop_for_outer): Update.
8754 (expand_omp_taskloop_for_inner): Update.
8755 * omp-simd-clone.c (simd_clone_adjust): Update.
8756 * optabs.c (expand_doubleword_shift): Update.
8757 (expand_abs): Update.
8758 (emit_cmp_and_jump_insn_1): Update.
8759 (expand_compare_and_swap_loop): Update.
8760 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
8761 * predict.c (predictable_edge_p): Update.
8762 (edge_probability_reliable_p): Update.
8763 (set_even_probabilities): Update.
8764 (combine_predictions_for_insn): Update.
8765 (combine_predictions_for_bb): Update.
8766 (propagate_freq): Update.
8767 (estimate_bb_frequencies): Update.
8768 (force_edge_cold): Update.
8769 * profile-count.c (profile_count::dump): Add missing space into dump.
8770 (profile_count::debug): Add newline.
8771 (profile_count::differs_from_p): Explicitly convert to unsigned.
8772 (profile_count::stream_in): Update.
8773 (profile_probability::dump): New member function.
8774 (profile_probability::debug): New member function.
8775 (profile_probability::differs_from_p): New member function.
8776 (profile_probability::differs_lot_from_p): New member function.
8777 (profile_probability::stream_in): New member function.
8778 (profile_probability::stream_out): New member function.
8779 * profile-count.h (profile_count_quality): Rename to ...
8780 (profile_quality): ... this one.
8781 (profile_probability): New.
8782 (profile_count): Update.
8783 * profile.c (compute_branch_probabilities): Update.
8784 * recog.c (peep2_attempt): Update.
8785 * sched-ebb.c (schedule_ebbs): Update.
8786 * sched-rgn.c (find_single_block_region): Update.
8787 (compute_dom_prob_ps): Update.
8788 (schedule_region): Update.
8789 * sel-sched-ir.c (compute_succs_info): Update.
8790 * stmt.c (struct case_node): Update.
8791 (do_jump_if_equal): Update.
8792 (get_outgoing_edge_probs): Update.
8793 (conditional_probability): Update.
8794 (emit_case_dispatch_table): Update.
8795 (expand_case): Update.
8796 (expand_sjlj_dispatch_table): Update.
8797 (emit_case_nodes): Update.
8798 * targhooks.c: Update.
8799 * tracer.c (better_p): Update.
8800 (find_best_successor): Update.
8801 * trans-mem.c (expand_transaction): Update.
8802 * tree-call-cdce.c: Update.
8803 * tree-cfg.c (gimple_split_edge): Upate.
8804 (move_sese_region_to_fn): Upate.
8805 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
8806 * tree-eh.c (lower_resx): Upate.
8807 (cleanup_empty_eh_move_lp): Upate.
8808 * tree-if-conv.c (version_loop_for_if_conversion): Update.
8809 * tree-inline.c (copy_edges_for_bb): Update.
8810 (copy_cfg_body): Update.
8811 * tree-parloops.c (gen_parallel_loop): Update.
8812 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
8813 (gimple_gen_time_profiler): Update.
8814 * tree-ssa-dce.c (remove_dead_stmt): Update.
8815 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8816 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8817 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8818 (unloop_loops): Update.
8819 (try_peel_loop): Update.
8820 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8821 * tree-ssa-loop-split.c (connect_loops): Update.
8822 (split_loop): Update.
8823 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
8824 (hoist_guard): Update.
8825 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8826 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8827 (value_replacement): Update.
8828 * tree-ssa-reassoc.c (branch_fixup): Update.
8829 * tree-ssa-tail-merge.c (replace_block_by): Update.
8830 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8831 (create_edge_and_update_destination_phis): Update.
8832 (compute_path_counts): Update.
8833 (recompute_probabilities): Update.
8834 (update_joiner_offpath_counts): Update.
8835 (freqs_to_counts_path): Update.
8836 (duplicate_thread_path): Update.
8837 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8838 (struct switch_conv_info): Update.
8839 (gen_inbound_check): Update.
8840 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8841 (vect_do_peeling): Update.
8842 (vect_loop_versioning): Update.
8843 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8844 (optimize_mask_stores): Update.
8845 * ubsan.c (ubsan_expand_null_ifn): Update.
8846 * value-prof.c (gimple_divmod_fixed_value): Update.
8847 (gimple_divmod_fixed_value_transform): Update.
8848 (gimple_mod_pow2): Update.
8849 (gimple_mod_pow2_value_transform): Update.
8850 (gimple_mod_subtract): Update.
8851 (gimple_mod_subtract_transform): Update.
8852 (gimple_ic): Update.
8853 (gimple_stringop_fixed_value): Update.
8854 (gimple_stringops_transform): Update.
8855 * value-prof.h: Update.
8857 2017-06-29 Carl Love <cel@us.ibm.com>
8859 * config/rs6000/rs6000-c.c: Add support for built-in functions
8860 vector signed int vec_signed (vector float);
8861 vector signed long long vec_signed (vector double);
8862 vector signed int vec_signed2 (vector double, vector double);
8863 vector signed int vec_signede (vector double);
8864 vector signed int vec_signedo (vector double);
8865 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
8866 instruction generator.
8867 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8868 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
8869 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
8871 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
8872 vunsignede_v2df): Add define_expands.
8873 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
8874 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
8875 VEC_UNSIGNEDO): Add definitions.
8876 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8877 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
8878 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
8879 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
8880 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
8881 * config/rs6000/altivec.h (vec_signed, vec_signed2,
8882 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
8883 vec_unsignede, vec_unsignedo): Add builtin defines.
8884 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
8886 * doc/extend.texi: Update the built-in documentation file for the
8887 new built-in functions.
8889 2017-06-29 Richard Biener <rguenther@suse.de>
8891 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
8892 reduction chains to LOOP_VINFO_REDUCTIONS.
8893 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
8894 SLP reductions after processing reduction chains.
8896 2017-06-29 Nathan Sidwell <nathan@acm.org>
8898 * builtins.c (fold_builtin_FUNCTION): Use
8899 lang_hooks.decl_printable_name.
8901 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
8904 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
8905 with only one label.
8906 * stmt.c (expand_case): Assert NCASES is greater than one.
8908 2017-06-29 Richard Biener <rguenther@suse.de>
8910 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
8912 (group_case_labels): Likewise.
8913 (find_taken_edge): Push sanity checking on val to workers...
8914 (find_taken_edge_cond_expr): ... here
8915 (find_taken_edge_switch_expr): ... and here, handle cases
8916 with just a default label.
8917 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
8918 (group_case_labels): Likewise.
8919 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
8920 group_case_labels does anything cleanup the CFG again.
8922 2017-06-29 Bin Cheng <bin.cheng@arm.com>
8924 PR tree-optimization/81196
8925 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
8926 exit condition comparing two IVs.
8928 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
8930 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
8931 profile to the dummy entry at the end of the list of architectures.
8932 * config/arm/arm-cpu-cdata.h: Regenerated.
8934 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8935 Michael Collison <michael.collison@arm.com>
8938 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
8940 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
8941 (*aarch64_reg_<mode>3_minus_mask): New pattern.
8942 (*aarch64_<optab>_reg_di3_mask2): New pattern.
8943 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
8944 of shift when the shift amount is masked with constant equal to
8945 the size of the mode.
8946 * config/aarch64/predicates.md (subreg_lowpart_operator): New
8949 2017-06-29 Martin Liska <mliska@suse.cz>
8951 * config/i386/i386.opt: Change range from [1,5] to [0,5].
8953 2017-06-29 Yury Gribov <tetra2005@gmail.com>
8956 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
8959 (edge_growth_cache_entry::edge_growth_cache_entry): New
8961 (reset_edge_growth_cache): Update to use constructor.
8963 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8965 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
8966 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
8967 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
8969 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
8971 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
8972 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
8974 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
8976 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
8977 (*-linux-uclibc*): Add t-uclibc tmake_file.
8978 * config/t-musl: New.
8979 * config/t-uclibc: New.
8981 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
8983 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
8985 (gen_comm_data): Emit architectural setting of arch_prof.
8986 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
8988 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
8989 (armv8-m.base, armv8-m.main): Likewise.
8990 * arm-protos.h (arm_build_target): Add profile field.
8991 (arch_option): Likewise.
8992 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
8994 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
8995 arm_active_target.profile.
8997 2017-06-28 Richard Biener <rguenther@suse.de>
9000 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
9001 TYPE_OVERFLOW_WRAPS.
9002 * match.pd (negate_expr_p): Likewise.
9003 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
9004 fold_build2, not fold_binary.
9006 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
9008 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
9009 Convert memory address to Pmode.
9010 (aarch64_print_operand): Assert MEM operands are always Pmode.
9012 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
9015 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
9016 Remove redundant if.
9017 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
9018 * config/arm/aarch-common-protos.h
9019 (aarch_forward_to_shift_is_not_shifted_re): Remove.
9020 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
9022 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
9025 * multiple_target.c (create_dispatcher_calls): Set the default
9026 clone to be static, not public.
9028 2017-06-28 Richard Biener <rguenther@suse.de>
9030 * tree-vect-loop.c (vectorizable_reduction): Move special
9031 cond reduction IV var creation ...
9032 (vect_create_epilog_for_reduction): ... here. Remove induction_index
9033 parameter. Use STMT_VINFO_VECTYPE.
9034 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
9037 2017-06-28 Martin Liska <mliska@suse.cz>
9040 * ipa-visibility.c (non_local_p): Handle visibility.
9042 2017-06-28 Martin Liska <mliska@suse.cz>
9045 * common.opt: Add IntegerRange to various options.
9046 * opt-functions.awk (integer_range_info): New function.
9047 * optc-gen.awk: Add integer_range_info to cl_options struct.
9048 * opts-common.c (decode_cmdline_option): Handle
9049 CL_ERR_INT_RANGE_ARG.
9050 (cmdline_handle_error): Likewise.
9051 * opts.c (print_filtered_help): Show valid interval in
9052 when --help is provided.
9053 * opts.h (struct cl_option): Add range_min and range_max fields.
9054 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
9056 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
9058 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
9059 (x * C EQ/NE y * C): New transformation.
9061 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
9063 * genmultilib (combination_space): Accept '+' in option names.
9065 2017-06-28 Martin Liska <mliska@suse.cz>
9068 * asan.c (instrument_derefs): Bail out inner references
9069 that are hard register variables.
9071 2017-06-28 Jakub Jelinek <jakub@redhat.com>
9074 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
9075 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
9077 2017-06-28 Richard Biener <rguenther@suse.de>
9079 * tree-vectorizer.h (vect_get_vec_defs): Remove.
9080 (vect_get_slp_defs): Adjust.
9081 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
9083 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
9085 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9086 get_initial_defs_for_reduction instead of vect_get_vec_defs.
9087 (vectorizable_reduction): Adjust.
9088 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
9090 (vect_get_slp_defs): Likewise.
9091 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
9092 (vectorizable_bswap): Adjust.
9093 (vectorizable_call): Likewise.
9094 (vectorizable_conversion): Likewise.
9095 (vectorizable_assignment): Likewise.
9096 (vectorizable_shift): Likewise.
9097 (vectorizable_operation): Likewise.
9098 (vectorizable_store): Likewise.
9099 (vectorizable_condition): Likewise.
9100 (vectorizable_comparison): Likewise.
9102 2017-06-28 Michael Collison <michael.collison@arm.com>
9105 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
9107 (arm_gen_movmemqi): Removed unused variable 'i'.
9108 Convert 'for' loop into 'while' loop.
9109 (arm_expand_prologue): Remove last unnecessary set of insn.
9110 (thumb_pop): Remove unused variable 'pushed_words'.
9111 (thumb_exit): Remove last unnecessary set of regs_to_pop.
9113 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9115 * config/s390/predicates.md: Use s390_rel_address_ok_p.
9116 * config/s390/s390-protos.h: Add prototype of
9117 s390_rel_address_ok_p.
9118 * config/s390/s390.c (s390_got_symbol): New function.
9119 (s390_rel_address_ok_p): New function.
9120 (legitimize_pic_address): Use s390_rel_address_ok_p.
9121 (s390_load_got): Use s390_got_symbol.
9122 (s390_option_override): Issue error if
9123 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
9124 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
9126 * config/s390/s390.opt: New option mpic-data-is-text-relative.
9128 2017-06-27 Andrew Pinski <apinski@cavium.com>
9130 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
9131 (X * copysign (1.0, X)): New pattern.
9132 (X * copysign (1.0, -X)): New pattern.
9133 (copysign (-1.0, CST)): New pattern.
9135 2017-06-27 Joseph Myers <joseph@codesourcery.com>
9137 * genmultilib (combination_space): Remove variable.
9138 Validate reuse rules against regular expression for any sequence
9139 of multilib options in any order.
9141 2017-06-27 Michael Collison <michael.collison@arm.com>
9143 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
9144 call aarch64_split_simd_combine.
9145 * (aarch64_combine_internal<mode>): Delete pattern.
9146 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
9147 Allow register and subreg operands.
9149 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9151 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
9152 specific need, just fallback on defaults.
9153 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
9155 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9156 Olivier Hainque <hainque@adacore.com>
9158 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
9160 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
9161 targets. Pick a default if no particular attempt applied.
9162 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
9165 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9167 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
9168 (x86_64-wrs-vxworks7): Likewise.
9170 2017-06-27 Marek Polacek <polacek@redhat.com>
9173 * ubsan.c (instrument_null): Check get_base_address's result for null.
9175 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
9177 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
9179 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
9181 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
9182 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
9183 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
9184 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
9186 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
9187 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
9188 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
9189 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
9190 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
9191 BUILT_IN_FEUPDATEENV): New builtins.
9192 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
9193 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
9194 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
9195 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
9197 (builtin_structptr_types): Adjust size.
9198 * tree.c (builtin_structptr_types): Add four entries.
9200 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9201 Olivier Hainque <hainque@adacore.com>
9203 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
9204 (TLS_SYM): New local macro, forcing reference to __tls__ on
9205 link command lines for VxWorks 7 RTPs, triggering initialization
9207 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
9208 OS features TLS support, true for RTPs on VxWorks 7.
9209 * config/vxworks.c (vxworks_override_options): Setup emutls
9212 2017-06-27 Jakub Jelinek <jakub@redhat.com>
9214 * predict.c (test_prediction_value_range): Use -1U instead of -1
9215 to avoid narrowing conversion warning.
9216 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
9217 to avoid narrowing conversion warning.
9218 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
9220 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
9222 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9224 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
9225 64bit configurations.
9226 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
9227 (SIZE_TYPE): Likewise.
9228 * config/vxworks.c (vxworks_emutls_var_fields): Use
9229 long_unsigned_type_node instead of unsigned_type_node as the offset
9230 field type, which is "pointer" mode in emutls.c.
9232 2017-06-27 Jakub Jelinek <jakub@redhat.com>
9235 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
9238 * gimple-fold.c (replace_call_with_call_and_fold): Handle
9239 gimple_vuse copying separately from gimple_vdef copying.
9241 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9243 * value-prof.c (free_hist): Remove call to memset and the enclosing if
9246 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
9247 Olivier Hainque <hainque@adacore.com>
9249 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
9250 for all vxworks7 targets.
9251 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
9252 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
9253 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
9254 variations for VX6/VX7 and 32/64bits later on in ...
9255 (VXWORKS_LIB_SPEC): Leverage new macros.
9256 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
9257 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
9259 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
9261 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
9262 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
9264 2017-06-26 Carl Love <cel@us.ibm.com>
9266 * config/rs6000/rs6000-c.c: Add support for built-in functions
9267 vector bool char vec_reve (vector bool char);
9268 vector signed char vec_reve (vector signed char);
9269 vector unsigned char vec_reve (vector unsigned char);
9270 vector bool int vec_reve (vector bool int);
9271 vector signed int vec_reve (vector signed int);
9272 vector unsigned int vec_reve (vector unsigned int);
9273 vector bool long long vec_reve (vector bool long long);
9274 vector signed long long vec_reve (vector signed long long);
9275 vector unsigned long long vec_reve (vector unsigned long long);
9276 vector bool short vec_reve (vector bool short);
9277 vector signed short vec_reve (vector signed short);
9278 vector double vec_reve (vector double);
9279 vector float vec_reve (vector float);
9280 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
9281 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
9282 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
9283 (altivec_vreve): New pattern.
9284 * config/rs6000/altivec.h (vec_reve): New define.
9285 * doc/extend.texi (vec_rev): Update the built-in documentation file
9286 for the new built-in functions.
9288 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9290 PR tree-optimization/71815
9291 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
9293 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
9295 (slsr_process_phi): Likewise.
9296 (replace_uncond_cands_and_profitable_phis): Don't replace a
9297 multiply candidate with a stride of 1 (copy or cast).
9298 (phi_incr_cost): Call uses_consumed_by_stmt rather than
9300 (lowest_cost_path): Likewise.
9301 (total_savings): Likewise.
9303 2017-06-26 Richard Biener <rguenther@suse.de>
9306 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
9307 Use def_builtin_pure for all gather builtins.
9309 2017-06-26 Richard Biener <rguenther@suse.de>
9311 PR tree-optimization/81203
9312 * tree-tailcall.c (find_tail_calls): Do not move stmts into
9315 2017-06-26 Marek Polacek <polacek@redhat.com>
9318 * doc/invoke.texi: Document -Wmultistatement-macros.
9320 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
9322 * doc/sourcebuild.texi (ARM-specific attributes): Document new
9323 arm_neon_ok_no_float_abi effective target.
9325 2017-06-26 Richard Biener <rguenther@suse.de>
9327 PR tree-optimization/80928
9328 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
9329 (copy_bbs): Set BB_DUPLICATED flag early.
9330 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
9332 (execute_on_shrinking_pred): Likewise.
9333 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
9334 BB_DUPLICATED blocks.
9335 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
9336 iterate over all PHIs considering removal of *gsi.
9338 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
9340 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
9343 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9345 * config/rs6000/rs6000-string.c: (expand_block_clear,
9346 do_load_for_compare, select_block_compare_mode,
9347 compute_current_alignment, expand_block_compare,
9348 expand_strncmp_align_check, expand_strn_compare,
9349 expand_block_move, rs6000_output_load_multiple)
9350 Move functions related to string/block move/compare
9352 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
9353 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
9354 for this function which is now used in two files.
9355 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
9356 * config.gcc: Add rs6000-string.o to extra_objs for
9357 targets powerpc*-*-* and rs6000*-*-*.
9359 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
9362 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
9363 32-bit, since indexed is not valid for DImode.
9364 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
9365 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
9366 (define_peephole2 for Altivec d-form load): Add 32-bit support.
9367 (define_peephole2 for Altivec d-form store): Likewise.
9370 * multiple_target.c (create_dispatcher_calls): Only create the
9371 dispatcher call if the function is the default clone of a
9374 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
9377 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
9378 a call, force the call to not be a tail call.
9380 2017-06-23 Jeff Law <law@redhat.com>
9382 * doc/contrib.texi: Add entry for Steven Pemberton's work on
9385 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
9387 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
9388 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
9389 handling for early expansion of vector shifts (sl,sr,sra,rl).
9390 (builtin_function_type): Add vector shift right instructions
9391 to the unsigned argument list.
9393 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9395 rtl-optimizatoin/79286
9396 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
9397 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
9398 trap. PIC register plus a const unspec without offset can never trap.
9400 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
9402 * tree.h (builtin_structptr_type): New type.
9403 (builtin_structptr_types): Declare new array.
9404 * tree.c (builtin_structptr_types): New array.
9405 (free_lang_data, build_common_tree_nodes): Use it.
9407 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
9410 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
9413 2017-06-22 Matt Turner <mattst88@gmail.com>
9415 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
9416 Lake models to skylake case. Assume skylake for unknown
9417 models with clflushopt.
9419 2017-06-22 Jeff Law <law@redhat.com>
9421 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
9422 frame sizes that do not satisfy aarch64_uimm12_shift.
9424 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
9426 * profile-count.h (apply_probability,
9427 apply_scale, probability_in): Fix checks for zero.
9429 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9431 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
9432 * doc/cppdiropts.texi (-I @var{dir}): Document it.
9434 2016-06-22 Richard Biener <rguenther@suse.de>
9436 * tree-vect-loop.c (vect_model_reduction_cost): Handle
9437 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
9438 REDUC_MAX_EXPR support.
9439 (vectorizable_reduction): Likewise.
9440 (vect_create_epilog_for_reduction): Likewise.
9442 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
9444 * match.pd (A / (1 << B) -> A >> B): New.
9445 * generic-match-head.c: Include optabs-tree.h.
9446 * gimple-match-head.c: Likewise.
9447 * optabs-tree.h (target_supports_op_p): New.
9448 * optabs-tree.c (target_supports_op_p): New.
9450 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9452 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
9453 $gcc_cv_ld --help output.
9454 (gcc_cv_ld_demangle): Likewise.
9455 (gcc_cv_ld_eh_frame_hdr): Likewise.
9456 (gcc_cv_ld_pie): Likewise.
9457 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
9458 (gcc_cv_ld_buildid): Likewise.
9459 (gcc_cv_ld_sysroot): Likewise.
9460 (ld_bndplt_support): Likewise.
9461 (ld_pushpopstate_support): Likewise.
9462 * configure: Regenerate.
9463 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
9465 2017-06-21 Jakub Jelinek <jakub@redhat.com>
9468 * config/i386/sse.md (round<mode>2): Renumber match_dup and
9469 operands indexes to avoid gap between operands and match_dups.
9471 2017-06-21 Andrew Pinski <apinski@cavium.com>
9473 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
9474 Increment Arith_shift and Arith_shift_reg by 1.
9475 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
9477 * config/aarch64/aarch64.c (thunderx_tunings): Enable
9478 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
9479 (aarch64_strip_extend): Add new argument and test for it.
9480 (aarch64_cheap_mult_shift_p): New function.
9481 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
9482 add a cost if it is true.
9483 Update calls to aarch64_strip_extend.
9484 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
9486 2017-06-21 Andrew Pinski <apinski@cavium.com>
9488 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
9490 (thunderxt88): Likewise.
9491 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
9492 (thunderx_prefetch_tune): New variable.
9493 (thunderx2t99_prefetch_tune): Update for the correct values.
9494 (thunderxt88_tunings): New variable.
9495 (thunderx_tunings): Use thunderx_prefetch_tune instead of
9496 generic_prefetch_tune.
9497 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
9499 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9501 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
9502 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
9503 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
9504 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
9505 (aarch64_atomic_cas<mode>, GPI): Likewise.
9507 2017-06-21 Martin Liska <mliska@suse.cz>
9509 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
9510 statements on cold and hot labels.
9511 * predict.c (tree_estimate_probability_bb): Remove the
9512 prediction from this place.
9514 2017-06-21 Martin Liska <mliska@suse.cz>
9516 PR tree-optimization/79489
9517 * gimplify.c (maybe_add_early_return_predict_stmt): New
9519 (gimplify_return_expr): Call the function.
9520 * predict.c (tree_estimate_probability_bb): Remove handling
9522 * predict.def: Update comment about early return predictor.
9523 * gimple-predict.h (is_gimple_predict): New function.
9524 * predict.def: Change default value of early return to 66.
9525 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
9527 * passes.def: Put pass_strip_predict_hints to the beginning of
9530 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
9532 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
9533 FUNCTION_DECL declarations.
9534 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
9536 (dwaf2out_decl): Likewise.
9537 * godump.c (go_early_global_decl): Skip call to the real debug hook
9538 for FUNCTION_DECL declarations.
9539 * passes.c (rest_of_decl_compilation): Skip call to the
9540 early_global_decl debug hook for FUNCTION_DECL declarations, unless
9541 -fdump-go-spec is passed.
9543 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
9545 * config/i386/i386.c (struct builtin_isa): New field pure_p.
9546 Reorder for compactness.
9547 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
9548 (def_builtin_pure, def_builtin_pure2): New functions.
9549 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
9551 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
9553 * match.pd (nop_convert): New predicate.
9554 ((A +- CST1) +- CST2): Allow some NOP conversions.
9556 2017-06-21 Jakub Jelinek <jakub@redhat.com>
9559 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
9560 with ctors/dtors if GOVD_SHARED is set.
9562 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9564 * config/aarch64/aarch64.md (movti_aarch64):
9565 Emit mov rather than orr.
9566 (movtf_aarch64): Likewise.
9567 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
9568 Emit mov rather than orr.
9570 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9572 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
9573 Swap alternatives, make integer dup more expensive.
9575 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9577 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9578 Return true for non-tls symbols.
9580 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
9582 * config/aarch64/aarch64-cores.def (cortex-a55): New.
9583 (cortex-a75): Likewise.
9584 (cortex-a75.cortex-a55): Likewise.
9585 * config/aarch64/aarch64-tune.md: Regenerate.
9586 * doc/invoke.texi (-mtune): Document new values for -mtune.
9588 2017-06-21 Tom de Vries <tom@codesourcery.com>
9590 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
9592 (Effective-Target Keywords, Other attributes): Suggest using
9593 dg-add-options stack_size feature to get stack limit in stack_size
9594 effective target documentation.
9596 2017-06-21 Julian Brown <julian@codesourcery.com>
9597 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9599 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
9600 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
9601 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
9603 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
9604 attribute type list for neon_multiply.
9605 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
9606 attribute type list for neon_multiply.
9607 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
9608 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
9609 attribute type list for neon_multiply.
9610 * config/arm/types.md (crypto_pmull): Add.
9611 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
9612 attribute type list.
9614 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
9616 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
9619 2017-06-20 Jakub Jelinek <jakub@redhat.com>
9621 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
9622 to make sure not to dereference a NULL cost_classes_ptr pointer.
9624 2017-06-20 Carl Love <cel@us.ibm.com>
9626 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9627 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9628 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9629 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9630 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9631 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9632 VMULOSW): New enum "unspec" values.
9633 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
9634 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
9635 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9636 altivec_vmulosw): New patterns.
9637 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9638 VMULOSW): Add definitions.
9640 2017-06-20 Julia Koval <julia.koval@intel.com>
9642 * config/i386/i386.c: Fix rounding expand for new pattern.
9643 * config/i386/subst.md: Fix pattern (parallel -> unspec).
9645 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9647 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
9648 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
9650 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9652 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
9655 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9657 * config/aarch64/aarch64-cores.def: Rearrange to sort by
9658 architecture, then by implementer ID.
9659 * config/aarch64/aarch64-tune.md: Regenerate.
9661 2017-06-20 Richard Biener <rguenther@suse.de>
9664 * fold-const.c (split_tree): Fold to type before negating.
9666 2017-06-20 David Malcolm <dmalcolm@redhat.com>
9668 * diagnostic-show-locus.c
9669 (selftest::test_fixit_deletion_affecting_newline): New function.
9670 (selftest::diagnostic_show_locus_c_tests): Call it.
9672 2017-06-20 Andreas Schwab <schwab@suse.de>
9675 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
9678 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9680 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
9681 __ARM_FEATURE_COPROC according to support.
9683 2017-06-20 Jakub Jelinek <jakub@redhat.com>
9685 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
9686 Rewritten to avoid overflow for > 32-bit pointers.
9689 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
9690 by removing enum keyword.
9691 (ubsan_type_descriptor): Likewise. Formatting fix.
9694 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
9695 splitter): Require TARGET_SSE2 in the condition.
9697 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
9700 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
9701 for doing vector set of SFmode on ISA 3.0.
9702 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
9703 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
9705 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
9706 SFmode value into a V4SF variable that was extracted from another
9707 V4SF variable without converting the element to double precision
9708 and back to single precision vector format.
9709 (vsx_insert_extract_v4sf_p9_2): Likewise.
9711 2017-06-19 Jakub Jelinek <jakub@redhat.com>
9713 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
9714 in UWHI to avoid undefined overflow.
9717 * ubsan.h (enum ubsan_encode_value_phase): New.
9718 (ubsan_encode_value): Change second argument to
9719 enum ubsan_encode_value_phase with default value of
9720 UBSAN_ENCODE_VALUE_GENERIC.
9721 * ubsan.c (ubsan_encode_value): Change second argument to
9722 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
9723 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
9724 create_tmp_var_raw instead of create_tmp_var and use a
9726 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
9727 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
9728 ubsan_encode_value callers.
9731 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
9732 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
9733 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
9735 2017-06-19 Richard Biener <rguenther@suse.de>
9738 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
9739 estimates if we changed anything.
9741 2017-06-19 Richard Biener <rguenther@suse.de>
9743 PR tree-optimization/80887
9744 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
9745 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
9746 simplified lookups, then reset mprts_hook.
9747 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
9749 (try_to_simplify): Likewise.
9751 2017-06-19 Martin Liska <mliska@suse.cz>
9754 * gimplify.c (gimplify_switch_expr):
9755 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
9757 2017-06-19 Martin Liska <mliska@suse.cz>
9759 * doc/install.texi: Document that PGO runs in 4 stages.
9761 2017-06-19 Martin Liska <mliska@suse.cz>
9764 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
9765 to dispatcher function name.
9766 * multiple_target.c (replace_function_decl): New function.
9767 (create_dispatcher_calls): Redirect both edges and references.
9769 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
9771 * profile-count.c (profile_count::dump): Dump quality.
9772 (profile_count::differs_from_p): Update for unsigned val.
9773 * profile-count.h (profile_count_quality): New enum.
9774 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
9776 2017-06-19 Richard Biener <rguenther@suse.de>
9778 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
9779 struct function as arg.
9780 (estimate_numbers_of_iterations): Export overload with loop arg.
9781 (free_numbers_of_iterations_estimates_loop): Use an overload of
9782 free_numbers_of_iterations_estimates instead.
9783 * tree-cfg.c (remove_bb): Adjust.
9784 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
9785 * tree-parloops.c (gen_parallel_loop): Likewise.
9786 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
9788 (tree_unroll_loops_completely): Likewise.
9789 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
9790 Use an overload instead and export.
9791 (estimated_loop_iterations): Adjust.
9792 (max_loop_iterations): Likewise.
9793 (likely_max_loop_iterations): Likewise.
9794 (estimate_numbers_of_iterations): Take struct function as arg
9796 (loop_exits_before_overflow): Adjust.
9797 (free_numbers_of_iterations_estimates_loop): Use an overload.
9798 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
9799 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
9801 2017-06-19 Richard Biener <rguenther@suse.de>
9804 * ipa-prop.c (find_constructor_constant_at_offset): Handle
9805 RANGE_EXPR conservatively.
9807 2017-06-16 Carl Love <cel@us.ibm.com>
9809 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9810 definitions for vec_float, vec_float2, vec_floato,
9811 vec_floate built-ins.
9812 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
9813 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
9815 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
9816 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
9817 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
9818 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
9819 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
9820 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
9821 vec_floato): Add builtin defines.
9822 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
9823 Update the built-in documentation file for the new built-in
9826 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9828 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
9829 (mthumb): Mark as the negative of -marm.
9831 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9833 * doc/invoke.texi (ARM Options, -mcpu): Document supported
9835 (ARM Options, -mtune): Document that this accepts the same
9836 extension options as -mcpu.
9837 (ARM Options, -mfpu): Document addition of -mfpu=auto.
9839 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9841 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
9842 permitted extensions.
9844 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9846 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
9847 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
9848 (armv8-m.main): Add option +nodsp.
9849 * config/arm/arm-cpu-cdata.h: Regenerated.
9851 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9853 * config/arm/t-fuchsia: New file.
9854 * config.gcc (arm*-*-fuchsia*): Use it.
9856 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9858 * config/arm/t-symbian: Rewrite for new option infrastructure.
9860 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9862 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
9863 (MULTILIB_REQUIRED): Likewise.
9865 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9867 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
9868 (MULTILIB_RESUE): Likewise.
9869 (MULTILIB_MATCHES): Likewise.
9870 (MULTLIB_REQUIRED): Likewise.
9872 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9874 * config/arm/t-rtems: Rewrite for new option framework.
9876 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9878 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
9879 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
9880 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
9881 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
9882 * config/arm/t-multilib: ... here.
9883 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
9884 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
9885 armv7-a and armv8*-a when A-profile libraries have not been built.
9886 * config/arm/t-rmprofile: Rewrite.
9888 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9890 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
9891 with a backslash. Remove the backslash after substituting unescaped
9893 * doc/fragments.texi (MULTILIB_REUSE): Document it.
9895 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9897 * config.gcc: (arm*-*-*): When building a-profile libraries, force
9898 the driver to pass through the default setting of -mfloat-abi.
9899 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
9901 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
9902 (all_feat_combs): New rule.
9903 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
9905 * config/arm/t-aprofile: Rewrite.
9907 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9909 * config/arm/arm.h (FPUTYPE_AUTO): Define.
9910 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
9911 fpu is not specified by the user/command-line.
9912 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
9913 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
9914 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
9915 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
9916 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
9917 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
9919 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9921 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
9922 * config/arm/t-arm-elf: Rewritten.
9924 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9926 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
9927 have some floating-point instructions.
9928 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
9929 (TARGET_MAYBE_HARD_FLOAT): New macro.
9930 * config/arm/arm-builtins.c (arm_init_builtins): Use
9931 TARGET_MAYBE_HARD_FLOAT.
9932 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
9934 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9936 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
9937 (configargs.h): Include it.
9938 (arm_print_hint_for_fpu_option): New function.
9939 (arm_parse_fpu_option): New function.
9940 (candidate_extension): New class.
9941 (arm_canon_for_multilib): New function.
9942 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
9943 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
9944 (ARCH_CANONICAL_SPECS): New macro.
9945 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
9947 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9949 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
9950 are set after handling multilib fragments. Set target_cpu_default2
9953 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9955 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
9957 (arm*-*-*): Set target_cpu_default2 to a quoted string.
9958 * config/arm/parsecpu.awk (check_cpu): Validate any extension
9960 (check_arch): Likewise.
9961 * config/arm/arm.c (arm_configure_build_target): Handle
9962 TARGET_CPU_DEFAULT being a string constant. Scan any feature
9963 options in the default.
9965 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9967 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
9968 when an option is an alias of another.
9969 * config/arm/parsecpu.awk (optalias): New parser token.
9970 (gen_comm_data): Mark non-alias options as such. Emit entries
9971 for extension aliases.
9972 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
9973 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
9974 (armv6kz, armv6zk, armv6t2): Likewise.
9975 (armv7): Make vfpv3-d16 an alias.
9976 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
9978 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
9979 Sort in canonical order.
9980 (armv8-a): Sort in canonical order.
9981 (armv8.1-a, armv8.2-a): Likewise.
9982 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
9984 (cortex-a9): Sort in canonical order.
9985 * config/arm/arm.c (selftests.h): Include it.
9986 (arm_test_cpu_arch_data): New function.
9987 (arm_run_self_tests): New function.
9988 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
9989 (targetm): Move declaration to the end of the file.
9990 * arm-cpu-cdata.h: Regenerated.
9992 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9994 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
9995 call to target_mode_check describing the type of option passed.
9996 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
9997 (arm_target_thumb_only): Use arm_parse_arch_option_name or
9998 arm_parse_cpu_option_name to match parameters against list of
10000 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
10001 arm_arch_core_flags data structure.
10002 * config/arm/arm-cpu_cdata.h: Regenerated.
10004 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10006 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
10008 (arm_print_hint_for_cpu_option): Likewise.
10009 (arm_print_hint_for_arch_option): Likewise.
10010 (arm_parse_cpu_option_name): Likewise.
10011 (arm_parse_arch_option_name): Likewise.
10012 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
10013 of entries in the all_fpus list.
10014 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
10015 (arm_parse_cpu_option_name): Declare.
10016 (arm_parse_arch_option_name): Declare.
10017 (arm_parse_option_features): Declare.
10018 (arm_intialize_isa): Declare.
10019 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
10021 (gen_comm_data): ... here. Make definitions non-static.
10022 * config/arm/arm-cpu-data.h: Regenerated.
10023 * config/arm/arm-cpu-cdata.h: Regenerated.
10025 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10027 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
10028 (cpu_arch_extension): New structure.
10029 (cpu_arch_option, arch_option, cpu_option): New structures.
10030 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
10031 architecture types.
10032 (gen_data): Generate new format data tables.
10033 * config/arm/arm.c (cpu_tune): New structure.
10034 (cpu_option, processors): Delete.
10035 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
10036 (arm_print_hint_for_cpu_option): ... this and ...
10037 (arm_print_hint_for_arch_option): ... this.
10038 (arm_parse_arch_cpu_name): Delete. Replace with ...
10039 (arm_parse_cpu_option_name): ... this and ...
10040 (arm_parse_arch_option_name): ... this.
10041 (arm_unrecognized_feature): Change type of target parameter to
10043 (arm_parse_arch_cpu_features): Delete. Replace with ...
10044 (arm_parse_option_features): ... this.
10045 (arm_configure_build_target): Rework to use new configuration data
10047 (arm_print_tune_info): Rework for new configuration data tables.
10048 * config/arm/arm-cpu-data.h: Regenerated.
10049 * config/arm/arm-cpu.h: Regenerated.
10051 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10053 * Makefile.in (OBJS): Move sbitmap.o from here ...
10054 (OBJS-libcommon): ... to here.
10056 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10058 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
10059 (ISA_ALL_CRYPTO): New macro.
10060 (ISA_ALL_SIMD): New macro
10061 (ISA_ALL_FP): New macro.
10062 * config/arm/arm.c (fpu_bitlist): Update initializer.
10063 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
10065 (arm9e): Add fpu. Add option for nofp
10066 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
10067 (arm926ej-s, arm1026ej-s): Likewise.
10068 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
10069 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
10070 neon-fp16, neon-vfpv4, nofp and nosimd.
10071 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
10072 (cortex-a8): Add fpu. Add option for nofp.
10073 (cortex-a9): Add fpu. Add options for nosimd and nofp.
10074 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
10075 (cortex-r4f): Add fpu.
10076 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
10077 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
10079 (cortex-r8): Likewise.
10080 (cortex-m4): Add fpu. Add option for nofp.
10081 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
10082 (cortex-a17.cortex-a7): Likewise.
10083 (cortex-a32): Add fpu. Add options for crypto and nofp.
10084 (cortex-a35, cortex-a53): Likewise.
10085 (cortex-a57): Add fpu. Add option for crypto.
10086 (cortex-a72, cortex-a73): Likewise.
10087 (exynos-m1): Likewise.
10088 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
10089 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
10090 (cortex-m33): Add fpu. Add option for nofp.
10091 * config/arm/arm-cpu-cdata.h: Regenerated
10092 * config/arm/arm-cpu-data.h: Regenerated.
10094 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10096 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
10097 (armv5te, armv5tej): Likewise.
10098 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
10099 (armv7): Add options fp and vfpv3-d16.
10100 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
10101 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
10103 (armv7ve): Likewise.
10104 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
10105 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
10106 (armv8-a): Add nocrypto option.
10107 (armv8.1-a, armv8.2-a): Likewise.
10108 (armv8-m.main): add options fp, fp.dp and nofp.
10110 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10112 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
10114 (armv8-a+crc): Delete.
10115 (armv8.1-a): Add options simd, crypto and nofp.
10116 (armv8.2-a): Add options fp16, simd, crypto and nofp.
10117 (armv8.2-a+fp16): Delete.
10118 (armv8-m.main): Add option dsp.
10119 (armv8-m.main+dsp): Delete.
10120 (cortex-a8): Add fpu. Add nofp option.
10121 (cortex-a9): Add fpu. Add nofp and nosimd options.
10122 * config/arm/parsecpu.awk (gen_data): Generate option tables and
10123 link to main cpu and architecture data structures.
10124 (gen_comm_data): Only put isa attributes from the main architecture
10126 (option): New statement for architecture and CPU entries.
10127 * arm.c (struct cpu_option): New structure.
10128 (struct processors): Add entry for options.
10129 (arm_unrecognized_feature): New function.
10130 (arm_parse_arch_cpu_name): Ignore any characters after the first
10132 (arm_parse_arch_cpu_feature): New function.
10133 (arm_configure_build_target): Separate out any CPU and architecture
10134 features and parse separately. Don't error out if -mfpu=auto is
10135 used with only an architecture string.
10136 (arm_print_asm_arch_directives): New function.
10137 (arm_file_start): Call it.
10138 * config/arm/arm-cpu-cdata.h: Regenerated.
10139 * config/arm/arm-cpu-data.h: Likewise.
10140 * config/arm/arm-tables.opt: Likewise.
10142 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10144 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
10145 assembler when it is not -mfpu=auto.
10147 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10149 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
10150 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
10151 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
10152 (ASM_CPU_SPEC): Rewrite.
10153 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
10154 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
10155 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
10156 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
10157 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
10158 copied string is NUL-terminated. Also strip any characters prefixed
10160 (arm_rewrite_selected_arch): New function.
10161 (arm_rewrite_march): New function.
10163 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10165 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
10166 (x_arm_cpu_string, x_arm_tune_string): Likewise.
10167 (march, mcpu, mtune): Convert to string-based options.
10168 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
10169 (arm_parse_arch_cpu_name): New function.
10170 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
10171 identify selected architecture or CPU.
10172 (arm_option_save): New function.
10173 (TARGET_OPTION_SAVE): Redefine.
10174 (arm_option_restore): Restore string options.
10175 (arm_option_print): Print string options.
10177 2017-06-16 Martin Sebor <msebor@redhat.com>
10179 PR tree-optimization/80933
10180 PR tree-optimization/80934
10181 * builtins.c (fold_builtin_3): Do not handle bcmp here.
10182 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
10183 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
10184 (gimple_fold_builtin): Call them.
10186 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10188 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
10189 as unlikely; update profile.
10191 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10193 * predict.c (force_edge_cold): Handle declaring edges impossible
10196 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10198 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
10200 (try_unroll_loop_completely): Fix reporting.
10202 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10204 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
10206 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
10209 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
10210 if given a non-constant argument for an intrinsic which requires a
10213 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10215 * profile.c (compare_freqs): New function.
10216 (branch_prob): Sort edge list.
10217 (find_spanning_tree): Assume that the list is priority sorted.
10219 2017-06-16 Richard Biener <rguenther@suse.de>
10221 PR tree-optimization/81090
10222 * passes.def (pass_record_bounds): Remove.
10223 * tree-pass.h (make_pass_record_bounds): Likewise.
10224 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
10225 make_pass_record_bounds): Likewise.
10226 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
10227 not free niter estimates at the beginning but at the end.
10228 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
10230 2017-06-16 Richard Biener <rguenther@suse.de>
10232 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
10233 initializer to workaround ICE in host GCC 4.8.
10235 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10237 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
10239 (clone_inlined_nodes): Update.
10241 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10243 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
10244 prefetch settings, and enable prefetching by default at -O3.
10246 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10248 * config/aarch64/aarch64.c (aarch64_override_options_internal):
10249 Set flag_prefetch_loop_arrays according to tuning data.
10251 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10253 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
10254 New tune structure.
10255 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
10256 [Unrelated to main purpose of the patch] Place the pointer field last
10257 to enable type checking errors when tune structure are wrongly merged.
10258 * config/aarch64/aarch64.c (generic_prefetch_tune,)
10259 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
10260 (thunderx2t99_prefetch_tune): New tune constants.
10261 (tune_params *_tunings): Update all tunings (no functional change).
10262 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
10263 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
10264 from tunings structures.
10266 2017-06-16 Jakub Jelinek <jakub@redhat.com>
10269 * ubsan.c (instrument_null): Add T argument, use it instead
10270 of computing it based on IS_LHS.
10271 (instrument_object_size): Likewise.
10272 (pass_ubsan::execute): Adjust instrument_null and
10273 instrument_object_size callers to pass gimple_get_lhs or
10274 gimple_assign_rhs1 result to it. Use instrument_null instead of
10275 calling get_base_address and instrument_mem_ref. Handle
10276 aggregate call arguments for object-size sanitization.
10278 2017-06-16 Yury Gribov <tetra2005@gmail.com>
10280 PR tree-optimization/81089
10281 * tree-vrp.c (is_masked_range_test): Validate operands of
10284 2017-06-15 Martin Sebor <msebor@redhat.com>
10287 * dumpfile.c (dump_register): Avoid calling memset to initialize
10288 a class with a default ctor.
10289 * gcc.c (struct compiler): Remove const qualification.
10290 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
10291 * hash-table.h: Ditto.
10292 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
10294 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
10295 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
10297 * params.h (struct param_info): Make struct members non-const.
10298 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
10299 with default initialization.
10300 * vec.h (vec_copy_construct, vec_default_construct): New helper
10302 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
10303 with vec_copy_construct.
10304 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
10305 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
10306 * doc/invoke.texi (-Wclass-memaccess): Document.
10308 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10310 * emit-rtl.h (is_leaf): Update comment about local
10311 register allocator.
10313 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10316 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
10317 for a variable to have a section before checking if the section has a
10319 Set section to.persistent if persistent attribute is set.
10320 Warn if .persistent attribute is used on an automatic variable.
10322 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
10324 PR rtl-optimization/80474
10325 * reorg.c (update_block): Do not ignore instructions in a delay slot.
10327 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
10329 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
10332 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
10334 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
10335 (casesi): Emit bounds checking as RTL.
10336 (casesi_internal_mips16_<mode>): Remove bounds checking.
10338 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
10340 * config/xtensa/xtensa.c (xtensa_option_override): Append
10341 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
10342 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
10343 xtensa_doloop_hooks): Define unconditionally.
10344 (xtensa_reorg_loops): Only call reorg_loops in the presence of
10346 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
10347 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
10348 for it in xtensa_option_override.
10349 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
10350 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
10352 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
10354 * doc/cppopts.texi: Document '-' special value to -MF.
10356 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
10358 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
10359 (cortex_a53_fconst): Likewise.
10360 (cortex_a53_fpmul): Likewise.
10361 (cortex_a53_f_load_64): Likewise.
10362 (cortex_a53_f_load_many): Likewise.
10363 (cortex_a53_advsimd_alu): Likewise.
10364 (cortex_a53_advsimd_alu_q): Likewise.
10365 (cortex_a53_advsimd_mul): Likewise.
10366 (cortex_a53_advsimd_mul_q): Likewise.
10367 (fpmac bypass): Add new bypass for fpmac-fpmac case.
10368 Add missing fmul, r2f_cvt and fconst cases.
10370 2017-06-14 Richard Biener <rguenther@suse.de>
10372 PR middle-end/81088
10373 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
10375 (fold_binary_loc): When associating do not treat pre-existing
10376 TREE_OVERFLOW on literal constants as a reason to allow
10377 TREE_OVERFLOW on associated literal constants.
10379 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
10381 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
10382 (MASK_FEATURES): New macro.
10383 * config/sparc/sparc.c (sparc_option_override): Remove the special
10384 handling of -mfpu and generalize it to all MASK_FEATURES switches.
10386 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
10388 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
10389 a division of 0 if non-call exceptions are enabled.
10391 2017-06-14 Andrew Pinski <apinski@cavium.com>
10392 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10395 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
10396 Improve vector initialization code gen for only variable case.
10398 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
10400 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
10402 2017-06-14 Richard Biener <rguenther@suse.de>
10404 PR tree-optimization/81083
10405 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
10408 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10410 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
10411 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
10412 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
10413 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
10414 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
10415 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
10417 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10419 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
10420 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
10422 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10424 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
10426 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10428 * config/rs6000/t-rtems: Don't handle SPE.
10430 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10432 * config/rs6000/t-linux: Don't handle SPE.
10434 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10436 * config/rs6000/eabispe.h: Delete file.
10438 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10440 * config/rs6000/t-spe: Delete file.
10442 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10444 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
10445 (rs6000_legitimate_offset_address_p): Return false for anything in
10446 V2SImode or V2SFmode.
10448 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10450 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
10451 except V2SF and V2SI. Rearrange the vector modes, and add comments.
10452 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
10454 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
10455 (rs6000_legitimate_offset_address_p): Ditto.
10456 (rs6000_emit_move): Ditto.
10457 (rs6000_init_builtins): Remove V4HI_type_node.
10459 2017-06-13 Martin Liska <mliska@suse.cz>
10462 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
10463 (gate_asan): Likewise.
10464 * asan.h (asan_no_sanitize_address_p): Remove the function.
10465 (sanitize_flags_p): New function.
10466 * builtins.def: Fix coding style.
10467 * common.opt: Use renamed enum value.
10468 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
10469 * doc/extend.texi: Document no_sanitize attribute.
10470 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
10471 to SANITIZE_UNDEFINED_NONDEFAULT.
10472 * gcc.c (sanitize_spec_function): Use the renamed enum value.
10473 * gimple-fold.c (optimize_atomic_compare_exchange_p):
10474 Use sanitize_flags_p.
10475 * gimplify.c (gimplify_function_tree): Likewise.
10476 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
10477 * opts.c (parse_no_sanitize_attribute): New function.
10478 (common_handle_option): Use renamed enum value.
10479 * opts.h (parse_no_sanitize_attribute): Declare.
10480 * tree.c (sanitize_flags_p): New function.
10481 * tree.h: Declared here.
10482 * tsan.c: Use sanitize_flags_p.
10483 * ubsan.c (ubsan_expand_null_ifn): Likewise.
10484 (instrument_mem_ref): Likewise.
10485 (instrument_bool_enum_load): Likewise.
10486 (do_ubsan_in_current_function): Remove the function.
10487 (pass_ubsan::execute): Use sanitize_flags_p.
10488 * ubsan.h: Remove do_ubsan_in_current_function
10489 * tree-cfg.c (print_no_sanitize_attr_value): New function.
10490 (dump_function_to_file): Use it here.
10492 2017-06-13 Martin Jambor <mjambor@suse.cz>
10494 PR tree-optimization/80803
10495 PR tree-optimization/81063
10496 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
10497 (propagate_subaccesses_across_link): Enqueue subtree whenever
10498 necessary instead of relying on the caller.
10500 2017-06-13 Martin Jambor <mjambor@suse.cz>
10502 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
10503 that have a first_link.
10504 (sort_and_splice_var_accesses): Do not check first_link before
10506 (subtree_mark_written_and_enqueue): Likewise.
10507 (propagate_all_subaccesses): Likewise and do not stop at first
10508 parent with a first_link.
10510 2017-06-13 Martin Jambor <mjambor@suse.cz>
10512 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
10515 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10517 * match.pd: New pattern.
10519 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10521 * tree-vrp.c (is_masked_range_test): New function.
10522 (register_edge_assert_for): Determine ranges for
10525 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10527 PR tree-optimization/67328
10528 * fold-const.c (maskable_range_p): New function.
10529 (build_range_check): Generate bittests if possible.
10531 2017-06-13 Martin Liska <mliska@suse.cz>
10533 * gimple-pretty-print.c (dump_probability): Add new argument.
10534 (dump_edge_probability): Dump both probability and count.
10535 (dump_gimple_label): Likewise.
10536 (dump_gimple_bb_header): Likewise.
10538 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
10541 * config/avr/avr-devices.c: Fix indentation.
10542 * config/avr/gen-avr-mmcu-specs.c: Dito.
10544 2017-06-13 Richard Biener <rguenther@suse.de>
10546 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
10547 instead get vector type from stmt_info.
10548 (vectorizable_reduction): Adjust. Remove dead code.
10550 2017-06-13 Richard Biener <rguenther@suse.de>
10552 PR middle-end/81065
10553 * fold-const.c (extract_muldiv_1): Remove bogus distribution
10554 case of C * (x * C2 + C3).
10555 (fold_addr_of_array_ref_difference): Properly fold index difference.
10557 2017-06-12 David S. Miller <davem@davemloft.net>
10560 * config/sparc/sparc.md (return expander): Emit frame blockage if
10561 function uses alloca.
10563 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
10565 * combine.c (make_field_assignment): Check len rather than the mode
10566 precision when calling force_to_mode.
10568 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
10570 Support multilibs and devices that see flash in RAM address range.
10573 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
10574 (avr_mcu_t) <flash_pm_offset>: New field.
10575 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
10576 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
10577 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
10578 (AVR_TINY_PM_OFFSET): Remove macro.
10579 * config/avr/avr.opt (-mshort-calls): New option.
10580 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
10581 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
10582 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
10583 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
10584 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
10585 instead of avr_arch->have_jmp_call.
10586 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
10587 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
10588 avr_arch->flash_pm_offset to define.
10589 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
10590 new field flash_pm_offset. Add entry for avrxmega3.
10591 (avr_texinfo): Add entry for avrxmega3.
10592 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
10593 attiny212, attiny214,
10594 attiny412, attiny414, attiny416, attiny417,
10595 attiny814, attiny816, attiny817,
10596 attiny1614, attiny1616, attiny1617,
10597 attiny3214, attiny3216, attiny3217.
10598 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
10599 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
10600 (avr_print_operand_address) [AVR_TINY]: Same.
10601 (avr_asm_init_sections) <readonly_data_section>: Only patch
10602 callback if avr_arch->flash_pm_offset = 0.
10603 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
10604 for rodata if avr_arch->flash_pm_offset != 0.
10605 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
10606 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
10607 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
10608 (m_options): Append opt_rcall.
10609 (m_dirnames): Append dir_rcall.
10610 * config/avr/t-multilib: Regenerate.
10612 * configure.ac [target=avr]: Check whether avrxmega3 default
10613 linker description file works as needed.
10614 * configure: Regenerate.
10615 * doc/avr-mmcu.texi: Regenerate.
10616 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
10617 <__AVR_ARCH__>: Document avrxmega3 and 103.
10618 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
10619 <__AVR_SHORT_CALLS__>: Document it.
10620 <__AVR_PM_BASE_ADDRESS__>: Document it.
10621 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
10622 (AVR Variable Attributes) <progmem>: Document this is
10623 not needed for avrxmega3.
10624 (AVR Named Address Spaces) <__flash>: Dito.
10626 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
10628 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
10630 2017-06-12 Doug Rupp <rupp@adacore.com>
10632 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
10633 Append vxworks-stdint.h to the tm_file list.
10634 * config/vxworks-stdint.h: New file.
10636 2017-06-12 Martin Liska <mliska@suse.cz>
10638 PR tree-optimization/81041
10639 * tree-profile.c (gimple_gen_ic_func_profiler):
10640 Create an extra BB in profile-generate
10641 (gimple_gen_time_profiler): Likewise.
10643 2017-06-12 Jakub Jelinek <jakub@redhat.com>
10645 PR tree-optimization/81003
10646 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
10647 (update_range_test): Use it instead of force_gimple_operand_gsi.
10649 2017-06-12 Richard Biener <rguenther@suse.de>
10651 PR tree-optimization/81053
10652 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
10653 with backedge value not defined in loop. Simplify def stmt
10656 2017-06-11 Tom de Vries <tom@codesourcery.com>
10659 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
10661 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
10662 nvptx_cannot_force_const_mem.
10664 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10666 * opts.c (finish_options): Move test for flag_split_stack after
10667 it has been initialized.
10669 2017-06-11 Jason Merrill <jason@redhat.com>
10671 * tree.h (id_equal): New.
10672 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
10673 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
10674 instead of strcmp of IDENTIFIER_POINTER.
10676 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10678 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
10679 (mark_all_inlined_calls_cdtor): Fix formating.
10680 (inline_transform): Rescale profile before inlining.
10682 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10684 * cgraph.h (cgraph_edge::clone): Update prototype.
10685 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
10686 (cgraph_node::create_clone): Update.
10687 (cgraph_node::create_version_clone): Update.
10688 * tree-inline.c (copy_bb): Update.
10689 (expand_call_inline): Update.
10691 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
10693 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
10694 factored out from ...
10695 (rs6000_emit_prologue): ... here.
10697 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
10699 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
10700 factored out from ...
10701 (rs6000_emit_prologue): ... here.
10703 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10705 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
10707 (handle_missing_profiles): Fix computation of tp_first_run.
10708 (counts_to_freqs): Do not touch freqs when count is 0.
10710 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10712 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
10715 2017-06-10 Tom de Vries <tom@codesourcery.com>
10717 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
10718 attributes): Document signal effective target.
10720 2017-06-10 Tom de Vries <tom@codesourcery.com>
10722 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
10723 Document effective target stack_size.
10725 2017-06-09 David Malcolm <dmalcolm@redhat.com>
10727 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
10728 to the edit_context if they can be auto-applied.
10730 2017-06-9 Ian Lance Taylor <iant@golang.org>
10732 * opts.c (finish_options): If -fsplit-stack, disable implicit
10733 -forder-blocks-and-partition.
10734 * doc/invoke.texi (Optimize Options): Document that when using
10735 -fsplit-stack -forder-blocks-and-partition is not implicitly
10738 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10740 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
10741 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
10742 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
10743 * builtins.def (abort, trap, unreachable): Declare cold.
10744 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
10745 * tree-core.h (ECF_COLD): New.
10746 * tree.c (set_call_expr_flags): Handle ECF_COLD.
10747 (build_common_builtin_nodes): Mark unreachable and abort as cold.
10749 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10751 * predict.c (unlikely_executed_stmt_p): Cleanup.
10753 2017-06-09 Richard Biener <rguenther@suse.de>
10755 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
10756 model if the ref is always written to.
10758 2017-06-09 Tamar Christina <tamar.christina@arm.com>
10760 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
10762 2017-06-09 Tamar Christina <tamar.christina@arm.com>
10764 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
10767 2017-06-09 Tom de Vries <tom@codesourcery.com>
10770 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
10771 "target cannot support label values" when encountering LABEL_REF.
10773 2017-06-09 Martin Liska <mliska@suse.cz>
10775 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
10776 (gimple_gen_ic_func_profiler): Emit direct comparison
10777 of __gcov_indirect_call_callee with NULL.
10778 (gimple_gen_time_profiler): Change probability from
10779 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
10781 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10783 * profile.c (edge_gcov_counts): Turn to pointer.
10784 (compute_branch_probabilities, compute_branch_probabilities): Update.
10785 (branch_prob): Do not clear edge_gcov_count.
10786 * profile.h (edge_gcov_counts): Turn to pointer.
10787 (edge_gcov_count): Update.
10789 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10791 * gimple.h (gimple_check_failed): Mark cold.
10793 2017-06-09 Richard Biener <rguenther@suse.de>
10795 PR tree-optimization/66623
10796 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
10797 refactor check_reduction into two parts, properly computing
10798 whether we have to check reduction validity for outer loop
10801 2017-06-09 Richard Biener <rguenther@suse.de>
10803 PR tree-optimization/79483
10804 * graphite-scop-detection.c (order): New global.
10805 (get_order): Compute bb to order mapping that satisfies code
10806 generation constraints.
10807 (cmp_pbbs): New helper.
10808 (build_scops): Start domwalk at entry block, sort generated
10811 2017-06-09 Richard Biener <rguenther@suse.de>
10813 PR middle-end/81007
10814 * ipa-polymorphic-call.c
10815 (ipa_polymorphic_call_context::restrict_to_inner_class):
10816 Skip FIELD_DECLs with error_mark_node type.
10817 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
10820 2017-06-09 Martin Liska <mliska@suse.cz>
10822 * predict.c (struct branch_predictor): New struct.
10823 (test_prediction_value_range): New test.
10824 (predict_c_tests): New function.
10825 * selftest-run-tests.c (selftest::run_tests): Run the function.
10826 * selftest.h: Declare new tests.
10828 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
10831 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
10832 gen_add3_insn did not fail.
10833 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
10834 r0, construct that number in a temporary reg and add that reg to r0.
10835 If asked to put the result in r0 as well, fail.
10837 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
10839 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10840 for early expansion of vec_eqv.
10842 2017-06-08 Jakub Jelinek <jakub@redhat.com>
10844 PR middle-end/81005
10845 * ubsan.c (instrument_null): Avoid pointless code temporary.
10846 (pass_ubsan::execute): Instrument aggregate arguments of calls.
10848 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
10852 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
10855 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
10858 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10860 * predict.c (unlikely_executed_edge_p): Move ahead.
10861 (probably_never_executed_edge_p): Use it.
10863 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10865 PR middle-end/79988
10866 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
10867 gimple_call_builtin_p call.
10869 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10871 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
10872 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
10873 rtl_check_failed_type2, rtl_check_failed_code1,
10874 rtl_check_failed_code2, rtl_check_failed_code_mode,
10875 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
10876 rtvec_check_failed_bounds, rtl_check_failed_flag,
10877 _fatal_insn_not_found, _fatal_insn): Likewise.
10878 * tree.h (tree_contains_struct_check_failed,
10879 tree_check_failed, tree_not_check_failed,
10880 tree_class_check_failed, tree_range_check_failed,
10881 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
10882 tree_vec_elt_check_failed, phi_node_elt_check_failed,
10883 tree_operand_check_failed, omp_clause_check_failed,
10884 omp_clause_operand_check_failed, omp_clause_range_check_failed):
10887 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10889 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
10890 flag_branch_probabilities.
10891 * ipa-inline.c (edge_badness): Likewise.
10892 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
10893 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
10894 * predict.c (maybe_hot_frequency_p): Likewise.
10895 (probably_never_executed): Likewise.
10896 * sched-ebb.c (schedule_ebbs): Likewise.
10897 * sched-rgn.c (find_single_block_region): Likewise.
10898 * tracer.c (tail_duplicate): Likewise.
10900 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10902 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
10903 longer requires x_flag_profile_use.
10905 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10907 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
10908 instead of flag_reorder_blocks_and_partition.
10909 * dbxout.c (dbxout_function_end): Likewise.
10910 * dwarf2out.c (gen_subprogram_die): Likewise.
10911 * haifa-sched.c (sched_create_recovery_edges): Likewise.
10912 * hw-doloop.c (reorg_loops): Likewise.
10913 * varasm.c (assemble_start_function,
10914 assemble_end_function): Likewise.
10915 (decide_function_section): Do not check for
10916 flag_reorder_blocks_and_partition.
10918 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10920 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10922 (chkp_get_hard_register_fake_addr_expr): Ditto.
10923 (chkp_build_addr_expr): Add check for hard reg case.
10924 (chkp_parse_array_and_component_ref): Ditto.
10925 (chkp_find_bounds_1): Ditto.
10926 (chkp_process_stmt): Don't generate bounds store for
10929 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10931 * predict.c (maybe_hot_bb_p): Do not check profile status.
10932 (maybe_hot_edge_p): Likewise.
10933 (probably_never_executed): Check for zero counts even if profile
10935 (unlikely_executed_edge_p): New function.
10936 (unlikely_executed_stmt_p): New function.
10937 (unlikely_executed_bb_p): New function.
10938 (set_even_probabilities): Use unlikely predicates.
10939 (combine_predictions_for_bb): Likewise.
10940 (predict_paths_for_bb): Likewise.
10941 (predict_paths_leading_to_edge): Likewise.
10942 (determine_unlikely_bbs): New function.
10943 (estimate_bb_frequencies): Use it.
10944 (compute_function_frequency): Use zero counts even if profile is
10946 * profile-count.h: Fix typo.
10948 2017-08-08 Julia Koval <julia.koval@intel.com>
10950 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
10951 _mm512_mask_cvtsepi16_storeu_epi8,
10952 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
10953 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
10954 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
10955 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
10956 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
10957 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
10958 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
10959 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
10960 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
10961 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
10962 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
10963 __builtin_ia32_pmovuswb256mem_mask,
10964 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
10965 __builtin_ia32_pmovwb512mem_mask): New builtins.
10967 2017-08-08 Julia Koval <julia.koval@intel.com>
10969 PR target/73350,80862
10970 * config/i386/subst.md (round): Fix round pattern.
10971 * config/i386/i386.c (ix86_erase_embedded_rounding):
10972 Fix erasing rounding for the fixed pattern.
10974 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10976 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
10978 2017-06-08 Martin Liska <mliska@suse.cz>
10980 PR gcov-profile/80911
10981 * gcov.c (block_info::block_info): New constructor.
10983 2017-06-07 Carl Love <cel@us.ibm.com>
10985 * config/rs6000/rs6000-c: The return type of the following
10986 built-in functions was implemented as int not long long. Fix sign
10987 of return value for the unsigned version of vec_mulo and vec_mule.
10988 vector unsigned long long vec_bperm (vector unsigned long long,
10989 vector unsigned char)
10990 vector signed long long vec_mule (vector signed int,
10992 vector unsigned long long vec_mule (vector unsigned int,
10993 vector unsigned int)
10994 vector signed long long vec_mulo (vector signed int,
10996 vector unsigned long long vec_mulo (vector unsigned int,
10997 vector unsigned int)
10998 * doc/extend.texi: Fix the documentation for the built-in
11001 2017-06-07 Carl Love <cel@us.ibm.com>
11004 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
11007 2017-06-07 Carl Love <cel@us.ibm.com>
11009 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
11010 support, Generate doublehv for signed int/float for BE case only.
11012 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
11014 * doc/invoke.texi (mcx16): Rewrite.
11016 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11018 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
11019 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
11020 *mov<mode>_softfloat, and an anonymous splitter): Use
11021 nonimmediate_operand instead of rs6000_nonimmediate_operand.
11023 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11025 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
11027 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
11028 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
11029 (rs6000_debug_reg_global): Adjust.
11030 (rs6000_init_hard_regno_mode_ok): Adjust.
11031 (rs6000_dbx_register_number): Adjust.
11032 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
11033 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
11034 Remove SPE_ACC and SPEFSCR.
11035 (REG_ALLOC_ORDER): Ditto.
11036 (FRAME_POINTER_REGNUM): Change to 111.
11037 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
11038 (REG_CLASS_NAMES): Ditto.
11039 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
11040 (REGISTER_NAMES): Ditto.
11041 (ADDITIONAL_REG_NAMES): Ditto.
11042 (rs6000_reg_names): Ditto.
11043 * config/rs6000/rs6000.md: Renumber some register number
11046 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11048 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
11050 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
11051 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
11053 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
11054 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
11055 Delete the SPE high registers.
11056 (REG_ALLOC_ORDER): Ditto.
11057 (enum reg_class): Remove SPE_HIGH_REGS.
11058 (REG_CLASS_NAMES): Ditto.
11059 (REG_CLASS_CONTENTS): Delete the SPE high registers.
11060 (REGISTER_NAMES): Ditto.
11061 (rs6000_reg_names): Ditto.
11062 * doc/tm.texi.in: Remove SPE as example.
11063 * doc/tm.texi: Regenerate.
11065 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11067 * config/rs6000/8540.md (ppc8540_brinc): Delete.
11068 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
11069 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
11070 * config/rs6000/rs6000.md (type): Remove "brinc".
11072 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11074 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
11075 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
11076 * config/rs6000/linuxspe.h: Delete file.
11077 * config/rs6000/rs6000.md: Don't include spe.md.
11078 * config/rs6000/spe.h: Delete file.
11079 * config/rs6000/spe.md: Delete file.
11080 * config/rs6000/t-rs6000: Remove spe.md.
11082 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11084 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
11085 (reg_or_none500mem_operand): Delete.
11086 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
11087 instead of reg_or_none500mem_operand.
11089 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11091 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
11092 handling of SPE flags.
11093 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
11095 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11097 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
11099 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
11100 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
11101 paired_divv2sf3): Similar.
11102 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
11103 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
11104 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
11106 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
11107 Rename the paired_* instruction patterns.
11108 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
11110 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
11111 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
11112 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
11113 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
11114 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
11115 PAIRED_VECTOR_MODE.
11116 (struct machine_function): Delete field spe_insn_chain_scanned_p.
11117 (spe_func_has_64bit_regs_p): Delete.
11118 (spe_expand_predicate_builtin): Delete.
11119 (spe_expand_evsel_builtin): Delete.
11120 (TARGET_DWARF_REGISTER_SPAN): Do not define.
11121 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
11122 (invalid_e500_subreg): Delete.
11123 (rs6000_legitimize_address): Always force_reg op2 as well, for
11124 paired single memory accesses.
11125 (rs6000_member_type_forces_blk): Delete.
11126 (rs6000_spe_function_arg): Delete.
11127 (rs6000_expand_unop_builtin): Delete SPE handling.
11128 (rs6000_expand_binop_builtin): Ditto.
11129 (spe_expand_stv_builtin): Delete.
11130 (bdesc_2arg_spe): Delete.
11131 (spe_expand_builtin): Delete.
11132 (spe_expand_predicate_builtin): Delete.
11133 (spe_expand_evsel_builtin): Delete.
11134 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
11135 (spe_init_builtins): Delete.
11136 (spe_func_has_64bit_regs_p): Delete.
11137 (savres_routine_name): Delete "info" parameter. Adjust callers.
11138 (rs6000_emit_stack_reset): Ditto.
11139 (rs6000_dwarf_register_span): Delete.
11140 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
11141 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
11142 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
11144 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
11146 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
11147 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
11148 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
11149 mulv2sf3, divv2sf3): Delete expanders.
11151 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11153 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
11155 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11157 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
11158 * config/rs6000/rs6000.c: Ditto.
11160 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11162 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
11163 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
11164 comparison_operator.
11166 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11168 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
11169 * config/rs6000/rs6000.opt: Ditto.
11170 * config/rs6000/t-rtems: Ditto.
11172 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11174 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
11175 TARGET_E500_SINGLE by 0, simplify.
11176 * config/rs6000/rs6000.c: Ditto.
11177 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
11178 (spe_build_register_parallel): Delete.
11179 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
11180 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
11181 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
11182 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
11183 (E500_CONVERT): Delete.
11184 * config/rs6000/spe.md: Remove many patterns and all define_constants.
11186 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11188 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
11189 * config/rs6000/dfp.md: Ditto.
11190 (negdd2, *negdd2_fpr): Merge.
11191 (absdd2, *absdd2_fpr): Merge.
11192 (negtd2, *negtd2_fpr): Merge.
11193 (abstd2, *abstd2_fpr): Merge.
11194 * config/rs6000/e500.h: Delete file.
11195 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
11196 TARGET_FPRS by 1 and simplify.
11197 * config/rs6000/rs6000-c.c: Ditto.
11198 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
11199 TARGET_DF_SPE by 0.
11200 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
11202 * config/rs6000/rs6000.md: Ditto.
11203 (floatdidf2, *floatdidf2_fpr): Merge.
11204 (move_from_CR_gt_bit): Delete.
11205 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
11206 (E500_CR_IOR_COMPARE): Delete.
11207 (All patterns that require !TARGET_FPRS): Delete.
11208 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
11210 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11212 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
11214 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11216 * graphds.c (add_edge): Intitialize edge's attached data.
11217 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
11218 pointer parameter. Call pointed function on each edge during
11219 graph traversing. Skip traversing the edge when the function
11221 (graphds_dfs, graphds_scc): Ditto.
11222 (for_each_edge): New parameter. Pass the new parameter to callback
11224 * graphds.h (skip_edge_callback): New function pointer type.
11225 (graphds_dfs, graphds_scc): New function pointer parameter.
11226 (graphds_edge_callback, for_each_edge): New parameter.
11228 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11230 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
11231 out code checking if runtime alias check is possible to below ...
11232 Call the new function.
11233 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
11234 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
11236 2017-06-07 Marek Polacek <polacek@redhat.com>
11239 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
11240 TYPE_OVERFLOW_WRAPS check.
11242 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11244 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
11245 if versioning is required.
11246 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
11247 peeling with the check for versioning.
11249 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11251 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
11252 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
11253 Set true to new parameter if new ssa variable is defined.
11254 (vect_gen_vector_loop_niters): Refactor. Set range information
11255 for the new vector loop bound variable.
11256 (vect_do_peeling): Ditto.
11258 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11260 * tree-affine.c (ssa.h): Include header file.
11261 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
11262 has wrapping overflow behavior.
11264 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11266 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
11268 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11270 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
11271 (tree_to_aff_combination): ... here.
11273 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11275 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
11276 reg_pressure model function.
11277 (ivopts_global_cost_for_size): Delete.
11278 (determine_set_costs, iv_ca_recount_cost): Call new model function
11279 ivopts_estimate_reg_pressure.
11281 2017-06-07 Tamar Christina <tamar.christina@arm.com>
11283 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
11284 expensive than udiv. Remove floating point cases from mod.
11286 2017-06-07 Tamar Christina <tamar.christina@arm.com>
11288 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
11289 Increase idiv cost.
11291 2017-06-07 Tamar Christina <tamar.christina@arm.com>
11293 * config/aarch64/aarch64.md
11294 (copysignsf3): Fix mask generation.
11296 2017-06-07 Jakub Jelinek <jakub@redhat.com>
11298 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
11300 (class dump_manager): Add register_dumps method.
11301 * dumpfile.c: Include langhooks.h.
11302 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
11303 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
11304 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
11305 (dump_manager::dump_register): Start with 512 entries instead of 32.
11306 (dump_manager::register_dumps): New method.
11307 * toplev.c (general_init): Instead of invoking register_dumps
11308 langhook, invoke register_dumps method on the dump manager.
11309 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
11312 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
11314 * doc/md.texi: Clarify the restrictions on a define_insn condition.
11315 Say that # requires an associated define_split to exist, and that
11316 the define_split must be suitable for use after register allocation.
11318 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11320 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
11321 (compute_outgoing_frequencies): Also initialize zero counts.
11322 (find_many_sub_basic_blocks): Do not produce uninitialized profile
11323 around loops; preserve more of profile when nothing changes.
11325 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
11327 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
11329 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
11330 * config/arm/arm-cpu-cdata.h: Regenerate.
11331 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
11332 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
11333 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
11334 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
11335 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
11337 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
11339 * config/arm/t-rmprofile: Likewise.
11340 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
11342 2017-06-06 David S. Miller <davem@davemloft.net>
11345 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
11346 blockage if function uses alloca.
11348 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11350 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
11351 New "uid" fields to hold pretty-print IDs of group and ref.
11352 Memory references are now identified as <group_id>:<ref_id>
11353 instead of using [random] addresses.
11354 (dump_mem_details): Simplify, no functional change.
11355 (dump_mem_ref): Simplify and make output more concise.
11356 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
11357 (find_or_create_group): Initialize group uid.
11358 (record_ref): Initialize ref uid. Improve debug output.
11359 (prune_group_by_reuse, should_issue_prefetch_p,)
11360 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
11361 (mark_nontemporal_store, determine_loop_nest_reuse):
11362 Improve debug output.
11364 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11366 * dbgcnt.def (prefetch): New debug counter.
11367 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
11368 (schedule_prefetches): Stop issueing prefetches if debug counter
11371 2017-06-06 Tom de Vries <tom@codesourcery.com>
11373 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
11374 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
11376 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11378 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
11379 Use aarch64_reg_or_zero predicate for operand 4.
11380 (aarch64_compare_and_swap<mode> define_insn_and_split):
11381 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
11382 (aarch64_store_exclusive<mode>): Likewise for operand 2.
11384 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
11386 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
11387 (arm_compute_save_core_reg_mask): This.
11388 (thumb1_compute_save_reg_mask): Rename into ...
11389 (thumb1_compute_save_core_reg_mask): This.
11390 (arm_compute_save_reg0_reg12_mask): Adapt comment.
11391 (arm_compute_frame_layout): Likewise.
11393 2017-06-06 Richard Biener <rguenther@suse.de>
11395 PR tree-optimization/80974
11396 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
11397 keep or clear leaders SSA info.
11399 2017-06-06 Tom de Vries <tom@codesourcery.com>
11401 * config/nvptx/nvptx.c (split_mode_p): New function.
11402 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
11404 2017-06-06 Tom de Vries <tom@codesourcery.com>
11406 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
11408 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11411 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
11414 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11416 * shrink-wrap.c (handle_simple_exit): Update profile.
11417 (try_shrink_wrapping): Upate profile.
11419 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11421 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
11422 (tree_guess_outgoing_edge_probabilities): New.
11423 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
11424 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
11426 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11428 * ipa-split.c (split_function): Initialize return bb profile.
11430 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11432 * profile.c (compute_branch_probabilities): Also initialize
11433 EXIT_BLOCK profile.
11435 2017-06-06 Richard Biener <rguenther@suse.de>
11437 PR tree-optimization/80928
11438 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
11439 (vect_analyze_loop_operations): Properly guard analysis for
11441 (vect_transform_loop): Likewise.
11442 (vect_analyze_loop_2): Also reset SLP type on PHIs.
11443 (vect_model_induction_cost): Do not cost for pure SLP.
11444 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
11445 of induction in inner loop vectorization.
11446 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
11447 (vect_get_and_check_slp_defs): Handle vect_induction_def.
11448 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
11450 (vect_analyze_slp_cost_1): Cost induction.
11451 (vect_detect_hybrid_slp_stmts): Handle PHIs.
11452 (vect_get_slp_vect_defs): Likewise.
11453 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
11454 (vect_transform_stmt): Handle SLP reductions.
11455 * tree-vectorizer.h (vectorizable_induction): Adjust.
11457 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
11459 * config/rs6000/rs6000.c (make_resolver_func): Update
11460 init_lowered_empty_function call.
11462 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
11464 * doc/invoke.texi: Document the -fprofile-abs-path option.
11465 * common.opt (fprofile-abs-path): New option.
11466 * gcov-io.h (gcov_write_filename): Declare.
11467 * gcov-io.c (gcov_write_filename): New function.
11468 * coverage.c (coverage_begin_function): Use gcov_write_filename.
11469 * profile.c (output_location): Likewise.
11471 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11473 * shring-wrap.c: Revert accidental commit.
11475 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
11477 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
11479 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11481 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
11483 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
11484 profile in callgraph edge.
11485 * profile-count.h (apply_probability): If THIS is 0, then result is 0
11486 (apply_scale): Likewise.
11487 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
11488 Also scale profile when inlining function with zero profile.
11489 (initialize_cfun): Update exit block profile even when it is zero.
11490 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
11491 when profile is read.
11493 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
11495 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
11496 (CLONE_*): New constants to define the processors we can generate
11497 code for with the target_clone attribute.
11498 (rs6000_clone_map): New array to identify which clone processors
11499 the current program is running on.
11500 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
11501 target_clone attribute.
11502 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
11503 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
11504 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
11505 (cpu_expand_builtin): Add support for target_clone attribute.
11506 (rs6000_valid_attribute_p): Allow "default" attribute.
11507 (get_decl_name): New debug function to simplify printing the
11508 current function name in debugging statements.
11509 (rs6000_clone_priority): New functions to support the target_clone
11510 attribute, and be able to generate code to switch between ISA 2.05
11511 through ISA 3.0 (power6 through power9).
11512 (rs6000_compare_version_priority): Likewise.
11513 (rs6000_get_function_versions_dispatcher): Likewise.
11514 (make_resolver_func): Likewise.
11515 (add_condition_to_bb): Likewise.
11516 (dispatch_function_versions): Likewise.
11517 (rs6000_generate_version_dispatcher_body): Likewise.
11518 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
11519 (fusion_gpr_load_p): Fix a spacing issue.
11520 * doc/extend.texi (Common Function Attributes): Document that the
11521 PowerPC supports the target_clone attribute.
11523 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
11525 * config/arm/arm.h: explain F symbol found in description of ARM
11526 register allocation in its legend.
11528 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11530 * config/mips/frame-header-opt.c: Include profile-count.h.
11531 * config/riscv/riscv.c: Include profile-count.h
11533 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11535 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
11537 (sm_set_flag_if_changed): Add bbs field.
11538 (execute_sm_if_changed_flag_set): Pass BBS.
11539 (execute_sm): Update.
11541 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11543 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
11546 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11548 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
11549 (peephole2): New peephole2 to emit the above.
11550 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
11552 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11554 * config/aarch64/aarch64.c (define_peephole2 above
11555 *sub_<shift>_<mode>): New peephole.
11557 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11559 * config/i386/i386.c (make_resolver_func): Update.
11560 * Makefile.in: Add profile-count.h and profile-count.o
11561 * auto-profile.c (afdo_indirect_call): Update to new API.
11562 (afdo_set_bb_count): Update.
11563 (afdo_propagate_edge): Update.
11564 (afdo_propagate_circuit): Update.
11565 (afdo_calculate_branch_prob): Update.
11566 (afdo_annotate_cfg): Update.
11567 * basic-block.h: Include profile-count.h
11568 (struct edge_def): Turn count to profile_count.
11569 (struct basic_block_def): Likewie.
11570 (REG_BR_PROB_BASE): Move to profile-count.h
11571 (RDIV): Move to profile-count.h
11572 * bb-reorder.c (max_entry_count): Turn to profile_count.
11573 (find_traces): Update.
11574 (rotate_loop):Update.
11575 (connect_traces):Update.
11576 (sanitize_hot_paths):Update.
11577 * bt-load.c (migrate_btr_defs): Update.
11578 * cfg.c (RDIV): Remove.
11579 (init_flow): Use alloc_block.
11580 (alloc_block): Uninitialize count.
11581 (unchecked_make_edge): Uninitialize count.
11582 (check_bb_profile): Update.
11583 (dump_edge_info): Update.
11584 (dump_bb_info): Update.
11585 (update_bb_profile_for_threading): Update.
11586 (scale_bbs_frequencies_int): Update.
11587 (scale_bbs_frequencies_gcov_type): Update.
11588 (scale_bbs_frequencies_profile_count): New.
11589 * cfg.h (update_bb_profile_for_threading): Update.
11590 (scale_bbs_frequencies_profile_count): Declare.
11591 * cfgbuild.c (compute_outgoing_frequencies): Update.
11592 (find_many_sub_basic_blocks): Update.
11593 * cfgcleanup.c (try_forward_edges): Update.
11594 (try_crossjump_to_edge): Update.
11595 * cfgexpand.c (expand_gimple_tailcall): Update.
11596 (construct_exit_block): Update.
11597 * cfghooks.c (verify_flow_info): Update.
11598 (dump_bb_for_graph): Update.
11599 (split_edge): Update.
11600 (make_forwarder_block): Update.
11601 (duplicate_block): Update.
11602 (account_profile_record): Update.
11603 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11604 (get_estimated_loop_iterations): Update.
11605 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11606 (single_likely_exit): Update.
11607 * cfgloopmanip.c (scale_loop_profile): Update.
11609 (set_zero_probability): Update.
11610 (lv_adjust_loop_entry_edge): Update.
11611 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
11612 (purge_dead_edges): Update.
11613 (rtl_account_profile_record): Update.
11614 * cgraph.c (cgraph_node::create): Uninitialize count.
11615 (symbol_table::create_edge): Uninitialize count.
11616 (cgraph_update_edges_for_call_stmt_node): Update.
11617 (cgraph_edge::dump_edge_flags): Update.
11618 (cgraph_node::dump): Update.
11619 (cgraph_edge::maybe_hot_p): Update.
11620 * cgraph.h: Include profile-count.h
11621 (create_clone), create_edge, create_indirect_edge): Update.
11622 (cgraph_node): Turn count to profile_count.
11623 (cgraph_edge0: Likewise.
11624 (make_speculative, clone): Update.
11625 (create_edge): Update.
11626 (init_lowered_empty_function): Update.
11627 * cgraphclones.c (cgraph_edge::clone): Update.
11628 (duplicate_thunk_for_node): Update.
11629 (cgraph_node::create_clone): Update.
11630 * cgraphunit.c (cgraph_node::analyze): Update.
11631 (cgraph_node::expand_thunk): Update.
11632 * final.c (dump_basic_block_info): Update.
11633 * gimple-streamer-in.c (input_bb): Update.
11634 * gimple-streamer-out.c (output_bb): Update.
11635 * graphite.c (print_global_statistics): Update.
11636 (print_graphite_scop_statistics): Update.
11637 * hsa-brig.c: Include basic-block.h.
11638 * hsa-dump.c: Include basic-block.h.
11639 * hsa-gen.c (T sum_slice): Update.
11640 (convert_switch_statements):Update.
11641 * hsa-regalloc.c: Include basic-block.h.
11642 * ipa-chkp.c (chkp_produce_thunks): Update.
11643 * ipa-cp.c (struct caller_statistics): Update.
11644 (init_caller_stats): Update.
11645 (gather_caller_stats): Update.
11646 (ipcp_cloning_candidate_p): Update.
11647 (good_cloning_opportunity_p): Update.
11648 (get_info_about_necessary_edges): Update.
11649 (dump_profile_updates): Update.
11650 (update_profiling_info): Update.
11651 (update_specialized_profile): Update.
11652 (perhaps_add_new_callers): Update.
11653 (decide_about_value): Update.
11654 (ipa_cp_c_finalize): Update.
11655 * ipa-devirt.c (struct odr_type_warn_count): Update.
11656 (struct decl_warn_count): Update.
11657 (struct final_warning_record): Update.
11658 (possible_polymorphic_call_targets): Update.
11659 (ipa_devirt): Update.
11660 * ipa-fnsummary.c (redirect_to_unreachable): Update.
11661 * ipa-icf.c (sem_function::merge): Update.
11662 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
11663 * ipa-inline.c (compute_uninlined_call_time): Update.
11664 (compute_inlined_call_time): Update.
11665 (want_inline_small_function_p): Update.
11666 (want_inline_self_recursive_call_p): Update.
11667 (edge_badness): Update.
11668 (lookup_recursive_calls): Update.
11669 (recursive_inlining): Update.
11670 (inline_small_functions): Update.
11671 (dump_overall_stats): Update.
11672 (dump_inline_stats): Update.
11673 * ipa-profile.c (ipa_profile_generate_summary): Update.
11674 (ipa_propagate_frequency): Update.
11675 (ipa_profile): Update.
11676 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11677 * ipa-utils.c (ipa_merge_profiles): Update.
11678 * loop-doloop.c (doloop_modify): Update.
11679 * loop-unroll.c (report_unroll): Update.
11680 (unroll_loop_runtime_iterations): Update.
11681 * lto-cgraph.c (lto_output_edge): Update.
11682 (lto_output_node): Update.
11683 (input_node): Update.
11684 (input_edge): Update.
11685 (merge_profile_summaries): Update.
11686 * lto-streamer-in.c (input_cfg): Update.
11687 * lto-streamer-out.c (output_cfg): Update.
11688 * mcf.c (create_fixup_graph): Update.
11689 (adjust_cfg_counts): Update.
11690 (sum_edge_counts): Update.
11691 * modulo-sched.c (sms_schedule): Update.
11692 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11693 * predict.c (maybe_hot_count_p): Update.
11694 (probably_never_executed): Update.
11695 (dump_prediction): Update.
11696 (combine_predictions_for_bb): Update.
11697 (propagate_freq): Update.
11698 (handle_missing_profiles): Update.
11699 (counts_to_freqs): Update.
11700 (rebuild_frequencies): Update.
11701 (force_edge_cold): Update.
11702 * predict.h: Include profile-count.h
11703 (maybe_hot_count_p, counts_to_freqs): UPdate.
11704 * print-rtl-function.c: Do not include cfg.h
11705 * print-rtl.c: Include basic-block.h
11706 * profile-count.c: New file.
11707 * profile-count.h: New file.
11708 * profile.c (is_edge_inconsistent): Update.
11709 (correct_negative_edge_counts): Update.
11710 (is_inconsistent): Update.
11711 (set_bb_counts): Update.
11712 (read_profile_edge_counts): Update.
11713 (compute_frequency_overlap): Update.
11714 (compute_branch_probabilities): Update; Initialize and deinitialize
11716 (branch_prob): Update.
11717 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
11718 (edge_gcov_count): New.
11719 (bb_gcov_count): New.
11720 * shrink-wrap.c (try_shrink_wrapping): Update.
11721 * tracer.c (better_p): Update.
11722 * trans-mem.c (expand_transaction): Update.
11723 (ipa_tm_insert_irr_call): Update.
11724 (ipa_tm_insert_gettmclone_call): Update.
11725 * tree-call-cdce.c: Update.
11726 * tree-cfg.c (gimple_duplicate_sese_region): Update.
11727 (gimple_duplicate_sese_tail): Update.
11728 (gimple_account_profile_record): Update.
11729 (execute_fixup_cfg): Update.
11730 * tree-inline.c (copy_bb): Update.
11731 (copy_edges_for_bb): Update.
11732 (initialize_cfun): Update.
11733 (freqs_to_counts): Update.
11734 (copy_cfg_body): Update.
11735 (expand_call_inline): Update.
11736 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11737 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
11738 (try_unroll_loop_completely): Update.
11739 (try_peel_loop): Update.
11740 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11741 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
11742 * tree-ssa-loop-split.c (connect_loops): Update.
11743 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11744 * tree-ssa-reassoc.c (branch_fixup): Update.
11745 * tree-ssa-tail-merge.c (replace_block_by): Update.
11746 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
11747 (compute_path_counts): Update.
11748 (update_profile): Update.
11749 (recompute_probabilities): Update.
11750 (update_joiner_offpath_counts): Update.
11751 (estimated_freqs_path): Update.
11752 (freqs_to_counts_path): Update.
11753 (clear_counts_path): Update.
11754 (ssa_fix_duplicate_block_edges): Update.
11755 (duplicate_thread_path): Update.
11756 * tree-switch-conversion.c (case_bit_test_cmp): Update.
11757 (struct switch_conv_info): Update.
11758 * tree-tailcall.c (decrease_profile): Update.
11759 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11760 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11761 * value-prof.c (check_counter): Update.
11762 (gimple_divmod_fixed_value): Update.
11763 (gimple_mod_pow2): Update.
11764 (gimple_mod_subtract): Update.
11765 (gimple_ic_transform): Update.
11766 (gimple_stringop_fixed_value): Update.
11767 * value-prof.h (gimple_ic): Update.
11769 2017-06-02 Carl Love <cel@us.ibm.com>
11771 * config/rs6000/rs6000-c: Add support for built-in functions
11772 vector double vec_doublee (vector signed int);
11773 vector double vec_doublee (vector unsigned int);
11774 vector double vec_doublee (vector float);
11775 vector double vec_doubleh (vector signed int);
11776 vector double vec_doubleh (vector unsigned int);
11777 vector double vec_doubleh (vector float);
11778 vector double vec_doublel (vector signed int);
11779 vector double vec_doublel (vector unsigned int);
11780 vector double vec_doublel (vector float);
11781 vector double vec_doubleo (vector signed int);
11782 vector double vec_doubleo (vector unsigned int);
11783 vector double vec_doubleo (vector float);.
11784 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
11785 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
11787 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
11788 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
11789 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
11791 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
11792 vec_doublel, vec_doubleh.
11793 * doc/extend.texi: Update the built-in documentation file for the
11794 new built-in functions.
11796 2017-06-02 David Malcolm <dmalcolm@redhat.com>
11799 * ipa-inline-analysis.c (free_growth_caches): Set
11800 edge_removal_hook_holder to NULL after removing it.
11802 2017-06-02 Sudakshina Das <sudi.das@arm.com>
11804 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
11805 comparision with zero.
11807 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
11808 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11809 for early expansion of vec_min and vec_max builtins.
11810 (builtin_function_type): Add min/max unsigned variants to those
11811 identified as having unsigned arguments.
11813 2017-06-02 Olivier Hainque <hainque@adacore.com>
11815 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
11817 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11819 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
11820 Use VALL_F16 iterator rather than VALL.
11822 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11824 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
11825 Emit CBNZ inside loop when doing a strong exchange and comparing
11826 against zero. Generate the CC flags after the loop.
11828 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
11830 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
11831 (dl_section_ref): New.
11832 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
11833 On AIX, append an expression to subtract the size of the
11834 section length to dl_section_ref.
11836 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
11838 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11839 for early expansion of vector absolute builtins.
11841 2017-06-02 Richard Biener <rguenther@suse.de>
11843 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
11844 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
11846 2017-06-02 Richard Biener <rguenther@suse.de>
11848 PR tree-optimization/80948
11849 * tree-tailcall.c (find_tail_calls): Track stmts to move in
11850 stmt order as well.
11852 2017-06-02 Richard Biener <rguenther@suse.de>
11854 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
11856 * tree-vect-stmts.c (process_use): Do not mark backedge defs
11857 for inductions as relevant.
11859 2017-06-02 Richard Biener <rguenther@suse.de>
11861 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
11862 (vectorizable_induction): ... this. Remove dead code.
11864 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
11866 * builtins. (expand_builtin_alloca): Remove second parameter and
11867 infer its value from the first parameter instead.
11868 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
11870 2017-06-02 Jakub Jelinek <jakub@redhat.com>
11872 PR rtl-optimization/80903
11873 * loop-doloop.c (add_test): Unshare sequence.
11875 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
11877 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
11879 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
11881 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
11883 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
11884 xlogue_layout::get_instance, logue_layout::xlogue_layout,
11885 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
11886 (xlogue_layout::get_stub_rtx): Make static.
11887 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
11888 (xlogue_layout::compute_stub_managed_regs): Rename to...
11889 (xlogue_layout::count_stub_managed_regs): ...this.
11890 (xlogue_layout::is_stub_managed_reg): New function.
11891 (xlogue_layout::m_stub_names): Rename to...
11892 (xlogue_layout::s_stub_names): ...this, make static.
11893 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
11894 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
11895 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
11896 xlogue_layout::s_stub_names): Instantiate statics.
11897 (stub_managed_regs): Remove.
11898 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
11899 (disable_call_ms2sysv_xlogues): Rename to...
11900 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
11901 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
11903 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
11904 change after reload_completed.
11905 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
11907 (ix86_expand_prologue): Likewise.
11908 (ix86_expand_epilogue): Likewise.
11909 (ix86_expand_split_stack_prologue): Likewise.
11910 (ix86_compute_frame_layout): Remove frame parameter ...
11911 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
11912 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
11914 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
11915 (ix86_frame): Move from here ...
11916 * config/i386/i386.h (ix86_frame): ... to here.
11917 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
11918 complete ix86_frame data structure instead. Remove some_ld_name.
11920 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
11922 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
11923 symbols that hold a DECL_VALUE_EXPR.
11925 2017-06-01 Martin Jambor <mjambor@suse.cz>
11927 PR tree-optimization/80898
11928 * tree-sra.c (process_subtree_disqualification): Removed.
11929 (disqualify_candidate): Do not acll
11930 process_subtree_disqualification.
11931 (subtree_mark_written_and_enqueue): New function.
11932 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
11933 RHS has been disqualified and re-queue LHS if necessary. Apart
11934 from that, ignore disqualified RHS.
11936 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11938 * config/s390/s390.c (s390_emit_epilogue): Disable early return
11939 address fetch for z10 or later.
11941 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11943 * config/arc/arc.md (tst_movb): Add guard when splitting.
11945 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11947 * config/arc/arc.c (arc_can_eliminate): Test against
11948 arc_frame_pointer_needed.
11950 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11952 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
11953 to prevent store reordering.
11954 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
11955 (type): Add block type.
11956 (stack_tie): Define special instruction to be used in
11959 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11961 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
11962 constraint. It is not valid for the pattern.
11963 (noncommutative_binary_comparison): Likewise.
11965 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11967 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
11970 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11972 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
11973 be used by the reg-alloc.
11975 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11977 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
11978 reg-alloc when having mul64 or mul32x16 instructions.
11979 (mulsidi3): Likewise.
11980 (umulsidi3): Likewise.
11981 (mulsi32x16): New pattern.
11982 (mulsi64): Likewise.
11983 (mulsidi64): Likewise.
11984 (umulsidi64): Likewise.
11985 (MUL32x16_REG): Define.
11986 (mul64_600): Use MUL32x16_REG.
11987 (mac64_600): Likewise.
11988 (umul64_600): Likewise.
11989 (umac64_600): Likewise.
11991 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11993 * config/arc/arc.md (mulsi3_700): Make it commutative.
11995 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
11997 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
11999 (*sign_extendsidi2_insn): Likewise for movstosw.
12001 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
12003 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
12004 the type of the input discriminant value. Convert the
12005 discriminant value of signedness vary.
12007 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
12009 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
12010 Add to -Wall section.
12012 2017-06-01 Richard Biener <rguenther@suse.de>
12014 PR middle-end/66313
12015 * fold-const.c (fold_plusminus_mult_expr): If the factored
12016 factor may be zero use a wrapping type for the inner operation.
12017 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
12018 and handle moved defs.
12019 (process_assignment): Properly guard the unary op case. Return a
12020 tri-state indicating that moving the stmt before the call may allow
12021 to continue. Pass through to_move.
12022 (find_tail_calls): Handle moving unrelated defs before
12025 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
12028 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
12029 splitter result in the canonical way.
12031 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
12033 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
12034 also for 32bit target. Update insn attributes.
12035 (zero-extendsidi2 splitter): Allow all registers for operand 1.
12037 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
12039 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
12040 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
12041 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
12042 (_mm_maskz_min_ss): New intrinsics.
12044 2017-05-31 Martin Liska <mliska@suse.cz>
12046 * tree-vect-loop.c (vect_create_epilog_for_reduction):
12047 Change comment style to one we normally use.
12048 (vectorizable_reduction): Likewise.
12049 (vectorizable_induction): Likewise.
12050 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12051 (vectorizable_call): Likewise.
12052 (vectorizable_simd_clone_call): Likewise.
12053 (vectorizable_conversion): Likewise.
12054 (vectorizable_assignment): Likewise.
12055 (vectorizable_shift): Likewise.
12056 (vectorizable_operation): Likewise.
12057 (vectorizable_store): Likewise.
12058 (vectorizable_load): Likewise.
12059 * tree-vectorizer.h: Likewise.
12061 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
12063 * passes.c (emergency_dump_function): New.
12064 * tree-pass.h (emergency_dump_function): Declare.
12065 * plugin.c (plugins_internal_error_function): Remove.
12066 * plugin.h (plugins_internal_error_function): Remove declaration.
12067 * toplev.c (internal_error_function): New static function. Use it...
12068 (general_init): ...here.
12070 2017-05-31 Graham Markall <graham.markall@embecosm.com>
12072 * config/arc/arc.c (arc_print_operand): Handle constant operands.
12073 (arc_rtx_costs): Add costs for new patterns.
12074 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
12075 * config/arc/predicates.md: Add _1_2_3_operand predicate.
12077 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
12079 * tree-ssa-strlen.c (get_next_strinfo): New function.
12080 (get_stridx_plus_constant): Use it.
12081 (zero_length_string): Likewise.
12082 (adjust_related_strinfos): Likewise.
12083 (adjust_last_stmt): Likewise.
12085 2017-05-31 Richard Biener <rguenther@suse.de>
12088 * config/i386/i386.c (ix86_expand_builtin): Remove assert
12089 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
12091 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
12093 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
12094 loop_vinfo argument and use of dependence distance vectors.
12095 Check instead whether the two references differ only in their
12096 initial value and assume that they have the same alignment if the
12097 difference is a multiple of the vector alignment.
12098 (vect_analyze_data_refs_alignment): Update call accordingly.
12100 2017-05-31 Martin Liska <mliska@suse.cz>
12103 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
12105 2017-05-31 Bin Cheng <bin.cheng@arm.com>
12107 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
12108 (create_intersect_range_checks): Move from ...
12109 * tree-data-ref.c (create_intersect_range_checks_index)
12110 (create_intersect_range_checks): ... to here.
12111 (create_runtime_alias_checks): New function factored from ...
12112 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
12113 here. Call above function.
12114 * tree-data-ref.h (create_runtime_alias_checks): New function.
12116 2017-05-31 Bin Cheng <bin.cheng@arm.com>
12118 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
12119 segment length for dr_b and compute it in wide_int.
12121 2017-05-31 Richard Biener <rguenther@suse.de>
12123 PR tree-optimization/80906
12124 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
12125 and pass through iv_map.
12126 (copy_bb_and_scalar_dependences): Adjust.
12127 (translate_pending_phi_nodes): Likewise.
12128 (copy_loop_close_phi_args): Handle code-generating IVs instead
12131 2017-05-30 David Malcolm <dmalcolm@redhat.com>
12133 * diagnostic-color.c (color_dict): Add "type-diff".
12134 (parse_gcc_colors): Update comment.
12135 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
12136 -fdiagnostics-show-template-tree and -fno-elide-type.
12137 (GCC_COLORS): Add type-diff to example.
12139 (-fdiagnostics-show-template-tree): New.
12140 (-fno-elide-type): New.
12141 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
12142 the pp_format_decoder callback. Call any m_format_postprocessor's
12144 (pretty_printer::pretty_printer): Initialize
12145 m_format_postprocessor.
12146 (pretty_printer::~pretty_printer): Delete any
12147 m_format_postprocessor.
12148 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
12149 (class format_postprocessor): New class.
12150 (struct pretty_printer::format_decoder): Document the new parameters.
12151 (struct pretty_printer::m_format_postprocessor): New field.
12152 * tree-diagnostic.c (default_tree_printer): Update for new
12153 bool and const char ** params.
12154 * tree-diagnostic.h (default_tree_printer): Likewise.
12156 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
12158 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
12159 (lwa_operand): Delete rs6000_gen_cell_microcode test.
12160 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
12161 rs6000_gen_cell_microcode code.
12162 (rs6000_final_prescan_insn): Delete.
12163 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
12164 "warn-cell-microcode" entries.
12165 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
12166 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
12167 throughout. Change cc_reg_not_micro_cr0_operand to
12168 cc_reg_not_cr0_operand throughout.
12169 (*extendhi<mode>2_noload): Delete.
12170 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
12171 (mwarn-cell-microcode): Delete.
12172 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
12173 -mgen-cell-microcode and -mwarn-cell-microcode.
12175 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
12178 * config/i386/constraints.md (Yd): New constraint.
12180 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
12181 and (?Yd, r) alternatives. Update insn attributes.
12182 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
12183 and (?*Yd, r) alternatives. Update insn attributes.
12184 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
12186 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
12188 * gimplify.c (gimplify_modify_expr): Don't create a
12189 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
12192 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
12194 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
12196 2017-05-30 Richard Biener <rguenther@suse.de>
12198 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
12199 and reduc_def fields.
12200 (STMT_VINFO_REDUC_TYPE): New define.
12201 (STMT_VINFO_REDUC_DEF): Likewise.
12202 (vect_force_simple_reduction): Adjust prototype.
12203 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
12204 (vect_is_simple_reduction): Remove check_reduction argument.
12205 (vect_force_simple_reduction): Adjust and set
12206 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
12207 (vectorizable_reduction): Do not re-do reduction analysis
12208 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
12209 * tree-parloops.c (gather_scalar_reductions): Adjust.
12211 2017-05-30 Richard Biener <rguenther@suse.de>
12213 PR middle-end/80901
12214 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
12217 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
12219 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
12220 Introduce unknown_misalignment parameter and remove vf.
12221 (vect_peeling_hash_get_lowest_cost):
12222 Pass unknown_misalignment parameter.
12223 (vect_enhance_data_refs_alignment):
12224 Fix unsupportable data ref treatment.
12226 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12228 * tree-vect-data-refs.c (vect_get_data_access_cost):
12229 Workaround for SLP handling.
12230 (vect_enhance_data_refs_alignment):
12231 Compute costs for doing no peeling at all, compare to the best
12232 peeling costs so far and avoid peeling if cheaper.
12234 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12236 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
12237 Return peeling info and set costs to zero for unlimited cost
12239 (vect_enhance_data_refs_alignment): Also inspect all datarefs
12240 with unknown misalignment. Compute and costs for unknown
12241 misalignment, compare them to the costs for known misalignment
12242 and choose the cheapest for peeling.
12244 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12246 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
12247 (vect_get_peeling_costs_all_drs): Create function.
12248 (vect_peeling_hash_get_lowest_cost):
12249 Use vect_get_peeling_costs_all_drs.
12250 (vect_peeling_supportable): Create function.
12251 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
12253 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12255 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
12256 DR_HAS_NEGATIVE_STEP.
12257 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
12258 (vect_enhance_data_refs_alignment): Use.
12259 (vect_duplicate_ssa_name_ptr_info): Use.
12260 * tree-vectorizer.h (dr_misalignment): Use.
12261 (known_alignment_for_access_p): Use.
12263 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12266 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
12268 (has_section_name): New function.
12270 2017-05-30 Martin Liska <mliska@suse.cz>
12273 * auto-profile.c (get_function_decl_from_block): Fix
12276 2017-05-30 Richard Biener <rguenther@suse.de>
12278 PR middle-end/80876
12279 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
12281 2017-05-30 Martin Liska <mliska@suse.cz>
12283 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
12284 * dumpfile.h (struct dump_file_info): Remove ctors.
12286 2017-05-30 Martin Liska <mliska@suse.cz>
12288 * predict.def: Fix GNU coding style.
12290 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
12292 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
12293 Mark 'to' argument with ATTRIBUTE_UNUSED.
12295 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
12297 * config/xtensa/xtensa.c (xtensa_emit_call): Use
12298 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
12299 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
12302 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
12304 * doc/install.texi (Options specification): Restore entry of
12305 --enable-sjlj-exceptions.
12307 2017-05-27 Michael Eager <eager@eagercon.com>
12310 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
12312 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
12314 * config/microblaze/microblaze.h
12315 (FIXED_REGISTERS): Update in macro.
12316 (CALL_USED_REGISTERS): Update in macro.
12318 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
12320 * doc/install.texi: Add links to macOS binary distributions.
12322 2017-05-27 Jakub Jelinek <jakub@redhat.com>
12326 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12328 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
12330 2017-05-26 Martin Liska <mliska@suse.cz>
12332 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
12334 2017-05-26 Martin Liska <mliska@suse.cz>
12336 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
12337 always leading ';; '.
12338 (dump_bb_info): Likewise.
12339 (brief_dump_cfg): Likewise.
12340 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
12341 * dumpfile.c: Remove usage of TDF_VERBOSE.
12342 * dumpfile.h (enum dump_kind): Likewise.
12343 (dump_gimple_bb_header): Do not use TDF_COMMENT.
12344 * print-tree.c (debug_verbose): Remove.
12345 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
12346 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
12347 * tree-diagnostic.c (default_tree_printer): Replace
12348 TDF_DIAGNOSTIC with TDF_SLIM.
12350 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12352 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
12353 in parameter loop, rather than loop_vinfo.
12354 (create_intersect_range_checks): Ditto.
12355 (vect_create_cond_for_alias_checks): Update call to above functions.
12357 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12359 PR tree-optimization/80815
12360 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
12361 for merging runtime alias checks. Handle negative DR_STEPs.
12363 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12365 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
12367 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
12368 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
12369 out code pruning runtime alias checks.
12370 * tree-data-ref.c (prune_runtime_alias_test_list): New function
12371 factored out from above.
12372 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
12374 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
12376 (prune_runtime_alias_test_list): New decalaration.
12378 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12380 * tree-vect-data-refs.c (compare_tree): Rename and move ...
12381 * tree-data-ref.c (data_ref_compare_tree): ... to here.
12382 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
12383 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
12384 (operator==, comp_dr_with_seg_len_pair): Ditto.
12385 (vect_prune_runtime_alias_test_list): Ditto.
12387 2017-05-26 Martin Liska <mliska@suse.cz>
12390 * params.def: Bound partial-inlining-entry-probability param.
12392 2017-05-26 Marek Polacek <polacek@redhat.com>
12395 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
12398 2017-05-26 Richard Biener <rguenther@suse.de>
12400 PR tree-optimization/80842
12401 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
12404 2017-05-26 Richard Biener <rguenther@suse.de>
12406 PR tree-optimization/80844
12407 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
12409 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
12411 * doc/md.texi (Machine Constraints): Update x86 family
12412 machine constraints section to match 'config/i386/constraints.md'.
12414 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
12416 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
12418 2017-05-25 Nathan Sidwell <nathan@acm.org>
12420 * doc/invoke.texi (--enable-languages): Update documentation.
12422 2017-05-25 Martin Liska <mliska@suse.cz>
12424 * dumpfile.c: Add TDF_FOLDING.
12425 * dumpfile.h (enum dump_kind): Likewise.
12426 * genmatch.c (dt_simplify::gen_1): Use it.
12428 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12430 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
12432 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12434 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
12435 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
12437 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12439 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
12440 * match.pd (X == C): Rewrite it here.
12441 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
12442 with_certain_nonzero_bits2): New predicates.
12443 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
12445 2017-05-24 Nathan Sidwell <nathan@acm.org>
12447 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
12450 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
12453 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
12455 * config/powerpcspe: New port. Files are copied from the rs6000
12456 port, with "rs6000" in filenames replaced by "powerpcspe".
12458 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
12460 PR rtl-optimization/80754
12461 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
12463 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
12465 * config/sparc/sparc.md (length): Return the correct value for -mflat
12466 sibcalls to match output_sibcall.
12468 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
12472 * config/rs6000/rs6000.c (struct machine_function): Add new field
12474 (rs6000_get_separate_components): Init that field, use it.
12475 (rs6000_components_for_bb): Use the field.
12477 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
12479 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
12481 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
12483 PR middle-end/80823
12484 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
12486 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12489 * config/s390/s390.c (s390_check_qrst_address): Check incoming
12490 address against address_operand predicate.
12491 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
12493 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
12495 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
12496 parameters passed indirectly.
12498 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
12500 * config/i386/i386.md (*movdi_internal): Remove SSE4
12501 alternative 18 (?r, *v). Update insn attributes.
12502 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
12503 Update insn attributes.
12504 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
12505 Update insn attributes.
12506 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
12507 alternative 1 (r, v). Remove isa attribute.
12508 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
12509 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
12510 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
12512 2017-05-23 Tom de Vries <tom@codesourcery.com>
12514 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
12517 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12519 * cgraphunit.c (symbol_table::process_new_functions): Update.
12520 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
12521 (inline_generate_summary): Rename to ...
12522 (ipa_fn_summary_generate): ... this one.
12523 (inline_read_summary): Rename to ...
12524 (ipa_fn_summary_read): ... this one.
12525 (inline_write_summary): Rename to ...
12526 (ipa_fn_summary_write): ... this one.
12527 (inline_free_summary): Rename to ...
12528 (ipa_free_fn_summary): ... this one.
12529 (pass_data_local_fn_summary, pass_local_fn_summary,
12530 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
12531 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
12532 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
12533 make_pass_ipa_fn_summary): New.
12534 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
12535 inline_write_summary, inline_free_summary): Remove.
12536 (ipa_free_fn_summary) : New.
12537 * ipa-inline.c (ipa_inline): Update.
12538 (pass_ipa_inline): Do not generate summaries.
12539 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
12541 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
12542 and add pass_ipa_fn_summary.
12543 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
12545 (make_pass_inline_parameters): Remove.
12547 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
12549 * omp-low.c (struct omp_context): Remove "default_kind" member.
12552 * omp-offload.c (execute_oacc_device_lower): Remove the
12553 parallelism dimensions function attributes for unparallelized
12554 OpenACC kernels constructs.
12556 2017-05-23 Martin Liska <mliska@suse.cz>
12558 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
12560 (cgraph_edge::make_speculative): Likewise.
12561 (cgraph_edge::resolve_speculation): Likewise.
12562 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
12563 (cgraph_node::dump): Likewise.
12564 * cgraph.h: Likewise.
12565 * cgraphunit.c (analyze_functions): Likewise.
12566 (symbol_table::compile): Likewise.
12567 * ipa-cp.c (print_all_lattices): Likewise.
12568 (determine_versionability): Likewise.
12569 (initialize_node_lattices): Likewise.
12570 (ipcp_verify_propagated_values): Likewise.
12571 (estimate_local_effects): Likewise.
12572 (update_profiling_info): Likewise.
12573 (create_specialized_node): Likewise.
12574 (perhaps_add_new_callers): Likewise.
12575 (decide_about_value): Likewise.
12576 (decide_whether_version_node): Likewise.
12577 (identify_dead_nodes): Likewise.
12578 (ipcp_store_bits_results): Likewise.
12579 * ipa-devirt.c (dump_targets): Likewise.
12580 (ipa_devirt): Likewise.
12581 * ipa-icf.c (sem_item::dump): Likewise.
12582 (sem_function::equals): Likewise.
12583 (sem_variable::equals): Likewise.
12584 (sem_item_optimizer::read_section): Likewise.
12585 (sem_item_optimizer::execute): Likewise.
12586 (congruence_class::dump): Likewise.
12587 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
12588 (dump_inline_summary): Likewise.
12589 (estimate_node_size_and_time): Likewise.
12590 (inline_analyze_function): Likewise.
12591 * ipa-inline-transform.c (inline_call): Likewise.
12592 * ipa-inline.c (report_inline_failed_reason): Likewise.
12593 (want_early_inline_function_p): Likewise.
12594 (edge_badness): Likewise.
12595 (update_edge_key): Likewise.
12596 (inline_small_functions): Likewise.
12597 * ipa-profile.c (ipa_profile): Likewise.
12598 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
12599 (ipa_make_edge_direct_to_target): Likewise.
12600 (remove_described_reference): Likewise.
12601 (ipa_impossible_devirt_target): Likewise.
12602 (propagate_controlled_uses): Likewise.
12603 (ipa_print_node_params): Likewise.
12604 (ipcp_transform_function): Likewise.
12605 * ipa-pure-const.c (pure_const_read_summary): Likewise.
12606 (propagate_pure_const): Likewise.
12607 * ipa-reference.c (generate_summary): Likewise.
12608 (read_write_all_from_decl): Likewise.
12609 (propagate): Likewise.
12610 (ipa_reference_read_optimization_summary): Likewise.
12611 * ipa-utils.c (ipa_merge_profiles): Likewise.
12612 * ipa.c (walk_polymorphic_call_targets): Likewise.
12613 (symbol_table::remove_unreachable_nodes): Likewise.
12614 (ipa_single_use): Likewise.
12615 * passes.c (execute_todo): Likewise.
12616 * predict.c (drop_profile): Likewise.
12617 * symtab.c (symtab_node::get_dump_name): New function.
12618 (symtab_node::dump_name): Likewise.
12619 (symtab_node::dump_asm_name): Likewise.
12620 (symtab_node::dump_references): Likewise.
12621 (symtab_node::dump_referring): Likewise.
12622 (symtab_node::dump_base): Likewise.
12623 (symtab_node::debug_symtab): Likewise.
12624 * tree-sra.c (convert_callers_for_node): Likewise.
12625 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
12626 * value-prof.c (init_node_map): Likewise.
12628 2017-05-23 Martin Liska <mliska@suse.cz>
12630 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
12631 and symtab_node::debug_symtab to symbol_table::debug.
12632 * cgraphunit.c (analyze_functions): Use the renamed function.
12633 (symbol_table::compile): Likewise.
12634 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
12635 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
12636 * passes.c (execute_todo): Likewise.
12637 * symtab.c (symbol_table::dump): New function.
12638 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
12640 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12642 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
12643 that nonconst implies exec.
12645 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12647 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
12648 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
12649 (inline_edge_summary_vec): Turn into ...
12650 (ipa_call_summaries): ... this one.
12651 (redirect_to_unreachable, edge_set_predicate,
12652 evaluate_properties_for_edge, inline_summary_alloc,
12653 reset_ipa_call_summary, reset_inline_summary,
12654 inline_summary_t::duplicate): Update.
12655 (inline_edge_duplication_hook): Turn to ...
12656 (ipa_call_summary_t::duplicate): ... this one.
12657 (inline_edge_removal_hook): Turn to ...
12658 (ipa_call_summary_t::remove): ... this one.
12659 (dump_inline_edge_summary): Turn to ...
12660 (dump_ipa_call_summary): ... this one.
12661 (estimate_function_body_sizes): Update.
12662 (inline_update_callee_summaries): Update.
12663 (remap_edge_change_prob): Update.
12664 (remap_edge_summaries): Update.
12665 (inline_merge_summary): Update.
12666 (do_estimate_edge_time): Update.
12667 (inline_generate_summary): Update.
12668 (inline_read_section): Update.
12669 (inline_read_summary): Update.
12670 (inline_free_summary): Update.
12671 * ipa-inline.c (can_inline_edge_p): Update.
12672 (compute_inlined_call_time): Update.
12673 (want_inline_small_function_p): Update.
12674 (edge_badness): Update.
12675 (early_inliner): Update.
12676 * ipa-inline.h (inline_edge_summary): Turn to ...
12677 (ipa_call_summary): ... this one.
12678 (ipa_call_summary_t): New class.
12679 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
12680 (ipa_call_summaries): New.
12681 (inline_edge_summary): Remove.
12682 (estimate_edge_growth): Update.
12683 * ipa-profile.c (ipa_propagate_frequency_1): Update.
12684 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12685 * ipa-split.c (execute_split_functions): Update.
12686 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
12688 2017-05-23 Tom de Vries <tom@codesourcery.com>
12690 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
12691 attributes): Document rdrand effective target.
12693 2017-05-23 Tom de Vries <tom@codesourcery.com>
12695 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
12696 attributes): Sort alphabetically.
12698 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
12700 * config/avr/genmultilib.awk: Use gsub instead of gensub.
12702 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
12705 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
12706 V2DF/V2DI splat into two separate patterns, one that handles
12707 registers, and the other that only handles memory. Drop support
12708 for splatting from a GPR on ISA 2.07 and then splitting the
12709 splat into direct move and splat.
12710 (vsx_splat_<mode>_reg): Likewise.
12711 (vsx_splat_<mode>_mem): Likewise.
12713 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
12715 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
12717 2017-05-22 Jakub Jelinek <jakub@redhat.com>
12719 PR middle-end/80809
12720 * omp-low.c (finish_taskreg_remap): New function.
12721 (finish_taskreg_scan): If unit size of ctx->record_type
12722 is non-constant, unshare the size expression and replace
12723 decls in it with possible outer var refs.
12725 PR middle-end/80809
12726 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
12727 GOVD_SHARED rather than GOVD_PRIVATE with it.
12728 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
12729 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
12731 PR middle-end/80853
12732 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
12733 as last argument to build_outer_var_ref for pointer bases of array
12734 section reductions.
12736 2017-05-19 Martin Sebor <msebor@redhat.com>
12738 * print-tree.c (print_node): Print DECL_READ_P flag.
12740 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12742 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
12743 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
12744 * cgraph.c: Likewise.
12745 * cgraphunit.c: Likewise.
12746 * gengtype.c: Likewise.
12747 * ipa-cp.c: Likewise.
12748 * ipa-devirt.c: Likewise.
12749 * ipa-icf.c: Likewise.
12750 * ipa-predicate.c: Likewise.
12751 * ipa-profile.c: Likewise.
12752 * ipa-prop.c: Likewise.
12753 * ipa-split.c: Likewise.
12755 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
12756 edge_predicate_pool, dump_inline_hints,
12757 inline_summary::account_size_time, redirect_to_unreachable,
12758 edge_set_predicate, set_hint_predicate,
12759 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
12760 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
12761 inline_summary_t::remove, remap_hint_predicate_after_duplication,
12762 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
12763 ipa_call_summary_t::remove, initialize_growth_caches,
12764 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
12765 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
12766 mark_modified, unmodified_parm_1, unmodified_parm,
12767 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
12768 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
12769 compute_bb_predicates, will_be_nonconstant_expr_predicate,
12770 will_be_nonconstant_predicate, record_modified_bb_info,
12771 get_minimal_bb, record_modified, param_change_prob,
12772 phi_result_unknown_predicate, predicate_for_phi_result,
12773 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
12774 estimate_function_body_sizes, compute_inline_parameters,
12775 compute_inline_parameters_for_curren, pass_data_inline_parameters,
12776 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
12777 inline_update_callee_summaries, remap_edge_change_prob,
12778 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
12779 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
12780 inline_analyze_function, inline_summary_t::insert,
12781 inline_generate_summary, read_ipa_call_summary, inline_read_section,
12782 inline_read_summary, write_ipa_call_summary, inline_write_summary,
12783 inline_free_summary): Move to ipa-fnsummary.h
12784 (predicate_t): Remove.
12785 * ipa-fnsummary.c: New file.
12786 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
12787 (enum inline_hints_vals, inline_hints, agg_position_info,
12788 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
12789 inline_summaries, ipa_call_summary, ipa_call_summary_t,
12790 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
12791 dump_inline_summary, dump_inline_hints, inline_generate_summary,
12792 inline_read_summary, inline_write_summary, inline_free_summary,
12793 inline_analyze_function, initialize_inline_failed,
12794 inline_merge_summary, inline_update_overall_summary,
12795 compute_inline_parameters): Move to ipa-fnsummary.h
12796 * ipa-fnsummary.h: New file.
12797 * ipa-inline-transform.h: Include ipa-inline.h.
12798 * ipa-inline.c: LIkewise.
12800 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12802 * ipa-inline.c (edge_badness): Use inlined_time instead of
12803 inline_summaries->get.
12805 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12807 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
12809 2017-05-22 Nathan Sidwell <nathan@acm.org>
12811 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
12812 (fdump-lang): Document 'raw' option.
12813 * dumpfile.h (TDI_tu): Delete.
12814 * dumpfile.c (dump_files): Remove translation-unit.
12815 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
12817 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
12819 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
12820 command option from $(AWK) call.
12821 * config/avr/genmultilib.awk: Simplify and rewrite so that it
12822 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
12823 [FORMAT]: Remove handling of variable.
12824 * config/avr/t-multilib: Regenerate.
12826 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12828 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
12830 (dump_inline_summary): Do not print self_time.
12831 (estimate_function_body_sizes): Do not set self_time.
12832 (compute_inline_parameters): Likewise.
12833 (inline_read_section, inline_write_summary): Do not stream self_time.
12834 * ipa-inline.h (inline_summary): Drop self_time.
12836 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12838 * ipa-inline-analysis.c (account_size_time): Rename to ...
12839 (inline_summary::account_size_time): ... this one.
12840 (reset_ipa_call_summary): Turn to ...
12841 (ipa_call_summary::reset): ... this one.
12842 (reset_inline_summary): Turn to ...
12843 (inline_summary::reset): ... this one.
12844 (inline_summary_t::remove): Update.
12845 (inline_summary_t::duplicate): Update.
12846 (ipa_call_summary_t::remove): Update.
12847 (dump_inline_summary): Update.
12848 (estimate_function_body_sizes): Update.
12849 (compute_inline_parameters): Update.
12850 (estimate_node_size_and_time): Update.
12851 (inline_merge_summary): Update.
12852 (inline_update_overall_summary): Update.
12853 (inline_read_section): Update.
12854 (inline_write_summary): Update.
12855 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
12856 add account_size_time and reset member functions.
12857 (ipa_call_summary): Add reset function.
12858 * ipa-predicate.h (predicate::operator &): Constify.
12860 2017-05-22 Richard Biener <rguenther@suse.de>
12862 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
12864 2017-05-19 Jason Merrill <jason@redhat.com>
12866 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
12868 2017-05-19 Marek Polacek <polacek@redhat.com>
12871 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
12872 TYPE_OVERFLOW_WRAPS checks.
12874 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
12876 * tree-core.h (enum omp_clause_default_kind): Add
12877 "OMP_CLAUSE_DEFAULT_PRESENT".
12878 * tree-pretty-print.c (dump_omp_clause): Handle it.
12879 * gimplify.c (enum gimplify_omp_var_data): Add
12880 "GOVD_MAP_FORCE_PRESENT".
12881 (gimplify_adjust_omp_clauses_1): Map it to
12882 "GOMP_MAP_FORCE_PRESENT".
12883 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
12885 * gimplify.c (oacc_default_clause): Clarify.
12887 2017-05-19 Nathan Sidwell <nathan@acm.org>
12889 LANG_HOOK_REGISTER_DUMPS
12890 * toplev.c (general_init): Call register dump lang hook.
12891 * doc/invoke.texi: Document -fdump-lang option family.
12892 * dumpfile.c (dump_files): Remove class dump here.
12893 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
12894 * dumpfile.h (tree_dump_index): Remove TDI_class.
12895 * langhooks-def.h (lhd_register_dumps): Declare.
12896 (LANG_HOOKS_REGISTER_DUMPS): Define.
12897 (LANG_HOOKS_INITIALIZER): Add it.
12898 * langhooks.c (lhd_register_dumps): Define.
12899 * langhooks.h (struct lang_hooks): Add register_dumps.
12901 2017-05-19 Nathan Sidwell <nathan@acm.org>
12903 * context.h (context::set_passes): New.
12904 * context.c (context::context): Do not create pass manager.
12905 * toplev.c (general_init): Create pass manager here.
12907 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
12909 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
12910 use this splitter if two add or or instructions would also work for
12911 the constant we want to generate.
12913 2017-05-19 Richard Biener <rguenther@suse.de>
12916 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
12917 predicate evaluation.
12919 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12921 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
12923 * ipa-inline.c (want_inline_small_function_p): Do not cast to
12926 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12928 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
12929 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
12930 (inline_edge_summary_vec): Turn into ...
12931 (ipa_call_summaries): ... this one.
12932 (redirect_to_unreachable, edge_set_predicate,
12933 evaluate_properties_for_edge, inline_summary_alloc,
12934 reset_ipa_call_summary, reset_inline_summary,
12935 inline_summary_t::duplicate): Update.
12936 (inline_edge_duplication_hook): Turn to ...
12937 (ipa_call_summary_t::duplicate): ... this one.
12938 (inline_edge_removal_hook): Turn to ...
12939 (ipa_call_summary_t::remove): ... this one.
12940 (dump_inline_edge_summary): Turn to ...
12941 (dump_ipa_call_summary): ... this one.
12942 (estimate_function_body_sizes): Update.
12943 (inline_update_callee_summaries): Update.
12944 (remap_edge_change_prob): Update.
12945 (remap_edge_summaries): Update.
12946 (inline_merge_summary): Update.
12947 (do_estimate_edge_time): Update.
12948 (inline_generate_summary): Update.
12949 (inline_read_section): Update.
12950 (inline_read_summary): Update.
12951 (inline_free_summary): Update.
12952 * ipa-inline.c (can_inline_edge_p): Update.
12953 (compute_inlined_call_time): Update.
12954 (want_inline_small_function_p): Update.
12955 (edge_badness): Update.
12956 (early_inliner): Update.
12957 * ipa-inline.h (inline_edge_summary): Turn to ...
12958 (ipa_call_summary): ... this one.
12959 (ipa_call_summary_t): New class.
12960 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
12961 (ipa_call_summaries): New.
12962 (inline_edge_summary): Remove.
12963 (estimate_edge_growth): Update.
12964 * ipa-profile.c (ipa_propagate_frequency_1): Update.
12965 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12966 * ipa-split.c (execute_split_functions): Update.
12967 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
12969 2017-05-19 Richard Biener <rguenther@suse.de>
12971 PR middle-end/80764
12972 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
12974 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
12976 * config/rs6000/rs6000.c (struct machine_function): Add field
12977 fpr_is_wrapped_separately.
12978 (rs6000_get_separate_components): Use 64 components. Handle the
12979 new FPR components.
12980 (rs6000_components_for_bb): Handle the FPR components.
12981 (rs6000_emit_prologue_components): Handle the FPR components.
12982 (rs6000_emit_epilogue_components): Handle the FPR components.
12983 (rs6000_set_handled_components): Handle the FPR components.
12984 (rs6000_emit_prologue): Don't output prologue code for those FPRs
12985 that are already separately shrink-wrapped.
12986 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
12987 that are already separately shrink-wrapped.
12989 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
12992 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
12995 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
12996 (define_peephole2 for Altivec d-form load): Add peepholes to catch
12997 cases where the register allocator uses a move and an offsettable
12998 memory operation to/from a FPR register on ISA 2.06/2.07.
12999 (define_peephole2 for Altivec d-form store): Likewise.
13001 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
13004 * config/i386/mmx.md (*mov<mode>_internal): Enable
13005 alternatives 11, 12, 13 and 14 also for 32bit targets.
13006 Remove alternatives 15, 16, 17 and 18.
13007 * config/i386/sse.md (vec_concatv2di): Change
13008 alternative (!x, *y) to (x, ?!*Yn).
13010 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
13012 * dumpfile.h (enum dump_kind): Remove stray comma.
13014 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
13016 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
13017 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
13018 predicate::num_conditions
13019 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
13020 (CHANGED): turn into predicate::changed.
13021 (agg_position_info): Move to ipa-predicate.h
13022 (add_condition, predicate::add_clause, predicate::operator &=,
13023 predicate::or_with, predicate::evaluate, predicate::probability,
13024 dump_condition, dump_clause, predicate::dump,
13025 predicate::remap_after_duplication, predicate::remap_after_inlining,
13026 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
13027 (evaluate_conditions_for_known_args): Update.
13028 (set_cond_stmt_execution_predicate): Update.
13029 * ipa-inline.h: Include ipa-predicate.h
13030 (condition, inline_param_summary, conditions, agg_position_info,
13031 predicate): Move to ipa-predicate.h
13032 * ipa-predicate.c: New file.
13033 * ipa-predicate.h: New file.
13035 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
13037 * final.c (leaf_function_p): Check we are not in a sequence.
13039 2017-05-18 Martin Liska <mliska@suse.cz>
13041 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
13042 * dumpfile.c (dump_register): Use new enum dump_kind.
13043 (get_dump_file_name): Likewise.
13044 (dump_enable_all): Likewise.
13045 (dump_switch_p_1): Likewise.
13046 (enable_rtl_dump_file): Remove usage of TDF_RTL.
13047 * dumpfile.h (enum dump_kind): New enum type.
13048 (struct dump_file_info): Create constructor and
13049 format fields and comments.
13050 * passes.c (pass_manager::register_one_dump_file):
13052 * statistics.c (statistics_early_init): Likewise.
13053 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
13054 TDF_TREE with TDF_SLIM.
13055 (gather_memory_references_ref): Likewise.
13057 2017-05-18 Martin Liska <mliska@suse.cz>
13059 * vec.h (struct vnull): Use it.
13061 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
13063 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
13064 (true_predicate, false_predicate, true_predicate_p,
13065 false_predicate_p): Remove.
13066 (single_cond_predicate, not_inlined_predicate): Turn to member function
13068 (add_condition): Update.
13069 (add_clause): Turn to...
13070 (predicate::add_clause): ... this one; update; allow passing NULL
13072 (and_predicates): Turn to ...
13073 (predicate::operator &=): ... this one.
13074 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
13075 (or_predicates): Turn to ...
13076 (predicate::or_with): ... this one.
13077 (evaluate_predicate): Turn to ...
13078 (predicate::evaluate): ... this one.
13079 (predicate_probability): Turn to ...
13080 (predicate::probability): ... this one.
13081 (dump_condition): Update.
13082 (dump_predicate): Turn to ...
13083 (predicate::dump): ... this one.
13084 (account_size_time): Update.
13085 (edge_set_predicate): Update.
13086 (set_hint_predicate): UPdate.
13087 (evaluate_conditions_for_known_args): Update.
13088 (evaluate_properties_for_edge): Update.
13089 (remap_predicate_after_duplication): Turn to...
13090 (predicate::remap_after_duplication): ... this one.
13091 (remap_hint_predicate_after_duplication): Update.
13092 (inline_summary_t::duplicate): UPdate.
13093 (dump_inline_edge_summary): Update.
13094 (dump_inline_summary): Update.
13095 (set_cond_stmt_execution_predicate): Update.
13096 (set_switch_stmt_execution_predicate): Update.
13097 (compute_bb_predicates): Update.
13098 (will_be_nonconstant_expr_predicate): Update.
13099 (will_be_nonconstant_predicate): Update.
13100 (phi_result_unknown_predicate): Update.
13101 (predicate_for_phi_result): Update.
13102 (array_index_predicate): Update.
13103 (estimate_function_body_sizes): Update.
13104 (estimate_node_size_and_time): Update.
13105 (estimate_ipcp_clone_size_and_time): Update.
13106 (remap_predicate): Rename to ...
13107 (predicate::remap_after_inlining): ... this one.
13108 (remap_hint_predicate): Update.
13109 (inline_merge_summary): Update.
13110 (inline_update_overall_summary): Update.
13111 (estimate_size_after_inlining): Update.
13112 (read_predicate): Rename to ...
13113 (predicate::stream_in): ... this one.
13114 (read_inline_edge_summary): Update.
13115 (write_predicate): Rename to ...
13116 (predicate::stream_out): ... this one.
13117 (write_inline_edge_summary): Update.
13118 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
13119 (clause_t): Turn to uint32_t
13120 (predicate): Turn to class; implement constructor and operators
13122 (size_time_entry): Update.
13123 (inline_summary): Update.
13124 (inline_edge_summary): Update.
13126 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
13128 * fold-const.c (fold_binary_loc): Move transformation...
13129 * match.pd (C - X CMP X): ... here.
13131 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
13133 * config/sparc/sparc.c (sparc_option_override): Set function
13134 alignment for -mcpu=niagara7 to 64 to match the I$ line.
13135 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
13137 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
13139 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
13141 2017-05-18 Marek Polacek <polacek@redhat.com>
13144 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
13145 (pass_ubsan::execute): Call gimple_assign_single_p instead of
13146 gimple_assign_load_p.
13148 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
13150 PR middle-end/80692
13151 * real.c (do_compare): Give decimal_do_compare preference over
13152 comparing just the signs.
13154 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
13156 * doc/md.texi (Canonicalization of Instructions): Describe the
13157 canonical form of instructions that inherently set a condition
13160 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
13162 PR middle-end/80775
13163 * tree-cfg.c: Move deletion of unreachable case statements to after
13164 the merging of consecutive case labels.
13166 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
13168 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
13169 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
13170 restoring of callee-saved registers.
13172 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
13174 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
13175 * config/visium/visium.c (single_set_and_flags): Likewise.
13176 * config/visium/visium.md (Substitutions): Likewise.
13178 2017-05-17 Martin Liska <mliska@suse.cz>
13180 * cfg.c: Introduce dump_flags_t type and
13181 use it instead of int type.
13183 * cfghooks.c: Likewise.
13184 * cfghooks.h (struct cfg_hooks): Likewise.
13185 * cfgrtl.c: Likewise.
13186 * cfgrtl.h: Likewise.
13187 * cgraph.c (cgraph_node::get_body): Likewise.
13188 * coretypes.h: Likewise.
13189 * domwalk.c: Likewise.
13190 * domwalk.h: Likewise.
13191 * dumpfile.c (struct dump_option_value_info): Likewise.
13192 (dump_enable_all): Likewise.
13193 (dump_switch_p_1): Likewise.
13194 (opt_info_switch_p): Likewise.
13195 * dumpfile.h (enum tree_dump_index): Likewise.
13196 (struct dump_file_info): Likewise.
13197 * genemit.c: Likewise.
13198 * generic-match-head.c: Likewise.
13199 * gengtype.c (open_base_files): Likewise.
13200 * gimple-pretty-print.c: Likewise.
13201 * gimple-pretty-print.h: Likewise.
13202 * graph.c (print_graph_cfg): Likewise.
13203 * graphite-scop-detection.c (dot_all_sese): Likewise.
13204 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
13205 * loop-unroll.c (report_unroll): Likewise.
13206 * passes.c (pass_manager::register_one_dump_file): Likewise.
13207 * print-tree.c: Likewise.
13208 * statistics.c: Likewise.
13209 * tree-cfg.c: Likewise.
13210 * tree-cfg.h: Likewise.
13211 * tree-dfa.c: Likewise.
13212 * tree-dfa.h: Likewise.
13213 * tree-dump.c (dump_function): Likewise.
13214 * tree-dump.h (struct dump_info): Likewise.
13215 * tree-pretty-print.c: Likewise.
13216 * tree-pretty-print.h: Likewise.
13217 * tree-ssa-live.c: Likewise.
13218 * tree-ssa-live.h: Likewise.
13219 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
13220 * tree-vect-loop.c: Likewise.
13221 * tree-vect-slp.c: Likewise.
13223 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
13224 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13226 PR tree-optimization/80457
13227 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
13228 of all arguments to a statement as scalar_to_vec operations.
13229 (vectorizable_call): Adjust call to vect_model_simple_cost for
13231 (vectorizable_conversion): Likewise.
13232 (vectorizable_assignment): Likewise.
13233 (vectorizable_shift): Likewise.
13234 (vectorizable_operation): Likewise.
13235 (vectorizable_comparison): Likewise.
13236 (vect_is_simple_cond): Record the def types for operands.
13237 (vectorizable_condition): Likewise, call vect_model_simple_cost.
13238 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
13239 for statement argument count.
13241 2017-05-16 Carl Love <cel@us.ibm.com>
13243 * config/rs6000/rs6000-c: Add support for built-in functions
13244 vector unsigned long long vec_bperm (vector unsigned long long,
13245 vector unsigned char)
13246 vector signed long long vec_mule (vector signed int,
13248 vector unsigned long long vec_mule (vector unsigned int,
13249 vector unsigned int)
13250 vector signed long long vec_mulo (vector signed int,
13252 vector unsigned long long vec_mulo (vector unsigned int,
13253 vector unsigned int)
13254 vector signed char vec_sldw (vector signed char,
13255 vector signed char,
13257 vector unsigned char vec_sldw (vector unsigned char,
13258 vector unsigned char,
13260 vector signed short vec_sldw (vector signed short,
13261 vector signed short,
13263 vector unsigned short vec_sldw (vector unsigned short,
13264 vector unsigned short,
13266 vector signed int vec_sldw (vector signed int,
13269 vector unsigned int vec_sldw (vector unsigned int,
13270 vector unsigned int,
13272 vector signed long long vec_sldw (vector signed long long,
13273 vector signed long long,
13275 vector unsigned long long vec_sldw (vector unsigned long long,
13276 vector unsigned long long,
13278 * config/rs6000/rs6000-c: Add support for built-in functions
13279 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
13280 * config/rs6000/altivec.h: Add defintion for vec_sldw.
13281 * doc/extend.texi: Update the built-in documentation for the
13282 new built-in functions.
13284 2017-05-16 Marek Polacek <polacek@redhat.com>
13288 * tree.c (save_expr): Don't fold the expression.
13290 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
13292 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
13293 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
13294 and (?*y,m). Update insn attributes.
13296 2017-05-16 Martin Liska <mliska@suse.cz>
13298 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
13299 flags argument of print_gimple_stmt, print_gimple_expr,
13300 print_generic_stmt and print_generic_expr.
13301 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
13302 * coretypes.h: Likewise.
13303 * except.c (dump_eh_tree): Likewise.
13304 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
13305 * gimple-pretty-print.h: Likewise.
13306 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
13307 (backprop::push_to_worklist): Likewise.
13308 (backprop::pop_from_worklist): Likewise.
13309 (backprop::process_use): Likewise.
13310 (backprop::intersect_uses): Likewise.
13311 (note_replacement): Likewise.
13312 * gimple-ssa-store-merging.c
13313 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
13314 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
13315 (pass_store_merging::execute): Likewise.
13316 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
13317 (ssa_base_cand_dump_callback): Likewise.
13318 (dump_incr_vec): Likewise.
13319 (replace_refs): Likewise.
13320 (replace_mult_candidate): Likewise.
13321 (create_add_on_incoming_edge): Likewise.
13322 (create_phi_basis): Likewise.
13323 (insert_initializers): Likewise.
13324 (all_phi_incrs_profitable): Likewise.
13325 (introduce_cast_before_cand): Likewise.
13326 (replace_one_candidate): Likewise.
13327 * gimplify.c (gimplify_expr): Likewise.
13328 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
13329 (set_rename): Likewise.
13330 (rename_uses): Likewise.
13331 (copy_loop_phi_nodes): Likewise.
13332 (add_close_phis_to_merge_points): Likewise.
13333 (copy_loop_close_phi_args): Likewise.
13334 (copy_cond_phi_args): Likewise.
13335 (graphite_copy_stmts_from_block): Likewise.
13336 (translate_pending_phi_nodes): Likewise.
13337 * graphite-poly.c (print_pdr): Likewise.
13338 (dump_gbb_cases): Likewise.
13339 (dump_gbb_conditions): Likewise.
13340 (print_scop_params): Likewise.
13341 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
13342 (build_cross_bb_scalars_use): Likewise.
13343 (gather_bbs::before_dom_children): Likewise.
13344 * hsa-dump.c (dump_hsa_immed): Likewise.
13345 * ipa-cp.c (print_ipcp_constant_value): Likewise.
13346 (get_replacement_map): Likewise.
13347 * ipa-inline-analysis.c (dump_condition): Likewise.
13348 (estimate_function_body_sizes): Likewise.
13349 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
13350 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
13351 * ipa-prop.c (ipa_dump_param): Likewise.
13352 (ipa_print_node_jump_functions_for_edge): Likewise.
13353 (ipa_modify_call_arguments): Likewise.
13354 (ipa_modify_expr): Likewise.
13355 (ipa_dump_param_adjustments): Likewise.
13356 (ipa_dump_agg_replacement_values): Likewise.
13357 (ipcp_modif_dom_walker::before_dom_children): Likewise.
13358 * ipa-pure-const.c (check_stmt): Likewise.
13359 (pass_nothrow::execute): Likewise.
13360 * ipa-split.c (execute_split_functions): Likewise.
13361 * omp-offload.c (dump_oacc_loop_part): Likewise.
13362 (dump_oacc_loop): Likewise.
13363 * trans-mem.c (tm_log_emit): Likewise.
13364 (tm_memopt_accumulate_memops): Likewise.
13365 (dump_tm_memopt_set): Likewise.
13366 (dump_tm_memopt_transform): Likewise.
13367 * tree-cfg.c (gimple_verify_flow_info): Likewise.
13368 (print_loop): Likewise.
13369 * tree-chkp-opt.c (chkp_print_addr): Likewise.
13370 (chkp_gather_checks_info): Likewise.
13371 (chkp_get_check_result): Likewise.
13372 (chkp_remove_check_if_pass): Likewise.
13373 (chkp_use_outer_bounds_if_possible): Likewise.
13374 (chkp_reduce_bounds_lifetime): Likewise.
13375 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
13376 (chkp_mark_completed_bounds): Likewise.
13377 (chkp_register_incomplete_bounds): Likewise.
13378 (chkp_mark_invalid_bounds): Likewise.
13379 (chkp_maybe_copy_and_register_bounds): Likewise.
13380 (chkp_build_returned_bound): Likewise.
13381 (chkp_get_bound_for_parm): Likewise.
13382 (chkp_build_bndldx): Likewise.
13383 (chkp_get_bounds_by_definition): Likewise.
13384 (chkp_generate_extern_var_bounds): Likewise.
13385 (chkp_get_bounds_for_decl_addr): Likewise.
13386 * tree-chrec.c (chrec_apply): Likewise.
13387 * tree-data-ref.c (dump_data_reference): Likewise.
13388 (dump_subscript): Likewise.
13389 (dump_data_dependence_relation): Likewise.
13390 (analyze_overlapping_iterations): Likewise.
13391 * tree-inline.c (expand_call_inline): Likewise.
13392 (tree_function_versioning): Likewise.
13393 * tree-into-ssa.c (dump_defs_stack): Likewise.
13394 (dump_currdefs): Likewise.
13395 (dump_names_replaced_by): Likewise.
13396 (dump_update_ssa): Likewise.
13397 (update_ssa): Likewise.
13398 * tree-object-size.c (pass_object_sizes::execute): Likewise.
13399 * tree-parloops.c (build_new_reduction): Likewise.
13400 (try_create_reduction_list): Likewise.
13401 (ref_conflicts_with_region): Likewise.
13402 (oacc_entry_exit_ok_1): Likewise.
13403 (oacc_entry_exit_single_gang): Likewise.
13404 * tree-pretty-print.h: Likewise.
13405 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
13406 (get_scalar_evolution): Likewise.
13407 (add_to_evolution): Likewise.
13408 (get_loop_exit_condition): Likewise.
13409 (analyze_evolution_in_loop): Likewise.
13410 (analyze_initial_condition): Likewise.
13411 (analyze_scalar_evolution): Likewise.
13412 (instantiate_scev): Likewise.
13413 (number_of_latch_executions): Likewise.
13414 (gather_chrec_stats): Likewise.
13415 (final_value_replacement_loop): Likewise.
13416 (scev_const_prop): Likewise.
13417 * tree-sra.c (dump_access): Likewise.
13418 (disqualify_candidate): Likewise.
13419 (create_access): Likewise.
13420 (reject): Likewise.
13421 (maybe_add_sra_candidate): Likewise.
13422 (create_access_replacement): Likewise.
13423 (analyze_access_subtree): Likewise.
13424 (analyze_all_variable_accesses): Likewise.
13425 (sra_modify_assign): Likewise.
13426 (initialize_constant_pool_replacements): Likewise.
13427 (find_param_candidates): Likewise.
13428 (decide_one_param_reduction): Likewise.
13429 (replace_removed_params_ssa_names): Likewise.
13430 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
13431 * tree-ssa-copy.c (dump_copy_of): Likewise.
13432 (copy_prop_visit_cond_stmt): Likewise.
13433 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
13434 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
13435 (record_equivalences_from_stmt): Likewise.
13436 * tree-ssa-dse.c (compute_trims): Likewise.
13437 (delete_dead_call): Likewise.
13438 (delete_dead_assignment): Likewise.
13439 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
13440 (forward_propagate_into_cond): Likewise.
13441 (pass_forwprop::execute): Likewise.
13442 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13443 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
13445 (move_computations_worker): Likewise.
13446 (execute_sm): Likewise.
13447 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
13448 (remove_exits_and_undefined_stmts): Likewise.
13449 (remove_redundant_iv_tests): Likewise.
13450 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
13451 (adjust_iv_update_pos): Likewise.
13452 * tree-ssa-math-opts.c (bswap_replace): Likewise.
13453 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
13454 (value_replacement): Likewise.
13455 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
13456 * tree-ssa-pre.c (print_pre_expr): Likewise.
13457 (get_representative_for): Likewise.
13458 (create_expression_by_pieces): Likewise.
13459 (insert_into_preds_of_block): Likewise.
13460 (eliminate_insert): Likewise.
13461 (eliminate_dom_walker::before_dom_children): Likewise.
13462 (eliminate): Likewise.
13463 (remove_dead_inserted_code): Likewise.
13464 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
13465 * tree-ssa-reassoc.c (get_rank): Likewise.
13466 (eliminate_duplicate_pair): Likewise.
13467 (eliminate_plus_minus_pair): Likewise.
13468 (eliminate_not_pairs): Likewise.
13469 (undistribute_ops_list): Likewise.
13470 (eliminate_redundant_comparison): Likewise.
13471 (update_range_test): Likewise.
13472 (optimize_range_tests_var_bound): Likewise.
13473 (optimize_vec_cond_expr): Likewise.
13474 (rewrite_expr_tree): Likewise.
13475 (rewrite_expr_tree_parallel): Likewise.
13476 (linearize_expr): Likewise.
13477 (break_up_subtract): Likewise.
13478 (linearize_expr_tree): Likewise.
13479 (attempt_builtin_powi): Likewise.
13480 (attempt_builtin_copysign): Likewise.
13481 (transform_stmt_to_copy): Likewise.
13482 (transform_stmt_to_multiply): Likewise.
13483 (dump_ops_vector): Likewise.
13484 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
13485 (print_scc): Likewise.
13486 (set_ssa_val_to): Likewise.
13487 (visit_reference_op_store): Likewise.
13488 (visit_use): Likewise.
13489 (sccvn_dom_walker::before_dom_children): Likewise.
13490 (run_scc_vn): Likewise.
13491 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
13493 (expr_hash_elt::print): Likewise.
13494 (const_and_copies::pop_to_marker): Likewise.
13495 (const_and_copies::record_const_or_copy_raw): Likewise.
13496 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
13497 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
13498 (dump_predicates): Likewise.
13499 (find_uninit_use): Likewise.
13500 (warn_uninitialized_phi): Likewise.
13501 (pass_late_warn_uninitialized::execute): Likewise.
13502 * tree-ssa.c (verify_vssa): Likewise.
13503 (verify_ssa): Likewise.
13504 (maybe_optimize_var): Likewise.
13505 * tree-vrp.c (dump_value_range): Likewise.
13506 (dump_all_value_ranges): Likewise.
13507 (dump_asserts_for): Likewise.
13508 (register_edge_assert_for_2): Likewise.
13509 (vrp_visit_cond_stmt): Likewise.
13510 (vrp_visit_switch_stmt): Likewise.
13511 (vrp_visit_stmt): Likewise.
13512 (vrp_visit_phi_node): Likewise.
13513 (simplify_cond_using_ranges_1): Likewise.
13514 (fold_predicate_in): Likewise.
13515 (evrp_dom_walker::before_dom_children): Likewise.
13516 (evrp_dom_walker::push_value_range): Likewise.
13517 (evrp_dom_walker::pop_value_range): Likewise.
13518 (execute_early_vrp): Likewise.
13520 2017-05-16 Richard Biener <rguenther@suse.de>
13522 * dwarf2out.c (loc_list_from_tree_1): Do not create
13523 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
13525 2017-05-16 Richard Biener <rguenther@suse.de>
13527 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
13529 (note_variable_value_in_expr): If we resolved the decl ref
13530 do not push to the stack.
13532 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
13534 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
13535 operations in fast-math mode.
13536 (vaddq_f16): Likewise.
13537 (vmul_f16): Likewise.
13538 (vmulq_f16): Likewise.
13539 (vsub_f16): Likewise.
13540 (vsubq_f16): Likewise.
13541 * config/arm/neon.md (add<mode>3): New.
13543 (fma:<VH:mode>3): New. Also remove outdated comment.
13546 2017-05-16 Martin Liska <mliska@suse.cz>
13550 * ipa-devirt.c (warn_types_mismatch): Fix typo.
13551 (odr_types_equivalent_p): Likewise.
13553 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
13555 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
13557 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
13560 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
13561 non-interunit SSE move alternatives with '?'.
13562 (zero-extendsidi peephole2): New peephole to skip intermediate
13563 general register in SSE zero-extend sequence.
13565 2017-05-15 Jeff Law <law@redhat.com>
13567 * reorg.c (relax_delay_slots): Create a new variable to hold
13568 the temporary target rather than clobbering TARGET_LABEL.
13570 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
13571 missing argument to extract_bit_field call.
13572 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
13574 2017-05-15 Martin Liska <mliska@suse.cz>
13577 * gcc.c (process_command): Do not allow empty argument of -o option.
13579 2017-05-15 Renlin Li <renlin.li@arm.com>
13581 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
13582 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
13583 * config/aarch64/constraints.md (Usf): Add long call check.
13584 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
13585 (call_value): Likewise.
13586 (sibcall): Likewise.
13587 (sibcall_value): Likewise.
13589 (call_value_insn): New.
13590 (sibcall_insn): Update rtx pattern.
13591 (sibcall_value_insn): Likewise.
13592 (call_internal): Remove.
13593 (call_value_internal): Likewise.
13594 (sibcall_internal): Likewise.
13595 (sibcall_value_internal): Likewise.
13596 (call_reg): Likewise.
13597 (call_symbol): Likewise.
13598 (call_value_reg): Likewise.
13599 (call_value_symbol): Likewise.
13601 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
13604 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
13606 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
13608 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
13609 compatible with CCGOCmode and with CCZmode.
13611 2017-05-14 Martin Sebor <msebor@redhat.com>
13613 PR middle-end/77671
13614 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
13615 (gimple_fold_builtin_snprintf): Same.
13616 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
13617 (gimple_fold_builtin_snprintf): Same.
13618 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
13619 of character types.
13620 (is_call_safe): New function.
13621 (try_substitute_return_value): Call it.
13622 (try_simplify_call): New function.
13623 (pass_sprintf_length::handle_gimple_call): Call it.
13625 2017-05-14 Martin Sebor <msebor@redhat.com>
13627 PR middle-end/80669
13628 * builtins.c (expand_builtin_stpncpy): Simplify.
13630 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
13632 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
13633 * config/i386/i386.h
13634 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
13635 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
13636 (struct machine_function): Add new members call_ms2sysv,
13637 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
13638 (struct machine_frame_state): New fields sp_realigned and
13639 sp_realigned_offset.
13640 * config/i386/i386.c
13641 (enum xlogue_stub): New enum.
13642 (enum xlogue_stub_sets): New enum.
13643 (class xlogue_layout): New class.
13644 (struct ix86_frame): New fields stack_realign_allocate_offset,
13645 stack_realign_offset and outlined_save_offset. Modify comments to
13646 detail stack layout when using out-of-line stubs.
13647 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
13648 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
13649 -mcall-ms2sysv-xlogues.
13650 (stub_managed_regs): New static variable.
13651 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
13652 registers managed by out-of-line stub.
13653 (disable_call_ms2sysv_xlogues): New function.
13654 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
13655 m->call_ms2sysv when appropriate and compute frame layout for
13657 (sp_valid_at, fp_valid_at): New inline functions.
13658 (choose_basereg): New function.
13659 (choose_baseaddr): Add align parameter, use choose_basereg and modify
13661 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
13662 Use align parameter of choose_baseaddr to generated aligned SSE movs
13664 (pro_epilogue_adjust_stack): Modify to track
13665 machine_frame_state::sp_realigned.
13666 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
13667 (ix86_nsaved_sseregs): Likewise.
13668 (ix86_emit_save_regs): Likewise.
13669 (ix86_emit_save_regs_using_mov): Likewise.
13670 (ix86_emit_save_sse_regs_using_mov): Likewise.
13671 (get_scratch_register_on_entry): Likewise.
13672 (gen_frame_set): New function.
13673 (gen_frame_load): Likewise.
13674 (gen_frame_store): Likewise.
13675 (emit_outlined_ms2sysv_save): Likewise.
13676 (emit_outlined_ms2sysv_restore): Likewise.
13677 (ix86_expand_prologue): Modify stack re-alignment code and call
13678 emit_outlined_ms2sysv_save when appropriate.
13679 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
13680 parameter rtx_insn *insn, which allows the function to be used to only
13681 generate the notes.
13682 (ix86_expand_epilogue): Modify validity checks of frame and stack
13683 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
13684 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
13685 * config/i386/predicates.md
13686 (save_multiple): New predicate.
13687 (restore_multiple): Likewise.
13688 * config/i386/sse.md
13689 (save_multiple<mode>): New pattern.
13690 (save_multiple_realign<mode>): Likewise.
13691 (restore_multiple<mode>): Likewise.
13692 (restore_multiple_and_return<mode>): Likewise.
13693 (restore_multiple_leave_return<mode>): Likewise.
13694 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
13696 2017-05-14 Julia Koval <julia.koval@intel.com>
13698 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
13699 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
13700 (__builtin_ia32_xsetbv): New builtins.
13701 * config/i386/i386.c (ix86_expand_special_args_builtin):
13703 (ix86_expand_builtin): Special expand for new intrinsics.
13704 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
13705 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
13706 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
13708 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13710 * cfganal.c (inverted_post_order_compute): Change argument type
13712 * cfganal.h (inverted_post_order_compute): Adjust prototype.
13713 * df-core.c (rest_of_handle_df_initialize): Adjust.
13714 (rest_of_handle_df_finish): Likewise.
13715 (df_analyze_1): Likewise.
13716 (df_analyze): Likewise.
13717 (loop_inverted_post_order_compute): Change argument to be a vec *.
13718 (df_analyze_loop): Adjust.
13719 (df_get_n_blocks): Likewise.
13720 (df_get_postorder): Likewise.
13721 * df.h (struct df_d): Change field to be a vec.
13722 * lcm.c (compute_laterin): Adjust.
13723 (compute_available): Likewise.
13724 * lra-lives.c (lra_create_live_ranges_1): Likewise.
13725 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
13726 * tree-ssa-pre.c (compute_antic): Likewise.
13728 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13730 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
13731 (depth_first_search::depth_first_search): Change structure init
13732 function to this constructor.
13733 (depth_first_search::add_bb): Rename function to this member.
13734 (depth_first_search::execute): Likewise.
13735 (flow_dfs_compute_reverse_finish): Adjust.
13737 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13739 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
13740 (longest_simple_path): Likewise.
13741 * shrink-wrap.c (spread_components): Likewise.
13742 (disqualify_problematic_components): Likewise.
13743 (emit_common_heads_for_components): Likewise.
13744 (emit_common_tails_for_components): Likewise.
13745 (insert_prologue_epilogue_for_components): Likewise.
13747 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13749 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
13752 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13754 * df-core.c (df_set_blocks): Start using auto_bitmap.
13755 (df_compact_blocks): Likewise.
13756 * df-problems.c (df_rd_confluence_n): Likewise.
13757 * df-scan.c (df_insn_rescan_all): Likewise.
13758 (df_process_deferred_rescans): Likewise.
13759 (df_update_entry_block_defs): Likewise.
13760 (df_update_exit_block_uses): Likewise.
13761 (df_entry_block_bitmap_verify): Likewise.
13762 (df_exit_block_bitmap_verify): Likewise.
13763 (df_scan_verify): Likewise.
13764 * lra-constraints.c (lra_constraints): Likewise.
13765 (undo_optional_reloads): Likewise.
13766 (lra_undo_inheritance): Likewise.
13767 * lra-remat.c (calculate_gen_cands): Likewise.
13768 (do_remat): Likewise.
13769 * lra-spills.c (assign_spill_hard_regs): Likewise.
13770 (spill_pseudos): Likewise.
13771 * tree-ssa-pre.c (bitmap_set_and): Likewise.
13772 (bitmap_set_subtract_values): Likewise.
13774 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13776 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
13777 management with auto_bitmap.
13778 (fix_inter_tick): Likewise.
13779 (fix_recovery_deps): Likewise.
13780 * ira.c (add_store_equivs): Likewise.
13781 (find_moveable_pseudos): Likewise.
13782 (split_live_ranges_for_shrink_wrap): Likewise.
13783 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
13784 (rtx_reuse_manager::seen_def_p): Likewise.
13785 (rtx_reuse_manager::set_seen_def): Likewise.
13786 * print-rtl.h (class rtx_reuse_manager): Likewise.
13788 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13790 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
13792 (migrate_btr_def): Likewise.
13793 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
13794 * df-core.c (loop_post_order_compute): Likewise.
13795 (loop_inverted_post_order_compute): Likewise.
13796 * hsa-common.h: Likewise.
13797 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
13798 * init-regs.c (initialize_uninitialized_regs): Likewise.
13799 * ipa-inline.c (resolve_noninline_speculation): Likewise.
13800 (inline_small_functions): Likewise.
13801 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
13802 * ira.c (combine_and_move_insns): Likewise.
13803 (build_insn_chain): Likewise.
13804 * loop-invariant.c (find_invariants): Likewise.
13805 * lower-subreg.c (propagate_pseudo_copies): Likewise.
13806 * predict.c (tree_predict_by_opcode): Likewise.
13807 (predict_paths_leading_to): Likewise.
13808 (predict_paths_leading_to_edge): Likewise.
13809 (estimate_loops_at_level): Likewise.
13810 (estimate_loops): Likewise.
13811 * shrink-wrap.c (try_shrink_wrapping): Likewise.
13812 (spread_components): Likewise.
13813 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
13814 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
13815 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
13816 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
13817 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
13818 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
13819 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
13820 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
13821 (mark_threaded_blocks): Likewise.
13822 (thread_through_all_blocks): Likewise.
13823 * tree-ssa.c (verify_ssa): Likewise.
13824 (execute_update_addresses_taken): Likewise.
13825 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
13827 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13829 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
13831 (post_order_compute): Likewise.
13832 (inverted_post_order_compute): Likewise.
13833 (pre_and_rev_post_order_compute_fn): Likewise.
13835 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13837 * genrecog.c (int_set::int_set): Explicitly construct our
13838 auto_vec base class.
13839 * vec.h (auto_vec::auto_vec): New constructor.
13841 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13843 * bitmap.h (class auto_bitmap): New constructor taking
13844 bitmap_obstack * argument.
13846 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13848 * bitmap.h (class auto_bitmap): Change type of m_bits to
13849 bitmap_head, and adjust ctor / dtor and member operators.
13851 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
13853 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
13854 when returned register mode doesn't match original mode.
13856 2017-05-12 Jeff Law <law@redhat.com>
13857 Jakub Jelinek <jakub@redhat.com>
13859 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
13860 we look for cc setter after the compare-elim changes.
13861 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
13862 within the vector to match what compare-elim now expects.
13863 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
13864 (xorsi3_flags, one_cmplsi2_flags): Likewise.
13866 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
13867 after the compare-elim changes.
13868 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
13869 the vector to match what compare-elim now expects.
13870 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
13871 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
13872 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
13873 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
13874 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
13876 * config/visium/visium.c (single_set_and_flags): Fix where
13877 we look for cc setter after the compare-elim changes.
13878 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
13879 with the vector to match what compare-elim now expects.
13880 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
13881 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
13882 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
13883 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
13884 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
13885 (neg<mode>2_insn_set_overflow): Likewise.
13887 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
13889 PR middle-end/79794
13890 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
13891 maybe_expand_insn call, set ops[0].target. If still set after call,
13892 set alt_rtl. Add extra arg to recursive calls.
13893 (extract_bit_field): Add alt_rtl argument. Pass to
13895 * expmed.h (extract_bit_field): Fix prototype.
13896 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
13897 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
13898 to extract_bit_field_calls.
13899 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
13900 Pass alt_rtl to extract_bit_field calls.
13901 * calls.c (store_unaligned_arguments_into_psuedos)
13902 load_register_parameters): Pass extra NULL to extract_bit_field calls.
13903 * optabs.c (maybe_legitimize_operand): Clear op->target when call
13905 * optabs.h (struct expand_operand): Add target bitfield.
13907 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13909 * compare-elim.c (try_eliminate_compare): Canonicalize
13910 operation with embedded compare to
13911 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
13912 (set (reg) (operation)].
13914 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
13916 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13919 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
13920 cost of adding a carry flag for ADC instruction.
13921 [case MINUS]: Ignore the cost of subtracting a carry flag
13922 for SBB instruction.
13924 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
13926 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
13928 * config/rs6000/bmiintrin.h: New file.
13929 * config/rs6000/bmi2intrin.h: New file.
13930 * config/rs6000/x86intrin.h: New file.
13932 2017-05-12 Jeff Law <law@redhat.com>
13934 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
13937 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
13939 PR middle-end/80707
13940 * tree-cfg.c: Remove cfg edges of unreachable case statements.
13942 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13944 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
13945 early expansion of vector divide builtins.
13946 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
13947 builtins identified as having unsigned arguments.
13949 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13951 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
13952 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
13953 expansion of vector logical operations (and, andc, or, xor,
13956 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13958 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
13959 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
13961 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13963 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
13964 early GIMPLE expansion of vector multiplies.
13966 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13968 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
13969 TARGET_HAVE_MOVT conditional.
13970 (movt splitter): Likewise.
13972 2017-05-12 Richard Biener <rguenther@suse.de>
13974 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
13975 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13976 Fold all stmts not inplace.
13978 2017-05-12 Richard Biener <rguenther@suse.de>
13980 PR tree-optimization/80713
13981 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
13982 inserted_exprs bit for not removed stmts.
13984 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
13986 PR middle-end/69921
13987 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
13988 parallelized" attribute for parallelized OpenACC kernels.
13989 * omp-offload.c (execute_oacc_device_lower): Use it.
13991 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
13992 Set "oacc kernels" attribute.
13993 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
13994 parameter. Adjust all users.
13995 (oacc_fn_attrib_kernels_p): Remove function.
13996 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
13997 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
13998 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
13999 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
14000 assert "oacc kernels" attribute is set.
14002 2017-05-11 Carl Love <cel@us.ibm.com>
14004 * config/rs6000/rs6000-c: Add support for built-in functions
14005 vector unsigned char vec_popcnt (vector signed char)
14006 vector unsigned char vec_popcnt (vector unsigned char)
14007 vector unsigned short vec_popcnt (vector signed short)
14008 vector unsigned short vec_popcnt (vector unsigned short)
14009 vector unsigned int vec_popcnt (vector signed int)
14010 vector unsigned int vec_popcnt (vector unsigned int)
14011 vector unsigned long long vec_popcnt (vector signed long long)
14012 vector unsigned long long vec_popcnt (vector unsigned long long)
14013 vector signed long long vec_slo (vector signed long long,
14014 vector signed char)
14015 vector signed long long vec_slo (vector signed long long,
14016 vector unsigned char)
14017 vector unsigned long long vec_slo (vector unsigned long long,
14018 vector signed char)
14019 vector unsigned long long vec_slo (vector unsigned long long,
14020 vector unsigned char)
14021 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
14022 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
14023 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
14024 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
14025 * doc/extend.texi: Update the built-in documentation file for the
14026 new built-in functions.
14028 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
14030 * attribs.h (sorted_attr_string): Move machine independent
14031 functions for target clone support from the i386 port to common
14032 code. Rename ix86_function_versions to common_function_versions.
14033 Rename make_name to make_unique_name.
14034 (common_function_versions): Likewise.
14035 (make_unique_name): Likewise.
14036 (make_dispatcher_decl): Likewise.
14037 (is_function_default_version): Likewise.
14038 * attribs.c (attr_strcmp): Likewise.
14039 (sorted_attr_string): Likewise.
14040 (common_function_versions): Likewise.
14041 (make_unique_name): Likewise.
14042 (make_dispatcher_decl): Likewise.
14043 (is_function_default_version): Likewise.
14044 * config/i386/i386.c (attr_strcmp): Likewise.
14045 (sorted_attr_string): Likewise.
14046 (ix86_function_versions): Likewise.
14047 (make_name): Likewise.
14048 (make_dispatcher_decl): Likewise.
14049 (is_function_default_version): Likewise.
14050 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
14052 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14055 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
14056 Account for direct move costs for vec_construct of integer
14059 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
14062 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
14063 (UNSPEC_STX_ATOMIC): Ditto.
14064 (loaddi_via_sse): New insn.
14065 (storedi_via_sse): Ditto.
14066 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
14067 Update corresponding peephole2 patterns.
14068 (atomic_storedi_fpu): Ditto.
14070 2017-05-11 Julia Koval <julia.koval@intel.com>
14072 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
14073 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
14075 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
14076 (__builtin_ia32_rsqrt14ss_mask): New builtins.
14077 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
14079 2017-05-11 Nathan Sidwell <nathan@acm.org>
14081 * graphite-poly.c: Include dumpfile.h.
14083 * dumpfle.h (dump_function): Declare here ...
14084 * tree-dump.h (dump_function): ... not here.
14085 * dumpfile.c: #include tree-cfg.h.
14086 (dump_function): Move here from ...
14087 * tree-dump.c (dump_function): ... here.
14088 * gimplify.c: #include splay-tree.h, not tree-dump.h.
14089 * graphite-poly.c: Don't include tree-dump.h.
14090 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
14091 * print-tree.c: Likewise.
14092 * stor-layout.c: Likewise.
14093 * tree-nested.c: Likewise.
14095 * dumpfile.c (dump_start): Use TDF_FLAGS.
14096 (dump_enable_all): Fix TDF_KIND check thinko.
14098 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
14100 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14101 array entries to represent two legal parameterizations of the
14102 overloaded __builtin_cmpb function, as represented by the
14103 P6_OV_BUILTIN_CMPB constant.
14104 (altivec_resolve_overloaded_builtin): Add special case handling
14105 for the __builtin_cmpb function, as represented by the
14106 P6_OV_BUILTIN_CMPB constant.
14107 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
14108 (BU_P6_64BIT_2): New macro.
14109 (BU_P6_OVERLOAD_2): New macro
14110 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
14111 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
14112 (CMPB): Add overload support to represent both 32-bit and 64-bit
14113 compare-bytes function.
14114 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14115 support for TARGET_CMPB.
14116 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
14117 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
14118 documentation of the __builtin_cmpb overloaded built-in function.
14120 2017-05-11 Richard Biener <rguenther@suse.de>
14122 PR tree-optimization/80705
14123 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
14124 bases are not vectorizable.
14126 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14128 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
14129 when counting register pressure.
14131 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14133 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
14134 (iv_ca_more_deps): Renamed to ...
14135 (iv_ca_compare_deps): ... this.
14136 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
14138 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14140 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
14142 (determine_group_iv_costs): ... here.
14143 (find_inv_vars_cb): Record inv var if it's not recorded before.
14145 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14147 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
14148 (get_shiftadd_cost): Ditto.
14150 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14152 * tree-ssa-address.c: Include header file.
14153 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
14155 (add_to_parts): Refactor.
14156 (addr_to_parts): New parameter. Update use of move_hint_to_base.
14157 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
14160 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14162 PR tree-optimization/53090
14163 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
14165 (extract_cond_operands): Detect condition with IV on both sides
14166 and return COMP_IV_EXPR_2.
14167 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
14168 (rewrite_use_compare): Simplify by removing call to function
14169 extract_cond_operands.
14171 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14173 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
14174 (extract_cond_operands): Detect condition comparing against non-
14175 invariant bound and return appropriate enum value.
14176 (find_interesting_uses_cond): Update use of extract_cond_operands.
14177 Handle its return value accordingly.
14178 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
14180 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14182 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
14183 nonlinear iv_use computation in loop invariant sensitive way.
14185 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14187 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
14188 (find_iv_candidates): Call relate_compare_use_with_all_cands.
14190 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14192 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
14193 (dump_cand): Support iv_cand.inv_exprs.
14194 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
14196 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
14199 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14201 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
14203 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
14204 as local function. Include necessary header files.
14205 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
14207 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14209 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
14211 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14213 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
14214 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
14215 RSHIFT_EXPR and BIT_NOT_EXPR.
14217 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14219 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
14220 (adjust_setup_cost): New parameter supporting round up adjustment.
14221 (struct address_cost_data): Delete.
14222 (force_expr_to_var_cost): Don't bound cost with spill_cost.
14223 (split_address_cost, ptr_difference_cost): Delete.
14224 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
14225 (struct ainc_cost_data): New struct.
14226 (get_address_cost_ainc): New function.
14227 (get_address_cost, get_computation_cost): Reimplement.
14228 (determine_group_iv_cost_address): Record inv_expr for all uses of
14230 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
14231 (iv_ca_has_deps): Reimplemented to ...
14232 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
14234 (iv_ca_extend): Call iv_ca_more_deps.
14236 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14238 * tree-ssa-address.c (struct mem_address): Move to header file.
14239 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
14240 * tree-ssa-address.h (struct mem_address): Move from C file.
14241 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
14243 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14245 * tree-affine.h (aff_combination_type): New interface.
14246 (aff_combination_zero_p): Remove static.
14247 (aff_combination_const_p): New interface.
14248 (aff_combination_singleton_var_p): New interfaces.
14250 2017-05-11 Richard Biener <rguenther@suse.de>
14252 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14253 Skip unreachable blocks and destinations.
14254 (eliminate): Move stmt removal and fixup ...
14255 (fini_eliminate): ... here. Skip inserted exprs.
14256 (pass_pre::execute): Move fini_pre after fini_eliminate.
14257 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
14258 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
14259 PRE to get rid of dead code that has invalid SSA form and
14260 split critical edges again.
14262 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14264 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14266 2017-05-11 Richard Biener <rguenther@suse.de>
14268 * passes.c (execute_function_todo): Verify loops if they are
14269 said to be up-to-date.
14270 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
14271 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
14273 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
14276 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
14277 handle calling assemble_external ourself.
14280 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
14281 modes with zero size. Enhance comment.
14283 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14285 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
14286 built-ins for vec_xl and vec_xst with short and char pointer
14289 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
14291 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
14292 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
14293 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
14294 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
14295 (_mm_maskz_min_round_ss): New intrinsics.
14296 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
14297 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
14298 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
14299 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
14300 (__builtin_ia32_minss_mask_round): New builtins.
14301 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14302 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
14303 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
14305 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
14306 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
14308 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
14311 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
14313 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
14314 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
14315 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
14316 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
14317 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
14318 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
14319 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
14320 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14321 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
14322 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
14323 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
14324 (__builtin_ia32_mulss_mask_round): New builtins.
14325 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14326 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
14327 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
14329 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
14330 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
14332 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
14335 2017-05-10 Julia Koval <julia.koval@intel.com>
14337 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
14338 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
14339 (_mm256_setr_m128i): New intrinsics.
14341 2017-05-10 Julia Koval <julia.koval@intel.com>
14343 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
14344 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
14345 (_mm_maskz_rcp14_ss): New intrinsics.
14346 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
14347 (__builtin_ia32_rcp14ss_mask): New builtins.
14348 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
14350 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
14352 PR tree-optimization/51513
14353 * tree-cfg.c (gimple_seq_unreachable_p): New function.
14354 (assert_unreachable_fallthru_edge_p): Use it.
14355 (group_case_labels_stmt): Likewise.
14356 * tree-cfg.h: Prototype it.
14357 * stmt.c: Include cfghooks.h and tree-cfg.h.
14358 (emit_case_dispatch_table) <gap_label>: New local variable.
14359 Use it to fill dispatch table gaps.
14360 Test for default_label before updating probabilities.
14361 (expand_case) <default_label>: Remove unneeded initialization.
14362 Test for unreachable default case statement and remove its edge.
14363 Set default_label accordingly.
14364 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
14366 2017-05-10 Carl Love <cel@us.ibm.com>
14368 * config/rs6000/rs6000-c: Add support for built-in functions
14369 vector signed char vec_neg (vector signed char)
14370 vector signed short int vec_neg (vector short int)
14371 vector signed int vec_neg (vector signed int)
14372 vector signed long long vec_neg (vector signed long long)
14373 vector float vec_neg (vector float)
14374 vector double vec_neg (vector double)
14375 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
14377 * config/rs6000/altivec.h: Add define for vec_neg
14378 * doc/extend.texi: Update the built-in documentation for the
14379 new built-in functions.
14381 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14383 PR tree-optimization/77644
14384 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
14386 2017-05-10 Nathan Sidwell <nathan@acm.org>
14388 * dumpfile.h (TDI_lang_all): New.
14389 (TDF_KIND): New. Renumber others
14390 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
14392 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
14394 (get_dump_file_name): Adjust suffix generation.
14395 (dump_enable_all): Use TDF_KIND.
14396 * doc/invoke.texi (-fdump-lang-all): Document.
14398 * dumpfile.h: Tabify.
14400 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
14403 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
14404 Move member access before delete.
14406 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
14408 * tree-inline.c (expand_call_inline): Split block at stmt
14411 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
14414 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
14415 are now unused after splitting mov{sf,sd}_hardfloat.
14416 (f32_lr2): Likewise.
14417 (f32_lm): Likewise.
14418 (f32_lm2): Likewise.
14419 (f32_li): Likewise.
14420 (f32_li2): Likewise.
14421 (f32_lv): Likewise.
14422 (f32_sr): Likewise.
14423 (f32_sr2): Likewise.
14424 (f32_sm): Likewise.
14425 (f32_sm2): Likewise.
14426 (f32_si): Likewise.
14427 (f32_si2): Likewise.
14428 (f32_sv): Likewise.
14429 (f32_dm): Likewise.
14430 (f32_vsx): Likewise.
14431 (f32_av): Likewise.
14432 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
14433 For movsf, order stores so the VSX stores occur before the GPR
14434 store which encourages the register allocator to use a traditional
14435 FPR instead of a GPR. For movsd, order the stores so that the GPR
14436 store comes before the VSX stores to allow the power6 to work.
14437 This is due to the power6 not having a 32-bit integer store
14438 instruction from a FPR.
14439 (movsf_hardfloat): Likewise.
14440 (movsd_hardfloat): Likewise.
14442 2017-05-09 Martin Sebor <msebor@redhat.com>
14444 PR translation/80280
14445 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
14448 PR translation/80280
14449 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
14450 data member added in r247778.
14451 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
14453 2017-05-09 Nathan Sidwell <nathan@acm.org>
14455 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
14457 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
14460 2017-05-09 Marek Polacek <polacek@redhat.com>
14462 * doc/invoke.texi: Fix typo.
14464 2017-05-09 Richard Biener <rguenther@suse.de>
14466 * tree-vrp.c (vrp_val_is_max): Adjust comment.
14467 (vrp_val_is_min): Likewise.
14468 (set_value_range_to_value): Likewise.
14469 (set_value_range_to_nonnegative): Likewise.
14470 (gimple_assign_nonzero_p): Likewise.
14471 (gimple_stmt_nonzero_p): Likewise.
14472 (vrp_int_const_binop): Likewise. Remove unreachable case.
14473 (adjust_range_with_scev): Adjust comments.
14474 (compare_range_with_value): Likewise.
14475 (extract_range_from_phi_node): Likewise.
14476 (test_for_singularity): Likewise.
14478 2017-05-09 Richard Biener <rguenther@suse.de>
14480 * tree-vrp.c (get_single_symbol): Add assert that we don't
14481 get overflowed constants as invariant part.
14482 (compare_values_warnv): Add comment before the TREE_NO_WARNING
14483 checks. Use wi::cmp instead of recursing for integer constants.
14484 (compare_values): Just ignore whether we assumed undefined
14485 overflow instead of failing the compare.
14486 (extract_range_for_var_from_comparison_expr): Add comment before the
14487 TREE_NO_WARNING sets.
14488 (test_for_singularity): Likewise.
14489 (extract_range_from_comparison): Do not disable optimization
14490 when we assumed undefined overflow.
14491 (extract_range_basic): Remove init of unused var.
14493 2017-05-09 Richard Biener <rguenther@suse.de>
14495 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
14496 (extract_range_from_multiplicative_op_1): Adjust.
14497 (extract_range_from_binary_expr_1): Use int_const_binop.
14499 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
14502 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
14503 rs6000_store_data_bypass_p in seven define_bypass directives and
14504 in several comments.
14505 * config/rs6000/rs6000-protos.h: Add prototype for
14506 rs6000_store_data_bypass_p function.
14507 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
14508 function implements slightly different (rs6000-specific) semantics
14509 than store_data_bypass_p, returning false rather than aborting
14510 with assertion error when arguments do not satisfy the
14511 requirements of store data bypass.
14512 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
14513 rs6000_store_data_bypass_p.
14515 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
14517 * config/xtensa/xtensa-protos.h
14518 (xtensa_initial_elimination_offset): New declaration.
14519 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
14520 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
14521 macro definition, add case for FRAME_POINTER_REGNUM when
14522 FRAME_GROWS_DOWNWARD.
14523 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
14524 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
14525 xtensa_initial_elimination_offset.
14527 2017-05-08 Nathan Sidwell <nathan@acm.org>
14529 * doc/invoke.texi: Alphabetize -fdump options.
14531 2017-05-08 Martin Sebor <msebor@redhat.com>
14533 PR translation/80280
14534 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
14536 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
14538 * target.def (compute_frame_layout): New optional target hook.
14539 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
14540 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
14541 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
14543 * reload1.c (verify_initial_elim_offsets): Likewise.
14544 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
14545 (use_simple_return_p): Call arm_compute_frame_layout if needed.
14546 (arm_get_frame_offsets): Split up into this ...
14547 (arm_compute_frame_layout): ... and this function.
14549 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
14551 * config/aarch64/constraints.md (Usa): New constraint.
14552 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
14554 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
14556 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
14557 with_multilib_list after it has been checked.
14559 2017-05-08 Richard Biener <rguenther@suse.de>
14561 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
14562 (bitmap_set_subtract_values): Likewise.
14564 2017-05-08 Richard Biener <rguenther@suse.de>
14566 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
14567 (gimple_assign_nonzero): ... this and remove strict_overflow_p
14569 (gimple_stmt_nonzero_warnv_p): Rename to ...
14570 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
14572 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
14573 (extract_range_basic): Adjust, do not disable propagation on
14574 strict overflow sensitive simplification.
14575 (vrp_visit_cond_stmt): Likewise.
14577 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
14579 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
14580 body size unconditionally.
14582 2017-05-07 Jeff Law <law@redhat.com>
14585 2017-05-06 Jeff Law <law@redhat.com>
14586 PR tree-optimization/78496
14587 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
14590 PR tree-optimization/78496
14591 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
14592 (simplify_stmt_using_ranges): Call it.
14593 (vrp_dom_walker::before_dom_children): Extract equivalences
14594 from an ASSERT_EXPR with an equality comparison against a
14597 2017-05-06 Jeff Law <law@redhat.com>
14599 PR tree-optimization/78496
14600 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
14603 PR tree-optimization/78496
14604 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
14605 (simplify_stmt_using_ranges): Call it.
14606 (vrp_dom_walker::before_dom_children): Extract equivalences
14607 from an ASSERT_EXPR with an equality comparison against a
14610 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
14612 * lra-constraints.c (lra_copy_reg_equiv): New function.
14613 (split_reg): Use it to copy equivalence information from the
14614 original register to the spill register.
14616 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
14618 PR rtl-optimization/75964
14619 * simplify-rtx.c (simplify_const_relational_operation): Remove
14620 invalid handling of comparisons of integer ABS.
14622 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
14624 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
14625 initialize to zero.
14626 (init_regs): Remove declaration.
14627 (function_arg_advance_32): Initialize error_p as boolean variable.
14629 2017-05-05 Nathan Sidwell <nathan@acm.org>
14631 * store-motion.c (remove_reachable_equiv_notes): Reformat long
14632 lines. Use for (;;).
14634 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14636 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
14637 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
14638 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
14639 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
14640 VF=2 that require versioning.
14642 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14644 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
14647 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14649 * diagnostic.h (diagnostic_override_option_index): Convert from
14650 macro to inline function.
14652 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14654 * diagnostic.c (last_module_changed_p): New function.
14655 (set_last_module): New function.
14656 (diagnostic_report_current_module): Convert macro usage to
14657 the above functions.
14658 * diagnostic.h (diagnostic_context::last_module): Strengthen
14659 from const line_map * to const line_map_ordinary *.
14660 (diagnostic_last_module_changed): Delete macro.
14661 (diagnostic_set_last_module): Delete macro.
14663 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14665 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
14666 with diagnostic_report_diagnostic.
14667 (diagnostic_n_impl_richloc): Likewise.
14668 * diagnostic.h (report_diagnostic): Delete macro.
14669 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
14670 with diagnostic_report_diagnostic.
14671 * substring-locations.c (format_warning_va): Likewise.
14673 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14675 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
14676 save/restor of format_spec. Move option-printing code to...
14677 (print_option_information): ...this new function, and
14678 reimplement by simply printing to the pretty_printer,
14679 rather than appending to the format string.
14681 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14683 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
14684 handling logic into...
14685 (update_effective_level_from_pragmas): ...this new function.
14687 2017-05-04 Andrew Waterman <andrew@sifive.com>
14689 * config/riscv/riscv.opt (mstrict-align): New option.
14690 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
14691 (SLOW_UNALIGNED_ACCESS): Define.
14692 (riscv_slow_unaligned_access): Declare.
14693 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
14695 (riscv_slow_unaligned_access): New variable.
14696 (rocket_tune_info): Set slow_unaligned_access to true.
14697 (optimize_size_tune_info): Set slow_unaligned_access to false.
14698 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
14699 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
14700 (riscv_option_override): Set riscv_slow_unaligned_access.
14701 * doc/invoke.texi: Add -mstrict-align to RISC-V.
14703 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
14705 * config/riscv/riscv.md: Unify indentation.
14707 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14712 * config/rs6000/rs6000.md (u code attribute): Add FIX and
14714 (extendsi<mode>2): Add support for doing sign extension via
14715 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
14716 don't have ISA 3.0 instructions.
14717 (extendsi<mode>2 splitter): Likewise.
14718 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
14719 generate the normal insns since SImode can now go in vector
14720 registers. Disallow the special UNSPECs needed for previous
14721 machines to hide SImode being used. Add new insns
14722 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
14723 (fix_trunc<mode>si2_stfiwx): Likewise.
14724 (fix_trunc<mode>si2_internal): Likewise.
14725 (fixuns_trunc<mode>si2): Likewise.
14726 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14727 (fctiw<u>z_<mode>_smallint): Likewise.
14728 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
14729 of floating point to 32-bit integer from doing a direct move to
14730 the GPR registers to do a store.
14731 (fctiwz_<mode>): Break long line.
14733 2017-05-05 Bin Cheng <bin.cheng@arm.com>
14735 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
14736 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
14737 (addr_list, addr_offset_valid_p): New.
14738 (split_address_groups): Check offset validity with above function.
14739 (gt-tree-ssa-loop-ivopts.h): Include header file.
14741 2017-05-05 Nathan Sidwell <nathan@acm.org>
14743 * config.gcc (arm*-*-*): Add missing 'fi'.
14745 2017-05-05 Steve Ellcey <sellcey@cavium.com>
14747 * doc/invoke.texi (-fopt-info): Explicitly say order of options
14748 included in -fopt-info does not matter.
14749 * doc/optinfo.texi (-fopt-info): Fix description of default
14750 behavour. Explicitly say order of options included in -fopt-info
14753 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
14755 * config.gcc: Allow combinations of aprofile and rmprofile values for
14756 --with-multilib-list.
14757 * config/arm/t-multilib: New file.
14758 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
14759 variables. Remove setting of ISA and floating-point ABI in
14760 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
14761 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
14762 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
14763 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
14765 * config/arm/t-rmprofile: Likewise except for the matches changes.
14766 * doc/install.texi (--with-multilib-list): Document the combination of
14767 aprofile and rmprofile values and warn about pitfalls in doing that.
14769 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
14771 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
14772 (movdi_aarch64): Likewise.
14774 2017-05-05 Jakub Jelinek <jakub@redhat.com>
14776 PR tree-optimization/80632
14777 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
14779 (build_arrays): Initialize it for virtual phis.
14780 (fix_phi_nodes): Use it for virtual phis.
14782 PR tree-optimization/80558
14783 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
14784 [x, y] op z into [x op, y op z] for op & or | if conditions
14787 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
14788 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14791 * config/arm/arm.md (use_literal_pool): Remove.
14792 (64-bit immediate split): No longer takes cost into consideration
14793 if arm_disable_literal_pool is enabled.
14794 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
14795 used when arm_disable_literal_pool is enabled.
14796 (arm_max_const_double_inline_cost): Remove use of
14797 arm_disable_literal_pool.
14798 (push_minipool_fix): Add assert.
14799 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
14800 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
14801 (no_literal_pool_sf_immediate): New.
14803 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14805 PR tree-optimization/80613
14806 * tree-ssa-dce.c (propagate_necessity): Remove cases for
14807 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
14809 2017-05-05 Richard Biener <rguenther@suse.de>
14811 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
14813 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
14815 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
14816 of this flag from insn conditions due to removal from r247495.
14818 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
14820 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
14822 (arm_early_store_addr_dep_ptr): Likewise.
14823 * config/arm/aarch-common-protos.h
14824 (arm_early_load_addr_dep_ptr): Add prototype.
14825 (arm_early_store_addr_dep_ptr): Likewise.
14826 * config/arm/cortex-a53.md: Add new bypasses.
14828 2017-05-05 Jakub Jelinek <jakub@redhat.com>
14830 * tree.c (next_type_uid): Change type to unsigned.
14831 (type_hash_canon): Decrement back next_type_uid if
14832 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
14833 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
14836 2017-05-04 Martin Sebor <msebor@redhat.com>
14838 * builtins.c: Fix a trivial typo in a comment.
14840 PR middle-end/79234
14841 * builtins.c (check_sizes): Adjust to handle reading past the end.
14842 Avoid printing excessive upper bound of ranges. Use %E to print
14843 tree nodes instead of converting them to %wu.
14844 (expand_builtin_memchr): New function.
14845 (compute_dest_size): Rename...
14846 (compute_objsize): ...to this.
14847 (expand_builtin_memcpy): Adjust.
14848 (expand_builtin_mempcpy): Adjust.
14849 (expand_builtin_strcat): Adjust.
14850 (expand_builtin_strcpy): Adjust.
14851 (check_strncat_sizes): Adjust.
14852 (expand_builtin_strncat): Adjust.
14853 (expand_builtin_strncpy): Adjust and simplify.
14854 (expand_builtin_memset): Adjust.
14855 (expand_builtin_bzero): Adjust.
14856 (expand_builtin_memcmp): Adjust.
14857 (expand_builtin): Handle memcmp.
14858 (maybe_emit_chk_warning): Check strncat just once.
14860 2017-05-04 Martin Sebor <msebor@redhat.com>
14862 PR preprocessor/79214
14863 PR middle-end/79222
14864 PR middle-end/79223
14865 * builtins.c (check_sizes): Add inlining context and issue
14866 warnings even when -Wno-system-headers is set.
14867 (check_strncat_sizes): Same.
14868 (expand_builtin_strncat): Same.
14869 (expand_builtin_memmove): New function.
14870 (expand_builtin_stpncpy): Same.
14871 (expand_builtin): Handle memmove and stpncpy.
14873 2017-05-04 Bin Cheng <bin.cheng@arm.com>
14875 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
14876 which is not used any more.
14878 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14880 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
14882 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14884 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
14885 (cortexa53_tunings): Likewise.
14886 (cortexa57_tunings): Likewise.
14887 (cortexa72_tunings): Likewise.
14888 (cortexa73_tunings): Likewise.
14890 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14892 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
14893 Set loop alignment to 8.
14895 2017-05-04 Martin Sebor <msebor@redhat.com>
14897 PR translation/80280
14898 * builtins.c (expand_builtin_object_size): Add missing quoting to
14899 %D and like directives.
14900 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
14901 (hsa_type_for_tree_type): Same.
14902 (verify_function_arguments): Same.
14903 * symtab.c (symbol_table::change_decl_assembler_name): Same.
14904 * varasm.c (get_section): Same.
14907 2017-05-04 Martin Sebor <msebor@redhat.com>
14909 PR translation/80280
14910 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
14912 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14914 * config/aarch64/aarch64.c (generic_addrcost_table):
14915 Change HI/TI mode setting.
14917 2017-05-04 Martin Jambor <mjambor@suse.cz>
14919 PR tree-optimization/80622
14920 * tree-sra.c (comes_initialized_p): New function.
14921 (build_accesses_from_assign): Only set write lazily when
14922 comes_initialized_p is false.
14923 (analyze_access_subtree): Use comes_initialized_p.
14924 (propagate_subaccesses_across_link): Assert !comes_initialized_p
14925 instead of testing for PARM_DECL.
14927 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14929 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
14930 constraint on operand 0 to allow more general addressing modes.
14931 Adjust output template.
14932 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
14934 * config/aarch64/aarch64-protos.h
14935 (aarch64_address_valid_for_prefetch_p): Declare prototype.
14936 * config/aarch64/constraints.md (Dp): New address constraint.
14937 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
14940 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
14942 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
14943 update use of estimate_ipcp_clone_size_and_time.
14944 (estimate_local_effects): Update use of
14945 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
14946 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
14947 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
14948 Return nonspecialized time.
14950 2017-05-04 Richard Biener <rguenther@suse.de>
14952 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
14953 for the last VUSE which def dominates the PHI. Directly call
14955 (get_continuation_for_phi_1): Remove.
14957 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
14959 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
14960 to explain the use of truncating division. Cap the number of
14961 iterations to the maximum given by nb_iterations_upper_bound,
14964 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14966 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
14967 * configure: Regenerate.
14968 * config.in: Regenerate.
14969 * config/i386/driver-mingw32.c: new file.
14970 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
14971 * config.host: Link driver-mingw32.o on MinGW host.
14972 * doc/install.texi: Document new --enable-mingw-wildcard configure
14975 2017-05-04 Marek Polacek <polacek@redhat.com>
14977 PR tree-optimization/80612
14978 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
14980 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14981 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
14983 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
14984 (movt splitter): Likewise.
14985 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
14986 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
14987 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
14988 block for Thumb-1 with MOVT.
14989 (thumb2_legitimate_address_p): Move code block ...
14990 (can_avoid_literal_pool_for_label_p): ... into this new function.
14991 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
14993 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
14994 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
14995 "M-profile targets with the MOVT instruction".
14997 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14999 * config/arm/arm-builtins.c (arm_init_builtins): Rename
15000 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
15001 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
15003 2017-05-04 Martin Liska <mliska@suse.cz>
15005 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
15006 variable cond_code.
15008 2017-05-04 Richard Biener <rguenther@suse.de>
15010 * tree.c (array_at_struct_end_p): Handle arrays at struct
15011 end with flexarrays more conservatively. Refactor and treat
15012 arrays of arrays or aggregates more strict. Fix
15013 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
15014 * tree.c (array_at_struct_end_p): Adjust prototype.
15015 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
15016 * gimple-fold.c (get_range_strlen): Likewise.
15017 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
15019 2017-05-04 Richard Biener <rguenther@suse.de>
15021 PR tree-optimization/31130
15022 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
15024 (supports_overflow_infinity): Likewise.
15025 (is_negative_overflow_infinity): Likewise.
15026 (is_positive_overflow_infinity): Likewise.
15027 (is_overflow_infinity): Likewise.
15028 (stmt_overflow_infinity): Likewise.
15029 (overflow_infinity_range_p): Likewise.
15030 (usable_range_p): Remove as always returning true.
15031 (make_overflow_infinity): Remove.
15032 (negative_overflow_infinity): Likewise.
15033 (positive_overflow_infinity): Likewise.
15034 (avoid_overflow_infinity): Likewise.
15035 (set_value_range): Adjust accordingly.
15036 (set_value_range_to_nonnegative): Likewise, remove now unused
15037 overflow_infinity arg.
15038 (vrp_operand_equal_p): Adjust.
15039 (update_value_range): Likewise.
15040 (range_int_cst_singleton_p): Likewise.
15041 (operand_less_p): Likewise.
15042 (compare_values_warnv): Likewise.
15043 (extract_range_for_var_from_comparison_expr): Likewise.
15044 (vrp_int_const_binop): Likewise.
15045 (zero_nonzero_bits_from_vr): Likewise.
15046 (extract_range_from_multiplicative_op_1): Likewise.
15047 (extract_range_from_binary_expr_1): Likewise.
15048 (extract_range_from_unary_expr): Likewise.
15049 (extract_range_from_comparison): Likewise.
15050 (extract_range_basic): Likewise.
15051 (adjust_range_with_scev): Likewise.
15052 (compare_ranges): Likewise.
15053 (compare_range_with_value): Likewise.
15054 (dump_value_range): Likewise.
15055 (test_for_singularity): Likewise, remove strict_overflow_p parameter
15057 (simplify_cond_using_ranges): Adjust.
15059 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
15061 * brig-builtins.def: Added a builtin for class_f64.
15062 * builtin-types.def: Added a builtin type needed by class_f64.
15064 2017-05-03 Jason Merrill <jason@redhat.com>
15066 * timevar.def: Add TV_CONSTEXPR.
15068 2017-05-03 David Malcolm <dmalcolm@redhat.com>
15070 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
15072 2017-05-03 Martin Jambor <mjambor@suse.cz>
15074 * ipa-prop.c (ipa_update_after_lto_read): Removed.
15075 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
15076 * ipa-cp.c (ipcp_propagate_stage): Do not call
15077 ipa_update_after_lto_read.
15078 * ipa-inline.c (ipa_inline): Likewise.
15080 2017-05-03 Martin Jambor <mjambor@suse.cz>
15082 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
15083 tag. Added a default constructor and a destructor.
15084 (ipa_edge_args_sum_t): New class;
15085 (ipa_edge_args_sum): Declare.
15086 (ipa_edge_args_vector): Remove declaration.
15087 (IPA_EDGE_REF): Use ipa_edge_args_sum.
15088 (ipa_free_edge_args_substructures): Remove declaration.
15089 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
15090 (ipa_edge_args_info_available_for_edge_p): Likewise.
15091 * ipa-prop.c (ipa_edge_args_vector): Removed.
15092 (edge_removal_hook_holder): Likewise.
15093 (edge_duplication_hook_holder): Likewise.
15094 (ipa_edge_args_sum): New variable.
15095 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
15096 ipa_edge_args_vector.
15097 (ipa_free_edge_args_substructures): Likewise.
15098 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
15099 ipa_edge_args_vector.
15100 (ipa_edge_removal_hook): Turned into method
15101 ipa_edge_args_sum_t::remove.
15102 (ipa_edge_duplication_hook): Turned into method
15103 ipa_edge_args_sum_t::duplicate.
15104 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
15105 registering edge hooks.
15106 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
15107 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
15108 ipa_edge_args_sum instead of ipa_edge_args_vector.
15109 * ipa-profile.c (ipa_profile): Likewise.
15111 2017-05-03 Martin Jambor <mjambor@suse.cz>
15113 * symbol-summary.h (function_summary): New method exists.
15114 (function_summary::symtab_removal): Deallocate through release.
15115 (call_summary): New class.
15116 (gt_ggc_mx): New overload.
15117 (gt_pch_nx): Likewise.
15118 (gt_pch_nx): Likewise.
15120 2017-05-03 Jeff Law <law@redhat.com>
15122 PR tree-optimization/78496
15123 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
15124 from simplify_cond_using_ranges. Split off code to walk
15125 backwards through casts into ...
15126 (simplify_cond_using_ranges_2): New function.
15127 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
15128 (execute_vrp): After identifying jump threads, call
15129 simplify_cond_using_ranges_2.
15131 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
15134 * ipa-inline.h (inline_summary): Add ctor.
15135 (create_ggc): Do not use ggc_cleared_alloc.
15137 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
15138 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15140 * gcc.c (handle_braces): Support escaping in switch matching
15142 * doc/invoke.texi (Spec Files): Document it.
15143 Remove superfluous @code markup in items.
15145 2017-05-03 David Malcolm <dmalcolm@redhat.com>
15147 * diagnostic-show-locus.c (struct column_range): New struct.
15148 (get_affected_columns): New function.
15149 (get_printed_columns): New function.
15150 (struct correction): New struct.
15151 (correction::ensure_capacity): New function.
15152 (correction::ensure_terminated): New function.
15153 (struct line_corrections): New struct.
15154 (line_corrections::~line_corrections): New dtor.
15155 (line_corrections::add_hint): New function.
15156 (layout::print_trailing_fixits): Reimplement in terms of the new
15158 (selftest::test_overlapped_fixit_printing): New function.
15159 (selftest::diagnostic_show_locus_c_tests): Call it.
15161 2017-05-03 Nathan Sidwell <nathan@acm.org>
15163 Canonicalize canonical type hashing
15164 * tree.h (type_hash_canon_hash): Declare.
15165 * tree.c (type_hash_list, attribute_hash_list): Move into
15166 type_hash_canon_hash.
15167 (build_type_attribute_qual_variant): Break out hash code calc into
15168 type_hash_canon_hash.
15169 (type_hash_canon_hash): New. Generic type hash computation.
15170 (build_range_type_1, build_array_type_1, build_function_type,
15171 build_method_type_directly, build_offset_type, build_complex_type,
15172 make_vector_type): Call it.
15174 2017-05-03 Richard Biener <rguenther@suse.de>
15176 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
15177 When all DRs have unknown misaligned do not always peel
15178 when there is a store but apply the same costing model as if
15179 there were only loads.
15181 2017-05-03 Richard Biener <rguenther@suse.de>
15184 PR tree-optimization/80492
15185 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
15186 compare_base_decls returning dont-know properly.
15188 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
15190 * config/arm/iterators.md (CCSI): New mode iterator.
15191 (arch): New mode attribute.
15192 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
15193 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
15194 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
15195 code iterator for success result mode.
15196 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
15197 the corresponding new insn generators.
15199 2017-05-03 Bin Cheng <bin.cheng@arm.com>
15202 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15203 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15205 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
15207 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
15208 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
15209 (DDR_A): Wrap DDR argument in brackets.
15210 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
15211 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
15212 (DDR_REVERSED_P): Likewise.
15214 2017-05-03 Jakub Jelinek <jakub@redhat.com>
15216 PR tree-optimization/79472
15217 * tree-switch-conversion.c (struct switch_conv_info): Add
15218 contiguous_range and default_case_nonstandard fields.
15219 (collect_switch_conv_info): Compute contiguous_range and
15220 default_case_nonstandard fields, don't clear final_bb if
15221 contiguous_range and only the default case doesn't have the required
15223 (check_all_empty_except_final): Set default_case_nonstandard instead
15224 of failing if contiguous_range and the default case doesn't have empty
15226 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
15227 and only the default case doesn't have the required constants. Skip
15229 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
15230 if default_case_nonstandard.
15231 (build_constructors): Build constant 1 just once. Assert that default
15232 values aren't inserted in between cases if contiguous_range. Skip
15234 (build_arrays): Skip virtual phis.
15235 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
15236 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
15237 Handle virtual phis.
15238 (gen_inbound_check): Handle default_case_nonstandard case.
15239 (process_switch): Adjust check_final_bb caller. Call
15240 gather_default_values with the first non-default case instead of
15241 default case if default_case_nonstandard.
15243 2017-05-02 Nathan Sidwell <nathan@acm.org>
15245 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
15246 check. Fix formatting.
15248 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
15250 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
15251 errors when comparing specialized and unspecialized times.
15253 2017-05-02 David Malcolm <dmalcolm@redhat.com>
15255 * diagnostic-show-locus.c
15256 (layout::should_print_annotation_line_p): Make private.
15257 (layout::print_annotation_line): Make private.
15258 (layout::annotation_line_showed_range_p): Make private.
15259 (layout::show_ruler): Make private.
15260 (layout::print_source_line): Make private. Pass in line and
15261 line_width, rather than calling location_get_source_line. Drop
15263 (layout::print_leading_fixits): New method.
15264 (layout::print_any_fixits): Rename to...
15265 (layout::print_trailing_fixits): ...this, and make private.
15266 Don't print newline fixits.
15267 (diagnostic_show_locus): Move logic for printing one row into...
15268 (layout::print_line): ...this new function. Move the
15269 location_get_source_line call and error-handling from
15270 print_source_line to here. Call print_leading_fixits, and rename
15271 print_any_fixits to print_trailing_fixits.
15272 (selftest::test_fixit_insert_containing_newline): Update now that
15273 newlines are partially supported.
15274 (selftest::test_fixit_insert_containing_newline_2): New test.
15275 (selftest::test_fixit_replace_containing_newline): Update comments.
15276 (selftest::diagnostic_show_locus_c_tests): Call the new test.
15277 * edit-context.c (class added_line): New class.
15278 (class edited_line): Describe newline handling in comment.
15279 (edited_line::actually_edited_p): New method.
15280 (edited_line::print_content): Delete redundant decl.
15281 (edited_line::m_predecessors): New field.
15282 (edited_file::print_content): Call edited_line::print_content.
15283 (edited_file::print_diff): Update to support newlines.
15284 (edited_file::print_diff_hunk): Likewise.
15285 (edited_file::print_run_of_changed_lines): New function.
15286 (edited_file::print_diff_line): Convert to...
15287 (print_diff_line): ...this.
15288 (edited_file::get_effective_line_count): New function.
15289 (edited_line::edited_line): Initialize new field m_predecessors.
15290 (edited_line::~edited_line): Clean up m_predecessors.
15291 (edited_line::apply_fixit): Handle newlines.
15292 (edited_line::get_effective_line_count): New function.
15293 (edited_line::print_content): New function.
15294 (edited_line::print_diff_lines): New function.
15295 (selftest::test_applying_fixits_insert_containing_newline): New
15297 (selftest::test_applying_fixits_replace_containing_newline): New
15299 (selftest::insert_line): New function.
15300 (selftest::test_applying_fixits_multiple_lines): Add example of
15302 (selftest::edit_context_c_tests): Call the new tests.
15304 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15306 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
15307 parameter cand. Update dump information.
15308 (get_computation_cost): Update uses.
15310 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15312 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
15313 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
15314 (get_computation_at, rewrite_use_address): Update use of
15315 get_computation_aff.
15317 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15319 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
15320 (get_computation): Delete.
15321 (get_computation_cost): Implement like get_computation_cost_at.
15322 Use get_computation_at.
15323 (get_computation_cost_at): Delete.
15324 (rewrite_use_nonlinear_expr): Use get_computation_at.
15325 (rewrite_use_compare, remove_unused_ivs): Ditto.
15327 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15329 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
15331 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15333 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
15334 (ivopts_global_cost_for_size): Rename parameter and update uses.
15335 (iv_ca_recount_cost): Update uses.
15336 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
15337 candidates seperately in n_invs and n_cands.
15338 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
15340 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15342 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
15343 (find_inv_vars_cb): New.
15344 (find_depends): Renamed to ...
15345 (find_inv_vars): ... this.
15346 (add_candidate_1, force_var_cost): Call find_inv_vars.
15347 (split_address_cost, determine_group_iv_cost_cond): Ditto.
15349 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15351 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
15352 inv_vars. Add inv_exprs.
15353 (struct iv_cand): Rename depends_on to inv_vars.
15354 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
15355 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
15356 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
15357 (dump_cand): Dump inv_vars.
15358 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
15359 (record_invariant, find_depends, add_candidate_1): Ditto.
15360 (set_group_iv_cost, force_var_cost): Ditto.
15361 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
15362 (get_computation_cost_at, get_computation_cost): Ditto.
15363 (determine_group_iv_cost_generic): Ditto.
15364 (determine_group_iv_cost_address): Ditto.
15365 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
15366 (determine_group_iv_costs): Ditto.
15367 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
15368 (iv_ca_set_remove_invariants): Renamed to ...
15369 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
15370 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
15371 (iv_ca_set_add_invariants): Renamed to ...
15372 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
15373 (iv_ca_set_cp): Use iv_ca_set_add_invs.
15374 (iv_ca_has_deps): Support inv_vars and inv_exprs.
15375 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
15376 (create_new_ivs): Remove useless dump.
15378 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15380 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
15382 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
15383 (iv_ca_set_no_cp, create_new_iv): Ditto.
15385 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15387 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15389 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
15391 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
15392 function tree_check2.
15394 2017-05-02 Martin Liska <mliska@suse.cz>
15396 * doc/gcov.texi: Add missing preposition.
15397 * gcov.c (function_info::function_info): Properly fill up
15398 all member variables.
15400 2017-05-02 Tamar Christina <tamar.christina@arm.com>
15402 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
15404 2017-05-02 Tamar Christina <tamar.christina@arm.com>
15406 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
15408 2017-05-02 Martin Liska <mliska@suse.cz>
15411 * lto-streamer-in.c (lto_read_tree_1): Remove
15412 LTO_STREAMER_DEBUG.
15413 * lto-streamer.c (struct tree_hash_entry): Likewise.
15414 (struct tree_entry_hasher): Likewise.
15415 (tree_entry_hasher::hash): Likewise.
15416 (tree_entry_hasher::equal): Likewise.
15417 (lto_streamer_init): Likewise.
15418 (lto_orig_address_map): Likewise.
15419 (lto_orig_address_get): Likewise.
15420 (lto_orig_address_remove): Likewise.
15421 * lto-streamer.h: Likewise.
15422 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
15423 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
15425 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
15427 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
15428 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
15429 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
15430 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
15431 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
15432 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
15433 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
15434 (mm_maskz_sub_ss): New intrinsics.
15435 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15436 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
15437 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
15438 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
15439 (__builtin_ia32_subss_mask_round): New builtins.
15440 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15441 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15442 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
15444 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
15445 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
15447 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
15450 2017-05-02 Martin Jambor <mjambor@suse.cz>
15452 PR tree-optimization/78687
15453 * tree-sra.c (access): New field parent.
15454 (process_subtree_disqualification): New function.
15455 (disqualify_candidate): Call it.
15456 (build_accesses_from_assign): Reset write flag if creating an
15458 (build_access_subtree): Fill in parent field and also prpagate
15459 down grp_write flag.
15460 (create_artificial_child_access): New parameter set_grp_write, set
15461 grp_write to its value.
15462 (propagate_subaccesses_across_link): Also propagate grp_write flag
15464 (propagate_all_subaccesses): Push the closest parent back to work
15465 queue if add_access_to_work_queue returned true.
15467 2017-05-02 Richard Biener <rguenther@suse.de>
15469 * common.opt (fstrict-overflow): Alias negative to fwrapv.
15470 * doc/invoke.texi (fstrict-overflow): Remove all traces of
15471 -fstrict-overflow documentation.
15472 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
15473 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
15474 flag_strict_overflow.
15475 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
15476 * lto-opts.c (lto_write_options): Do not stream it.
15477 * lto-wrapper.c (merge_and_complain): Do not handle it.
15478 * opts.c (default_options_table): Do not set -fstrict-overflow.
15479 (finish_options): Likewise do not clear it when sanitizing.
15480 * simplify-rtx.c (simplify_const_relational_operation): Do not
15481 test flag_strict_overflow.
15483 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
15485 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
15486 using enabled attribute.
15487 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
15488 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
15489 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
15490 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
15491 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
15492 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
15493 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
15494 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
15495 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
15496 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
15498 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
15500 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
15502 2017-05-02 Richard Biener <rguenther@suse.de>
15504 PR tree-optimization/80591
15506 2017-04-10 Richard Biener <rguenther@suse.de>
15508 * tree-ssa-structalias.c (find_func_aliases): Properly handle
15511 2017-05-02 Richard Biener <rguenther@suse.de>
15513 PR tree-optimization/80549
15514 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
15515 (cleanup_tree_cfg_noloop): Create forwarders to known loop
15516 headers if they do not have a preheader.
15518 2017-05-02 Martin Liska <mliska@suse.cz>
15521 * common.opt: Fix typo.
15522 * doc/invoke.texi: Likewise.
15524 2017-05-01 Jan Beulich <jbeulich@suse.com>
15526 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
15527 swapping, add (x,x,m,x,n) alternative.
15529 2017-05-01 Nathan Sidwell <nathan@acm.org>
15531 * calls.c (combine_pending_stack_adjustment_and_call): Remove
15532 unnecessary unadjusted_alignment check.
15534 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
15537 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
15539 * gimplify.c (gimplify_cilk_detach): New function.
15540 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
15541 * tree-core.h: Document EXPR_CILK_SPAWN.
15542 * tree.h (EXPR_CILK_SPAWN): Define.
15544 2017-05-01 David Malcolm <dmalcolm@redhat.com>
15546 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
15547 to use new fixit_hint representation, using the "replace" logic.
15548 (get_line_span_for_fixit_hint): Likewise.
15549 (layout::print_any_fixits): Likewise.
15550 (selftest::test_one_liner_many_fixits): Rename to...
15551 (selftest::test_one_liner_many_fixits_1): ...this, and update
15552 comment and expected output to reflect that the multiple fix-it
15553 hints are now consolidated into one insertion.
15554 (selftest::test_one_liner_many_fixits_2): New test.
15555 (selftest::test_diagnostic_show_locus_one_liner): Update for
15557 (selftest::test_fixit_consolidation): Update for fix-it API
15559 * diagnostic.c (print_parseable_fixits): Likewise.
15560 * edit-context.c (edited_line::m_line_events): Convert from
15561 auto_vec <line_event *> to auto_vec <line_event>.
15562 (class line_event): Convert from abstract base class to a concrete
15563 class, taking over the role of replace_event.
15564 (class insert_event): Delete.
15565 (class replace_event): Rename to class line_event. Convert to
15567 (edit_context::add_fixits): Reimplement.
15568 (edit_context::apply_insert): Delete.
15569 (edit_context::apply_replace): Rename to...
15570 (edit_context::apply_fixit): ...this. Convert to half-open range.
15571 (edited_file::apply_insert): Delete.
15572 (edited_file::apply_replace): Rename to...
15573 (edited_file::apply_fixit): ...this.
15574 (edited_line::~edited_line): Drop deletion of events.
15575 (edited_line::apply_insert): Delete.
15576 (edited_line::apply_replace): Rename to...
15577 (edited_line::apply_fixit): ...this. Convert to half-open range.
15578 Update for change to type of m_line_events.
15579 * edit-context.h (edit_context::apply_insert): Delete.
15580 (edit_context::apply_replace): Rename to...
15581 (edit_context::apply_fixit): ...this.
15583 2017-05-01 Martin Sebor <msebor@redhat.com>
15585 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
15588 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
15591 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
15592 __get_cpuid_max returns 0.
15593 (__get_cpuid_count): Ditto.
15595 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
15597 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
15598 replacement expression is another instance of one of its arguments.
15600 2017-05-01 Jakub Jelinek <jakub@redhat.com>
15603 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
15604 check for stack push/pop autoinc.
15605 * config/i386/i386.c (ix86_agi_dependent): Return false
15606 if the only reason why modified_in_p returned true is that
15607 addr is SP based and set_insn is a push or pop.
15609 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
15611 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
15614 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
15617 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
15618 (account_size_time): Use two predicates - exec_pred and
15620 (evaluate_conditions_for_known_args): Compute both clause and
15622 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
15623 (inline_summary_t::duplicate): Update.
15624 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
15626 (compute_inline_parameters): Likewise.
15627 (estimate_edge_size_and_time): Update caluclation of time.
15628 (estimate_node_size_and_time): Compute both time and nonspecialized
15630 (estimate_ipcp_clone_size_and_time): Update.
15631 (inline_merge_summary): Update.
15632 (do_estimate_edge_time): Update.
15633 (do_estimate_edge_size): Update.
15634 (do_estimate_edge_hints): Update.
15635 (inline_read_section, inline_write_summary): Stream both new predicates.
15636 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
15638 (compute_inlined_call_time): Cleanup.
15639 (big_speedup_p): Update.
15640 (edge_badness): Update.
15641 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
15642 (size_time_entry): Replace predicate by exec_predicate and
15643 nonconst_predicate.
15644 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
15645 (estimate_edge_time): Return also nonspec_time.
15646 (reset_edge_growth_cache): Update.
15648 2017-04-29 Jakub Jelinek <jakub@redhat.com>
15650 PR rtl-optimization/80491
15651 * ifcvt.c (noce_process_if_block): When looking for x setter
15652 with missing else_bb, don't check only the insn right before
15653 cond_earliest, but look for the last insn that x is modified in
15654 within the same bb.
15656 PR rtl-optimization/80491
15657 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
15659 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
15661 PR tree-optimization/80487
15662 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
15664 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15666 PR tree-optimization/79697
15667 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
15668 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
15669 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
15671 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
15672 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
15674 2017-04-28 Martin Sebor <msebor@redhat.com>
15676 PR tree-optimization/80523
15677 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
15678 (init_target_to_host_charmap, target_to_host, target_strtol10): New
15680 (maybe_warn, format_directive, parse_directive): Use new functions.
15681 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
15683 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
15685 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
15687 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
15689 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
15690 target_header_dir): Set correctly.
15691 * configure: Regenerated.
15692 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15693 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15694 instead of SYSTEM_HEADER_DIR.
15696 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
15698 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
15699 (estimate_local_effects): Likewise.
15700 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
15701 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
15702 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
15703 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
15704 do_estimate_edge_time, estimate_edge_time): Likewise.
15705 * ipa-inline-analysis.c (estimate_node_size_and_time,
15706 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
15707 (estimate_time_after_inlining): Remove.
15709 2017-04-28 Martin Liska <mliska@suse.cz>
15711 * doc/gcov.texi: Enhance documentation of gcov.
15713 2017-04-28 Martin Liska <mliska@suse.cz>
15715 * doc/gcov.texi: Sort options in alphabetic order.
15716 * doc/gcov-dump.texi: Likewise.
15717 * doc/gcov-tool.texi: Likewise.
15718 * gcov.c (print_usage): Likewise.
15719 * gcov-dump.c (print_usage): Likewise.
15720 * gcov-tool.c (print_merge_usage_message): Likewise.
15721 (print_rewrite_usage_message): Likewise.
15722 (print_overlap_usage_message): Likewise.
15724 2017-04-28 Martin Liska <mliska@suse.cz>
15726 PR gcov-profile/53915
15727 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
15729 2017-04-28 Martin Liska <mliska@suse.cz>
15731 PR gcov-profile/79891
15732 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
15733 is marked by compiler as living on a line.
15734 (get_cycles_count): Remove usage of the union.
15735 (output_intermediate_file): Likewise.
15736 (find_source): Fix GNU coding style.
15737 (accumulate_line_counts): Remove old non-all block mode.
15738 (output_lines): Remove usage of the union.
15739 * profile.c (output_location): Include all BBs, even if
15740 belonging to a same line (and file) as a previous BB.
15742 2017-04-28 Martin Liska <mliska@suse.cz>
15744 * gcov.c (process_args): Handle new argument 'w'.
15745 (read_graph_file): Assign ID to BBs.
15746 (output_branch_count): Display BB # if verbose flag is set.
15747 (output_lines): Likewise for arcs.
15748 (print_usage): Add '--verbose' option help.
15749 * doc/gcov.texi: Document --verbose (-w) option.
15751 2017-04-28 Martin Liska <mliska@suse.cz>
15753 * gcov.c (struct block_location_info): New struct.
15754 (process_file): Fill up the new structure.
15755 (read_graph_file): Replace usage of encoding by the newly added
15757 (add_line_counts): Likewise.
15758 (accumulate_line_counts): Remove usage of the union.
15759 (function_info::function_info): New function.
15760 (function_info::~function_info): Likewise.
15761 (process_file): Call delete instead of release_function.
15762 (release_function): Release the function.
15763 (release_structures): Call delete instead of release_function.
15764 (solve_flow_graph): Replace usage of num_blocks.
15765 (find_exception_blocks): Likewise.
15766 (output_lines): Fix GNU coding style.
15768 2017-04-28 Martin Liska <mliska@suse.cz>
15771 * coverage.c (coverage_remove_note_file): New function.
15772 * coverage.h: Declare the function.
15773 * toplev.c (finalize): Clean if an error has been seen.
15775 2017-04-28 Martin Liska <mliska@suse.cz>
15777 PR gcov-profile/80031
15778 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
15779 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
15780 * gcov.c (read_graph_file): Read just number of blocks.
15781 * profile.c (branch_prob): Do not stream 0 flags per a basic
15784 2017-04-28 Martin Liska <mliska@suse.cz>
15786 * gcov-dump.c (tag_*): Add new argument to declarations.
15787 (dump_gcov_file): Likewise.
15788 (tag_blocks): Add and use new argument depth.
15789 (tag_arcs): Likewise.
15790 (tag_lines): Likewise.
15791 (tag_counters): Likewise.
15792 (tag_summary): Likewise.
15793 (dump_working_sets): Use depth to do a proper indentation.
15795 2017-04-28 Jakub Jelinek <jakub@redhat.com>
15798 * cgraph.h (symtab_node::debug_symtab): No longer inline.
15799 * symtab.c (symtab_node::debug_symtab): Move definition here.
15801 2017-04-28 Richard Biener <rguenther@suse.de>
15803 * lto-streamer.h (LTO_major_version): Bump to 7.
15805 2017-04-28 Richard Biener <rguenther@suse.de>
15807 * tree-vrp.c (assert_info): New struct.
15808 (add_assert_info): New helper.
15809 (register_edge_assert_for_2): Refactor to add asserts to a vector
15811 (register_edge_assert_for_1): Likewise.
15812 (register_edge_assert_for): Likewise.
15813 (finish_register_edge_assert_for): New helper actually registering
15814 asserts where live on edge.
15815 (find_conditional_asserts): Adjust.
15816 (find_switch_asserts): Likewise.
15817 (evrp_dom_walker::try_find_new_range): Generalize.
15818 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
15820 2017-04-27 Marek Polacek <polacek@redhat.com>
15823 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
15824 arg10 and arg11 to itype.
15826 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
15828 * doc/extend.texi (Object Size Checking): Improve grammar.
15830 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
15833 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
15834 that the logic for permitting reciprocal estimates matches that
15837 2017-04-27 Jakub Jelinek <jakub@redhat.com>
15840 * tree.c (type_cache_hasher::equal): Only compare
15841 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
15842 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
15843 non-aggregate element types.
15844 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
15845 about the flag on ARRAY_TYPEs in the comment, formatting fix.
15847 2017-04-27 Richard Biener <rguenther@suse.de>
15849 PR middle-end/80533
15850 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
15851 stripping ARRAY_REFs from MEM_EXPR make sure we're not
15852 keeping a reference to a trailing array.
15854 2017-04-27 Richard Biener <rguenther@suse.de>
15856 PR middle-end/80539
15857 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
15858 being in loop-closed SSA form conservatively.
15859 (chrec_fold_multiply_poly_poly): Likewise.
15861 2017-04-27 Tamar Christina <tamar.christina@arm.com>
15863 PR middle-end/79665
15864 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
15865 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
15867 2017-04-27 Jakub Jelinek <jakub@redhat.com>
15870 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
15871 (aarch64_function_arg_alignment): Return unsigned int again, but still
15872 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
15873 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
15874 Don't emit -Wpsabi note.
15875 (aarch64_function_arg_boundary): Likewise.
15876 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
15879 2017-04-26 Nathan Sidwell <nathan@acm.org>
15881 * tree.h (crc32_unsigned_n): Declare.
15882 (crc32_unsigned, crc32_unsigned): Make inline.
15883 * tree.c (crc32_unsigned_bits): Replace with ...
15884 (crc32_unsigned_n): ... this.
15885 (crc32_unsigned, crc32_byte): Remove.
15886 (crc32_string): Remove unnecessary braces.
15888 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15890 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
15891 * ipa-inline-analysis.c (MAX_TIME): Remove.
15892 (account_size_time): Use sreal for time.
15893 (dump_inline_summary): Update.
15894 (estimate_function_body_sizes): Update.
15895 (estimate_edge_size_and_time): Update.
15896 (estimate_calls_size_and_time): Update.
15897 (estimate_node_size_and_time): Update.
15898 (inline_merge_summary): Update.
15899 (inline_update_overall_summary): Update.
15900 (estimate_time_after_inlining): Update.
15901 (inline_read_section): Update.
15902 (inline_write_summary): Update.
15903 * ipa-inline.c (compute_uninlined_call_time): Update.
15904 (compute_inlined_call_time): Update.
15905 (recursive_inlining): Update.
15906 (inline_small_functions): Update.
15907 (dump_overall_stats): Update.
15908 * ipa-inline.h: Include sreal.h.
15909 (size_time_entry): Turn time to sreal.
15910 (inline_summary): Turn self_time nad time to sreal.
15912 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15914 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
15916 (sreal::stream_out, sreal::stream_in): New.
15917 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
15919 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15921 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
15924 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
15927 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
15929 (dimode_scalar_chain::compute_convert_gain): Ditto.
15930 (dimode_scalar_chain::make_vector_copies): Ditto.
15931 (dimode_scalar_chain::convert_reg): Ditto.
15932 (dimode_scalar_chain::convert_insn): Ditto.
15933 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
15934 (VI248_AVX512BW_1): New mode iterator.
15935 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
15936 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
15939 2017-04-25 Martin Sebor <msebor@redhat.com>
15941 PR tree-optimization/80497
15942 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
15943 constants are representable in HOST_WIDE_INT.
15944 (parse_directive): Ditto.
15946 2017-04-25 Martin Sebor <msebor@redhat.com>
15949 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
15950 (new_zero_array): Adjust signature.
15951 (dom_info::dom_init): Used unsigned rather that size_t.
15952 (dom_info::dom_info): Same.
15954 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15955 Jakub Jelinek <jakub@redhat.com>
15958 * config/arm/arm.c: Include gimple.h.
15959 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
15960 returns negative, increment ncrn only if it returned positive.
15961 (arm_needs_doubleword_align): Return int instead of bool,
15962 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
15963 members, but if there is any such non-FIELD_DECL
15964 > PARM_BOUNDARY aligned decl, return -1 instead of false.
15965 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
15966 returns negative, increment nregs only if it returned positive.
15967 (arm_setup_incoming_varargs): Likewise.
15968 (arm_function_arg_boundary): Emit -Wpsabi note if
15969 arm_needs_doubleword_align returns negative, return
15970 DOUBLEWORD_ALIGNMENT only if it returned positive.
15972 2017-04-25 Marek Polacek <polacek@redhat.com>
15975 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
15976 first argument to type.
15978 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
15981 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
15982 type checks to test for compatibility instead of equality.
15984 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15985 Jakub Jelinek <jakub@redhat.com>
15988 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
15990 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
15991 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
15992 the alignment computation, but return their maximum in warn_alignment.
15993 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
15994 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
15996 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
15997 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
16000 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16002 * config/arc/simdext.md (dmpyh): Fix typo.
16004 2017-04-25 Richard Biener <rguenther@suse.de>
16006 PR tree-optimization/80492
16007 * alias.c (compare_base_decls): Handle registers with asm
16008 specification conservatively.
16009 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16010 compare_base_decls returning dont-know properly.
16012 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16014 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
16015 (legitimate_offset_address_p): New function.
16016 (arc_legitimate_address_p): Use above function.
16018 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16020 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
16022 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16024 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
16025 ACCH registers whenever they are available.
16027 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16029 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
16030 double regs fix when not used.
16032 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16034 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
16036 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
16037 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
16039 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16041 * config/arc/arc.c (arc_output_addsi): Check for h-register class
16042 when emitting short ADD instructions.
16044 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16046 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
16048 (cmpsi_cc_c_insn): Likewise.
16049 (cbranchsi4_scratch): Compute proper instruction length using
16050 compact_hreg_operand.
16051 * config/arc/predicates.md (compact_hreg_operand): New predicate.
16053 2017-04-25 Richard Biener <rguenther@suse.de>
16055 PR middle-end/80509
16056 * passes.c (pass_manager::pass_manager): Initialize
16057 m_name_to_pass_map.
16059 2017-04-25 Richard Biener <rguenther@suse.de>
16061 PR tree-optimization/79201
16062 * tree-ssa-sink.c (statement_sink_location): Handle calls.
16064 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16067 * config/s390/vector.md: Split MEM->GPR vector moves for
16068 non-s_operand addresses.
16070 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16073 * config/s390/predicates.md (reload_const_wide_int_operand): New
16075 * config/s390/s390.md ("movti"): Remove d/P alternative.
16076 ("movti_bigconst"): New pattern definition.
16078 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16081 * s390-protos.h (s390_expand_cs_hqi): Removed.
16082 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
16083 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
16084 modes as well as CCZ1mode and CCZmode.
16085 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
16086 signature of s390_emit_compare_and_swap.
16087 (s390_expand_cs_hqi): Likewise, make static.
16088 (s390_expand_cs_tdsi): Generate an explicit compare before trying
16089 compare-and-swap, in some cases.
16090 (s390_expand_cs): Wrapper function.
16091 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
16093 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
16094 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
16095 patterns for small and large integers. Forbid symref memory operands.
16096 Move expander to s390.c. Require cc register.
16097 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
16098 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
16099 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
16100 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
16101 symref memory operands. Remove CC mode and call s390_match_ccmode
16103 ("atomic_exchange<mode>"): Allow and implement all integer modes.
16105 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16107 * config/s390/s390.md (define_peephole2): New peephole to help
16108 combining the load-and-test pattern with volatile memory.
16110 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16112 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
16113 with CCZmode for TARGET_Z196.
16115 2017-04-25 Jakub Jelinek <jakub@redhat.com>
16117 PR rtl-optimization/80501
16118 * combine.c (make_compound_operation_int): Set subreg_code to SET
16119 even for AND with mask of the sign bit of mode.
16121 PR rtl-optimization/80500
16122 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
16123 sum's initial value.
16125 2017-04-25 Julian Brown <julian@codesourcery.com>
16126 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16128 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
16130 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
16132 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
16134 2017-04-25 Julian Brown <julian@codesourcery.com>
16135 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16137 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
16138 (thunderx2t99_sha): New Reservation.
16140 2017-04-25 Julian Brown <julian@codesourcery.com>
16141 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16143 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
16144 type for 1-element load.
16146 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
16148 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
16150 2017-04-24 Martin Jambor <mjambor@suse.cz>
16152 PR tree-optimization/80293
16153 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
16154 char arrays not totally scalarizable if it is false.
16155 (analyze_all_variable_accesses): Pass correct value in the new
16156 parameter. Add a statistics counter.
16158 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
16160 PR middle-end/79931
16161 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
16163 2017-04-24 Richard Biener <rguenther@suse.de>
16165 PR tree-optimization/80494
16166 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
16167 out for complex types.
16169 2017-04-24 Richard Biener <rguenther@suse.de>
16171 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
16172 * tree-ssa-sccvn.c (print_scc): Print SCC size.
16173 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
16174 (DFS): Adjust and never fail.
16175 (sccvn_dom_walker::fail): Remove.
16176 (sccvn_dom_walker::before_dom_children): Adjust.
16177 (run_scc_vn): Likewise and never fail.
16178 * tree-ssa-pre.c (pass_pre::execute): Adjust.
16179 (pass_fre::execute): Likewise.
16181 2017-04-24 Richard Biener <rguenther@suse.de>
16183 PR tree-optimization/79725
16184 * tree-ssa-sink.c (statement_sink_location): Return whether
16185 failure reason was zero uses. Move that check later.
16186 (sink_code_in_bb): Deal with zero uses by removing the stmt
16189 2017-04-24 Richard Biener <rguenther@suse.de>
16192 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
16193 pointer-based references.
16195 2017-04-24 Richard Biener <rguenther@suse.de>
16198 * pass_manager.h (pass_manager::operator new): Remove.
16199 (pass_manager::operator delete): Likewise.
16200 * passes.c (pass_manager::operator new): Remove.
16201 (pass_manager::operator delete): Likewise.
16202 (pass_manager::pass_manager): Zero individual pass members.
16204 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
16207 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
16208 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
16209 Check "XEXP (src, 1)" operand here.
16210 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
16211 Check "XEXP (src, 1)" operand here.
16212 (dimode_scalar_chain::make_vector_copies): Detect count register
16213 of a shift instruction. Zero extend count register from QImode
16214 to DImode to satisfy vector shift pattern count operand predicate.
16215 Substitute vector shift count operand with a DImode copy.
16216 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
16219 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
16221 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
16222 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
16223 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
16224 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
16225 (UNSPEC_NOREX_MEM): Remove definition.
16227 2017-04-21 Richard Biener <rguenther@suse.de>
16229 PR tree-optimization/79547
16230 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
16231 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
16232 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
16233 without any constraints.
16235 2017-04-21 Richard Biener <rguenther@suse.de>
16237 PR tree-optimization/78847
16238 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
16240 2017-04-21 Richard Biener <rguenther@suse.de>
16242 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
16243 (build_distinct_type_copy): Likewise.
16244 (build_variant_type_copy): Likewise.
16245 * tree.c (build_qualified_type): Pass down mem-stat info.
16246 (build_distinct_type_copy): Likewise.
16247 (build_variant_type_copy): Likewise.
16249 2017-04-21 Richard Biener <rguenther@suse.de>
16251 PR tree-optimization/80237
16252 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
16254 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
16255 for a simplified result.
16257 2016-04-21 Richard Biener <rguenther@suse.de>
16259 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
16260 sth as strict as a simple_iv but a chrec without symbols and an
16261 operand defined in the loop we are peeling (and not some subloop).
16262 (propagate_constants_for_unrolling): Propagate all constants.
16264 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
16267 * config/i386/i386.c (print_reg): Remove assert for disalowed
16268 regno values, call output_operand_lossage instead.
16270 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
16273 * config/i386/constraints.md (Yc): New register constraint.
16274 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
16275 Use Yc constraint for alternative 2 of operand 0. Remove
16276 preferred_for_speed attribute.
16278 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
16280 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
16281 lastprivate clauses in SIMT case.
16283 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
16285 * doc/invoke.texi (-Wextra-semi): Document new warning option.
16287 2017-04-20 Richard Biener <rguenther@suse.de>
16289 PR tree-optimization/57796
16290 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
16291 as N scalar stores.
16292 (vect_model_load_cost): Cost gathers as N scalar loads.
16294 2017-04-20 Richard Biener <rguenther@suse.de>
16296 * ggc-page.c (ggc_allocated_p): Rename to ...
16297 (safe_lookup_page_table_entry): ... this and return the lookup
16299 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
16301 2017-04-20 Richard Biener <rguenther@suse.de>
16303 PR tree-optimization/80453
16304 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
16305 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
16306 from the conditions.
16307 (vn_phi_eq): Pass them down.
16308 (vn_phi_lookup): Record them.
16309 (vn_phi_insert): Likewise.
16311 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
16313 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
16314 uninitialized variable warning to avoid buffer overrun.
16316 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
16319 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
16320 is suppressed for '{ 0 }' in C.
16322 2017-04-20 Jakub Jelinek <jakub@redhat.com>
16324 * BASE-VER: Set to 8.0.0.
16326 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
16328 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
16329 priority .init_array and .fini_array section with SECTION_NOTYPE
16332 2017-04-20 Jakub Jelinek <jakub@redhat.com>
16334 PR middle-end/80423
16335 * tree.h (build_array_type): Add typeless_storage default argument.
16336 * tree.c (type_cache_hasher::equal): Also compare
16337 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
16338 (build_array_type): Add typeless_storage argument, set
16339 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
16341 (build_nonshared_array_type): Adjust build_array_type_1 caller.
16342 (build_array_type): Likewise. Add typeless_storage argument.
16344 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
16345 Jakub Jelinek <jakub@redhat.com>
16347 PR tree-optimization/80426
16348 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
16349 operation on symbolic operands, also compute the overflow for the
16350 invariant part when the operation degenerates into a negation.
16352 2017-04-19 Jakub Jelinek <jakub@redhat.com>
16355 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
16356 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
16359 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
16361 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
16364 * config/avr/avr.c (tree.h): Include it.
16365 (cgraph.h): Include it.
16366 (avr_encode_section_info): Don't warn for uninitialized progmem
16367 variable if it's just an alias.
16369 2017-04-19 Richard Biener <rguenther@suse.de>
16372 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
16373 when needed by AutoPGO.
16375 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
16378 * doc/lto.texi: Remove an extra 'that'.
16380 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
16382 PR rtl-optimization/80429
16383 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
16384 are only used in debug insns.
16386 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
16387 Vladimir Makarov <vmakarov@redhat.com>
16389 * config/sparc/predicates.md (input_operand): Add comment. Return
16390 true for any memory operand when LRA is in progress.
16391 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
16393 2017-04-18 Jeff Law <law@redhat.com>
16396 * mips.md ({return,simple_return}_internal): Do not overwrite
16399 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16401 PR tree-optimization/80443
16402 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
16403 instead of adding 1, subtract -1 and similarly instead of subtracting
16406 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
16408 PR rtl-optimization/80357
16409 * haifa-sched.c (tmp_bitmap): New variable.
16410 (model_recompute): Handle duplicate use records.
16411 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
16412 (free_global_sched_pressure_data): Free it.
16414 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
16417 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
16418 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
16419 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
16420 instead of SYSTEM_HEADER_DIR.
16422 2017-04-18 Jeff Law <law@redhat.com>
16424 PR middle-end/80422
16425 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
16426 predecessors after walking up the insn chain.
16428 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16431 * dwarf2out.c (modified_type_die): Try harder not to emit internal
16432 sizetype type into debug info.
16434 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
16437 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
16438 unneeded test for TARGET_UPPER_REGS_SF.
16439 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
16441 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16444 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
16445 instead of gsi_after_labels.
16447 2017-04-18 Jeff Law <law@redhat.com>
16449 * regcprop.c (maybe_mode_change): Avoid creating copies of the
16453 2017-04-13 Jeff Law <law@redhat.com>
16454 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
16455 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
16457 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
16460 * config/avr/avr.c (intl.h): Include it.
16461 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
16463 2017-04-18 Martin Liska <mliska@suse.cz>
16465 PR gcov-profile/78783
16466 * gcov-tool.c (gcov_output_files): Validate that destination
16467 file is either removed by the tool or by a user.
16469 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
16470 Guy Benyei <guybe@mellanox.com>
16472 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
16473 block, and do not negate it, the stored id is already negative.
16475 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
16477 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
16479 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
16482 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
16483 masks of options that should be turned off if the VSX vector
16484 options are turned off.
16485 (OTHER_P8_VECTOR_MASKS): Likewise.
16486 (OTHER_VSX_VECTOR_MASKS): Likewise.
16487 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
16488 rs6000_disable_incompatible_switches to validate no type switches
16490 (rs6000_incompatible_switch): New function to disallow turning on
16491 other vector options if -mno-vsx, -mno-power8-vector, or
16492 -mno-power9-vector are specified.
16494 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16496 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
16498 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16500 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
16501 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
16502 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
16503 (ARG_POINTER_CFA_OFFSET): Likewise.
16505 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16507 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
16508 conditions to take advantage of various optimizations.
16510 2017-04-13 Jeff Law <law@redhat.com>
16512 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
16513 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
16514 (zero_extendsidi2_dext): Likewise.
16516 2017-04-13 Jakub Jelinek <jakub@redhat.com>
16519 * fold-const.c (fold_ternary_loc): Revert
16520 use op0 instead of fold_convert_loc (loc, type, arg0) part of
16523 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
16525 PR rtl-optimization/80343
16526 * lra-remat.c (update_scratch_ops): Assign original hard reg to
16527 new scratch pseudo.
16529 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
16532 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
16533 to ubsan_encode_value.
16535 2017-04-13 Jeff Law <law@redhat.com>
16537 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
16538 appearing in DEBUG_INSNs.
16540 2017-04-13 Martin Liska <mliska@suse.cz>
16542 PR gcov-profile/80413
16543 * gcov-io.c (gcov_write_string): Copy to buffer just when
16544 allocated size is greater than zero.
16546 2017-04-13 Jakub Jelinek <jakub@redhat.com>
16549 * dwarf2out.c (decls_for_scope): Ignore declarations of
16550 current_function_decl in BLOCK_NONLOCALIZED_VARS.
16552 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
16555 * ipa-visibility.c (non_local_p): Fix typos.
16556 (localize_node): When localizing symbol in same comdat group,
16557 dissolve the group only when we know external symbols are going
16559 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
16561 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16563 PR tree-optimization/79390
16564 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
16565 order does not result in usable sequence, retry with reversed operand
16571 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
16572 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
16573 op0 instead of fold_convert_loc (loc, type, arg0).
16575 2017-04-12 Jeff Law <law@redhat.com>
16577 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
16578 has a delay slot in the generated code.
16580 * config/cris/cris.md (cris_preferred_reload_class): Return
16581 GENNONACR_REGS rather than GENERAL_REGS.
16583 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16586 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
16587 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
16588 signedness of the result type.
16590 2017-04-12 Richard Biener <rguenther@suse.de>
16591 Jeff Law <law@redhat.com>
16593 PR tree-optimization/80359
16594 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
16595 trim stores to TARGET_MEM_REFs.
16597 2017-04-12 Richard Biener <rguenther@suse.de>
16599 PR tree-optimization/79390
16600 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
16601 threading case even more.
16603 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
16606 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
16607 for quad_address_p for TImode, instead of just not indexed_address.
16609 2017-04-12 Richard Biener <rguenther@suse.de>
16610 Bernd Edlinger <bernd.edlinger@hotmail.de>
16612 PR middle-end/79671
16613 * alias.c (component_uses_parent_alias_set_from): Handle
16614 TYPE_TYPELESS_STORAGE.
16615 (get_alias_set): Likewise.
16616 * tree-core.h (tree_type_common): Add typeless_storage flag.
16617 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
16618 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
16619 for types containing members with TYPE_TYPELESS_STORAGE.
16620 (place_field): Likewise.
16621 (layout_type): Likewise for ARRAY_TYPE.
16622 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
16623 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
16624 TYPE_TYPELESS_STORAGE.
16625 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
16627 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16630 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
16631 first argument to type.
16633 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16637 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
16638 CONST0_RTX (mode) rather than const0_rtx where appropriate.
16639 (rs6000_expand_binop_builtin): Likewise.
16640 (rs6000_expand_ternop_builtin): Likewise; also add missing
16641 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
16642 vshasigma built-ins.
16643 * doc/extend.texi: Document that vec_xxpermdi's third argument
16644 must be a constant.
16646 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
16648 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
16649 Use shift_const cost parameter when calculating gain of STV shifts.
16651 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
16653 PR rtl-optimization/70478
16654 * lra-constraints.c (process_alt_operands): Check memory for
16655 disfavoring memory insn operand.
16657 2017-04-11 Jakub Jelinek <jakub@redhat.com>
16659 PR middle-end/80100
16660 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
16661 left shift in unsigned HOST_WIDE_INT type.
16663 PR rtl-optimization/80385
16664 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
16665 (not (neg X)) into (plus X -1) for complex or non-integral modes.
16668 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
16669 if they have any depend clauses.
16671 2017-04-11 Martin Liska <mliska@suse.cz>
16674 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
16675 * ipa-split.c (split_function): Create a local comdat symbol
16676 if caller is in a comdat group.
16678 2017-04-11 Martin Liska <mliska@suse.cz>
16681 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
16684 2017-04-11 Martin Sebor <msebor@redhat.com>
16686 PR middle-end/80364
16687 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
16688 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
16690 (directive::set_width, directive::set_precision, format_character):
16692 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
16695 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
16698 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
16699 conflict, set target->arch_name instead of target->cpu_name.
16701 2017-04-11 Richard Biener <rguenther@suse.de>
16703 PR tree-optimization/80374
16704 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
16705 build_zero_cst, remove fold_convertible_p check again.
16707 2017-04-11 Martin Liska <mliska@suse.cz>
16710 * ubsan.c (instrument_object_size): Do not instrument register
16713 2017-04-11 Jakub Jelinek <jakub@redhat.com>
16716 * config/i386/i386-builtin-types.def
16717 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
16718 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
16719 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
16720 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
16721 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
16722 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
16723 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
16724 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
16725 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
16726 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
16727 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
16728 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
16729 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
16730 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
16731 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
16732 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
16733 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
16734 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
16735 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
16736 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
16737 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
16738 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
16739 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
16740 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
16741 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
16742 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
16743 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
16744 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
16745 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
16746 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
16747 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
16748 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
16749 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
16750 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
16751 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
16752 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
16753 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
16754 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
16755 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
16756 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
16757 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
16758 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
16759 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
16760 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
16761 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
16762 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
16764 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
16765 flag to second_arg_count, handle 4 argument function type _COUNT
16766 aliases, handle second_arg_count on second argument rather than last.
16768 2017-04-10 Jeff Law <law@redhat.com>
16770 PR tree-optimization/80374
16771 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
16772 record anything if we can not convert integer_zero_node to the
16775 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
16778 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16779 Enhance special handling given to the TARGET_P9_MINMAX option in
16780 relation to certain other options.
16782 2017-04-10 Bin Cheng <bin.cheng@arm.com>
16784 PR tree-optimization/80153
16785 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
16786 remove POINTER_PLUS_EXPR's base part directly, rather than through
16789 2017-04-10 Richard Biener <rguenther@suse.de>
16790 Bin Cheng <bin.cheng@arm.com>
16792 PR tree-optimization/80153
16793 * tree-affine.c (aff_combination_to_tree): Get base pointer from
16794 the first element of pointer type aff_tree. Build result expr in
16796 (add_elt_to_tree): Convert to type unconditionally. Remove other
16797 fold_convert calls.
16798 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
16799 (rewrite_use_nonlinear_expr): Check invariant using iv information.
16801 2017-04-10 Richard Biener <rguenther@suse.de>
16803 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16806 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
16808 PR rtl-optimization/70478
16809 * lra-constraints.c (curr_small_class_check): New.
16810 (update_and_check_small_class_inputs): New.
16811 (process_alt_operands): Update curr_small_class_check. Disfavor
16812 alternative insn memory operands. Check available regs for small
16815 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
16818 * config/mips/mips.opt (-mvirt): Update description.
16819 * doc/invoke.texi (-mvirt): Likewise.
16821 2017-04-10 Richard Biener <rguenther@suse.de>
16823 PR middle-end/80362
16824 * fold-const.c (fold_binary_loc): Look at unstripped ops when
16825 looking for NEGATE_EXPR in -A / -B to A / B folding.
16827 2017-04-10 Martin Liska <mliska@suse.cz>
16829 PR gcov-profile/80224
16830 * gcov.c (print_usage): Fix usage string.
16831 (get_gcov_intermediate_filename): Remove.
16832 (output_gcov_file): Use both for normal and intermediate format.
16833 (generate_results): Do not initialize special file for
16834 intermediate format.
16836 2017-04-10 Richard Biener <rguenther@suse.de>
16838 PR tree-optimization/80304
16839 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
16842 2017-04-10 Nathan Sidwell <nathan@acm.org>
16845 * config/rs6000/rs6000.c (rs6000_vector_type): New.
16846 (rs6000_init_builtins): Use it.
16848 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16850 * config/arm/arm.md (<mrc>): Add mode to SET source.
16851 (<mrrc>): Likewise.
16853 2017-04-10 Richard Biener <rguenther@suse.de>
16855 PR middle-end/80344
16856 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
16858 2017-04-10 Jakub Jelinek <jakub@redhat.com>
16861 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
16862 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
16863 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
16864 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
16865 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
16866 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
16867 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
16868 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
16869 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
16870 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
16871 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
16872 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
16873 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
16874 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
16875 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
16876 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
16877 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
16878 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
16879 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
16880 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
16881 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
16882 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
16883 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
16885 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
16887 PR rtl-optimization/70478
16888 * lra-constraints.c: Reverse the last patch.
16890 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
16892 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
16893 Add comment for WCHAR_T.
16895 2017-04-08 Martin Liska <mliska@suse.cz>
16898 2017-04-07 Martin Liska <mliska@suse.cz>
16901 * ipa-split.c (split_function): Add function part to a same comdat
16904 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16907 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
16909 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
16911 * rs6000/rs6000.c (vec_load_pendulum): Rename...
16912 (vec_pairing): ...to this.
16913 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
16914 (rs6000_sched_init): Adjust for name change.
16915 (struct rs6000_sched_context): Likewise.
16916 (rs6000_init_sched_context): Likewise.
16917 (rs6000_set_sched_context): Likewise.
16919 2017-04-07 Jakub Jelinek <jakub@redhat.com>
16925 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
16927 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
16928 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
16929 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
16931 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
16933 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
16935 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
16937 PR rtl-optimization/70703
16938 * ira-color.c (update_conflict_hard_regno_costs): Use
16939 int64_t instead of HOST_WIDE_INT.
16941 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
16943 PR rtl-optimization/70478
16944 * lra-constraints.c (process_alt_operands): Disfavor alternative
16945 insn memory operands.
16947 2017-04-07 Jeff Law <law@redhat.com>
16949 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
16950 CALL and NOTE_INSN_CALL_ARG_LOCATION.
16952 2017-04-07 Martin Liska <mliska@suse.cz>
16955 * config/aarch64/aarch64.c (aarch64_process_target_attr):
16956 Show error message instead of an ICE.
16958 2017-04-07 Martin Liska <mliska@suse.cz>
16961 * ipa-split.c (split_function): Add function part to a same comdat
16964 2017-04-07 Richard Biener <rguenther@suse.de>
16966 PR middle-end/80341
16967 * tree.c (get_unwidened): Also handle ! for_type case for
16969 * convert.c (do_narrow): Split out from ...
16970 (convert_to_integer_1): ... here. Do not pass final truncation
16971 type to get_unwidened for TRUNC_DIV_EXPR.
16973 2017-04-07 Richard Biener <rguenther@suse.de>
16975 * tree-affine.c (wide_int_ext_for_comb): Take type rather
16977 (aff_combination_const): Adjust.
16978 (aff_combination_scale): Likewise.
16979 (aff_combination_add_elt): Likewise.
16980 (aff_combination_add_cst): Likewise.
16981 (aff_combination_convert): Likewise.
16982 (add_elt_to_tree): Likewise. Remove unused argument.
16983 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
16985 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
16987 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
16989 * config/arm/arm.c (arm_default_short_enums): Use
16990 ARM_DEFAULT_SHORT_ENUMS.
16991 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
16993 2017-04-06 Jakub Jelinek <jakub@redhat.com>
16996 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
16997 members with redundant out-of-class redeclaration.
16999 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
17002 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
17003 * config/i386/i386.md (*zero_extendsidi2):
17004 Add (?*x,*x) and (?*v,*v) alternatives.
17006 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
17009 * config/i386/i386.c (ix86_expand_builtin)
17010 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
17011 mode from insn data. Convert operands to insn operand mode.
17012 Copy operands that don't satisfy insn predicate to a register.
17014 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
17016 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
17019 2017-04-06 Richard Biener <rguenther@suse.de>
17021 PR tree-optimization/80334
17022 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
17023 preserve alignment of accesses.
17025 2017-04-06 Richard Biener <rguenther@suse.de>
17027 PR tree-optimization/80262
17028 * tree-sra.c (build_ref_for_offset): Preserve address-space
17030 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
17031 Drop useless address-space information on MEM_REF offsets.
17033 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
17035 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
17037 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
17039 PR rtl-optimization/70703
17040 * ira-color.c (update_conflict_hard_regno_costs): Use
17041 HOST_WIDE_INT instead of long.
17043 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
17046 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
17047 not defined for x86_64 target. Add -mmmx target option when __SSE2__
17049 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
17050 for x86_64 target. Handle -m3dnowa option.
17052 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
17054 PR rtl-optimization/70703
17055 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
17056 (update_conflict_hard_regno_costs): Use long instead of unsigned
17057 arithmetic for cost calculation.
17059 2017-04-05 Jakub Jelinek <jakub@redhat.com>
17060 Bernd Edlinger <bernd.edlinger@hotmail.de>
17063 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
17066 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
17069 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
17070 ptr_mode with Pmode throughout.
17071 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
17072 into probe_stack_range and use DImode.
17074 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
17077 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
17078 call_eh_return is true.
17080 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17082 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
17083 Initialize last_match_fntype_index.
17085 2017-04-05 Jakub Jelinek <jakub@redhat.com>
17088 * tree-nvr.c: Include internal-fn.h.
17089 (pass_return_slot::execute): Ignore internal calls without
17092 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17093 Richard Biener <rguenther@suse.de>
17096 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
17097 captures used multiple times, except for the last use.
17098 * generic-match-head.c: Include gimplify.h.
17100 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17102 PR tree-optimization/79390
17103 * target.h (struct noce_if_info): Declare.
17104 * targhooks.h (default_noce_conversion_profitable_p): Declare.
17105 * target.def (noce_conversion_profitable_p): New target hook.
17106 * ifcvt.h (struct noce_if_info): New type, moved from ...
17107 * ifcvt.c (struct noce_if_info): ... here.
17108 (noce_conversion_profitable_p): Renamed to ...
17109 (default_noce_conversion_profitable_p): ... this. No longer
17111 (noce_try_store_flag_constants, noce_try_addcc,
17112 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
17113 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
17114 instead of noce_conversion_profitable_p.
17115 * config/i386/i386.c: Include ifcvt.h.
17116 (ix86_option_override_internal): Don't override
17117 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
17118 (ix86_noce_conversion_profitable_p): New function.
17119 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
17120 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
17121 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
17122 * doc/tm.texi: Regenerated.
17124 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17126 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
17129 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17132 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
17133 instructions for small multiply cores.
17135 2017-04-04 Jeff Law <law@redhat.com>
17137 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
17139 (mips_expand_vec_perm_const): Initialize elements in orig_perm
17140 that are not set by the loop over the elements.
17142 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17145 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
17146 int mode, convert_modes it to mode as unsigned, otherwise use
17147 lowpart_subreg to mode rather than SImode.
17148 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
17149 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
17150 Use DImode instead of SImode for the shift count operand.
17151 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
17154 2017-04-04 Richard Biener <rguenther@suse.de>
17156 PR middle-end/80281
17157 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
17158 arithmetic done for the negate or the plus. Simplify.
17159 (A - (-B) -> A + B): Likewise.
17160 * fold-const.c (split_tree): Make sure to not negate pointers.
17162 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
17164 PR rtl-optimization/60818
17165 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
17166 a compare of comparisons with the thing compared if this results
17167 in a different machine mode.
17169 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
17171 * alias.c (base_alias_check): Fix typo in comment.
17172 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
17173 * cgraphunit.c (symbol_table::compile): Likewise.
17174 * collect2.c (maybe_run_lto_and_relink): Likewise.
17175 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
17176 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
17177 * config/avr/avr.c (avr_map_op_t): Likewise.
17178 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
17179 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
17180 * config/epiphany/epiphany.md (movcc): Likewise.
17181 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
17182 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
17184 * config/mips/mips.c (mips_save_restore_reg): Likewise.
17185 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
17186 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
17187 * config/sh/sh.c (sh_rtx_costs): Likewise.
17188 * fold-const.c (fold_truth_andor): Likewise.
17189 * genautomata.c (collapse_flag): Likewise.
17190 * gengtype.h (struct type::u::s): Likewise.
17191 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
17192 * input.c (FORMAT_AMOUNT): Likewise.
17193 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
17194 (known_aggs_to_agg_replacement_list): Likewise.
17195 * ipa-inline-analysis.c: Likewise.
17196 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
17197 * ipa-polymorphic-call.c
17198 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
17199 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
17200 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
17202 * modulo-sched.c (apply_reg_moves): Likewise.
17203 * omp-expand.c (build_omp_regions_1): Likewise.
17204 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
17205 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
17206 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
17207 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
17208 * value-prof.c: Likewise.
17209 * var-tracking.c (val_reset): Likewise.
17211 2017-04-03 Richard Biener <rguenther@suse.de>
17213 PR tree-optimization/80275
17214 * fold-const.c (split_address_to_core_and_offset): Handle
17217 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
17219 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
17220 descriptors is at least equal to that of functions.
17222 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
17224 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
17226 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
17229 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
17230 (mov<IMOD4:mode>): New expander.
17231 (*mov<IMOD4:mode>_internal): New insn and split pattern.
17233 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
17235 PR rtl-optimization/79405
17236 * fwprop.c (propagations_left): New variable.
17237 (forward_propagate_into): Decrement it.
17238 (fwprop_init): Initialize it.
17239 (fw_prop): If the variable has reached zero, stop propagating.
17240 (fwprop_addr): Ditto.
17242 2017-03-31 Jakub Jelinek <jakub@redhat.com>
17245 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
17246 a FUNCTION_DECL, pass it as decl instead of origin to
17249 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
17251 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
17254 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
17257 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
17258 TARGET_VSX_SMALL_INTEGER.
17260 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17262 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17263 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
17265 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
17267 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
17268 extraction from odd-numbered MSA register.
17270 2017-03-31 Jakub Jelinek <jakub@redhat.com>
17272 PR middle-end/80173
17273 * expmed.c (store_bit_field_1): Don't attempt to create
17274 a word subreg out of hard registers wider than word if they
17275 have HARD_REGNO_NREGS of 1 for their mode.
17277 PR middle-end/80163
17278 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
17279 conversions to integer types wider than word and pointer.
17282 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
17283 (rtx_equal_for_cselib_p): Pass 0 to it.
17284 * cselib.c (cselib_hasher::equal): Likewise.
17285 (rtx_equal_for_cselib_1): Add depth argument. If depth
17286 is 128, don't look up VALUE locs and punt. Increment
17287 depth in recursive calls when walking VALUE locs.
17289 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
17291 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
17292 (make_gcov_file_name): Use the canonical path name for generating
17294 (read_line): Fix handling of files with ascii null bytes.
17296 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
17298 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
17299 to initialise a vector register instead
17300 of using a const_int.
17302 2017-03-30 Jakub Jelinek <jakub@redhat.com>
17304 PR translation/80189
17305 * gimplify.c (omp_default_clause): Use %qs instead of %s in
17306 diagnostic messages.
17308 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
17311 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
17312 (dfp_diex_<mode>): Update mode of operand 1.
17313 * doc/extend.texi (dxex, dxexq): Document change to return type.
17314 (diex, diexq): Document change to argument type.
17316 2017-03-30 Martin Jambor <mjambor@suse.cz>
17319 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
17320 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
17321 it reflects the signature changes performed at the callee side.
17322 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
17323 to cgraph_build_function_type_skip_args.
17324 (build_function_decl_skip_args): Adjust call to the above function.
17326 2017-03-30 Jakub Jelinek <jakub@redhat.com>
17329 * config/i386/sse.md
17330 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
17331 register as dest whenever it is a MEM not rtx_equal_p to the
17332 corresponding dup operand, and when forcing into reg move the
17333 reg into the memory afterwards.
17334 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
17335 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
17336 for the force_reg mode.
17337 (avx512vl_vextractf128<mode>): Use register as dest either
17338 always when a MEM, or when it is a MEM not rtx_equal_p to the
17339 corresponding dup operand, or even not when it is a CONST_VECTOR
17340 depending on the mode and lo vs. hi.
17341 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
17343 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
17344 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
17345 Likewise. Require that operands[2] is even.
17346 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
17347 Remove extraneous parens. Require that operands[2] is a multiple
17349 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
17350 operands[0] is a MEM if <mask_applied>, the predicates/constraints
17351 disallow memory then.
17353 2017-03-30 Richard Biener <rguenther@suse.de>
17355 PR tree-optimization/77498
17356 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
17357 to non-constants over backedges.
17359 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
17361 PR rtl-optimization/80233
17362 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
17363 as last_combined_insn. Do not test for BARRIER_P separately.
17365 2017-03-29 Andreas Schwab <schwab@suse.de>
17368 * calls.c (prepare_call_address): Convert funexp to Pmode before
17369 copying to temp reg.
17371 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17373 PR tree-optimization/80158
17374 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
17375 Handle possible future case of more than one alternate
17377 (replace_rhs_if_not_dup): Likewise.
17378 (replace_one_candidate): Likewise.
17380 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
17382 PR rtl-optimization/80193
17383 * ira.c (ira): Do not check allocation for LRA.
17385 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
17387 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
17388 (nvptx_output_simt_exit): Declare.
17389 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
17390 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
17391 (init_softstack_frame): Move initialization of crtl->is_leaf to...
17392 (nvptx_declare_function_name): ...here. Emit declaration of local
17393 memory space buffer for omp_simt_enter insn.
17394 (nvptx_output_unisimt_switch): New.
17395 (nvptx_output_softstack_switch): New.
17396 (nvptx_output_simt_enter): New.
17397 (nvptx_output_simt_exit): New.
17398 * config/nvptx/nvptx.h (struct machine_function): New fields
17399 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
17400 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
17401 (UNSPECV_SIMT_EXIT): Ditto.
17402 (omp_simt_enter_insn): New insn.
17403 (omp_simt_enter): New expansion.
17404 (omp_simt_exit): New insn.
17405 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
17407 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
17408 (expand_GOMP_SIMT_ENTER_ALLOC): New.
17409 (expand_GOMP_SIMT_EXIT): New.
17410 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
17411 (GOMP_SIMT_ENTER_ALLOC): Ditto.
17412 (GOMP_SIMT_EXIT): Ditto.
17413 * target-insns.def (omp_simt_enter): New insn.
17414 (omp_simt_exit): Ditto.
17415 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
17417 (lower_rec_simd_input_clauses): Implement SIMT privatization.
17418 (lower_rec_input_clauses): Likewise.
17419 (lower_lastprivate_clauses): Handle SIMT privatization.
17421 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
17422 (ompdevlow_adjust_simt_enter): New.
17423 (find_simtpriv_var_op): New.
17424 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
17425 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
17427 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
17428 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
17429 (copy_decl_for_dup_finish): Ditto.
17431 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
17433 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
17436 * config/i386/i386.c (ix86_option_override_internal): Always
17437 allow -mpreferred-stack-boundary=3 for 64-bit targets.
17439 2017-03-28 Bin Cheng <bin.cheng@arm.com>
17441 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
17443 2017-03-28 Bin Cheng <bin.cheng@arm.com>
17445 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
17446 mark new edge's irreducible flag accordign to it.
17447 (vect_do_peeling): Check loop preheader edge's irreducible flag
17448 and pass it to function slpeel_add_loop_guard.
17450 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
17452 PR tree-optimization/80218
17453 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
17454 Update block frequencies and counts.
17456 2017-03-28 Richard Biener <rguenther@suse.de>
17458 PR tree-optimization/78644
17459 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
17460 of a simplification result we may not use it at all.
17462 2017-03-28 Richard Biener <rguenther@suse.de>
17465 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
17466 without arguments, generate default definition of a SSA name.
17468 2017-03-28 Richard Biener <rguenther@suse.de>
17470 PR middle-end/80222
17471 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
17472 TYPE_REF_CAN_ALIAS_ALL references.
17473 * fold-const.c (fold_indirect_ref_1): Likewise.
17475 2017-03-28 Martin Liska <mliska@suse.cz>
17478 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
17479 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
17481 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
17482 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
17484 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
17485 (EXTRA_SPECS): Define.
17486 (SUBTARGET_EXTRA_SPECS): Likewise.
17487 (SUBTARGET_CPP_SPEC): Likewise.
17488 * config/arc/elf.h (EXTRA_SPECS): Renamed to
17489 SUBTARGET_EXTRA_SPECS.
17490 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
17492 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
17494 * config/arc/simdext.md (vst64_insn): Update pattern.
17495 (vld32wh_insn): Likewise.
17496 (vld32wl_insn): Likewise.
17497 (vld64_insn): Likewise.
17498 (vld32_insn): Likewise.
17500 2017-03-28 Marek Polacek <polacek@redhat.com>
17503 * fold-const.c (fold_comparison): Use protected_set_expr_location
17504 instead of SET_EXPR_LOCATION.
17506 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
17508 * tree.c (add_expr): Avoid name lookup warning.
17510 2017-03-27 Jeff Law <law@redhat.com>
17512 PR tree-optimization/80216
17513 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
17514 function name. Limit recursion depth.
17515 (record_temporary_equivalences): Corresponding changes.
17517 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
17519 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
17522 2017-03-27 Jakub Jelinek <jakub@redhat.com>
17525 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
17527 * cfgcleanup.c (reg_note_cfa_p): New array.
17528 (insns_have_identical_cfa_notes): New function.
17529 (old_insns_match_p): Don't cross-jump in between /f
17530 and non-/f instructions. If both i1 and i2 are frame related,
17531 verify all CFA notes, their order and content.
17533 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
17536 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
17537 HImode and SImode with zero extend to DImode to one insn.
17538 (bswap<mode>2_extenddi): Likewise.
17539 (bswapsi2_extenddi): Likewise.
17540 (bswaphi2_extendsi): Likewise.
17541 (bswaphi2): Combine bswap HImode and SImode into one insn.
17542 Separate memory insns from swapping register.
17543 (bswapsi2): Likewise.
17544 (bswap<mode>2): Likewise.
17545 (bswaphi2_internal): Delete, no longer used.
17546 (bswapsi2_internal): Likewise.
17547 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
17548 store, and gpr<-gpr swap insns.
17549 (bswap<mode>2_store): Likewise.
17550 (bswaphi2_reg): Register only splitter, combine with the splitter.
17551 (bswaphi2 splitter): Likewise.
17552 (bswapsi2_reg): Likewise.
17553 (bswapsi2 splitter): Likewise.
17554 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
17555 the insns into load, store, and register/register insns.
17556 (bswapdi2_ldbrx): Likewise.
17557 (bswapdi2_load): Likewise.
17558 (bswapdi2_store): Likewise.
17559 (bswapdi2_reg): Likewise.
17561 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
17563 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
17564 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
17566 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
17569 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
17571 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
17572 special handling for target option conflicts between dform
17573 options (-mpower9-dform, -mpower9-dform-vector,
17574 -mpower9-dform-scalar) and -mno-direct-move.
17576 2017-03-27 Richard Biener <rguenther@suse.de>
17578 PR tree-optimization/80181
17579 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
17581 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17583 * config/arc/predicates.md (move_double_src_operand): Replace the
17584 call to move_double_src_operand with a call to address_operand.
17586 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17588 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
17589 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
17590 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
17592 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17594 * config/arc/predicates.md (long_immediate_loadstore_operand):
17595 Consider scaled addresses cases.
17597 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17599 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
17600 restored when in interrupt.
17601 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
17602 doesn't have delay slot.
17604 2017-03-27 Richard Biener <rguenther@suse.de>
17607 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
17608 inlined thunk clones.
17610 2017-03-27 Jakub Jelinek <jakub@redhat.com>
17613 * asan.c (instrument_derefs): Copy over last operand from
17614 original COMPONENT_REF to the new COMPONENT_REF with
17615 DECL_BIT_FIELD_REPRESENTATIVE.
17616 * ubsan.c (instrument_object_size): Likewise.
17618 2017-03-27 Richard Biener <rguenther@suse.de>
17620 PR tree-optimization/80170
17621 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
17622 sure DR/SCEV didnt fold in constants we do not see when looking
17623 at the reference base alignment.
17625 2017-03-27 Richard Biener <rguenther@suse.de>
17627 PR middle-end/80171
17628 * gimple-fold.c (fold_ctor_reference): Properly guard against
17629 NULL return value from canonicalize_constructor_val.
17631 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
17634 * config/i386/i386.c (ix86_expand_builtin)
17635 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
17636 flags reg setting and flags reg using instructions.
17637 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
17638 clobbering instructions to zero extend op2.
17640 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
17642 * doc/install.texi (Configuration) <--with-aix-soname>:
17643 Update link to AIX ld.
17645 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
17647 PR rtl-optimization/80160
17648 PR rtl-optimization/80159
17649 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
17650 reg_alternate_class into account.
17652 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
17655 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
17656 to consider in curr_insn_transform.
17658 2017-03-24 Jakub Jelinek <jakub@redhat.com>
17660 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
17661 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
17662 and emit_mode_inner.
17664 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17666 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
17667 argument to the overloaded builtin variants. Use the new flag to
17668 deprecate certain builtin variants.
17669 * config/s390/s390-builtin-types.def: Add new builtin types.
17670 * config/s390/s390-builtins.h: Support new flags field for
17671 overloaded builtins.
17672 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
17673 (s390_macro_to_expand): Enable vector float data type.
17674 (s390_cpu_cpp_builtins_internal): Indicate support of the new
17675 builtins by incrementing the __VEC__ version number.
17676 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
17678 (s390_resolve_overloaded_builtin): Emit error messages depending
17679 on the builtin flags.
17680 * config/s390/s390.c (s390_expand_builtin): Support additional
17681 flags argument. Change error message to match the messages
17682 emitted in s390-c.c.
17683 * config/s390/s390.md: New UNSPEC_* constants.
17684 (op_type): Add new instruction types.
17685 * config/s390/vecintrin.h: Add new builtins and test data class
17687 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
17688 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
17689 (VEC_INEXACT, VEC_NOINEXACT): New constants.
17690 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
17691 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
17692 ("vec_mergel<mode>"): V_HW -> VEC_HW.
17694 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
17695 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
17696 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
17697 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
17699 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
17700 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
17701 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
17702 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
17704 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
17705 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
17706 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
17707 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
17708 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
17709 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
17710 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
17712 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
17713 ("vec_scatter_element<V_HW_4:mode>_DI")
17714 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
17715 ("vec_fpint<mode>", "vflls")
17716 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
17717 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
17718 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
17719 ("*vec_cmphe<mode>_cc"): ... these.
17721 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
17722 mode constant instead of magic value.
17724 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17726 * config/s390/s390.c (s390_expand_vec_compare): Support other
17727 vector floating point modes than just V2DF.
17728 (s390_expand_vcond): Likewise.
17729 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
17730 (s390_cannot_change_mode_class): Prevent mode changes between TF
17731 and V1TF in vector registers.
17732 * config/s390/s390.md (DF, SF): New mode attributes.
17733 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
17734 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
17735 SFmode support for VRs.
17736 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
17738 (VFT, VF_HW): New mode iterators.
17739 (vw, sdx): New mode attributes.
17740 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
17741 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
17742 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
17743 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
17744 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
17745 also the new vector floating point modes. Renaming to ...
17747 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
17748 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
17749 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
17750 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
17751 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
17752 ("vec_unordered<mode>"): ... these.
17754 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
17755 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
17756 ("*vec_extendv2df"): New insn definitions.
17758 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17760 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
17761 ("mulditi3_2", "*muldi3_sign"): New patterns.
17762 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
17763 rename the pattern definition.
17765 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17767 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
17769 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
17771 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17773 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
17774 instruction if possible.
17775 * config/s390/vector.md (vec_halfnumelts): New mode
17777 ("*vec_vllezlf<mode>"): New pattern.
17779 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17781 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
17782 ("popcountv4si2", "popcountv2di2"): Rename to ...
17783 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
17784 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
17786 ("popcount<mode>2_vxe"): New pattern.
17788 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17790 * common/config/s390/s390-common.c (processor_flags_table): Add
17792 * config.gcc: Add arch12.
17793 * config/s390/driver-native.c (s390_host_detect_local_cpu):
17794 Default to arch12 for unknown CPU model numbers.
17795 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
17796 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
17797 PROCESSOR_max sanity check.
17798 * config/s390/s390-opts.h (enum processor_type): Add
17800 * config/s390/s390.c (processor_table): Add arch12.
17801 (s390_expand_builtin): Add check for B_VXE flag.
17802 (s390_issue_rate): Add PROCESSOR_ARCH12.
17803 (s390_get_sched_attrmask): Likewise.
17804 (s390_get_unit_mask): Likewise.
17805 (s390_sched_score): Enable z13 scheduling for arch12.
17806 (s390_sched_reorder): Likewise.
17807 (s390_sched_variable_issue): Likewise.
17808 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
17810 (s390_tune_attr): Use z13 scheduling also for arch12.
17811 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
17812 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
17813 (TARGET_VXE_P): New macros.
17814 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
17815 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
17816 * config/s390/s390.opt: Add arch12 as processor_type.
17818 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17820 * config/s390/s390.md
17821 ("fixuns_truncdddi2", "fixuns_trunctddi2")
17822 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
17823 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
17825 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
17826 Rename expanders to ...
17828 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
17829 ("fixuns_truncdddi2_emu"): ... these.
17831 ("fixuns_trunc<mode>si2_emu"): New expander.
17833 ("*fixuns_truncdfdi2_z13"): Rename to ...
17834 ("*fixuns_truncdfdi2_vx"): ... this.
17836 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17838 * config/s390/2964.md: Remove the single element vector compare
17839 instructions which are no longer used.
17840 * config/s390/s390.c (s390_select_ccmode): Remove handling of
17842 (s390_canonicalize_comparison): Remove handling of DFmode
17844 (s390_expand_vec_compare_scalar): Remove function.
17845 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
17846 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
17848 ("*cmp<mode>_ccs"): Add wfcdb instruction.
17850 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17852 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
17854 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
17855 will anyway by matched by mov<mode>_64dfp.
17857 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17859 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
17860 vlef/vstef. Add missing operand to vleif.
17862 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17864 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
17865 pair for all vector types with 64 bit elements.
17866 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
17867 * config/s390/vector.md (V_HW_64): ... here.
17868 (V_128_NOSINGLE): New mode iterator.
17869 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
17870 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
17871 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
17872 ("*vec_load_pairv2di"): Change to ...
17873 ("*vec_load_pair<mode>"): ... this one.
17875 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17877 * config/s390/constraints.md: Add comments.
17878 (jKK): Reject element sizes > 8 bytes.
17879 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
17881 * config/s390/s390.md: Add the s_operand checks formerly in
17882 s390_split_ok_p to various splitters where they are still
17884 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
17885 for 128 bit vectors. Plus two splitters.
17887 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17889 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
17892 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17895 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
17896 error if the boundary argument is not constant.
17898 2017-03-24 Jakub Jelinek <jakub@redhat.com>
17900 PR rtl-optimization/80112
17901 * loop-doloop.c (doloop_condition_get): Don't check condition
17902 if cmp isn't SET with IF_THEN_ELSE src.
17904 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17906 PR tree-optimization/80158
17907 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
17908 replacing a candidate statement, also replace it for the
17909 candidate's alternate interpretation.
17910 (replace_rhs_if_not_dup): Likewise.
17911 (replace_one_candidate): Likewise.
17913 2017-03-24 Richard Biener <rguenther@suse.de>
17915 PR tree-optimization/80167
17916 * graphite-isl-ast-to-gimple.c
17917 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
17919 (translate_isl_ast_to_gimple::get_rename): Likewise.
17921 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
17923 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
17924 handling of certain combinations of target options, including the
17925 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
17926 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
17928 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17931 * config/arm/arm.md (*load_multiple): Add reload_completed to
17932 matching condition.
17934 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17935 Richard Biener <rguenth@suse.de>
17937 PR tree-optimization/79908
17938 PR tree-optimization/80136
17939 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
17940 been cast away, gimplify_and_add suffices.
17942 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
17944 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
17946 2017-03-23 Richard Biener <rguenther@suse.de>
17948 PR tree-optimization/80032
17949 * gimplify.c (gimple_push_cleanup): Forced unconditional
17950 cleanups still have to go to the conditional_cleanups
17953 2017-03-22 Jakub Jelinek <jakub@redhat.com>
17955 PR tree-optimization/80072
17956 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
17958 (next_operand_entry_id): Change type to unsigned int.
17959 (sort_by_operand_rank): Make sure to return the right return value
17960 even if unsigned fields are bigger than INT_MAX.
17961 (struct oecount): Change cnt and id type to unsigned int.
17962 (oecount_hasher::equal): Formatting fix.
17963 (oecount_cmp): Make sure to return the right return value
17964 even if unsigned fields are bigger than INT_MAX.
17965 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
17968 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
17969 TREE_READONLY on result if writing it more than once.
17972 * doc/invoke.texi (-fsanitize=thread): Document that with
17973 -fnon-call-exceptions atomics are not able to throw
17977 * tsan.c: Include tree-eh.h.
17978 (instrument_builtin_call): Call maybe_clean_eh_stmt or
17979 maybe_clean_or_replace_eh_stmt where needed.
17980 (instrument_memory_accesses): Add cfg_changed argument.
17981 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
17982 if it returned true.
17983 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
17985 PR rtl-optimization/63191
17986 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
17987 wrapper function, moved the whole old content into ...
17988 (ix86_delegitimize_address_1): ... this. New inline function.
17989 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
17990 true as last argument instead of ix86_delegitimize_address.
17992 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
17994 * config/aarch64/aarch64.c (generic_branch_cost): Copy
17995 cortexa57_branch_cost.
17997 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
17999 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
18001 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18004 * doc/md.texi (Constraints): Document wA constraint.
18005 * config/rs6000/constraints.md (wA): New.
18006 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
18007 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
18008 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
18009 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
18011 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
18014 * gimplify.c (is_oacc_declared): New function.
18015 (oacc_default_clause): Use it to set default flags for acc declared
18016 variables inside parallel regions.
18017 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
18018 declared variables.
18019 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
18020 declare attribute to any decl as necessary.
18022 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
18025 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
18026 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
18027 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
18028 (arm_arch_lpae): This.
18029 * config/arm/arm.c (arm_arch7ve): Rename into ...
18030 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
18031 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
18034 2017-03-22 Martin Liska <mliska@suse.cz>
18037 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
18038 error message instead of an ICE.
18040 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18042 * doc/extend.texi (6.11 Additional Floating Types): Revise.
18044 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
18046 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
18048 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18051 2017-03-21 Martin Sebor <msebor@redhat.com>
18053 * doc/extend.texi: Use "cannot" instead of "can't."
18054 * doc/hostconfig.texi: Same.
18055 * doc/install.texi: Same.
18056 * doc/invoke.texi: Same.
18057 * doc/loop.texi: Same.
18058 * doc/md.texi: Same.
18059 * doc/objc.texi: Same.
18060 * doc/rtl.texi: Same.
18061 * doc/tm.texi: Same.
18062 * doc/tm.texi.in: Same.
18063 * doc/trouble.texi: Same.
18065 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
18068 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
18069 (collect_checksum_attributes): Set it.
18070 (die_checksum_ordered): Use it.
18072 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18074 PR tree-optimization/79908
18075 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
18076 change: For a VA_ARG whose LHS has been cast away, use
18077 force_gimple_operand to construct the side effects.
18079 2017-03-21 David Malcolm <dmalcolm@redhat.com>
18081 PR translation/80001
18082 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
18083 more amenable to translation.
18084 (oacc_loop_auto_partitions): Likewise.
18086 2017-03-21 Marek Polacek <polacek@redhat.com>
18087 Martin Sebor <msebor@redhat.com>
18089 PR tree-optimization/80109
18090 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
18091 on INTEGRAL_TYPE_P.
18093 2017-03-21 Jakub Jelinek <jakub@redhat.com>
18094 Segher Boessenkool <segher@kernel.crashing.org>
18097 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
18098 check reg_used_between_p between insn and one of succ or succ2
18099 depending on if succ is artificial insn not inserted into insn
18102 2017-03-21 Martin Liska <mliska@suse.cz>
18104 PR gcov-profile/80081
18105 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
18106 * doc/gcc.texi: Include gcov-dump stuff.
18107 * doc/gcov-dump.texi: New file.
18109 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
18111 PR rtl-optimization/79150
18112 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
18113 conditional jump, if the jump is the last insn of the loop.
18115 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18116 Richard Biener <rguenth@suse.de>
18118 PR tree-optimization/79908
18119 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
18120 been cast away, use force_gimple_operand to construct the side
18123 2017-03-21 Martin Liska <mliska@suse.cz>
18125 PR libfortran/79956
18126 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
18129 2017-03-21 Brad Spengler <spender@grsecurity.net>
18132 * plugin.c (htab_hash_plugin): New function.
18133 (add_new_plugin): Use it and adjust.
18134 (parse_plugin_arg_opt): Adjust.
18135 (init_one_plugin): Likewise.
18137 2017-03-21 Richard Biener <rguenther@suse.de>
18139 PR tree-optimization/80032
18140 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
18141 if set force the cleanup to happen unconditionally.
18142 (gimplify_target_expr): Push inserted clobbers with force_uncond
18143 to avoid them being removed by control-dependent DCE.
18145 2017-03-21 Richard Biener <rguenther@suse.de>
18147 PR tree-optimization/80122
18148 * tree-inline.c (copy_bb): Do not expans va-arg packs or
18149 va_arg_pack_len when the inlined call stmt requires pack
18151 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
18153 2017-03-21 Jakub Jelinek <jakub@redhat.com>
18156 * tsan.c (instrument_builtin_call): If the memory model argument
18157 is not a constant, assume it is valid.
18160 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
18163 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
18165 PR rtl-optimization/79910
18166 * combine.c (can_combine_p): Do not allow combining an I0 or I1
18167 if its dest is used by an insn before I2 (other than the combined
18168 insns themselves, which are properly handled already).
18170 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
18173 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
18175 * combine.c (record_used_regs): New static function.
18176 (try_combine): Handle situations where there is an additional
18177 instruction between I2 and I3 which needs to have a LOG_LINK
18181 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
18183 * combine.c (try_combine): Delete redundant i1 test. Call
18184 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
18186 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18189 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
18190 alternatives 13/14.
18192 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18194 PR tree-optimization/80054
18195 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
18196 the optimization if a PHI or any of its arguments is not dominated
18197 by the candidate's basis. Use gphi* rather than gimple* as
18199 (replace_profitable_candidates): Clean up a gimple* variable that
18200 should be a gphi* variable.
18202 2017-03-20 Martin Sebor <msebor@redhat.com>
18205 * doc/extend.texi (attribute constructor): Document present limitation.
18207 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
18210 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
18211 __POWER9_VECTOR__ #ifdef control, change template definition to
18212 use Power9-specific built-in function.
18213 (vec_any_eq): Likewise.
18214 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
18215 to control outcomes from this test.
18216 (vector_ae_<mode>p): For VEC_F modes, likewise.
18218 2017-03-20 Ian Lance Taylor <iant@google.com>
18220 * config/i386/i386.c (ix86_function_regparm): Save an extra
18221 register for -fsplit-stack with DECL_STATIC_CHAIN.
18223 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
18226 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
18227 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
18229 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
18231 * config/riscv/riscv.c (riscv_print_operand): Use "fence
18233 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
18236 2017-03-20 Marek Polacek <polacek@redhat.com>
18239 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
18241 2017-03-20 Richard Biener <rguenther@suse.de>
18243 PR tree-optimization/80113
18244 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
18245 allocate extra SSA name for PHI def.
18246 (add_close_phis_to_outer_loops): Likewise.
18247 (add_close_phis_to_merge_points): Likewise.
18248 (copy_loop_close_phi_args): Likewise.
18249 (copy_cond_phi_nodes): Likewise.
18251 2017-03-20 Martin Liska <mliska@suse.cz>
18253 PR middle-end/79753
18254 * tree-chkp.c (chkp_build_returned_bound): Do not build
18255 returned bounds for a LHS that's not a BOUNDED_P type.
18257 2017-03-20 Martin Liska <mliska@suse.cz>
18261 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
18262 COMPLEX_CST and VECTOR_CST.
18264 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18267 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
18268 target operand. A new splitter adds the clobber statement in case
18269 the target operand is dead anyway.
18271 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
18273 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
18274 to age-old versions of binutils and glibc.
18276 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
18278 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
18280 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
18282 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
18284 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
18286 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
18287 requirement for binutils 2.13.
18289 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
18291 * combine.c (try_combine): Delete redundant i1 test. Call
18292 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
18294 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
18296 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
18297 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
18299 <riscv64-*-elf>: Re-arrange section
18300 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
18301 <riscv32-*-linux>: Likewise.
18302 <riscv64-*-elf>: Likewise
18303 <riscv64-*-linux>: Likewise.
18305 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
18308 * aarch64.opt(verbose-cost-dump): Fix typo.
18310 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
18313 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
18314 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
18316 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
18318 * reload.c (find_reloads): When reloading a nonoffsettable address,
18319 use RELOAD_OTHER for it and its address reloads.
18321 PR rtl-optimization/79910
18322 * combine.c (record_used_regs): New static function.
18323 (try_combine): Handle situations where there is an additional
18324 instruction between I2 and I3 which needs to have a LOG_LINK
18327 2017-03-17 Jeff Law <law@redhat.com>
18329 PR tree-optimization/71437
18330 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
18331 conditional in the hash table first.
18332 (vrp_dom_walker::before_dom_children): Extract condition from
18333 ASSERT_EXPR. Record condition, its inverion and any implied
18334 conditions as well.
18336 2017-03-17 Marek Polacek <polacek@redhat.com>
18337 Markus Trippelsdorf <markus@trippelsdorf.de>
18339 PR tree-optimization/80079
18340 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
18343 2017-03-17 Richard Biener <rguenther@suse.de>
18345 PR middle-end/80075
18346 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
18347 Properly verify the LHS before the RHS possibly claims to be
18349 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
18352 2017-03-17 Martin Jambor <mjambor@suse.cz>
18354 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
18355 (List of -O2 options): Likewise.
18356 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
18359 2017-03-17 Tom de Vries <tom@codesourcery.com>
18361 * gcov-dump.c (print_usage): Print bug_report_url.
18363 2017-03-17 Richard Biener <rguenther@suse.de>
18365 PR middle-end/80050
18366 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
18367 (parser::peek): Likewise.
18369 2017-03-17 Richard Biener <rguenther@suse.de>
18371 PR tree-optimization/80048
18372 * sese.c (free_sese_info): Properly release rename_map and
18373 copied_bb_map elements.
18375 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
18377 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
18378 Add linked-list forward and backlinks. Insert on
18379 construction, remove on destruction.
18380 (class pass_store_merging): Add m_stores_head field.
18381 (pass_store_merging::terminate_and_process_all_chains):
18382 Iterate over m_stores_head list.
18383 (pass_store_merging::terminate_all_aliasing_chains):
18385 (pass_store_merging::execute): Check for debug stmts first.
18386 Push new chains onto the m_stores_head stack.
18388 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
18391 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
18392 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
18393 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
18395 2017-03-16 Jeff Law <law@redhat.com>
18397 PR tree-optimization/71437
18398 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
18399 member function. Implementation moved into after_dom_children
18400 member function and into the threader's thread_outgoing_edges
18402 (dom_opt_dom_walker::after_dom_children): Simplify by moving
18403 some code into new thread_outgoing_edges.
18404 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
18405 definition. Simplify marker handling (do it here). Assume we always
18406 have the available expression and the const/copies tables.
18407 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
18409 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
18410 * tree-vrp.c (equiv_stack): No longer file scoped.
18411 (vrp_dom_walker): New class.
18412 (vrp_dom_walker::before_dom_children): New member function.
18413 (vrp_dom_walker::after_dom_children): Likewise.
18414 (identify_jump_threads): Setup domwalker. Use it rather than
18415 walking edges in a random order by hand. Simplify setup/finalization.
18416 (finalize_jump_threads): Remove.
18417 (vrp_finalize): Do not call identify_jump_threads here.
18418 (execute_vrp): Do it here instead and call thread_through_all_blocks
18421 PR tree-optimization/71437
18422 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
18424 (simplify_stmt_for_jump_threading): Add basic_block argument. All
18426 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
18427 (dom_opt_dom_walker::thread_across_edge): Remove
18428 handle_dominating_asserts argument. All callers changed.
18429 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
18430 changes. Remove calls to lhs_of_dominating_assert. Other
18431 uses of handle_dominating_asserts turn into unconditional code
18432 (simplify_control_stmt_condition_1): Likewise.
18433 (simplify_control_stmt_condition): Likewise.
18434 (thread_through_normal_block, thread_across_edge): Likewise.
18435 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
18436 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
18437 object if it is not an SSA_NAME.
18438 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
18439 before calling into the VRP specific simplifiers.
18440 (identify_jump_threads): Remove handle_dominating_asserts
18443 2017-03-16 Jakub Jelinek <jakub@redhat.com>
18446 * tree-diagnostic.c (default_tree_printer): No longer static.
18447 * tree-diagnostic.h (default_tree_printer): New prototype.
18449 2017-03-16 Tamar Christina <tamar.christina@arm.com>
18451 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
18452 Change ins into fmov.
18454 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18456 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
18457 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
18458 Use h_con constraint for operand 1.
18459 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
18460 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
18462 2017-03-15 Jeff Law <law@redhat.com>
18464 PR tree-optimization/71437
18465 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
18466 (record_temporary_equivalences): Use it.
18468 PR tree-optimization/71437
18469 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
18470 tree-ssa-scopedtables.
18471 (lookup_avail_expr, build_and_record_new_cond): Likewise.
18472 (record_conditions, record_cond, vuse_eq): Likewise.
18473 (record_edge_info): Adjust to API tweak of record_conditions.
18474 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
18475 (record_temporary_equivalences, optimize_stmt): Likewise.
18476 (eliminate_redundant_computations): Likewise.
18477 (record_equivalences_from_stmt): Likewise.
18478 * tree-ssa-scopedtables.c: Include options.h and params.h.
18479 (vuse_eq): New function, moved from tree-ssa-dom.c
18480 (build_and_record_new_cond): Likewise.
18481 (record_conditions): Likewise. Accept vector of conditions rather
18482 than edge_equivalence structure for first argument.
18483 for the first argument.
18484 (avail_exprs_stack::lookup_avail_expr): New member function, moved
18485 from tree-ssa-dom.c.
18486 (avail_exprs_stack::record_cond): Likewise.
18487 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
18488 from tree-ssa-dom.c.
18489 (avail_exprs_stack): Add new member functions lookup_avail_expr
18491 (record_conditions): Declare.
18493 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
18496 * lra-constraints.c (process_alt_operands): Increase reject for
18497 reloading an input/output operand.
18499 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
18502 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
18503 insns to convert from signed/unsigned char/short to IEEE 128-bit
18505 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
18507 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
18510 * config/i386/i386.c (ix86_vector_duplicate_value): Create
18511 subreg of inner mode for values already in registers.
18513 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
18515 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
18516 iteration reg is used after the loop.
18518 2017-03-14 Martin Sebor <msebor@redhat.com>
18520 PR tree-optimization/79800
18521 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
18522 precision in negative-positive range.
18523 (format_floating): Call non-const overload with adjusted precision.
18525 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
18528 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
18531 2017-03-14 Martin Sebor <msebor@redhat.com>
18533 PR middle-end/80020
18534 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
18535 * builtins.def (aligned_alloc): Use it.
18538 * Makefile.in (GTFILES): Add calls.c.
18539 * calls.c: Include "gt-calls.h".
18541 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
18543 PR rtl-optimization/79728
18544 * regs.h (struct target_regs): New field
18545 x_contains_allocatable_regs_of_mode.
18546 (contains_allocatable_regs_of_mode): New macro.
18547 * reginfo.c (init_reg_sets_1): Initialize it, and change
18548 contains_reg_of_mode so it includes global regs as well.
18549 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
18550 rather than contains_regs_of_mode.
18552 2017-03-14 Martin Liska <mliska@suse.cz>
18554 * doc/invoke.texi: Document options that can't be combined with
18555 -fcheck-pointer-bounds.
18557 2017-03-14 Martin Liska <mliska@suse.cz>
18559 PR middle-end/79831
18560 * doc/invoke.texi (-Wchkp): Document the option.
18562 2017-03-14 Martin Liska <mliska@suse.cz>
18564 * Makefile.in: Install gcov-dump.
18566 2017-03-14 Martin Liska <mliska@suse.cz>
18568 * multiple_target.c (expand_target_clones): Bail out for
18569 an invalid attribute.
18571 2017-03-14 Richard Biener <rguenther@suse.de>
18573 * alias.c (struct alias_set_entry): Pack properly.
18574 * cfgloop.h (struct loop): Likewise.
18575 * cse.c (struct set): Likewise.
18576 * ipa-utils.c (struct searchc_env): Likewise.
18577 * loop-invariant.c (struct invariant): Likewise.
18578 * lra-remat.c (struct cand): Likewise.
18579 * recog.c (struct change_t): Likewise.
18580 * rtl.h (struct address_info): Likewise.
18581 * symbol-summary.h (function_summary): Likewise.
18582 * tree-loop-distribution.c (struct partition): Likewise.
18583 * tree-object-size.c (struct object_size_info): Likewise.
18584 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
18585 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
18586 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
18587 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
18588 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
18589 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
18590 (struct _stmt_vec_info): Likewise.
18592 2017-03-14 Martin Liska <mliska@suse.cz>
18595 * multiple_target.c (create_dispatcher_calls): Check that
18596 a target can create a function dispatcher.
18598 2017-03-14 Martin Liska <mliska@suse.cz>
18601 * multiple_target.c (expand_target_clones): Drop local.local
18602 flag for default implementation.
18604 2017-03-14 Richard Biener <rguenther@suse.de>
18606 PR tree-optimization/80030
18607 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
18609 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
18611 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
18612 gcc_fallthrough() instead of __attribute__((fallthrough));
18614 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
18616 * doc/gcc.texi: Remove "up" link to (DIR).
18617 * doc/gccint.texi: Ditto.
18619 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
18621 * doc/install.texi (Specific) <avr>: Remove reference to
18624 2017-03-13 Jeff Law <law@redhat.com>
18626 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
18627 attribute rather than comments.
18629 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
18630 match_scratch operand is highest.
18632 2017-03-13 Martin Liska <mliska@suse.cz>
18634 PR middle-end/78339
18635 * ipa-pure-const.c (warn_function_noreturn): If the declarations
18636 is a CHKP clone, use original declaration.
18638 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18640 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
18641 (arc_conditional_register_usage): Use a different allocation order
18642 when optimizing for size.
18643 * common/config/arc/arc-common.c (arc_option_optimization_table):
18644 Section anchors default on when optimizing for size.
18646 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18648 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
18650 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18652 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
18653 * config/arc/arc.md (cpu_facility): Add cd variant.
18654 (*movqi_insn): Add code density variant.
18655 (*movhi_insn): Likewise.
18656 (*movqi_insn): Likewise.
18657 (*addsi3_mixed): Likewise.
18658 (subsi3_insn): Likewise.
18660 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18662 * config/arc/arc.md (movsi_cond_exec): Update constraint.
18664 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18666 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
18667 expressions with MINUS and UNARY ops.
18669 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18672 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
18674 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
18675 between vec_select and vector argument.
18676 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
18677 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
18678 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
18679 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
18680 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
18681 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
18683 2017-03-13 Richard Biener <rguenther@suse.de>
18686 * params.def (vect-max-peeling-for-alignment): Fix typo.
18688 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18690 * doc/install.texi (Specific) <mips-*-*>: Remove description of
18691 issue that only occurred with binutils below 2.18.
18693 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18695 * doc/install.texi (Specific) <cris-axis-elf>: No longer
18696 refer to binutils 2.11/2.12 minimum.
18698 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18700 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
18701 ftp.kernel.org and simplify binutils requirement.
18703 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
18705 * doc/invoke.texi (Warning Options): Fix spelling of link-time
18707 (Optimize Options): Ditto. Also remove redundancy.
18709 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18711 PR translation/79848
18712 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
18714 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
18715 to G_ to avoid double translation.
18717 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18719 PR translation/79923
18720 * auto-profile.c (get_combined_location): Convert leading
18721 character of diagnostics to lower case and remove trailing period.
18722 (read_profile): Likewise for various diagnostics.
18723 * config/arm/arm.c (arm_option_override): Remove trailing period
18724 from various diagnostics.
18725 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
18726 (msp430_expand_delay_cycles): Likewise.
18728 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18731 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
18732 full command-line argument, rather than just "str".
18733 (aarch64_validate_march): Likewise.
18734 (aarch64_validate_mtune): Likewise.
18736 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
18738 PR rtl-optimization/78911
18739 * lra-assigns.c (must_not_spill_p): New function.
18740 (spill_for): Use it.
18742 2017-03-10 Jakub Jelinek <jakub@redhat.com>
18744 PR tree-optimization/79981
18745 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
18746 ATOMIC_COMPARE_EXCHANGE ifn result.
18747 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
18748 IFN_ATOMIC_COMPARE_EXCHANGE.
18750 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18753 * opts.c (parse_sanitizer_options): Add missing question mark to
18754 "did you mean" message.
18756 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18758 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
18760 (VMULEUH_UNS): Likewise.
18761 (VMULOUB_UNS): Likewise.
18762 (VMULOUH_UNS): Likewise.
18763 * config/rs6000/rs6000.c (builtin_function_type): Remove
18764 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
18766 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18769 * read-rtl-function.c (function_reader::read_rtx_operand): Update
18770 x with result of extra_parsing_for_operand_code_0.
18771 (function_reader::extra_parsing_for_operand_code_0): Convert
18772 return type from void to rtx, returning x. When reading
18773 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
18774 larger size containing struct block_symbol.
18776 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
18778 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
18779 -mfloat128-hardware without -m64.
18781 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
18784 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
18785 entries to the case statement that marks unsigned arguments to
18786 overloaded functions.
18788 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18790 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
18791 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
18793 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
18796 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
18797 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
18799 2017-03-10 Martin Liska <mliska@suse.cz>
18803 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
18804 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
18807 2017-03-10 Olivier Hainque <hainque@adacore.com>
18809 * tree-switch-conversion (array_value_type): Start by resetting
18810 candidate type to it's main variant.
18812 2017-03-10 Jakub Jelinek <jakub@redhat.com>
18814 PR rtl-optimization/79909
18815 * combine.c (try_combine): Use simplify_replace_rtx on individual
18816 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
18817 of the whole CALL_INSN_FUNCTION_USAGE.
18819 PR tree-optimization/79972
18820 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
18821 get_range_info on SSA_NAMEs. Formatting fixes.
18823 2017-03-10 Richard Biener <rguenther@suse.de>
18824 Jakub Jelinek <jakub@redhat.com>
18826 PR tree-optimization/77975
18827 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
18828 edge to be constant.
18829 (get_val_for): For constant x return it. Formatting fix.
18830 (loop_niter_by_eval): Avoid pointless looping if the next iteration
18831 would use the same bases as the current one.
18833 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18835 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
18836 instead of vec_select for V1TImode.
18837 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
18839 (VSX_LE_128): Add V1TI to this mode iterator.
18840 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
18841 (*vsx_le_perm_store_<mode>): Likewise.
18842 (pre-reload splitter for VSX stores): Likewise.
18843 (post-reload splitter for VSX stores): Likewise.
18844 (*vsx_xxpermdi2_le_<mode>): Likewise.
18845 (*vsx_lxvd2x2_le_<mode>): Likewise.
18846 (*vsx_stxvd2x2_le_<mode>): Likewise.
18848 2017-03-09 Michael Eager <eager@eagercon.com>
18850 Correct failures with --enable-checking=yes,rtl.
18852 * config/microblaze/microblaze.c (microblaze_expand_shift):
18853 Replace GET_CODE test with CONST_INT_P and INTVAL test with
18854 test for const0_rtx.
18855 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
18856 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
18858 2017-03-09 Richard Biener <rguenther@suse.de>
18860 PR tree-optimization/79977
18861 * graphite-scop-detection.c (scop_detection::merge_sese):
18862 Handle the case of extra exits to blocks dominating the entry.
18864 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
18866 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
18869 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
18871 PR rtl-optimization/79949
18872 * lra-constraints.c (process_alt_operands): Check memory when
18873 trying to predict a cycle. Print about the overall increase.
18875 2017-03-09 Richard Biener <rguenther@suse.de>
18877 PR middle-end/79971
18878 * gimple-expr.c (useless_type_conversion_p): Preserve
18879 TYPE_SATURATING for fixed-point types.
18881 2017-03-09 Richard Biener <rguenther@suse.de>
18884 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
18885 alignment of BLKmode params.
18887 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18890 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
18891 (VALL_NO_V2Q): Likewise.
18893 * config/aarch64/aarch64-simd.md
18894 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
18896 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
18897 VALL_NO_V2Q mode iterator.
18898 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
18900 2017-03-09 Martin Liska <mliska@suse.cz>
18902 PR tree-optimization/79631
18903 * tree-chkp-opt.c (chkp_is_constant_addr): Call
18904 tree_int_cst_sign_bit just for INTEGER constants.
18906 2017-03-09 Martin Liska <mliska@suse.cz>
18910 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
18913 2017-03-09 Marek Polacek <polacek@redhat.com>
18916 * tree.c (inchash::add_expr): Handle TREE_VEC.
18918 2017-03-09 Martin Liska <mliska@suse.cz>
18921 (chkp_narrow_size_and_offset): New function.
18922 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
18923 (void chkp_parse_bit_field_ref): New function.
18924 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
18925 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
18927 2017-03-09 Martin Liska <mliska@suse.cz>
18930 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
18931 (chkp_find_bounds_1): Remove gcc_unreachable.
18933 2017-03-09 Jakub Jelinek <jakub@redhat.com>
18936 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
18937 BUILT_IN_SYNC*, determine the access type from the size suffix and
18938 always build a MEM_REF with that type. Handle forgotten
18939 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
18942 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
18943 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
18944 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
18945 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
18946 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
18947 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
18948 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
18949 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
18950 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
18951 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
18952 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
18953 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
18954 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
18955 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
18956 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
18957 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
18958 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
18959 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
18960 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
18961 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
18962 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
18963 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
18964 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
18965 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
18966 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
18967 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
18968 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
18969 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
18970 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
18971 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
18972 definitions outside of __OPTIMIZE__ guarded section.
18975 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
18976 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
18977 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
18978 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
18981 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18983 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
18984 ("vfenez<mode>"): Add missing constraints.
18986 2017-03-08 Martin Sebor <msebor@redhat.com>
18989 * config/nds32/nds32.c (nds32_option_override):
18990 Fix misspelled diagnostic.
18992 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18995 * gimplify.c (gimplify_omp_for): Replace index var in outer
18996 taskloop statement with an artificial variable and add
18997 OMP_CLAUSE_PRIVATE clause for it.
18999 2017-03-08 Richard Biener <rguenther@suse.de>
19001 PR tree-optimization/79955
19002 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
19003 for accesses that are completely outside of the variable.
19005 2017-03-08 Andrew Haley <aph@redhat.com>
19007 PR tree-optimization/79943
19008 * tree-ssa-loop-split.c (compute_new_first_bound): When
19009 calculating the new upper bound, (END-BEG) should be added, not
19012 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19014 * config/avr/avr.md (setmemhi): Make sure match_dup
19015 operand number comes before match_scratch.
19017 2017-03-08 Richard Biener <rguenther@suse.de>
19019 PR tree-optimization/79920
19020 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
19021 with ncopies == 1 to ...
19022 (vect_transform_slp_perm_load): ... here. Properly compute
19023 all element loads by iterating VF times over the group. Do
19024 not handle ncopies (computed in a broken way) in
19025 vect_create_mask_and_perm.
19027 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19030 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
19031 is a uniform vector, use uniform_vector_p return value instead of
19032 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
19034 2017-03-07 Marek Polacek <polacek@redhat.com>
19036 PR middle-end/79809
19037 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
19038 (alloca_call_type): Likewise.
19040 2017-03-07 Martin Liska <mliska@suse.cz>
19042 * gcov.c (process_args): Put comment to correct location.
19044 2017-03-07 Martin Liska <mliska@suse.cz>
19046 PR middle-end/68270
19047 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
19048 Use array_at_struct_end_p instead of DECL_CHAIN (field).
19049 (chkp_narrow_bounds_for_field): Likewise.
19050 (chkp_parse_array_and_component_ref): Pass one more argument to
19053 2017-03-07 Richard Biener <rguenther@suse.de>
19055 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
19058 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
19060 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
19061 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
19063 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19066 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
19067 to end of description.
19068 (PARAM_MAX_STORES_TO_MERGE): Likewise.
19070 2017-03-07 Jakub Jelinek <jakub@redhat.com>
19072 PR rtl-optimization/79901
19073 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
19075 (*avx512f_<code><mode>3<mask_name>): ... this.
19076 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
19077 iterator instead of VI8_AVX2_AVX512BW.
19079 PR rtl-optimization/79901
19080 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
19081 min/max expander, expand it using expand_vec_cond_expr.
19084 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
19087 2017-03-06 Jakub Jelinek <jakub@redhat.com>
19090 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
19091 to void * for PCH reasons.
19092 * dwarf2out.c (output_loc_operands, output_die): Cast
19093 v.val_vec.array to unsigned char *.
19095 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
19098 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
19101 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
19103 PR rtl-optimization/79571
19104 * lra-constraints.c (process_alt_operands): Calculate static
19105 reject and subtract it from overall when only addresses will be
19108 2017-03-06 Julia Koval <julia.koval@intel.com>
19111 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
19112 incoming stack boundary to 128 for 64-bit targets.
19114 2017-03-06 Richard Biener <rguenther@suse.de>
19116 PR tree-optimization/79894
19117 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
19118 to NULL after folding it.
19120 2017-03-06 Richard Biener <rguenther@suse.de>
19122 PR tree-optimization/79824
19123 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
19124 check disabling peeling for gaps.
19126 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
19128 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
19129 attributes): Document gettimeofday.
19131 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
19133 * config/s390/s390.c (s390_option_override_internal): Set
19134 PARAM_MIN_VECT_LOOP_BOUND
19136 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
19138 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
19139 * config/s390/s390.md: Likewise.
19141 2017-03-06 Jakub Jelinek <jakub@redhat.com>
19144 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
19145 (<avx2_avx512>_perm<mode>): Rename to ...
19146 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
19148 (<avx512>_perm<mode>_mask): Rename to ...
19149 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
19151 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
19152 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
19153 instead of VI8F_256_512.
19154 (avx512f_perm<mode>): New define_expand.
19155 (avx512f_perm<mode>_mask): Likewise.
19156 (avx512f_perm<mode>_1<mask_name>): New define_insn.
19157 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
19159 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
19161 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
19162 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
19164 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
19166 2017-03-06 Martin Liska <mliska@suse.cz>
19169 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
19170 when having a SSA NAME w/o VAR_DECL assigned to it.
19172 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
19174 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
19175 msa_dpsub_<su>_d): Fix MODE for vec_select.
19177 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
19179 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
19181 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
19183 2017-03-06 Richard Biener <rguenther@suse.de>
19185 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
19186 * plugin.c (register_plugin_info): Likewise.
19187 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
19189 2017-03-05 Jakub Jelinek <jakub@redhat.com>
19191 * config/i386/sse.md (sse_storehps, sse_storelps,
19192 avx_<castmode><avxsizesuffix>_<castmode>,
19193 avx512f_<castmode><avxsizesuffix>_<castmode>,
19194 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
19195 in condition that at least one operand is not a MEM.
19197 2017-03-03 Jakub Jelinek <jakub@redhat.com>
19199 PR middle-end/79805
19200 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
19201 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
19203 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
19204 gimple_call_nothrow_p flag based on whether original builtin can throw.
19205 If it can, emit following stmts on the fallthrough edge.
19206 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
19207 don't create new bb if inserting just debug stmts on the edge, try to
19208 insert them on the fallthru bb or just reset debug stmts.
19210 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
19213 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
19214 restore recog_data (including the operand rtxes inside it) around
19215 the call to get_insn_template.
19217 2017-03-03 Martin Sebor <msebor@redhat.com>
19219 PR tree-optimization/79699
19220 * context.c (context::~context): Free MPFR caches to avoid
19221 a memory leak on program exit.
19223 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19225 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
19226 Use wide_int::ulow () instead of .elt (0).
19228 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
19230 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
19231 (*pushxf): Limit oF constraint to 32bit targets and add oC
19232 constraint for 64bit targets.
19233 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
19234 (*pushdf): Change rmF constraint to rmC.
19236 2017-03-03 Martin Liska <mliska@suse.cz>
19238 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
19239 Remove unused variable.
19241 2017-03-03 Jakub Jelinek <jakub@redhat.com>
19244 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
19245 is a memory operand, increase num_memory.
19246 (ix86_expand_args_builtin): Likewise.
19248 2017-03-03 Jan Hubicka <jh@suse.cz>
19251 * ipa-devirt.c (maybe_record_node): Properly handle
19252 __cxa_pure_virtual visibility.
19254 2017-03-03 Martin Liska <mliska@suse.cz>
19256 PR tree-optimization/79803
19257 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
19259 (pass_loop_prefetch::execute): Disabled optimization if an
19260 assumption about L1 cache size is not met.
19262 2017-03-03 Martin Liska <mliska@suse.cz>
19264 PR rtl-optimization/79574
19265 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
19266 (hash_scan_set): Likewise.
19267 (dump_hash_table): Likewise.
19268 (hoist_code): Likewise.
19270 2017-03-03 Richard Biener <rguenther@suse.de>
19272 * fixed-value.c (fixed_from_string): Restore use of elt (1)
19273 in place of uhigh ().
19274 (fixed_convert_from_real): Likewise.
19276 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
19279 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
19281 2017-03-03 Richard Biener <rguenther@suse.de>
19283 PR middle-end/79818
19284 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
19285 TYPE_OVERFLOW_UNDEFINED check.
19287 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19289 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
19291 (vector_ae_<mode>_p): Likewise.
19292 (vector_nez_<mode>_p): Likewise.
19293 (vector_ne_v2di_p): Likewise.
19294 (vector_ae_v2di_p): Likewise.
19295 (vector_ne_<mode>_p): Likewise.
19296 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
19298 (vsx_tsqrt<mode>2_fe): Likewise.
19300 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
19303 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
19305 2017-03-02 Jakub Jelinek <jakub@redhat.com>
19307 PR rtl-optimization/79780
19308 * cprop.c (one_cprop_pass): When second and further conditional trap
19309 in a single basic block is turned into an unconditional trap, turn it
19310 into a deleted note to avoid RTL verification failures.
19312 2017-03-02 Richard Biener <rguenther@suse.de>
19314 * fold-const.c (const_binop): Use ulow () instead of elt (0).
19316 2017-03-02 Richard Biener <rguenther@suse.de>
19318 PR tree-optimization/79345
19320 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
19321 param and abort the walk, returning -1 if it is hit.
19322 (walk_aliased_vdefs): Take a limit param and pass it on.
19323 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
19324 defaulting to 0 and return a signed int.
19325 * tree-ssa-uninit.c (struct check_defs_data): New struct.
19326 (check_defs): New helper.
19327 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
19328 about uninitialized memory.
19329 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
19330 bogus uninitialized warning.
19331 (fixed_convert_from_real): Likewise.
19333 2017-03-02 Bin Cheng <bin.cheng@arm.com>
19335 PR tree-optimization/66768
19336 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
19337 iv_use if base object can't be determined.
19339 2017-03-02 Jakub Jelinek <jakub@redhat.com>
19341 PR tree-optimization/79345
19342 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
19343 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
19344 (get_pattern_stats): Initialize it.
19345 * genemit.c (gen_expand): Verify match_scratch numbers come after
19346 match_operand/match_dup numbers.
19347 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
19348 match_scratch numbers.
19349 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
19351 * config/s390/s390.md (trunctdsd2): Likewise.
19353 2017-03-02 Richard Biener <rguenther@suse.de>
19355 * wide-int.h (wide_int_storage::operator=): Implement in terms
19358 2017-03-02 Richard Biener <rguenther@suse.de>
19360 PR tree-optimization/79777
19361 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
19362 the to insert expression to sth existing.
19364 2017-03-01 Martin Sebor <msebor@redhat.com>
19366 PR middle-end/79692
19367 * gimple-ssa-sprintf.c
19368 (directive::known_width_and_precision): New function.
19369 (format_integer): Use it.
19370 (get_mpfr_format_length): Consider the full range of precision
19371 when computing %g output with the # flag. Set the likely byte
19372 count to 3 rather than 1 when precision is indeterminate.
19373 (format_floating): Correct the lower bound of precision.
19375 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19377 * doc/invoke.texi: Document default code model for 64-bit Linux.
19379 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19382 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
19383 udiv rather than div since input pattern is unsigned.
19385 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
19387 * config/i386/i386.c (print_reg): Warn for values of
19388 unsupported size in integer register.
19390 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
19393 * config/rs6000/predicates.md (current_file_function_operand): Do
19394 not allow self calls to be local if the function is replaceable.
19396 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
19399 * config/rs6000/altivec.h (vec_ctz and others): Change the
19400 preprocessor macro that controls conditional compilation from
19401 _ARCH_PWR9 to __POWER9_VECTOR__.
19402 (vec_all_ne): Change parameterization of __altivec_scalar_pred
19403 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
19404 control (instead of _ARCH_PWR9 control) so that template
19405 definition uses power9-specific function.
19406 (vec_any_eq): Likewise.
19407 (vec_all_ne): Change macro definition to use a power9-specific
19408 expansion under #ifdef __POWER9_VECTOR__ control (instead of
19409 _ARCH_PWR9 control).
19410 (vec_any_eq) Likewise.
19411 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
19412 expansion for CMPNEF to remove support for xvcmpnesp instruction.
19413 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
19414 support for xvcmpnedp instruction.
19415 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
19416 macro expansion so that Power9 implementation of vec_all_ne does
19417 not use the AltiVec predicate framework.
19418 (VCMPNEH_P): Likewise.
19419 (VCMPNEW_P): Likewise.
19420 (VCMPNED_P): Likewise.
19421 (VCMPNEFP_P): Likewise.
19422 (VCMPNEDP_P): Likewise.
19423 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
19424 implementation of vec_any_eq to not use AltiVec predicate
19426 (VCMPAEH_P): Likewise.
19427 (VCMPAEW_P): Likewise.
19428 (VCMPAED_P): Likewise.
19429 (VCMPAEFP_P): Likewise.
19430 (VCMPAEDP_P): Likewise.
19431 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
19432 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
19433 not use the AltiVec predicate framework.
19434 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
19435 of vec_any_eq to not use AltiVec predicate framework.
19436 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19437 support for predefined __POWER9_VECTOR__ macro to indicate that
19438 Power9 instruction selection is enabled.
19439 (altivec_overloaded_builtins): Remove extraneous
19440 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
19441 function argument types RS6000_BTI_bool_V16QI and
19442 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
19443 entry for overloaded function argument types RS6000_BTI_bool_V4SI
19444 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
19445 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
19446 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
19447 Power9 for implementations of vec_cmpne. Change the signature for
19448 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
19449 (representing vec_all_ne) to remove the previously described first
19450 argument of type RS6000_BTI_INTSI, as this was an artifact of
19451 reliance on the AltiVec predicate framework, which is no longer
19452 used in the implementation of these functions. Add
19453 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
19454 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
19455 since, unlike the AltiVec predicate framework implementation, we
19456 do not share function descriptors between vec_alle and vec_anyeq.
19457 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
19458 set of modes that receive special treatment even when
19459 TARGET_P9_VECTOR is true. The special treatment emits code that
19460 does not depend on Power9 instructions.
19461 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
19462 define_expand to not rely on AltiVec predicate framework.
19463 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
19465 (vector_ne_v2di_p): Change this define_expand to not rely on
19466 AltiVec predicate framework.
19467 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
19469 (vector_ne_<mode>_p): Change this define_expand to not rely on
19470 AltiVec predicate framework.
19471 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
19473 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
19474 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
19475 define_insn pattern.
19476 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
19477 define_insn pattern because the xvcmpne<VSs>. instruction is not
19479 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
19480 instruction is not supported.
19482 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19484 * config/nvptx/nvptx.c: Include intl.h.
19486 2017-03-01 Martin Jambor <mjambor@suse.cz>
19489 * ipa-prop.h (ipa_bits): Removed field known.
19490 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
19491 to pointers. Adjusted their comments to warn about their sharing.
19492 (ipcp_transformation_summary): Change bits to a vector of pointers.
19493 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
19494 (ipa_get_ipa_bits_for_value): Declare.
19495 * tree-vrp.h (value_range): Mark as GTY((for_user)).
19496 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
19497 (ipa_bits_hash_table): Likewise.
19498 (ipa_vr_ggc_hash_traits): Likewise.
19499 (ipa_vr_hash_table): Likewise.
19500 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
19501 being pointers and vr_known being removed.
19502 (ipa_set_jf_unknown): Likewise.
19503 (ipa_get_ipa_bits_for_value): New function.
19504 (ipa_set_jfunc_bits): Likewise.
19505 (ipa_get_value_range): New overloaded functions.
19506 (ipa_set_jfunc_vr): Likewise.
19507 (ipa_compute_jump_functions_for_edge): Use the above functions to
19508 construct bits and vr parts of jump functions.
19509 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
19510 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
19512 (ipcp_grow_transformations_if_necessary): Also allocate
19513 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
19515 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
19516 them. Fix too long lines.
19517 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
19518 vr_known being removed.
19519 (ipa_read_jump_function): Use new setter functions to construct bits
19520 and vr parts of jump functions or set them to NULL.
19521 (write_ipcp_transformation_info): Adjust for bits being pointers.
19522 (read_ipcp_transformation_info): Likewise.
19523 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
19525 Include gt-ipa-prop.h.
19526 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
19528 (ipcp_store_bits_results): Likewise.
19529 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
19530 Do not write to existing jump functions but use a temporary instead.
19532 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19535 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
19536 attempt to use its first operand as BIT_FIELD_REF base.
19538 2017-03-01 Richard Biener <rguenther@suse.de>
19540 PR middle-end/79721
19541 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
19542 interpolating formula in wrapping arithmetic.
19543 (chrec_apply): Convert chrec_evaluate return value to wanted type.
19545 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19547 PR tree-optimization/79734
19548 * tree-vect-generic.c (expand_vector_condition): Optimize
19549 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
19550 Handle VEC_COND_EXPR where comparison has different inner width from
19551 type's inner width.
19553 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
19555 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
19556 markup, and similar issues. Remove @opindex entries for things
19557 that aren't options. Add missing -mmpy-option entries.
19559 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19561 PR tree-optimization/79737
19562 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
19563 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
19564 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
19565 instead of byte_size. Formatting fix.
19566 (shift_bytes_in_array_right): Formatting fix.
19568 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
19571 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
19572 condition on optimize for the leaf function test.
19574 2017-02-28 Martin Liska <mliska@suse.cz>
19577 * read-rtl-function.c (function_reader::handle_unknown_directive):
19578 Bail out when one uses -flto.
19580 2017-02-28 Martin Liska <mliska@suse.cz>
19582 * common.opt: Replace space with tabular for options of <number>
19584 * config/i386/i386.opt: Show <number> value for
19585 -mlarge-data-threshold.
19586 * opts.c (print_filtered_help): Do not display number in hexadecimal
19589 2017-02-28 Martin Liska <mliska@suse.cz>
19591 * common.opt: Fix --help=option -Q for options which are of
19594 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
19596 * config/i386/i386.c (print_reg): Error out for values
19597 of 8-bit size in invalid integer register.
19599 2017-02-28 Martin Sebor <msebor@redhat.com>
19601 PR tree-optimization/79691
19602 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
19604 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19607 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
19608 gcc_unreachable with output_operand_lossage.
19610 2017-02-28 Richard Biener <rguenther@suse.de>
19612 PR tree-optimization/79740
19613 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
19615 (visit_nary_op): Insert the nary into the hashtable if we
19616 pattern-matched sth.
19617 * tree-ssa-pre.c (eliminate_insert): Robustify.
19619 2017-02-28 Richard Biener <rguenther@suse.de>
19621 PR middle-end/79731
19622 * fold-const.c (decode_field_reference): Reject out-of-bound
19625 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19627 * config/i386/i386.c: Include intl.h.
19628 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
19629 instead of just cond ? "..." : "...".
19630 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
19631 * coverage.c (read_counts_file): Likewise.
19632 * omp-offload.c: Include intl.h.
19633 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
19634 of just cond ? "..." : "...".
19635 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
19636 of just cond ? "..." : "...".
19638 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
19641 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
19643 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
19644 'tune for' CPU name.
19645 * config/arm/arm-cpu-data.h: Regenerated.
19647 2017-02-28 Richard Biener <rguenther@suse.de>
19649 PR tree-optimization/79732
19650 * tree-inline.c (expand_call_inline): Do not shadow var.
19652 2017-02-28 Richard Biener <rguenther@suse.de>
19654 PR tree-optimization/79723
19655 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
19656 address-space properly.
19658 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
19660 * doc/optinfo.texi (Optimization groups): Fix option used for
19662 * doc/invoke.texi (-fopt-info): Document "omp".
19663 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
19664 (OPTGROUP_ALL): Add OPTGROUP_OMP.
19665 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
19666 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
19667 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
19669 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
19671 * dumpfile.c (optgroup_options): Instead of "openmp", associate
19672 OPTGROUP_OMP with "omp".
19674 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
19677 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
19678 for arithmetic shift of unsigned V2DI.
19680 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
19682 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
19683 arc/linux.h headers.
19684 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
19685 (LINK_SPEC): Likewise.
19686 (ARC_TLS_EXTRA_START_SPEC): Likewise.
19687 (EXTRA_SPECS): Likewise.
19688 (STARTFILE_SPEC): Likewise.
19689 (ENDFILE_SPEC): Likewise.
19690 (LIB_SPEC): Likewise.
19691 (TARGET_SDATA_DEFAULT): Likewise.
19692 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
19693 (MULTILIB_DEFAULTS): Likewise.
19694 (DWARF2_UNWIND_INFO): Likewise.
19695 * config/arc/big.h: New file.
19696 * config/arc/elf.h: Likewise.
19697 * config/arc/linux.h: Likewise.
19698 * config/arc/t-uClibc: Remove.
19700 2017-02-27 Bin Cheng <bin.cheng@arm.com>
19702 PR tree-optimization/77536
19703 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
19704 (tree_transform_and_unroll_loop): Use above function to compute the
19705 estimated niter of unrolled loop and use it when scaling profile.
19706 Also use count info rather than frequency if it's non-zero.
19707 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
19708 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
19709 (vect_transform_loop): Call above function.
19711 2017-02-27 Richard Biener <rguenther@suse.de>
19713 PR tree-optimization/45397
19714 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
19715 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
19716 (visit_nary_op): Add pattern matching for CSEing sign-changed
19717 or truncated operations with wider ones.
19719 2017-02-27 Richard Biener <rguenther@suse.de>
19721 PR tree-optimization/79690
19722 * tree-vect-stmts.c (vectorizable_store): Use vector type
19723 built from the DR with address-space.
19725 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
19727 * doc/invoke.texi (Optimize Options): Refine the description
19728 of asan-use-after-return.
19730 2017-02-25 Alan Modra <amodra@gmail.com>
19732 PR rtl-optimization/79584
19733 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
19734 base, not ad->base_term, the reg within base. Remove assertion
19735 that ad->base == ad->base_term. Replace gen_int_mode using
19736 bogus mode with const0_rtx.
19738 2017-02-25 Jakub Jelinek <jakub@redhat.com>
19740 PR middle-end/79396
19741 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
19742 FMA_EXPR like tcc_binary or tcc_unary.
19744 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
19747 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
19749 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
19750 (output_loc_operands): Handle DW_OP_call_ref and
19751 DW_OP_GNU_variable_value.
19752 (struct variable_value_struct): New type.
19753 (struct variable_value_hasher): Likewise.
19754 (variable_value_hash): New variable.
19755 (string_types): Remove.
19756 (copy_loc_descr): New function.
19757 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
19758 (prepend_loc_descr_to_each): New function.
19759 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
19760 instead of add_loc_descr_to_each if the first argument is single
19761 location list and the second has multiple.
19762 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
19763 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
19764 when looking for variable value which doesn't have other location info.
19765 (loc_list_from_tree): Formatting fix.
19766 (gen_array_type_die): Simplify DW_AT_string_length handling.
19767 (adjust_string_types): Remove.
19768 (gen_subprogram_die): Don't call adjust_string_types nor test/set
19769 string_types. Call resolve_variable_values.
19770 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
19771 (resolve_addr_in_expr): Likewise. Add A argument.
19772 (copy_deref_exprloc): Remove deref argument. Adjust for the
19773 original expression being DW_OP_GNU_variable_value with optionally
19774 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
19775 optionally after it.
19776 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
19777 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
19778 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
19779 (variable_value_hasher::hash, variable_value_hasher::equal): New
19781 (resolve_variable_value_in_expr, resolve_variable_value,
19782 resolve_variable_values, note_variable_value_in_expr,
19783 note_variable_value): New functions.
19784 (dwarf2out_early_finish): Call note_variable_value on all toplevel
19787 2017-02-24 Jakub Jelinek <jakub@redhat.com>
19790 * opts.h (handle_generated_option): Add GENERATED_P argument.
19791 * opts-common.c (handle_option): Adjust function comment.
19792 (handle_generated_option): Add GENERATED_P argument, pass it to
19794 (control_warning_option): Pass false to handle_generated_option
19796 * opts.c (maybe_default_option): Pass true to handle_generated_option
19798 * optc-gen.awk: Likewise.
19800 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19802 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
19803 a REG, look at the REG it is a SUBREG of.
19804 (splitter for cmpeqsi_t): Ditto.
19806 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19808 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
19809 the special USEs with the pattern of the insn, not the insn itself.
19811 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
19814 * doc/invoke.texi: Document -mload-store-pairs.
19816 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19817 Sandra Loosemore <sandra@codesourcery.com>
19819 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
19820 argument isn't a CONST_INT.
19821 (nios2_alternate_compare_const): Assert op is a CONST_INT.
19822 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
19823 (nios2_validate_compare): Bypass alternate compare logic if *op2
19824 is not a CONST_INT.
19825 (ldstwm_operation_p): Return false if first_base is not a REG or
19826 if first_offset is not a CONST_INT.
19828 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19830 * config/cris/cris.md: Use correct operand in a define_peephole2.
19832 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19834 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
19836 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19838 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
19839 this_insn if it is an INSN or JUMP_INSN.
19840 (force_offsettable): Look at base, not at addr.
19841 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
19842 on things that aren't necessarily CONST_INTs.
19844 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
19846 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
19847 -mfpmath=sse is the default also for x86-32 targets with SSE2
19848 instruction set when @option{-ffast-math} is enabled
19850 2017-02-24 Jeff Law <law@redhat.com>
19852 PR rtl-optimizatoin/79286
19853 * ira.c (update_equiv_regs): Drop may_trap_p exception to
19856 2017-02-24 Richard Biener <rguenther@suse.de>
19858 PR tree-optimization/79389
19859 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
19862 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
19864 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
19865 function comment to reflect reality.
19866 (loop_exits_before_overflow): Fix typo in function description.
19868 2017-02-24 Richard Biener <rguenther@suse.de>
19870 PR tree-optimization/79389
19871 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
19872 properly that a threading opportunity exists. Detect conditional
19873 copy/constant propagation opportunities.
19875 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
19877 * config/visium/visium.md (type): Add trap.
19878 (b): New mode attribute.
19879 (*btst): Rename into...
19880 (*btst<mode>): ...this and adjust.
19881 (*cbranchsi4_btst_insn): Rename into...
19882 (*cbranch<mode>4_btst_insn): ...this and adjust.
19883 (trap): New define_insn.
19885 2017-02-23 Jakub Jelinek <jakub@redhat.com>
19887 PR tree-optimization/79389
19888 * ifcvt.c (struct noce_if_info): Add rev_cond field.
19889 (noce_reversed_cond_code): New function.
19890 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
19891 reversed_comparison_code. Formatting fix.
19892 (noce_try_store_flag): Test rev_cond != NULL in addition to
19893 reversed_comparison_code.
19894 (noce_try_store_flag_constants): Likewise.
19895 (noce_try_store_flag_mask): Likewise.
19896 (noce_try_addcc): Use rev_cond if non-NULL instead of
19897 reversed_comparison_code.
19898 (noce_try_cmove_arith): Likewise. Formatting fixes.
19899 (noce_try_minmax, noce_try_abs): Clear rev_cond.
19900 (noce_find_if_block): Initialize rev_cond.
19901 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
19902 instead of false as last argument never attempt to reverse it
19905 2017-02-23 Bin Cheng <bin.cheng@arm.com>
19907 PR tree-optimization/79663
19908 * tree-predcom.c (combine_chains): Process refs in reverse order
19909 only for ZERO length chains, and add explaining comment.
19911 2017-02-23 Jeff Law <law@redhat.com>
19913 PR tree-optimization/79578
19914 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
19915 in call to operand_equal_p.
19917 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
19920 * config/i386/cpuid.h: Fix another undefined behavior.
19922 2017-02-23 Richard Biener <rguenther@suse.de>
19924 PR tree-optimization/79683
19925 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
19926 vector types for data-refs.
19928 2017-02-23 Martin Liska <mliska@suse.cz>
19930 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
19932 2017-02-23 Jakub Jelinek <jakub@redhat.com>
19934 PR middle-end/79665
19935 * internal-fn.c (get_range_pos_neg): Moved to ...
19936 * tree.c (get_range_pos_neg): ... here. No longer static.
19937 * tree.h (get_range_pos_neg): New prototype.
19938 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
19939 are known to be in between 0 and signed maximum inclusive, try to
19940 expand both unsigned and signed divmod and use the cheaper one from
19943 2017-02-22 Jeff Law <law@redhat.com>
19945 PR tree-optimization/79578
19946 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
19947 to compare base operands.
19949 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
19952 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
19953 gpc_reg_operand instead of fpr_reg_operand.
19955 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
19957 * config/mips/mips.c (mips_return_in_memory): Force FP
19958 vector types to be returned in memory for o32 ABI.
19960 2017-02-22 Jakub Jelinek <jakub@redhat.com>
19962 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
19963 instead of DW_TAG_member for static data member declarations and don't
19964 set no_linkage_name for static inline data members.
19965 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
19968 2017-02-22 Martin Liska <mliska@suse.cz>
19970 * doc/invoke.texi: Replace inequality signs with square brackets
19973 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19975 PR tree-optimization/68644
19976 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
19978 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
19981 * lra-constraints.c (simplify_operand_subreg): Handle
19982 WORD_REGISTER_OPERATIONS targets.
19984 2017-02-22 Jakub Jelinek <jakub@redhat.com>
19987 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
19988 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
19989 elimination by swapping fld*.
19991 2017-02-22 Richard Biener <rguenther@suse.de>
19993 PR tree-optimization/79673
19994 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
19995 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
19996 irrelevant address-space qualifiers and avoiding a
19997 ADDR_SPACE_CONVERT_EXPR from fold_convert.
19999 2017-02-22 Richard Biener <rguenther@suse.de>
20001 PR tree-optimization/79666
20002 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
20003 to not symbolically negate if that may introduce undefined
20006 2017-02-22 Martin Liska <mliska@suse.cz>
20009 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
20010 * data-streamer-out.c (streamer_write_gcov_count_stream):
20012 * value-prof.c (stream_out_histogram_value): Make assert more
20013 precise based on type of counter.
20015 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
20018 * config/i386/i386.md (standard_x87sse_constant_load splitter):
20019 Use nonimmediate_operand instead of memory_operand for operand 1.
20020 (float-extend standard_x87sse_constant_load splitter): Ditto.
20022 2017-02-21 Jeff Law <law@redhat.com>
20024 PR tree-optimization/79621
20025 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
20026 blocks with edges to themselves.
20028 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20031 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
20032 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
20033 Use gimple_call_builtin_p.
20036 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
20037 on temporarily removed DEBUG_INSNs.
20039 PR tree-optimization/79649
20040 * tree-loop-distribution.c (classify_partition): Give up on
20041 non-generic address space loads/stores.
20043 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
20045 * doc/loop.texi (Loop manipulation): Remove nonexistent
20046 tree_ssa_loop_version from the documentation.
20047 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
20049 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20052 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
20053 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
20054 * config/rs6000/rs6000.c: Include except.h.
20055 (rs6000_expand_split_stack_prologue): Call
20056 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
20058 2017-02-21 Martin Jambor <mjambor@suse.cz>
20061 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
20062 have been analyzed.
20064 2017-02-21 Martin Jambor <mjambor@suse.cz>
20066 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
20067 for backward compatibility only.
20068 * doc/invoke.texi (Option Summary): Remove all references to
20069 -fipa-cp-alignment.
20071 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
20075 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20077 * lra-constraints.c (curr_insn_transform): Handle
20078 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
20080 2017-02-21 Martin Liska <mliska@suse.cz>
20082 * config/i386/i386.opt: Replace -masm-dialect with -masm.
20084 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
20086 PR translation/79638
20087 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
20089 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
20092 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
20093 (arm_function_ok_for_sibcall): Return false for an indirect call by
20094 descriptor if all the argument registers are used.
20095 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
20096 alignment of the function.
20098 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20100 PR tree-optimization/61441
20101 * simplify-rtx.c (simplify_const_unary_operation): For
20102 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
20103 the sNaN unmodified.
20105 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
20107 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20108 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20109 instead of SYSTEM_HEADER_DIR.
20111 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
20112 Martin Liška <mliska@suse.cz>
20114 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
20115 Fix typos and grammar, use active voice, and clarify.
20117 2017-02-20 Marek Polacek <polacek@redhat.com>
20119 PR middle-end/79537
20120 * gimplify.c (gimplify_expr): Handle unused *&&L;.
20123 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
20125 2017-02-20 Jakub Jelinek <jakub@redhat.com>
20128 * config/i386/i386.c (ix86_expand_builtin): Handle
20129 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
20130 ix86_builtins_isa[fcode].isa as a requirement of those
20131 flags and any other flag in the bitmask.
20132 (ix86_init_mmx_sse_builtins): Use 0 instead of
20133 ~OPTION_MASK_ISA_64BIT as mask.
20134 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
20135 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
20136 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
20137 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
20139 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20142 * lra-constraints.c (split_reg): Check requested split mode
20143 is supported by the register.
20145 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20147 * lra-constraints.c (simplify_operand_subreg): Remove early
20150 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20153 * lra-constraints.c (curr_insn_transform): Tighten condition
20154 for converting SUBREG reloads from OP_OUT to OP_INOUT.
20156 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20159 * lra-constraints.c (curr_insn_transform): Handle
20160 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
20162 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
20165 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
20167 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
20169 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
20172 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
20175 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20177 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
20180 2017-02-18 Jakub Jelinek <jakub@redhat.com>
20182 * final.c (last_columnnum, override_columnnum): New variables.
20183 (final_start_function): Set last_columnnum, pass it to begin_prologue
20184 hook and pass 0 to dwarf2out_begin_prologue.
20185 (final_scan_insn): Update override_columnnum. Pass last_columnnum
20186 to source_line debug hook.
20187 (notice_source_line): Compute last_columnnum and for debug_column_info
20188 return true on column changes.
20189 * debug.h (struct gcc_debug_hooks): Add column argument to
20190 source_line and begin_prologue hooks.
20191 (debug_nothing_int_charstar_int_bool): Remove prototype.
20192 (debug_nothing_int_int_charstar,
20193 debug_nothing_int_int_charstar_int_bool): New prototypes.
20194 (dwarf2out_begin_prologue): Add column argument.
20195 * debug.c (do_nothing_debug_hooks): Adjust source_line and
20196 begin_prologue hooks.
20197 (debug_nothing_int_charstar_int_bool): Remove.
20198 (debug_nothing_int_int_charstar,
20199 debug_nothing_int_int_charstar_int_bool): New functions.
20200 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
20201 through to dwarf2out_source_line.
20202 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
20203 (dwarf2out_source_line): Add column argument, emit it if requested.
20204 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
20206 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
20207 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
20208 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
20209 through to dwarf2out_begin_prologue.
20210 (vmsdbgout_source_line): Add column argument, pass it through to
20211 dwarf2out_source_line.
20212 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
20213 dbxout_source_line caller.
20214 (dbxout_source_line): Add column argument.
20216 * common.opt (gno-column-info, gcolumn-info): New options.
20217 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
20218 (check_die): Also test for multiple DW_AT_decl_column attributes.
20219 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
20220 DW_AT_decl_column if requested.
20221 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
20223 (gen_variable_die): Likewise.
20224 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
20225 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
20228 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
20229 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
20230 (ix86_handle_option): Handle OPT_m3dnowa.
20231 * doc/invoke.texi (-m3dnowa): Document.
20232 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
20233 -m3dnowa instead of -m3dnow -march=athlon.
20236 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
20237 instead of gcc_assert for K, r and R code checks. Formatting fixes.
20239 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20242 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
20243 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
20244 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
20245 generator for vsx_xxpermdi_<mode>_be.
20246 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
20247 force big-endian semantics.
20248 (vsx_xxpermdi_<mode>_be): New define_expand with same
20249 implementation as previous version of vsx_xxpermdi_<mode>.
20251 2017-02-17 Jakub Jelinek <jakub@redhat.com>
20253 PR tree-optimization/79327
20254 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
20255 variable, its initialization and use.
20257 2017-02-17 Julia Koval <julia.koval@intel.com>
20259 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
20260 (OPTION_MASK_ISA_PKU_UNSET): New.
20261 (ix86_handle_option): Handle -mrdpid.
20262 * config/i386/cpuid.h (bit_RDPID): New.
20263 * config/i386/driver-i386.c (host_detect_local_cpu):
20264 Detect RDPID feature.
20265 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
20266 * config/i386/i386-c.c (ix86_target_macros_internal):
20268 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
20269 (ix86_valid_target_attribute_inner_p): Add "rdpid".
20270 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
20271 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
20272 * config/i386/i386.md (define_insn "rdpid"): New.
20273 * config/i386/i386.opt Add -mrdpid.
20274 * config/i386/immintrin.h (_rdpid_u32): New.
20276 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
20278 PR rtl-optimization/79541
20279 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
20280 instead of transforming it into USE.
20282 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
20284 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
20285 If HONOR_SNANS (SFmode) force the input to a register.
20286 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
20287 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
20288 an frsp or similar insn.
20290 2017-02-17 Martin Liska <mliska@suse.cz>
20292 PR rtl-optimization/79577
20293 * params.def (selsched-max-sched-times): Increase minimum to 1.
20295 2017-02-17 Martin Liska <mliska@suse.cz>
20297 PR rtl-optimization/79574
20298 * gcse.c (want_to_gcse_p): Prevent integer overflow.
20300 2017-02-17 Martin Liska <mliska@suse.cz>
20302 PR tree-optimization/79529
20303 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
20304 ssa_defined_default_def_p to handle cases which are implicitly
20306 * tree-ssa.c (ssa_defined_default_def_p): New function.
20307 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
20308 which are implicitly defined.
20309 * tree-ssa.h (ssa_defined_default_def_p): Declare.
20311 2017-02-17 Richard Biener <rguenther@suse.de>
20313 PR middle-end/79576
20314 * params.def (max-ssa-name-query-depth): Limit to 10.
20316 2017-02-17 Richard Biener <rguenther@suse.de>
20318 PR tree-optimization/79552
20319 * tree-ssa-structalias.c (visit_loadstore): Properly verify
20322 2017-02-17 Richard Biener <rguenther@suse.de>
20325 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
20327 2017-02-17 Marek Polacek <polacek@redhat.com>
20329 PR middle-end/79536
20330 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
20331 (fold_negate_expr): New wrapper.
20333 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
20335 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
20336 Correct terminology and de-emphasize pre-standard behavior.
20338 2017-02-16 Alan Modra <amodra@gmail.com>
20340 PR rtl-optimization/79286
20341 * ira.c (def_dominates_uses): New function.
20342 (update_equiv_regs): Don't create an equivalence for insns that
20343 may trap where the register def does not dominate the use.
20345 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
20347 PR rtl-optimization/78127
20348 * lra.c (lra): Call lra_eliminate before finish the loop after
20351 2017-02-16 Richard Biener <rguenther@suse.de>
20353 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
20355 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
20356 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
20357 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
20358 (isl_val_int_from_wi): New function.
20359 (extract_affine_gmp): Rename to ...
20360 (extract_affine_wi): ... this, take a widest_int.
20361 (extract_affine_int): Just wrap extract_affine_wi.
20362 (add_param_constraints): Use isl_val_int_from_wi.
20363 (add_loop_constraints): Likewise, and extract_affine_wi.
20365 2017-02-15 Jeff Law <law@redhat.com>
20367 PR middle-end/79521
20368 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
20369 ira_init_register_move_cost_if_necessary.
20371 2017-02-15 Martin Sebor <msebor@redhat.com>
20373 PR middle-end/32003
20374 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
20375 removed in a prior commit.
20377 2017-02-15 Bin Cheng <bin.cheng@arm.com>
20379 PR tree-optimization/79347
20380 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
20381 counters during peeling.
20383 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
20385 * Makefile.in (site.exp): Remove "set ISLVER".
20387 2017-02-15 Jakub Jelinek <jakub@redhat.com>
20390 * real.c (real_from_integer): Call real_convert even for decimal.
20392 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
20395 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
20397 2017-02-14 Andrew Pinski <apinski@cavium.com>
20399 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
20400 cores and change the partno/implementer to be correct.
20401 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
20402 the 'B" as the implementer.
20403 * config/aarch64/aarch64-tune.md: Regenerate.
20405 2017-02-14 Carl Love <cel@us.ibm.com>
20407 * config/rs6000/rs6000.c: Add case statement entry to make the
20408 xvcvuxdsp built-in argument unsigned.
20409 * config/rs6000/vsx.md: Fix the source and return operand types so they
20410 match the instruction definitions from the ISA document. Fix typo
20411 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
20414 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
20417 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
20418 member early_clobber_alts.
20419 * lra-lives.c (reg_early_clobber_p): New.
20420 (process_bb_lives): Use it.
20421 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
20422 (debug_operand_data): Initialize early_clobber_alts.
20423 (setup_operand_alternative): Set up early_clobber_alts.
20424 (collect_non_operand_hard_regs): Ditto. Pass early clobber
20425 alternatives to new_insn_reg.
20426 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
20428 (lra_update_insn_regno_info): Pass the new arg.
20430 2017-02-14 Jakub Jelinek <jakub@redhat.com>
20432 PR middle-end/79505
20433 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
20434 (new_oacc_loop_raw): Don't clear already cleared fields.
20437 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
20438 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
20439 _mm512_prefetch_i64gather_ps): New inline functions and macros.
20441 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
20444 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
20446 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
20449 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
20450 the extra instruction to the right place to store 128-bit constant
20453 2017-02-14 Martin Sebor <msebor@redhat.com>
20455 PR middle-end/79448
20456 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
20457 warning for strings of unknown length.
20459 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
20461 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
20463 2017-02-14 Jeff Law <law@redhat.com>
20466 * ira-costs.c (scan_one_insn): Initialize register move costs
20467 for pseudos seen in USE/CLOBBER insns.
20469 PR tree-optimization/79095
20470 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
20471 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
20472 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
20473 if the operands are known to be not equal, then the resulting range
20475 (intersect_ranges): If the new range is ~[0,0] and the old range is
20476 wide, then prefer ~[0,0].
20477 * tree-vrp.c (overflow_comparison_p_1): New function.
20478 (overflow_comparison_p): New function.
20479 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
20480 if NAME is used in an overflow test.
20481 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
20482 overflow check that can be expressed as an equality test, then adjust
20483 ops to be that equality test.
20485 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20487 * config/s390/s390-builtin-types.def: Remove flags argument.
20488 * config/s390/s390.c (s390_init_builtins): Likewise.
20490 2017-02-14 Martin Liska <mliska@suse.cz>
20492 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
20493 vector. Fix trailing white spaces.
20495 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
20497 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
20500 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20502 PR rtl-optimization/68664
20503 * config/arm/arm.c (arm_sched_can_speculate_insn):
20504 New function. Declare prototype.
20505 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
20507 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20509 PR rtl-optimization/68664
20510 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
20512 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
20514 2017-02-14 Amit Pawar <amit.pawar@amd.com>
20516 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
20517 max skip bytes for function, loop and jump.
20519 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20521 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
20522 ABS_EXPR for gimple dump.
20524 2017-02-14 Jakub Jelinek <jakub@redhat.com>
20527 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
20529 PR tree-optimization/79408
20530 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
20531 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
20532 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
20533 also if rhs1 is INTEGER_CST.
20535 2017-02-14 Richard Biener <rguenther@suse.de>
20537 PR middle-end/79432
20538 * tree-into-ssa.c (insert_phi_nodes): When the function can
20539 have abnormal edges rewrite SSA names with broken use-def
20540 dominance out of SSA and register them for PHI insertion.
20542 2017-02-13 Martin Sebor <msebor@redhat.com>
20544 PR middle-end/79496
20545 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
20546 clearing info.nowrite flag when snprintf size argument is a range.
20548 2017-02-13 Jakub Jelinek <jakub@redhat.com>
20550 * cprop.c (cprop_jump): Add missing space in string literal.
20551 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
20552 (get_constraint_for_component_ref): Likewise.
20553 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
20554 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
20555 * lra-constraints.c (process_alt_operands): Likewise.
20556 * ipa-inline.c (inline_small_functions): Likewise.
20557 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
20558 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
20559 * trans-mem.c (diagnose_tm_1_op): Likewise.
20560 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
20561 (grid_parallel_clauses_gridifiable): Likewise.
20563 * config/nvptx/mkoffload.c (process): Add space in between
20566 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
20567 "MOD4_SSE_REGS" and "ALL_REGS".
20569 * spellcheck.c (test_data): Add , in between "foo" and "food".
20571 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20574 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
20575 boundary crossing check and subsequent code generation agree.
20577 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20579 * config/aarch64/aarch64.c (has_memory_op): Delete.
20580 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
20583 2017-02-13 Jakub Jelinek <jakub@redhat.com>
20585 PR rtl-optimization/79388
20586 PR rtl-optimization/79450
20587 * combine.c (distribute_notes): When removing TEM_INSN for which
20588 corresponding dest has last value recorded, invalidate that last
20591 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20593 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
20594 of explicit '@'. Add missing assembly comment marker on branch costs
20597 2017-02-13 Nathan Sidwell <nathan@acm.org>
20599 * gengtype-lex.l (<in_struct>): Add '/'.
20601 2017-02-13 Martin Liska <mliska@suse.cz>
20604 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
20606 2017-02-13 Richard Biener <rguenther@suse.de>
20608 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
20610 * configure: Re-generate.
20611 * config.in: Likewise.
20612 * graphite-dependences.c: Simplify as if
20613 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
20614 * graphite-isl-ast-to-gimple.c: Likewise.
20615 * graphite-optimize-isl.c: Likewise.
20616 * graphite-poly.c: Likewise.
20617 * graphite-sese-to-poly.c: Likewise.
20618 * graphite.h: Likewise.
20619 * toplev.c: Include isl/version.h and use isl_version () for
20620 printing the ISL version.
20621 * doc/install.texi: Update ISL requirement.
20623 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
20625 * doc/standards.texi (Standards): Update reference to
20628 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
20630 * doc/extend.texi (Named Address Spaces): sourceware.org now
20632 * doc/install.texi (Binaries): Ditto.
20635 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
20637 * doc/cpp.texi: Replace "stringify"/"stringification" with C
20638 standard terminology "stringize"/"stringizing" throughout.
20639 * doc/cppinternals.texi: Likewise.
20641 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
20643 * doc/extend.texi: Fix some spelling mistakes and typos.
20644 * doc/invoke.texi: Likewise.
20646 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
20649 * params.def (inline-min-speedup) Change from 10 to 8.
20651 2017-02-11 Jakub Jelinek <jakub@redhat.com>
20653 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
20656 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
20659 * ipa-inline-analysis.c (get_minimal_bb): New function.
20660 (record_modified): Use it.
20661 (remap_edge_change_prob): Handle also ancestor functions.
20663 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
20665 * doc/contrib.texi (Contributors): Remove broken link into
20666 the Mauve CVS repository.
20668 2017-02-11 Jakub Jelinek <jakub@redhat.com>
20670 PR middle-end/79454
20671 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
20672 result computation whenever lhs doesn't have vector mode, not
20673 just when it has BLKmode.
20675 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
20677 * doc/makefile.texi (profiledbootstrap): Refer to the
20678 installation instructions only in textual form.
20680 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20683 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
20685 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
20687 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
20688 (Specific): Update mingw-w64 reference.
20690 (Specific): Remove broken link to Renesas RX processor.
20692 2017-02-10 Richard Biener <rguenther@suse.de>
20694 * toplev.c (process_options): Do not mention obsolete graphite
20695 options when printing sorry message about missing graphite support.
20696 Mention -floop-nest-optimize.
20698 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
20700 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
20701 (vtst_p16): Likewise.
20702 (vtstq_p8): Likewise.
20703 (vtstq_p16): Likewise.
20705 (vtstq_p64): Likewise.
20706 * config/arm/arm_neon.h (vgetq_lane_p64): New.
20707 (vset_lane_p64): New.
20708 (vsetq_lane_p64): New.
20710 2017-02-10 Jakub Jelinek <jakub@redhat.com>
20712 PR tree-optimization/79411
20713 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
20714 stmt operands are SSA_NAMEs used in abnormal phis.
20715 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
20718 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
20721 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
20724 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
20726 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
20728 2017-02-09 Jakub Jelinek <jakub@redhat.com>
20730 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
20734 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
20735 not arbitrary TREE_CONSTANT.
20738 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
20739 "omp declare target link" attribute unless is_global_var.
20740 * omp-offload.c (find_link_var_op): Likewise.
20742 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
20743 Chung-Lin Tang <cltang@codesourcery.com>
20745 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
20747 (gimplify_adjust_omp_clauses): Don't delete TILE.
20748 (gimplify_omp_for): Deal with TILE.
20749 * internal-fn.c (expand_GOACC_TILE): New function.
20750 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
20752 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
20753 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
20755 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
20756 avoid DIV for outermost collapse var.
20757 (expand_oacc_for): Insert tile element loop as needed. Adjust.
20758 Remove out of date comments, fix whitespace.
20759 * omp-general.c (omp_extract_for_data): Deal with tiling.
20760 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
20761 adjust OLF_DIM_BASE value.
20762 (struct omp_for_data): Add tiling field.
20763 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
20764 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
20765 for auto loops. Remove default auto determining, moved to
20766 oacc_loop_fixed_partitions.
20767 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
20768 stmts, add e_mask field.
20769 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
20770 (oacc_thread_numbers): Use oacc_dim_call.
20771 (oacc_xform_tile): New.
20772 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
20773 (finish_oacc_loop): Adjust for ifns vector.
20774 (oacc_loop_discover_walk): Append loop abstraction sites to list,
20775 add case for GOACC_TILE fns.
20776 (oacc_loop_xform_loop): Delete.
20777 (oacc_loop_process): Iterate over call list directly, and add
20778 handling for GOACC_TILE fns.
20779 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
20781 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
20782 vector partitioning to outer loops. Assign 2 partitions to loops
20783 when available. Add TILE handling.
20784 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
20785 (execite_oacc_device_lower): Process GOACC_TILE fns,
20786 ignore unknown specs.
20787 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
20788 * tree.c (omp_clause_num_ops): Adjust TILE ops.
20789 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
20791 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
20793 * configure.ac (ACX_BUGURL): Update.
20794 * configure: Regenerate.
20796 2017-02-09 Richard Biener <rguenther@suse.de>
20798 PR tree-optimization/69823
20799 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
20800 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
20802 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
20804 * config/arc/arc-c.def: Add __NPS400__ definition.
20805 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
20806 (TARGET_NPS400): Define.
20808 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
20810 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
20812 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
20813 pointer, arch_info.
20814 (arc_cpu_types): Fill the arch_info field with a pointer into the
20815 arc_arch_types table.
20816 (arc_selected_cpu): Declare.
20817 * config/arc/arc.c (arc_selected_cpu): Make global.
20818 (arc_selected_arch): Delete.
20819 (arc_base_cpu): Delete.
20820 (arc_override_options): Remove references to deleted variables,
20821 update access to arch information.
20822 (ARC_OPT): Update access to arch information.
20823 (ARC_OPTX): Likewise.
20824 * config/arc/arc.h (arc_base_cpu): Remove declaration.
20825 (TARGET_ARC600): Update access to arch information.
20826 (TARGET_ARC601): Likewise.
20827 (TARGET_ARC700): Likewise.
20828 (TARGET_EM): Likewise.
20829 (TARGET_HS): Likewise.
20830 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
20833 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
20836 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
20837 condition/operands for integer GE/LE/GEU/LEU operations.
20839 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
20841 PR translation/79397
20842 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
20845 2017-02-08 Martin Jambor <mjambor@suse.cz>
20848 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
20849 whether allocation happened.
20850 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
20851 nothing was allocated.
20853 2017-02-08 Jakub Jelinek <jakub@redhat.com>
20855 PR tree-optimization/79408
20856 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
20857 constant, but SSA_NAME with a known integer range, use the minimum
20858 of that range instead of op1 to determine if modulo can be replaced
20859 with its first operand.
20861 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20863 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
20865 2017-02-08 Richard Biener <rguenther@suse.de>
20867 PR tree-optimization/71824
20868 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
20869 Check all loops contained in the merged region.
20871 2017-02-07 Andrew Pinski <apinski@cavium.com>
20873 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
20875 2017-02-07 Andrew Pinski <apinski@cavium.com>
20877 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
20878 (thunderxt88): Likewise.
20879 (thunderxt81): Disable LSE and change v8.1 to v8.
20880 (thunderxt83): Likewise.
20882 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20883 Richard Biener <rguenther@suse.de>
20885 PR middle-end/79399
20886 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
20887 type from int to size_t.
20888 * ira-costs.c (struct_costs_size): Change type from int to size_t.
20890 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20892 PR rtl-optimization/79386
20893 * cprop.c (bypass_conditional_jumps): Initialize
20894 bypass_last_basic_block already before splitting bbs after
20895 unconditional traps...
20896 (bypass_conditional_jumps): ... rather than here.
20899 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
20900 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
20901 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
20902 fix -masm=intel patterns.
20904 2017-02-07 Richard Biener <rguenther@suse.de>
20906 PR tree-optimization/79256
20907 PR middle-end/79278
20908 * builtins.c (get_object_alignment_2): Use min_align_of_type
20909 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
20910 and ADJUST_FIELD_ALIGN.
20912 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
20914 * doc/tm.texi: Regenerate.
20915 * stor-layout.c (layout_decl): Adjust.
20916 (update_alignment_for_field): Likewise.
20917 (place_field): Likewise.
20918 (min_align_of_type): Likewise.
20919 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
20920 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
20921 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
20922 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
20923 * config/frv/frv.c (frv_adjust_field_align): Likewise.
20924 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
20925 * config/i386/i386.c (x86_field_alignment): Likewise.
20926 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
20927 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
20928 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
20929 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
20930 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
20931 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
20935 2017-01-30 Richard Biener <rguenther@suse.de>
20937 PR tree-optimization/79256
20938 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
20939 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
20942 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
20944 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
20945 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
20946 builtins to SImode and emit a zero-extend, if necessary.
20948 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20950 * docs/invoke.texi (RISC-V Options): Alphabetize.
20952 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20954 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
20957 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20959 * config/riscv/riscv.c: New file.
20960 * common/config/riscv/riscv-common.c: Likewise.
20961 * config.gcc: Likewise.
20962 * config/riscv/constraints.md: Likewise.
20963 * config/riscv/elf.h: Likewise.
20964 * config/riscv/generic.md: Likewise.
20965 * config/riscv/linux.h: Likewise.
20966 * config/riscv/multilib-generator: Likewise.
20967 * config/riscv/peephole.md: Likewise.
20968 * config/riscv/pic.md: Likewise.
20969 * config/riscv/predicates.md: Likewise.
20970 * config/riscv/riscv-builtins.c: Likewise.
20971 * config/riscv/riscv-c.c: Likewise.
20972 * config/riscv/riscv-ftypes.def: Likewise.
20973 * config/riscv/riscv-modes.def: Likewise.
20974 * config/riscv/riscv-opts.h: Likewise.
20975 * config/riscv/riscv-protos.h: Likewise.
20976 * config/riscv/riscv.h: Likewise.
20977 * config/riscv/riscv.md: Likewise.
20978 * config/riscv/riscv.opt: Likewise.
20979 * config/riscv/sync.md: Likewise.
20980 * config/riscv/t-elf-multilib: Likewise.
20981 * config/riscv/t-linux: Likewise.
20982 * config/riscv/t-linux-multilib: Likewise.
20983 * config/riscv/t-riscv: Likewise.
20984 * configure.ac: Likewise.
20985 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
20986 Waterman as RISC-V maintainers.
20987 * doc/install.texi: Add RISC-V entries.
20988 * doc/invoke.texi: Add RISC-V options section.
20989 * doc/md.texi: Add RISC-V constraints section.
20990 * configure: Regenerated.
20992 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
20995 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
20996 false values to be constant vectors with all 0 or all 1 bits set.
20997 (vcondu<mode><mode>): Likewise.
20998 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
21000 (fpmask_comparison_operator): Update comment.
21001 (vecint_comparison_operator): New predicate.
21002 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
21003 vector conditionals when the true and false values are constant
21004 vectors with all 0 bits or all 1 bits set.
21006 2017-02-06 Martin Sebor <msebor@redhat.com>
21008 PR tree-optimization/79376
21009 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
21011 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
21013 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
21014 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
21015 to simplify split condition.
21017 2017-02-06 Jakub Jelinek <jakub@redhat.com>
21019 * omp-expand.c (oxpand_omp_atomic_fetch_op,
21020 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
21023 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
21025 PR rtl-optimization/68664
21026 * target.def (can_speculate_insn): New hook.
21027 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
21028 * doc/tm.texi: Regenerate.
21029 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
21030 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
21031 (rs6000_sched_can_speculate_insn): New function.
21033 2017-02-06 Jakub Jelinek <jakub@redhat.com>
21035 PR tree-optimization/79284
21036 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
21037 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
21038 vectorizable_mask_load_store, vectorizable_operation,
21039 vect_is_simple_cond, get_same_sized_vectype): Use it instead
21040 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
21041 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
21042 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
21043 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
21044 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
21045 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
21046 is_gimple_assign (stmt). Replace another such test with
21047 is_gimple_assign (stmt).
21049 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
21052 * config/avr/avr.c (rtl-iter.h): Include it.
21053 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
21054 (avr_legitimate_combined_insn): ...and implementation.
21056 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
21058 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
21059 * config/s390/s390.c (s390_const_operand_ok)
21060 (s390_canonicalize_comparison, s390_extract_part)
21061 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
21062 (s390_contiguous_bitmask_p, s390_rtx_costs)
21063 (legitimize_pic_address): Likewise.
21064 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
21065 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
21066 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
21067 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
21068 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
21070 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
21072 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
21073 REGNO($0) == REGNO($1).
21075 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21077 * config/s390/linux.h(SIZE_TYPE): Add comment.
21079 2017-02-06 Julian Brown <julian@codesourcery.com>
21080 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21081 Virendra Pathak <virendra.pathak@broadcom.com>
21083 * config/aarch64/aarch64-cores.def: Change the scheduler
21085 * config/aarch64/aarch64.md: Include thunderx2t99.md.
21086 * config/aarch64/thunderx2t99.md: New file.
21088 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
21090 * doc/standards.texi (Go Language): Update link to language
21093 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
21095 * tree-eh.c (lower_resx): Sanitize profile.
21096 (cleanup_empty_eh_move_lp): Likewise.
21098 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
21101 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
21103 * cfgloopmanip.h (loop_version): Update prototype.
21104 * modulo-sched.c (sms_schedule): Update call of loop_version.
21105 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
21106 * tree-parloops.c (gen_parallel_loop): Likewise.
21107 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
21108 * tree-ssa-loop-split.c (split_loop): Likewise.
21109 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
21110 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
21112 2017-02-05 Martin Liska <mliska@suse.cz>
21115 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
21117 (print_operand_address): Initialize a struct to zero.
21119 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
21121 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
21122 garbage collector only in textual form.
21124 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
21126 * doc/extend.texi (x86 specific memory model extensions for
21127 transactional memory): Simplify a phrase.
21129 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
21132 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
21133 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
21134 (atomic_storedi_1): Likewise.
21136 2017-02-04 Jakub Jelinek <jakub@redhat.com>
21138 PR tree-optimization/79338
21139 * tree-parloops.c (gather_scalar_reductions): Don't call
21140 vect_analyze_loop_form for loop->inner before destroying loop's
21143 2017-02-03 Martin Sebor <msebor@redhat.com>
21145 PR tree-optimization/79327
21146 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
21147 when precision has resulted in leading zeros.
21148 (format_integer): Adjust the likely counter to assume an unknown
21149 argument that may be zero is non-zero.
21151 2017-02-03 Jason Merrill <jason@redhat.com>
21154 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
21155 avoid copying non-taken branch.
21157 2017-02-03 Jakub Jelinek <jakub@redhat.com>
21159 PR tree-optimization/79340
21160 * tree-vect-loop.c (vectorizable_reduction): Release
21161 vec_defs elements after safe_splicing them into other vectors.
21164 PR tree-optimization/79327
21165 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
21166 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
21168 (format_integer): Use wide_int_to_tree instead of build_int_cst
21169 + to_?hwi. If argmin is NULL, just set argmin and argmax to
21170 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
21171 of shortest and longest sequence.
21173 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
21175 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
21176 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
21178 2017-02-03 Walter Lee <walt@tilera.com>
21181 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
21182 after initial stackframe link reg save.
21183 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
21185 2017-02-03 Jakub Jelinek <jakub@redhat.com>
21188 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
21189 wu for stxssp alternative.
21191 2017-02-03 Martin Sebor <msebor@redhat.com>
21193 PR tree-optimization/79352
21194 * gimple-fold.c (get_range_strlen): Add argument.
21195 (get_range_strlen): Change return type to bool.
21196 (get_maxval_strlen): Pass in a dummy argument.
21197 * gimple-fold.h (get_range_strlen): Change return type to bool.
21198 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
21199 * tree.h (array_at_struct_end_p): Add argument.
21200 * tree.c (array_at_struct_end_p): Handle it.
21202 2017-02-03 Martin Liska <mliska@suse.cz>
21205 * multiple_target.c (create_dispatcher_calls): Redirect edge
21206 from a caller of a dispatcher.
21207 (expand_target_clones): Make the clones local.
21208 (ipa_target_clone): Do both target clones and resolvers.
21209 (ipa_dispatcher_calls): Remove the pass.
21210 (pass_dispatcher_calls::gate): Likewise.
21211 (make_pass_dispatcher_calls): Likewise.
21212 * passes.def (pass_target_clone): Put as very first IPA early
21215 2017-02-03 Martin Liska <mliska@suse.cz>
21217 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
21218 in case of a function with ifunc attribute.
21220 2017-02-03 Martin Liska <mliska@suse.cz>
21222 * cgraph.c (cgraph_node::dump): Dump function version info.
21223 * symtab.c (symtab_node::dump_base): Add missing new line.
21225 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
21227 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
21228 (ifcombine_ifandif): Use it.
21230 2017-02-03 Martin Liska <mliska@suse.cz>
21232 * doc/invoke.texi: Document default value for
21233 use-after-scope-direct-emission-threshold.
21235 2017-02-03 Martin Liska <mliska@suse.cz>
21237 PR tree-optimization/79339
21238 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
21239 (format_floating): Likewise.
21241 2017-02-03 Martin Liska <mliska@suse.cz>
21244 * ipa-prop.c (ipa_node_params_t::insert): Remove current
21246 (ipa_node_params_t::remove): Likewise.
21247 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
21248 initialization from removed ipa_node_params_t::insert.
21249 (ipa_node_params::~ipa_node_params): Move from removed
21250 ipa_node_params_t::release.
21251 * symbol-summary.h (symbol_summary::m_released): New member.
21252 Do not release a summary twice. Do not allow to call finalizer
21253 for types of a summary that live in GGC memory.
21255 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21257 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
21260 2017-02-02 Martin Sebor <msebor@redhat.com>
21262 PR middle-end/79275
21263 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
21264 (format_string): Tighten up the range of output for non-constant
21265 strings and correct the expected range for wide non-constant strings.
21267 2017-02-02 Martin Sebor <msebor@redhat.com>
21269 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
21271 PR middle-end/32003
21272 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
21274 (-fdump-tree-@var): Add to index and document how to come up
21275 with pass-specific option and dump file names.
21276 (-fdump-passes): Clarify where to look for output.
21278 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
21280 PR middle-end/77445
21281 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
21282 statistics of the analyzed path; allow threading for speed when
21283 any of BBs along the path are optimized for speed.
21285 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
21287 PR middle-end/78468
21288 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
21289 settings of the virtual registers.
21292 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
21294 * explow.c (get_dynamic_stack_size): Take known alignment of stack
21295 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
21298 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21300 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
21301 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
21303 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21305 * config/s390/s390.md: Add missing comments with the expanded
21307 * config/s390/vector.md: Likewise.
21308 * config/s390/vx-builtins.md: Likewise.
21310 2017-02-02 Jakub Jelinek <jakub@redhat.com>
21313 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
21314 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
21315 conditions on a single line.
21317 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21319 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
21320 __S390_VX__ to __VX__.
21322 2017-02-01 Andrew Pinski <apinski@cavium.com>
21324 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
21325 stmt_info to record_stmt_cost.
21326 (vect_get_known_peeling_cost): Pass stmt_info if known to
21328 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
21329 cpu_vector_cost field into
21330 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
21331 field into vec_int_stmt_cost and vec_fp_stmt_cost.
21332 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
21333 splitting of scalar_stmt_cost and vec_stmt_cost.
21334 (thunderx_vector_cost): Likewise.
21335 (cortexa57_vector_cost): LIkewise.
21336 (exynosm1_vector_cost): Likewise.
21337 (xgene1_vector_cost): Likewise.
21338 (thunderx2t99_vector_cost): Improve after the splitting of the two
21340 (aarch64_builtin_vectorization_cost): Update for the splitting of
21341 scalar_stmt_cost and vec_stmt_cost.
21343 2017-02-01 Torvald Riegel <triegel@redhat.com>
21344 Richard Henderson <rth@redhat.com>
21346 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
21347 conditional on existance of a fast atomic load.
21348 * optabs-query.c (can_atomic_load_p): New function.
21349 * optabs-query.h (can_atomic_load_p): Declare it.
21350 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
21351 no fast atomic load is available for the particular size of access.
21352 (expand_atomic_compare_and_swap): Likewise.
21353 (expand_atomic_load): Likewise.
21354 (expand_atomic_store): Likewise.
21355 (expand_atomic_fetch_op): Likewise.
21356 * testsuite/lib/target-supports.exp
21357 (check_effective_target_sync_int_128): Remove x86 because it provides
21358 no fast atomic load.
21359 (check_effective_target_sync_int_128_runtime): Likewise.
21361 2017-02-01 Richard Biener <rguenther@suse.de>
21363 * graphite.c: Include tree-vectorizer.h for find_loop_location.
21364 (graphite_transform_loops): Provide opt-info for optimized nests.
21365 * tree-parloop.c (parallelize_loops): Provide opt-info for
21366 parallelized loops.
21368 2017-02-01 Richard Biener <rguenther@suse.de>
21370 PR middle-end/79315
21371 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
21372 was not set before.
21374 2017-02-01 Richard Biener <rguenther@suse.de>
21376 PR tree-optimization/71824
21377 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
21378 Verify the loops are valid in the merged SESE region.
21379 (scop_detection::can_represent_loop_1): Check analyzing the
21380 evolution of the number of iterations in the region succeeds.
21382 2017-01-31 Ian Lance Taylor <iant@golang.org>
21384 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
21385 REG_ARGS_SIZE note to 32-bit push insns and call insn.
21387 2017-01-31 David Malcolm <dmalcolm@redhat.com>
21389 PR preprocessor/79210
21390 * input.c (get_substring_ranges_for_loc): Replace line_width
21391 assertion with error-handling.
21393 2017-01-31 Richard Biener <rguenther@suse.de>
21395 PR tree-optimization/77318
21396 * graphite-sese-to-poly.c (extract_affine): Fix assert.
21397 (create_pw_aff_from_tree): Take loop parameter.
21398 (add_condition_to_pbb): Pass loop of the condition to
21399 create_pw_aff_from_tree.
21401 2017-01-31 Jakub Jelinek <jakub@redhat.com>
21403 * config/s390/s390.c (s390_asan_shadow_offset): New function.
21404 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
21406 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
21410 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
21412 (convert_int_to_float128): Likewise.
21413 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
21414 (convert_int_to_float128): Likewise.
21415 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
21416 (UNSPEC_IEEE128_CONVERT): Likewise.
21417 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
21418 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
21419 Use local variables for IBM extended format.
21420 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
21421 (fix_trunc<mode>si2_fprs): Likewise.
21422 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
21423 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
21424 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
21425 to know that we can now have integers of all sizes in vector
21427 (fix<uns>_<mode>di2_hw): Likewise.
21428 (float<uns>_<mode>si2_hw): Likewise.
21429 (fix_<mode>si2_hw): Likewise.
21430 (fixuns_<mode>si2_hw): Likewise.
21431 (float<uns>_<mode>di2_hw): Likewise.
21432 (float_<mode>di2_hw): Likewise.
21433 (float_<mode>si2_hw): Likewise.
21434 (floatuns_<mode>di2_hw): Likewise.
21435 (floatuns_<mode>si2_hw): Likewise.
21436 (xscvqp<su>wz_<mode>): Delete, no longer used.
21437 (xscvqp<su>dz_<mode>): Likewise.
21438 (xscv<su>dqp_<mode>): Likewise.
21439 (ieee128_mfvsrd_64bit): Likewise.
21440 (ieee128_mfvsrd_32bit): Likewise.
21441 (ieee128_mfvsrwz): Likewise.
21442 (ieee128_mtvsrw): Likewise.
21443 (ieee128_mtvsrd_64bit): Likewise.
21444 (ieee128_mtvsrd_32bit): Likewise.
21446 2017-01-31 Martin Liska <mliska@suse.cz>
21449 * ipa-prop.c (ipa_free_all_node_params): Call release method
21450 instead of ~sumbol_summary to not to trigger double times
21453 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
21455 PR tree-optimization/71691
21456 * bitmap.h (class auto_bitmap): New.
21457 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
21458 is_maybe_undefined instead of ssa_undefined_value_p.
21460 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21462 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
21463 __S390_ARCH_LEVEL__ to __ARCH__.
21465 2017-01-31 Jakub Jelinek <jakub@redhat.com>
21467 PR tree-optimization/79267
21468 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
21469 if should_remove_lhs_p is true.
21471 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
21474 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
21475 (add_alignment_attribute): New.
21476 (base_type_die): Add alignment attribute.
21477 (subrange_type_die): Likewise.
21478 (modified_type_die): Likewise.
21479 (gen_array_type_die): Likewise.
21480 (gen_descr_array_type_die: Likewise.
21481 (gen_enumeration_type_die): Likewise.
21482 (gen_subprogram_die): Likewise.
21483 (gen_variable_die): Likewise.
21484 (gen_field_die): Likewise.
21485 (gen_ptr_to_mbr_type_die): Likewise.
21486 (gen_struct_or_union_type_die): Likewise.
21487 (gen_subroutine_type_die): Likewise.
21488 (gen_typedef_die): Likewise.
21489 (base_type_cmp): Compare alignment attribute.
21491 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21494 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
21495 (setb_unsigned) New pattern for setb with CCUNS.
21496 * config/rs6000/rs6000.c (expand_block_compare): Use a different
21497 subfc./subfe sequence to avoid overflow problems. Generate a
21498 shorter sequence with cmpld/setb for power9.
21499 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
21500 for generating subfc. instruction.
21501 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
21502 now uses this instruction.
21504 2017-01-30 Ian Lance Taylor <iant@google.com>
21507 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
21508 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
21510 2017-01-30 Martin Sebor <msebor@redhat.com>
21512 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
21513 Move constant to the right of a relational operator.
21514 (get_mpfr_format_length, format_character, format_string): Ditto.
21515 (should_warn_p, maybe_warn): Same.
21517 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
21519 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
21522 * asan.c (get_translation_unit_decl): Remove function.
21523 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
21525 2017-01-30 Martin Liska <mliska@suse.cz>
21527 PR gcov-profile/79259
21528 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
21529 -fprofile-generate.
21531 2017-01-30 Martin Liska <mliska@suse.cz>
21534 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
21535 Initialize variables with NULL value.
21537 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
21540 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
21542 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
21544 2017-01-30 Richard Biener <rguenther@suse.de>
21546 PR tree-optimization/79276
21547 * tree-vrp.c (process_assert_insertions): Properly adjust common
21548 when removing a duplicate.
21550 * gcc.dg/torture/pr79276.c: New testcase.
21552 2017-01-30 Richard Biener <rguenther@suse.de>
21554 PR tree-optimization/79256
21555 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
21556 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
21558 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
21560 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
21563 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
21564 ("*r<noxa>sbg_<mode>_sll_bitmask")
21565 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
21566 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
21567 Use contiguous_bitmask_nowrap_operand.
21569 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21572 * config/rs6000/altivec.h (vec_xl): Revise #define.
21573 (vec_xst): Likewise.
21575 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
21577 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
21579 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
21581 PR rtl-optimization/79194
21582 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
21583 traps before call to bypass_conditional_jumps.
21585 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
21587 PR tree-optimization/71374
21588 * lra-constraints.c (check_conflict_input_operands): New.
21589 (match_reload): Use it.
21591 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
21594 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
21595 account to calculate conflict_set.
21597 2017-01-27 Bin Cheng <bin.cheng@arm.com>
21599 PR rtl-optimization/78559
21600 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
21601 other_insn in combine.
21603 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21605 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
21606 uint16_type_node for BT_UINT16.
21608 2017-01-27 David Malcolm <dmalcolm@redhat.com>
21610 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
21611 "RTL Tests" to menu.
21612 (GIMPLE Tests): New node.
21613 (RTL Tests): New node.
21615 2017-01-27 Richard Biener <rguenther@suse.de>
21617 PR tree-optimization/79245
21618 * tree-loop-distribution.c (distribute_loop): Apply cost
21619 modeling also to detected patterns.
21621 2017-01-27 Richard Biener <rguenther@suse.de>
21623 PR tree-optimization/71433
21624 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
21625 (compare_assert_loc): New function.
21626 (process_assert_insertions): Sort and optimize assert locations
21627 to remove duplicates and push down identical assertions on
21628 edges to their destination block.
21630 2017-01-27 Richard Biener <rguenther@suse.de>
21632 PR tree-optimization/79244
21633 * tree-vrp.c (remove_range_assertions): Forcefully propagate
21634 out SSA names even if abnormal.
21636 2017-01-27 Jakub Jelinek <jakub@redhat.com>
21638 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
21639 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
21640 instead of MPFR_RNDN.
21642 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
21645 * arm.c (arm_option_override): Don't call build_target_option_node
21646 until after doing all option overrides.
21647 (arm_valid_target_attribute_tree): Likewise.
21649 2017-01-27 Martin Liska <mliska@suse.cz>
21651 * doc/invoke.texi (-fprofile-arcs): Document profiling support
21652 for {cd}tors and C++ {cd}tors.
21654 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
21656 * config/s390/s390.md ("*setmem_long_and")
21657 ("*setmem_long_and_31z"): Use zero_extend instead of and.
21659 2017-01-26 Martin Sebor <msebor@redhat.com>
21661 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
21664 2017-01-26 Martin Sebor <msebor@redhat.com>
21666 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
21667 HAVE_DFmode before using XFmode or DFmode.
21668 (parse_directive): Avoid using the z length modifier to avoid
21669 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
21671 PR middle-end/78703
21672 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
21673 to accept adjustment as an array.
21674 (get_int_range): New function.
21675 (struct directive): Make width and prec arrays.
21676 (directive::set_width, directive::set_precision): Call get_int_range.
21677 (format_integer, format_floating): Handle width and precision ranges.
21678 (format_string, parse_directive): Same.
21680 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21683 * dwarf2out.c (generate_skeleton_bottom_up): For children with
21684 comdat_type_p set, just clone them, but keep the children in the
21688 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
21689 which have direct callers with -fvar-tracking-assignments enabled
21691 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
21692 inside of type units.
21694 2017-01-26 Martin Sebor <msebor@redhat.com>
21696 PR middle-end/78703
21697 * gimple-ssa-sprintf.c (struct result_range): Add likely and
21699 (struct format_result): Replace number_chars, number_chars_min,
21700 and number_chars_max with a single member of struct result_range.
21702 (format_result::operator+=): Adjust.
21703 (struct fmtresult): Remove bounded. Handle likely and unlikely
21705 (fmtresult::adjust_for_width_or_precision): New function.
21706 (fmtresult:type_max_digits): New function.
21707 (bytes_remaining): Handle likely and unlikely counters.
21708 (min_bytes_remaining): Remove.
21709 (format_percent): Simplify.
21710 (format_integer, format_floating): Set likely and unlikely counters.
21711 (get_string_length, format_character, format_string): Same.
21712 (format_plain, should_warn_p): New function.
21713 (maybe_warn): Call should_warn_p. Update diagnostic messages
21714 and handle those for all directives, including plain strings.
21715 (format_directive): Handle likely and unlikely counters.
21716 Remove unnecessary quoting from diagnostics. Add an informational
21718 (add_bytes): Remove.
21719 (pass_sprintf_length::compute_format_length): Simplify.
21720 (try_substitute_return_value): Handle likely and unlikely counters.
21722 2017-01-26 Carl Love <cel@us.ibm.com>
21724 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
21725 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
21727 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
21730 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
21731 endianess for subregs into account.
21732 * lra-constraints.c (lra_constraints): Do risky transformations
21733 always on the first iteration.
21734 * lra-lives.c (check_pseudos_live_through_calls): Add arg
21735 last_call_used_reg_set.
21736 (process_bb_lives): Define and use last_call_used_reg_set.
21737 * lra.c (lra): Always continue after lra_constraints on the first
21740 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
21742 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
21744 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
21746 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21748 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
21749 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
21750 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
21751 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
21752 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
21753 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
21754 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
21755 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
21756 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
21758 2017-01-26 Marek Polacek <polacek@redhat.com>
21761 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
21762 for the third operand.
21764 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21766 PR middle-end/79236
21767 * omp-low.c (struct omp_context): Add simt_stmt field.
21768 (scan_omp_for): Return omp_context *.
21769 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
21770 context to the _simt_ SIMD stmt.
21771 (lower_omp_for): For combined SIMD with sibling _simt_
21772 SIMD, make sure to use the same decls in _looptemp_
21773 clauses as in the sibling.
21775 2017-01-26 David Sherwood <david.sherwood@arm.com>
21777 PR middle-end/79212
21778 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
21781 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21784 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
21785 emit fld b; fld a; if possible.
21787 * brig-builtins.def: Update copyright years.
21788 * config/arm/arm_acle_builtins.def: Update copyright years.
21790 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
21793 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
21794 constraint instead of o for the stxsd instruction.
21796 2017-01-25 Carl Love <cel@us.ibm.com>
21798 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
21799 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
21801 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
21803 * doc/invoke.texi (C++ Dialect Options): Fix typo.
21805 2017-01-25 Richard Biener <rguenther@suse.de>
21807 PR tree-optimization/69264
21808 * target.def (vector_alignment_reachable): Improve documentation.
21809 * doc/tm.texi: Regenerate.
21810 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
21812 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
21813 earlier changes with respect to TYPE_USER_ALIGN.
21814 (vector_alignment_reachable_p): Likewise. Improve dumping.
21816 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21819 * config/arm/arm.md (xordi3): Force constant operand into a register
21822 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21824 * doc/invoke.texi (-fstore-merging): Correct default optimization
21825 levels at which it is enabled.
21826 (-O): Move -fstore-merging from list to...
21829 2017-01-25 Richard Biener <rguenther@suse.de>
21832 * omp-expand.c: Include debug.h.
21833 (expand_omp_taskreg): Make sure to generate early debug before
21834 outlining anything from a function.
21835 (expand_omp_target): Likewise.
21836 (grid_expand_target_grid_body): Likewise.
21838 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
21841 * asan.c (get_translation_unit_decl): New function.
21842 (asan_add_global): Extract modules file name from globals
21843 TRANSLATION_UNIT_DECL name.
21845 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
21848 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
21849 for long calls with APCS frame and VFP.
21851 2017-01-24 David Malcolm <dmalcolm@redhat.com>
21853 * cfg.c (original_copy_tables_initialized_p): New function.
21854 * cfg.h (original_copy_tables_initialized_p): New decl.
21855 * cfgrtl.c (relink_block_chain): Guard the call to
21856 free_original_copy_tables with a call to
21857 original_copy_tables_initialized_p.
21858 * cgraph.h (symtab_node::native_rtl_p): New decl.
21859 * cgraphunit.c (symtab_node::native_rtl_p): New function.
21860 (symtab_node::needed_p): Don't assert for early assembly output
21861 for __RTL functions.
21862 (cgraph_node::finalize_function): Set "force_output" for __RTL
21864 (cgraph_node::analyze): Bail out early for __RTL functions.
21865 (analyze_functions): Update assertion to support __RTL functions.
21866 (cgraph_node::expand): Bail out early for __RTL functions.
21867 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
21869 * function.h (struct function): Update comment for field
21871 * gimple-expr.c: Include "tree-pass.h".
21872 (gimple_has_body_p): Return false for __RTL functions.
21873 * Makefile.in (OBJS): Add run-rtl-passes.o.
21874 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
21876 (gcc::pass_manager::get_clean_slate): New accessor.
21877 * passes.c: Include "insn-addr.h".
21878 (should_skip_pass_p): Add logging. Update logic for running
21879 "expand" to be compatible with both __GIMPLE and __RTL. Guard
21880 property-provider override so it is only done for gimple passes.
21882 (skip_pass): New function.
21883 (execute_one_pass): Call skip_pass when skipping passes.
21884 * read-md.c (md_reader::read_char): Support filtering
21885 the input to a subset of line numbers.
21886 (md_reader::md_reader): Initialize fields
21887 m_first_line and m_last_line.
21888 (md_reader::read_file_fragment): New function.
21889 * read-md.h (md_reader::read_file_fragment): New decl.
21890 (md_reader::m_first_line): New field.
21891 (md_reader::m_last_line): New field.
21892 * read-rtl-function.c (function_reader::create_function): Only
21893 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
21894 curr_properties. Set DECL_INITIAL to a dummy block.
21895 (read_rtl_function_body_from_file_range): New function.
21896 * read-rtl-function.h (read_rtl_function_body_from_file_range):
21898 * run-rtl-passes.c: New file.
21899 * run-rtl-passes.h: New file.
21901 2017-01-24 Jeff Law <law@redhat.com>
21903 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
21906 2017-01-24 Bin Cheng <bin.cheng@arm.com>
21908 PR tree-optimization/79159
21909 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
21910 (record_nonwrapping_iv): Improve boundary using above function if no
21911 value range information.
21913 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
21914 Martin Jambor <mjambor@suse.cz>
21916 * brig-builtins.def: New file.
21917 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
21918 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
21919 (DEF_HSAIL_SAT_BUILTIN): Likewise.
21920 (DEF_HSAIL_INTR_BUILTIN): Likewise.
21921 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
21922 * builtin-types.def (BT_INT8): New.
21923 (BT_INT16): Likewise.
21924 (BT_UINT8): Likewise.
21925 (BT_UINT16): Likewise.
21926 (BT_FN_ULONG): Likewise.
21927 (BT_FN_UINT_INT): Likewise.
21928 (BT_FN_UINT_ULONG): Likewise.
21929 (BT_FN_UINT_LONG): Likewise.
21930 (BT_FN_UINT_PTR): Likewise.
21931 (BT_FN_ULONG_PTR): Likewise.
21932 (BT_FN_INT8_FLOAT): Likewise.
21933 (BT_FN_INT16_FLOAT): Likewise.
21934 (BT_FN_UINT32_FLOAT): Likewise.
21935 (BT_FN_UINT16_FLOAT): Likewise.
21936 (BT_FN_UINT8_FLOAT): Likewise.
21937 (BT_FN_UINT64_FLOAT): Likewise.
21938 (BT_FN_UINT16_UINT32): Likewise.
21939 (BT_FN_UINT32_UINT16): Likewise.
21940 (BT_FN_UINT16_UINT16_UINT16): Likewise.
21941 (BT_FN_INT_PTR_INT): Likewise.
21942 (BT_FN_UINT_PTR_UINT): Likewise.
21943 (BT_FN_LONG_PTR_LONG): Likewise.
21944 (BT_FN_ULONG_PTR_ULONG): Likewise.
21945 (BT_FN_VOID_UINT64_UINT64): Likewise.
21946 (BT_FN_UINT8_UINT8_UINT8): Likewise.
21947 (BT_FN_INT8_INT8_INT8): Likewise.
21948 (BT_FN_INT16_INT16_INT16): Likewise.
21949 (BT_FN_INT_INT_INT): Likewise.
21950 (BT_FN_UINT_FLOAT_UINT): Likewise.
21951 (BT_FN_FLOAT_UINT_UINT): Likewise.
21952 (BT_FN_ULONG_UINT_UINT): Likewise.
21953 (BT_FN_ULONG_UINT_PTR): Likewise.
21954 (BT_FN_ULONG_ULONG_ULONG): Likewise.
21955 (BT_FN_UINT_UINT_UINT): Likewise.
21956 (BT_FN_VOID_UINT_PTR): Likewise.
21957 (BT_FN_UINT_UINT_PTR: Likewise.
21958 (BT_FN_UINT32_UINT64_PTR): Likewise.
21959 (BT_FN_INT_INT_UINT_UINT): Likewise.
21960 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
21961 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
21962 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
21963 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
21964 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
21965 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
21966 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
21967 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
21968 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
21969 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
21970 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
21971 * doc/frontends.texi: List BRIG FE.
21972 * doc/install.texi (Testing): Add BRIG tesring requirements.
21973 * doc/invoke.texi (Overall Options): Mention BRIG.
21974 * doc/standards.texi (Standards): Doucment BRIG HSA version.
21976 2017-01-24 Richard Biener <rguenther@suse.de>
21978 PR translation/79208
21979 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
21981 2017-01-24 Martin Jambor <mjambor@suse.cz>
21984 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
21985 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
21986 and known_contexts.
21988 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
21990 PR middle-end/79123
21991 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
21992 casts from signed to unsigned really don't have a range.
21994 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
21996 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
21997 GMP_RNDx for compatiblity.
21999 2017-01-24 Martin Liska <mliska@suse.cz>
22002 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
22003 that would prevent us to call alloca with -1 as argument.
22005 2017-01-24 Jakub Jelinek <jakub@redhat.com>
22007 * dwarf2out.c (output_compilation_unit_header, output_file_names):
22008 Avoid -Wformat-security warning.
22010 2017-01-23 Andrew Pinski <apinski@cavium.com>
22012 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
22015 2017-01-23 Martin Sebor <msebor@redhat.com>
22017 PR middle-end/78703
22018 * gimple-ssa-sprintf.c (warn_level): New global.
22019 (format_integer): Use it here and throughout the rest of the file.
22020 Use the same switch to compute sign as base.
22021 (maybe_warn): New function.
22022 (format_directive): Factor out warnings into maybe_warn.
22023 Add debugging output. Use warn_level.
22024 (add_bytes): Use warn_level.
22025 (pass_sprintf_length::compute_format_length): Add debugging output.
22026 (try_substitute_return_value): Same.
22027 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
22029 PR middle-end/78703
22030 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
22031 (struct fmtresult, format_integer, format_floating): Adjust.
22032 (fmtresult::fmtresult): Set max correctly in two argument ctor.
22033 (get_string_length, format_string,format_directive): Same.
22034 (pass_sprintf_length::compute_format_length): Same.
22035 (try_substitute_return_value): Simplify slightly.
22037 PR middle-end/78703
22038 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
22039 (fmtresult::operator+=): Outlined.
22040 (struct fmtresult): Add ctors.
22041 (struct conversion_spec): Rename...
22042 (struct directive): ...to this. Add and remove data members.
22043 (directive::set_width, directive::set_precision): New functions.
22044 (format_percent): Use fmtresult ctor.
22045 (get_width_and_precision): Remove.
22046 (format_integer): Make naming changes. Avoid computing width and
22048 (format_floating): Same. Adjust indentation.
22049 (format_character, format_none): New functions.
22050 (format_string): Moved character handling to format_character.
22051 (format_directive): Remove arguments, change return type.
22052 (parse_directive): New function.
22053 (pass_sprintf_length::compute_format_length): Move directive
22054 parsing to parse_directive.
22056 2017-01-23 Jakub Jelinek <jakub@redhat.com>
22058 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
22059 (assign_assembler_name_if_needed): ... this.
22060 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
22061 (assign_assembler_name_if_needed): ... this.
22062 (free_lang_data_in_cgraph): Adjust callers.
22063 * cgraphunit.c (cgraph_node::analyze): Likewise.
22064 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
22067 2017-01-23 Richard Biener <rguenther@suse.de>
22069 PR tree-optimization/79088
22070 PR tree-optimization/79188
22071 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
22072 resetting loop bounds after last path deletion. Reset loop
22073 bounds of the target loop, make code match the comments.
22074 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
22075 Make sure loops need no fixups.
22077 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
22079 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
22080 exponent support with double type for first argument.
22081 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
22082 type returned by __builtin_vec_extract_sig,
22083 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
22084 functions from "vector int" to "vector unsigned int" or from
22085 "vector long long int" to "vector unsigned long long int".
22086 Changed type returned by __builtin_vec_extract_exp,
22087 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
22088 functions from "vector int" to "vector unsigned int" or from
22089 "vector long long int" to "vector unsigned long long int".
22090 Changed return type of __builtin_vec_test_data_class,
22091 __builtin_vec_test_data_class_sp, and
22092 __builtin_vec_test_data_class_dp from "vector int" to
22093 "vector bool int" or from "vector long long int" to "vector bool
22094 long long int" and changed second argument type from "unsigned
22095 int" to "int". Added new overloaded function forms "vector float
22096 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
22097 "vector float __builtin_vec_insert_exp_sp (vector float, vector
22098 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
22099 double, vector unsigned long long int)" and "vector double
22100 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
22101 long int)". Changed return type of
22102 __builtin_scalar_test_data_class and
22103 __builtin_scalar_test_data_class_sp and
22104 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
22105 int" and changed second argument from "unsigned int" to "int".
22106 Changed type returned by __builtin_scalar_test_neg,
22107 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
22108 from "int" to "bool int". Added new overloaded function form
22109 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
22110 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
22111 exponent double-precision with floating point first argument.
22112 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
22113 documentation of scalar_test_data_class, scalar_test_neg,
22114 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
22115 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
22116 vec_test_data_class built-in functions to reflect refinements in
22117 their type signatures.
22119 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
22121 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
22123 (aarch64_elf_asm_destructor): Likewise.
22125 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
22127 PR rtl-optimization/78634
22128 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
22129 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
22130 * ifcvt.c (noce_try_cmove): Add missing cost check.
22132 PR rtl-optimization/71724
22133 * combine.c (if_then_else_cond): Look for situations where it is
22134 beneficial to undo the work of one of the recursive calls.
22136 2017-01-23 Bin Cheng <bin.cheng@arm.com>
22138 PR tree-optimization/70754
22139 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
22140 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
22141 combined stmt before it if not NULL.
22142 (combine_chains): Process refs reversely and compute dominance point
22145 2017-01-23 Martin Liska <mliska@suse.cz>
22147 PR tree-optimization/79196
22148 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
22149 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
22151 (strlen_optimize_stmt): Call the renamed function.
22153 2017-01-23 Michael Matz <matz@suse.de>
22155 PR tree-optimization/78384
22156 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
22158 2017-01-23 Richard Biener <rguenther@suse.de>
22160 PR tree-optimization/79186
22161 * tree-vrp.c (register_new_assert_for): Make sure we've seen
22162 both incoming edges before moving an assert.
22164 2017-01-23 Martin Jambor <mjambor@suse.cz>
22166 * ipa-prop.c (load_from_param_1): Removed.
22167 (load_from_unmodified_param): Bits from load_from_param_1 put back
22169 (load_from_param): Removed.
22170 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
22171 with stmt. Reverted back to use of load_from_unmodified_param.
22173 2017-01-23 Martin Jambor <mjambor@suse.cz>
22176 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
22177 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
22178 field a pointer to garbage collected vector, mark lattices and
22179 ipcp_orig_node with GTY((skip)).
22180 (ipa_get_param_count): Adjust to descriptors being a pointer.
22181 (ipa_get_param): Likewise.
22182 (ipa_get_type): Likewise.
22183 (ipa_get_param_move_cost): Likewise.
22184 (ipa_set_param_used): Likewise.
22185 (ipa_get_controlled_uses): Likewise.
22186 (ipa_set_controlled_uses): Likewise.
22187 (ipa_is_param_used): Likewise.
22188 (ipa_node_params_t): Move into garbage collector. New methods insert
22190 (ipa_node_params_sum): Annotate wth GTY(()).
22191 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
22193 (ipa_load_from_parm_agg): Adjust declaration.
22194 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
22195 * ipa-profile.c (ipa_profile): Likewise.
22196 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
22197 (ipa_populate_param_decls): Make descriptors parameter garbage
22199 (ipa_dump_param): Adjust to descriptors being a pointer.
22200 (ipa_alloc_node_params): Likewise.
22201 (ipa_initialize_node_params): Likewise.
22202 (load_from_param_1): Make descriptors parameter garbage collected.
22203 (load_from_unmodified_param): Likewise.
22204 (load_from_param): Likewise.
22205 (ipa_load_from_parm_agg): Likewise.
22206 (ipa_node_params::~ipa_node_params): Removed.
22207 (ipa_free_all_node_params): Remove call to delete operator.
22208 (ipa_node_params_t::insert): New.
22209 (ipa_node_params_t::remove): Likewise.
22210 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
22211 copy known_csts and known_contexts vectors.
22212 (ipa_read_node_info): Adjust to descriptors being a pointer.
22213 (ipcp_modif_dom_walker): Make m_descriptors field garbage
22215 (ipcp_transform_function): Make descriptors variable garbage
22218 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
22220 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
22221 * config/i386/avx512dqintrin.h: Ditto.
22222 * config/i386/avx512fintrin.h: Ditto.
22223 * config/i386/i386.c: Handle new builtins.
22224 * config/i386/i386-builtin.def: Add new builtins.
22225 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
22226 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
22228 2017-01-23 Jakub Jelinek <jakub@redhat.com>
22229 Martin Liska <mliska@suse.cz>
22231 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
22232 * asan.c (asan_expand_poison_ifn): Support stores and use
22233 appropriate ASAN report function.
22234 * internal-fn.c (expand_ASAN_POISON_USE): New function.
22235 * internal-fn.def (ASAN_POISON_USE): Declare.
22236 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
22237 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
22238 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
22239 ASAN_POISON calls w/o LHS.
22240 * tree-ssa.c (execute_update_addresses_taken): Create clobber
22241 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
22242 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
22243 * gimplify.c (asan_poison_variables): Add attribute
22244 use_after_scope_memory to variables that really needs to live
22246 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
22247 having the attribute.
22249 2017-01-23 Martin Liska <mliska@suse.cz>
22251 * asan.c (create_asan_shadow_var): New function.
22252 (asan_expand_poison_ifn): Likewise.
22253 * asan.h (asan_expand_poison_ifn): New declaration.
22254 * internal-fn.c (expand_ASAN_POISON): Likewise.
22255 * internal-fn.def (ASAN_POISON): New builtin.
22256 * sanopt.c (pass_sanopt::execute): Expand
22257 asan_expand_poison_ifn.
22258 * tree-inline.c (copy_decl_for_dup_finish): Make function
22260 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
22261 * tree-ssa.c (is_asan_mark_p): New function.
22262 (execute_update_addresses_taken): Rewrite local variables
22263 (identified just by use-after-scope as addressable) into SSA.
22265 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
22267 * doc/install.texi (Specific): opensource.apple.com uses https
22268 now. Remove trailing slash.
22270 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
22272 * README.Portability: Remove note on an Irix compatibility issue.
22274 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
22276 * gcov.c (INCLUDE_ALGORITHM): Define.
22277 (INCLUDE_VECTOR): Define.
22278 No longer include <vector> and <algorithm> directly.
22280 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
22282 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
22284 * doc/invoke.texi (Code Gen Options): Ditto.
22286 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
22289 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
22291 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
22293 rtl-optimization/79125
22294 * cprop.c (local_cprop_pass): Handle cases where we make an
22295 unconditional trap.
22297 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
22301 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
22302 read from, for big endian.
22304 2017-01-20 Jiong Wang <jiong.wang@arm.com>
22306 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
22307 register pauth builtins for LP64 only.
22309 2017-01-20 Marek Polacek <polacek@redhat.com>
22312 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
22315 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
22317 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
22319 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
22320 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
22321 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
22323 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22326 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
22327 in big-endian mode when they are not a single duplicated value.
22329 2017-01-20 Richard Biener <rguenther@suse.de>
22331 * BASE-VER: Bump to 7.0.1.
22333 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
22335 * omp-low.c (omplow_simd_context): New struct. Use it...
22336 (lower_rec_simd_input_clauses): ...here and...
22337 (lower_rec_input_clauses): ...here to hold common data. Adjust all
22338 references to idx, lane, max_vf, is_simt.
22340 2017-01-20 Graham Markall <graham.markall@embecosm.com>
22342 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
22345 2017-01-20 Martin Jambor <mjambor@suse.cz>
22347 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
22348 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
22350 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
22351 (GTFILES): Rename hsa.c to hsa-common.c.
22352 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
22353 * hsa-dump.c: Likewise.
22354 * hsa-gen.c: Likewise.
22355 * hsa-regalloc.c: Likewise.
22356 * ipa-hsa.c: Likewise.
22357 * omp-expand.c: Likewise.
22358 * omp-low.c: Likewise.
22359 * toplev.c: Likewise.
22361 2017-01-20 Marek Polacek <polacek@redhat.com>
22364 * doc/invoke.texi: Document -Wduplicated-branches.
22365 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
22366 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
22367 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
22368 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
22369 return 0 only when not OEP_LEXICOGRAPHIC.
22370 (fold_build_cleanup_point_expr): Use the expression
22371 location when building CLEANUP_POINT_EXPR.
22372 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
22373 * tree.c (add_expr): Handle error_mark_node.
22375 2017-01-20 Martin Liska <mliska@suse.cz>
22378 * tree-profile.c (init_ic_make_global_vars): Do not call
22380 (gimple_init_gcov_profiler): Likewise.
22382 2017-01-20 Martin Liska <mliska@suse.cz>
22385 * cgraph.h (maybe_create_reference): Remove argument and
22387 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
22389 * ipa-cp.c (create_specialized_node): Likewise.
22390 * symtab.c (symtab_node::maybe_create_reference): Handle
22391 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
22393 2017-01-20 Martin Liska <mliska@suse.cz>
22395 * read-rtl-function.c (function_reader::create_function): Use
22396 build_decl instread of build_decl_stat.
22398 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
22400 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
22401 * config/i386/avx512dqintrin.h: Ditto.
22402 * config/i386/avx512fintrin.h: Ditto.
22403 * config/i386/i386-builtin-types.def: Add new types.
22404 * config/i386/i386.c: Handle new types.
22405 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
22406 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
22407 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
22408 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
22409 (__builtin_ia32_kshiftridi): New.
22410 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
22412 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
22416 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
22417 define to rs6000_init_stack_protect_guard.
22418 (rs6000_init_stack_protect_guard): New function.
22420 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
22421 Yunqiang Su <yunqiang.su@imgtec.com>
22423 * config.gcc (supported_defaults): Add madd4.
22424 (with_madd4): Add validation.
22425 (all_defaults): Add madd4.
22426 * config/mips/mips.opt (mmadd4): New option.
22427 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
22429 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
22431 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
22432 (ISA_HAS_FUSED_MADD4): Likewise.
22433 * doc/invoke.texi (-mmadd4): Document the new option.
22434 * doc/install.texi (--with-madd4): Document the new option.
22436 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22438 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
22439 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
22440 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
22441 (aarch64_init_pauth_hint_builtins): New.
22442 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
22443 (aarch64_expand_builtin): Expand new builtins.
22445 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22447 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
22448 * combine-stack-adj.c (no_unhandled_cfa): Handle
22449 REG_CFA_TOGGLE_RA_MANGLE.
22450 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
22451 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
22452 info for return address signing.
22453 (aarch64_expand_epilogue): Likewise.
22455 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22457 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
22458 * config/aarch64/aarch64-protos.h
22459 (aarch64_return_address_signing_enabled): New declaration.
22460 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
22462 (aarch64_expand_prologue): Sign return address before it's pushed onto
22464 (aarch64_expand_epilogue): Authenticate return address fetched from
22466 (aarch64_override_options): Sanity check for ILP32 and ISA level.
22467 (aarch64_attributes): New function attributes for "sign-return-address".
22468 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
22469 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
22470 ("*do_return"): Generate combined instructions according to key index.
22471 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
22472 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
22474 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
22475 * config/aarch64/aarch64.opt (msign-return-address=): New.
22476 * doc/extend.texi (AArch64 Function Attributes): Documents
22477 "sign-return-address=".
22478 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
22480 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
22482 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
22483 overall option summary.
22485 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22487 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
22488 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
22489 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
22490 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
22492 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
22494 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
22495 -mpower9-minmax by default for -mcpu=power9.
22496 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
22497 128-bit floating point.
22499 2017-01-20 Alan Modra <amodra@gmail.com>
22501 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
22502 optimizing for size.
22504 2017-01-20 Alan Modra <amodra@gmail.com>
22507 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
22508 for strcmp and strncmp from corresponding builtin decl.
22510 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
22512 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
22513 instead of i386/rtems-64.h.
22514 * config/i386/rtems-64.h: Remove.
22516 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
22520 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
22522 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
22524 2017-01-19 Tamar Christina <tamar.christina@arm.com>
22526 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
22527 Change int to HOST_WIDE_INT.
22528 * config/aarch64/aarch64-protos.h
22529 (aarch64_simd_gen_const_vector_dup): Likewise.
22530 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
22532 2017-01-19 David Malcolm <dmalcolm@redhat.com>
22534 * langhooks-def.h (lhd_type_for_size): New decl.
22535 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
22536 * langhooks.c (lhd_type_for_size): New function, taken from
22539 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
22541 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
22542 define_bypass for CR latency.
22543 (power9-cracked-alu): Update bypass latency and remove power9-branch.
22544 (power9-alu2): Add define_bypass for CR latency.
22546 (power9-mul): Update insn latency.
22547 (power9-mul-compare): Update insn latency, bypass latency and remove
22550 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22552 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
22554 * config/aarch64/aarch64.md
22555 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
22556 aarch64_nopcrelative_literal_loads.
22557 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
22559 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
22561 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
22562 TARGET_LOONGSON_3A.
22563 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
22565 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
22568 * config.gcc (supported_defaults): Add lxc1-sxc1.
22569 (with_lxc1_sxc1): Add validation.
22570 (all_defaults): Add lxc1-sxc1.
22571 * config/mips/mips.opt (mlxc1-sxc1): New option.
22572 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
22574 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
22575 __mips_no_lxc1_sxc1.
22576 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
22577 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
22578 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
22580 2017-01-19 Richard Biener <rguenther@suse.de>
22582 PR tree-optimization/72488
22583 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
22584 sure to restore SSA info.
22585 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
22587 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
22589 PR rtl-optimization/79121
22590 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
22591 of the inner type when shifting an extended value.
22593 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
22596 * symtab.c (symtab_node::equal_address_to): Fix comparing of
22597 interposable aliases.
22599 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
22602 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
22603 Use the evmergelohi instruction.
22604 (mov_si<mode>_e500_subreg4_2_le): Likewise.
22605 (mov_sitf_e500_subreg8_2_be): Likewise.
22606 (mov_sitf_e500_subreg12_2_le): Likewise.
22607 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
22608 (mov_si<mode>_e500_subreg4_2_be): Likewise.
22609 (mov_sitf_e500_subreg8_2_le): Likewise.
22610 (mov_sitf_e500_subreg12_2_be): Likewise.
22612 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22614 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
22615 attribute from vecsimple to vecperm.
22616 (altivec_vbpermq2): Likewise.
22618 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22621 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
22623 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22624 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
22625 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
22626 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
22627 case where N arg is SIZE_MAX.
22628 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
22629 (cmpstrsi): Add pattern.
22631 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22633 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22634 __builtin_vec_revb builtins.
22635 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
22636 built-in functions to support generation of the ISA 3.0 XXBR<x>
22637 vector byte reverse instructions.
22638 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
22639 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
22640 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
22641 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
22642 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
22643 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
22644 (P9V_BUILTIN_VEC_REVB): Likewise.
22645 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
22646 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
22647 (p9_xxbrq_v16qi): Likewise.
22648 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
22649 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
22650 (p9_xxbrh_v8hi): Likewise.
22651 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
22652 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
22653 vec_revb built-in functions.
22655 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
22657 PR rtl-optimization/78952
22658 * config/i386/i386.md (any_extract): New code iterator.
22659 (*insvqi_2): Use any_extract for source operand.
22660 (*insvqi_3): Use any_shiftrt for source operand.
22662 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
22664 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
22666 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
22668 2017-01-18 Matthias Klose <doko@ubuntu.com>
22670 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
22672 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22674 * config/rs6000/altivec.h (vec_bperm): Change #define.
22675 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
22676 (altivec_vbpermq2): New define_insn.
22677 (altivec_vbpermd): Likewise.
22678 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
22679 function interface.
22680 (VBPERMD): Likewise.
22681 (VBPERM): New polymorphic function interface.
22682 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
22683 Add entries for P9V_BUILTIN_VEC_VBPERM.
22684 * doc/extend.texi: Add interfaces for vec_bperm.
22686 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22688 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
22689 first letter of error messages.
22690 (s390_resolve_overloaded_builtin): Likewise.
22691 * config/s390/s390.c (s390_expand_builtin): Likewise.
22692 (s390_invalid_arg_for_unprototyped_fn): Likewise.
22693 (s390_valid_target_attribute_inner_p): Likewise.
22694 * config/s390/s390.md ("tabort"): Likewise.
22696 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
22698 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
22699 (ISA_AVOID_DIV_HILO): New macro.
22700 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
22701 (ISA_HAS_DDIV): Likewise.
22703 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22705 * doc/invoke.texi (fabi-version): Correct number of occurrences.
22707 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22709 * doc/invoke.texi (fabi-version): Spelling fix.
22711 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22714 * doc/invoke.texi (fabi-version): Mention mangling fix for
22717 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22720 * doc/invoke.texi (fabi-version): Document discriminator mangling.
22722 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
22725 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
22726 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
22728 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
22730 (stack_protect_test): Ditto.
22731 * config/rs6000/rs6000.opt (mstack-protector-guard=,
22732 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
22734 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
22735 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
22736 -mstack-protector-guard-offset=.
22737 (RS/6000 and PowerPC Options): Ditto.
22739 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
22741 * config/i386/i386.h (MASK_CLASS_P): New define.
22742 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
22743 there are no registers from different register sets also when
22744 mask registers are used. Update function comment.
22745 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
22746 to (*k/*r) and (*k/*km) alternatives.
22748 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
22750 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
22751 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
22752 (EH_RETURN_HANDLER_RTX): New define.
22753 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
22754 Force frame pointer in EH return functions.
22755 (aarch64_expand_epilogue): Add barrier for eh_return.
22756 (aarch64_final_eh_return_addr): Remove.
22757 (aarch64_eh_return_handler_rtx): New function.
22758 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
22760 (aarch64_eh_return_handler_rtx): New prototype.
22762 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22764 * config/rs6000/altivec.h (vec_rlmi): New #define.
22765 (vec_vrlnm): Likewise.
22766 (vec_rlnm): Likewise.
22767 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
22768 (UNSPEC_VRLNM): Likewise.
22769 (VIlong): New mode iterator.
22770 (altivec_vrl<VI_char>mi): New define_insn.
22771 (altivec_vrl<VI_char>nm): Likewise.
22772 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
22774 (VRLDNM): Likewise.
22775 (RLNM): New polymorphic function entry.
22776 (VRLWMI): New monomorphic function entry.
22777 (VRLDMI): Likewise.
22778 (RLMI): New polymorphic function entry.
22779 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
22780 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
22781 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
22784 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22787 * dwarf2out.c (field_byte_offset): Restore the
22788 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
22789 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
22790 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
22793 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
22796 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
22798 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22801 * dwarf2out.c (add_data_member_location_attribute): For constant
22802 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
22803 instead of DW_AT_data_member_location, DW_AT_bit_offset and
22804 DW_AT_byte_size attributes.
22806 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
22808 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
22809 after forcing to constant memory when the code model is medium.
22811 2017-01-17 Julia Koval <julia.koval@intel.com>
22814 * config/i386/avx512fintrin.h
22815 (_mm512_i32gather_ps): Change __addr type to void const*.
22816 (_mm512_mask_i32gather_ps): Ditto.
22817 (_mm512_i32gather_pd): Ditto.
22818 (_mm512_mask_i32gather_pd): Ditto.
22819 (_mm512_i64gather_ps): Ditto.
22820 (_mm512_mask_i64gather_ps): Ditto.
22821 (_mm512_i64gather_pd): Ditto.
22822 (_mm512_mask_i64gather_pd): Ditto.
22823 (_mm512_i32gather_epi32): Ditto.
22824 (_mm512_mask_i32gather_epi32): Ditto.
22825 (_mm512_i32gather_epi64): Ditto.
22826 (_mm512_mask_i32gather_epi64): Ditto.
22827 (_mm512_i64gather_epi32): Ditto.
22828 (_mm512_mask_i64gather_epi32): Ditto.
22829 (_mm512_i64gather_epi64): Ditto.
22830 (_mm512_mask_i64gather_epi64): Ditto.
22831 (_mm512_i32scatter_ps): Change __addr type to void*.
22832 (_mm512_mask_i32scatter_ps): Ditto.
22833 (_mm512_i32scatter_pd): Ditto.
22834 (_mm512_mask_i32scatter_pd): Ditto.
22835 (_mm512_i64scatter_ps): Ditto.
22836 (_mm512_mask_i64scatter_ps): Ditto.
22837 (_mm512_i64scatter_pd): Ditto.
22838 (_mm512_mask_i64scatter_pd): Ditto.
22839 (_mm512_i32scatter_epi32): Ditto.
22840 (_mm512_mask_i32scatter_epi32): Ditto.
22841 (_mm512_i32scatter_epi64): Ditto.
22842 (_mm512_mask_i32scatter_epi64): Ditto.
22843 (_mm512_i64scatter_epi32): Ditto.
22844 (_mm512_mask_i64scatter_epi32): Ditto.
22845 (_mm512_i64scatter_epi64): Ditto.
22846 (_mm512_mask_i64scatter_epi64): Ditto.
22847 * config/i386/avx512pfintrin.h
22848 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
22849 (_mm512_mask_prefetch_i32gather_ps): Ditto.
22850 (_mm512_mask_prefetch_i64gather_pd): Ditto.
22851 (_mm512_mask_prefetch_i64gather_ps): Ditto.
22852 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
22853 (_mm512_prefetch_i32scatter_ps): Ditto.
22854 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
22855 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
22856 (_mm512_prefetch_i64scatter_pd): Ditto.
22857 (_mm512_prefetch_i64scatter_ps): Ditto.
22858 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
22859 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
22860 * config/i386/avx512vlintrin.h
22861 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
22862 (_mm_mmask_i32gather_ps): Ditto.
22863 (_mm256_mmask_i32gather_pd): Ditto.
22864 (_mm_mmask_i32gather_pd): Ditto.
22865 (_mm256_mmask_i64gather_ps): Ditto.
22866 (_mm_mmask_i64gather_ps): Ditto.
22867 (_mm256_mmask_i64gather_pd): Ditto.
22868 (_mm_mmask_i64gather_pd): Ditto.
22869 (_mm256_mmask_i32gather_epi32): Ditto.
22870 (_mm_mmask_i32gather_epi32): Ditto.
22871 (_mm256_mmask_i32gather_epi64): Ditto.
22872 (_mm_mmask_i32gather_epi64): Ditto.
22873 (_mm256_mmask_i64gather_epi32): Ditto.
22874 (_mm_mmask_i64gather_epi32): Ditto.
22875 (_mm256_mmask_i64gather_epi64): Ditto.
22876 (_mm_mmask_i64gather_epi64): Ditto.
22877 (_mm256_i32scatter_ps): Change __addr type to void*.
22878 (_mm256_mask_i32scatter_ps): Ditto.
22879 (_mm_i32scatter_ps): Ditto.
22880 (_mm_mask_i32scatter_ps): Ditto.
22881 (_mm256_i32scatter_pd): Ditto.
22882 (_mm256_mask_i32scatter_pd): Ditto.
22883 (_mm_i32scatter_pd): Ditto.
22884 (_mm_mask_i32scatter_pd): Ditto.
22885 (_mm256_i64scatter_ps): Ditto.
22886 (_mm256_mask_i64scatter_ps): Ditto.
22887 (_mm_i64scatter_ps): Ditto.
22888 (_mm_mask_i64scatter_ps): Ditto.
22889 (_mm256_i64scatter_pd): Ditto.
22890 (_mm256_mask_i64scatter_pd): Ditto.
22891 (_mm_i64scatter_pd): Ditto.
22892 (_mm_mask_i64scatter_pd): Ditto.
22893 (_mm256_i32scatter_epi32): Ditto.
22894 (_mm256_mask_i32scatter_epi32): Ditto.
22895 (_mm_i32scatter_epi32): Ditto.
22896 (_mm_mask_i32scatter_epi32): Ditto.
22897 (_mm256_i32scatter_epi64): Ditto.
22898 (_mm256_mask_i32scatter_epi64): Ditto.
22899 (_mm_i32scatter_epi64): Ditto.
22900 (_mm_mask_i32scatter_epi64): Ditto.
22901 (_mm256_i64scatter_epi32): Ditto.
22902 (_mm256_mask_i64scatter_epi32): Ditto.
22903 (_mm_i64scatter_epi32): Ditto.
22904 (_mm_mask_i64scatter_epi32): Ditto.
22905 (_mm256_i64scatter_epi64): Ditto.
22906 (_mm256_mask_i64scatter_epi64): Ditto.
22907 (_mm_i64scatter_epi64): Ditto.
22908 (_mm_mask_i64scatter_epi64): Ditto.
22909 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
22910 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
22911 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
22912 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
22913 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
22914 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
22915 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
22916 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
22917 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
22918 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
22919 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
22920 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
22921 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
22922 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
22923 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
22924 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
22925 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
22926 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
22927 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
22928 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
22929 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
22930 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
22931 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
22932 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
22933 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
22934 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
22935 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
22936 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
22937 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
22938 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
22939 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
22940 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
22941 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
22942 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
22943 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
22944 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
22945 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
22946 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
22947 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
22948 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
22949 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
22950 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
22951 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
22952 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
22953 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
22954 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
22955 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
22956 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
22957 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
22958 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
22959 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
22960 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
22961 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
22962 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
22963 definitions accordingly.
22965 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
22966 Kuan-Lin Chen <kuanlinchentw@gmail.com>
22969 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
22972 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
22975 * ira-conflicts.c (ira_build_conflicts): Update total conflict
22976 hard regs for inner regno.
22978 2017-01-17 Martin Liska <mliska@suse.cz>
22981 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
22982 assumption and add comment.
22984 2017-01-17 Nathan Sidwell <nathan@acm.org>
22986 * ipa-visibility.c (localize_node): New function, broken out of ...
22987 (function_and_variable_visibility): ... here. Call it.
22989 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
22991 PR middle-end/77445
22992 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
22993 correctly set frequency of oudgoing edge.
22994 (duplicate_thread_path): Fix profile updating.
22996 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22999 * configure.ac: Add GCC_BASE_VER.
23000 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
23001 version from BASE-VER file.
23002 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
23003 (gcc.o): Depend on $(BASEVER).
23004 * common.opt (dumpfullversion): New option.
23005 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
23006 * doc/invoke.texi: Document -dumpfullversion.
23007 * doc/install.texi: Document --with-gcc-major-version-only.
23008 * configure: Regenerated.
23010 2017-01-17 Richard Biener <rguenther@suse.de>
23012 PR tree-optimization/71433
23013 * tree-vrp.c (register_new_assert_for): Merge same asserts
23014 on all incoming edges.
23015 (process_assert_insertions_for): Handle insertions at the
23018 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
23020 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
23021 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
23023 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
23026 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
23029 2017-01-17 Alan Modra <amodra@gmail.com>
23032 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
23033 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
23034 symbolic stack limit when pic.
23036 2017-01-16 Martin Sebor <msebor@redhat.com>
23038 PR tree-optimization/78608
23039 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
23041 2017-01-16 Jeff Law <law@redhat.com>
23044 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
23045 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
23046 for several include directories that may be relative to sysroot.
23047 * config/i386/x-mingw32 (gplus_includedir): Define.
23048 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
23049 (native_system_includedir): Likewise.
23050 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
23051 override if TARGET_SYSTEM_ROOT is defined.
23052 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
23054 PR tree-optimization/79090
23055 PR tree-optimization/33562
23056 PR tree-optimization/61912
23057 PR tree-optimization/77485
23058 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
23059 and computed trims into the dump file.
23061 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
23063 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
23065 2017-01-16 Jakub Jelinek <jakub@redhat.com>
23068 * gimplify.c (gimplify_init_constructor): If want_value and
23069 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
23073 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
23074 sequence. Formatting fixes.
23075 (doloop_optimize): Formatting fixes.
23078 * gcc.c (debug_level_greater_than_spec_func): New function.
23079 (static_spec_functions): Add debug-level-gt spec function.
23080 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
23082 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
23083 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
23084 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
23085 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
23086 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
23087 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
23089 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
23091 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
23092 QImode fixups to general and mask registers only.
23094 2017-01-16 Carl Love <cel@us.ibm.com>
23096 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
23097 for built-in functions
23098 vector signed char vec_nabs (vector signed char)
23099 vector signed short vec_nabs (vector signed short)
23100 vector signed int vec_nabs (vector signed int)
23101 vector signed long long vec_nabs (vector signed long long)
23102 vector float vec_nabs (vector float)
23103 vector double vec_nabs (vector double)
23104 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
23106 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
23107 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
23108 * doc/extend.texi: Update the documentation file for the new built-in
23111 2017-01-16 Martin Sebor <msebor@redhat.com>
23113 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
23116 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23118 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
23119 UNSPEC_VSX__XXSPLTD to require special splat handling.
23121 2017-01-16 David Malcolm <dmalcolm@redhat.com>
23124 * system.h: Poison strndup.
23126 2017-01-16 Alan Modra <amodra@gmail.com>
23129 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
23132 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
23134 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
23136 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
23138 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
23139 call recog here. Assert that INSN_CODE (insn) is non-negative.
23141 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
23144 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
23146 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
23147 in the currently scheduled RTL fragment.
23149 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
23151 PR rtl-optimization/78751
23152 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
23155 2017-01-14 Jeff Law <law@redhat.com>
23157 PR tree-optimization/79090
23158 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
23159 variable length stores.
23160 (compute_trims): Delete dead assignment to *trim_tail.
23161 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
23164 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
23166 PR rtl-optimization/78626
23167 PR rtl-optimization/78727
23168 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
23169 of a block, and split such blocks after everything else is finished.
23171 2017-01-14 Alan Modra <amodra@gmail.com>
23174 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
23175 target legitimate_combined_insn.
23176 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
23177 (rs6000_legitimate_combined_insn): New function.
23178 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
23180 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
23181 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
23182 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
23184 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
23186 * doc/frontends.texi (G++ and GCC): Remove references to Java.
23188 2017-01-13 Jeff Law <law@redhat.com>
23190 PR tree-optimization/33562
23191 PR tree-optimization/61912
23192 PR tree-optimization/77485
23193 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
23195 (delete_dead_assignment): Likewise.
23196 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
23197 statement to delete_dead_call and delete_dead_assignment.
23199 2017-01-13 David Malcolm <dmalcolm@redhat.com>
23202 * substring-locations.c (format_warning_va): Strengthen case 1 so
23203 that both endpoints of the substring must be within the format
23204 range for just the substring to be printed.
23206 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
23208 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
23209 * config/i386/i386.c (ix86_target_string): Add missing options
23210 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
23211 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
23212 flags_other and ix86_target_other to flags2_other. Display unknown
23214 (ix86_valid_target_attribute_inner_p): Add missing options and
23215 reorder options by implied ISAs, as in ix86_target_string.
23217 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
23219 * hash-table.h (hash_table::too_empty_p): New function.
23220 (hash_table::expand): Use it.
23221 (hash_table::traverse): Likewise.
23222 (hash_table::empty_slot): Use sizeof (value_type) instead of
23223 sizeof (PTR) to convert bytes to elements. Shrink the table
23224 if the current size is excessive for the current number of
23227 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
23229 * ira-costs.c (record_reg_classes): Break from the inner loop
23230 early once alt_fail is known to be true. Update outer loop
23231 handling accordingly.
23233 2017-01-13 Jeff Law <law@redhat.com>
23235 * tree-ssa-dse.c (decrement_count): New function.
23236 (increment_start_addr, maybe_trim_memstar_call): Likewise.
23237 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
23238 when we know the partially dead statement is a mem* function.
23240 PR tree-optimization/61912
23241 PR tree-optimization/77485
23242 * tree-ssa-dse.c: Include expr.h.
23243 (maybe_trim_constructor_store): New function.
23244 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
23246 PR tree-optimization/33562
23247 PR tree-optimization/61912
23248 PR tree-optimization/77485
23249 * doc/invoke.texi: Document new dse-max-object-size param.
23250 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
23251 * tree-ssa-dse.c: Include params.h.
23252 (dse_store_status): New enum.
23253 (initialize_ao_ref_for_dse): New, partially extracted from
23255 (valid_ao_ref_for_dse, normalize_ref): New.
23256 (setup_live_bytes_from_ref, compute_trims): Likewise.
23257 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
23258 (maybe_trim_partially_dead_store): Likewise.
23259 (maybe_trim_complex_store): Likewise.
23260 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
23261 Track what bytes live from the original store. Return tri-state
23262 for dead, partially dead or live.
23263 (dse_dom_walker): Add constructor, destructor and new private members.
23264 (delete_dead_call, delete_dead_assignment): New extracted from
23266 (dse_optimize_stmt): Make a member of dse_dom_walker.
23267 Use initialize_ao_ref_for_dse.
23269 PR tree-optimization/33562
23270 PR tree-optimization/61912
23271 PR tree-optimization/77485
23272 * sbitmap.h (bitmap_count_bits): Prototype.
23273 (bitmap_clear_range, bitmap_set_range): Likewise.
23274 * sbitmap.c (bitmap_clear_range): New function.
23275 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
23277 2017-01-13 Martin Liska <mliska@suse.cz>
23280 * function.c (set_cfun): Add new argument force.
23281 * function.h (set_cfun): Likewise.
23282 * ipa-inline-transform.c (inline_call): Use the function when
23283 strict alising from is dropped for function we inline to.
23285 2017-01-13 Richard Biener <rguenther@suse.de>
23287 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
23288 for dumping GIMPLE INTEGER_CSTs.
23290 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23292 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
23293 to 201112L since C++17.
23295 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23298 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
23299 if -fsanitize=kernel-address is present.
23301 2017-01-13 Richard Biener <rguenther@suse.de>
23303 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
23304 as _Literal ( type ) number in case usual suffixes do not
23305 preserve all information.
23307 2017-01-13 Richard Biener <rguenther@suse.de>
23309 PR tree-optimization/77283
23310 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
23311 and ssa-iterators.h.
23312 (is_feasible_trace): Implement a cost model based on joiner
23315 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23318 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
23319 char or short to __float128/_Float128 directly.
23321 2017-01-12 Martin Sebor <msebor@redhat.com>
23323 to -Wformat-overflow.
23324 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
23325 (min_bytes_remaining): Same.
23326 (get_string_length): Same.
23327 (format_string): Same.
23328 (format_directive): Same.
23330 (pass_sprintf_length::handle_gimple_call): Same.
23332 2017-01-12 Jakub Jelinek <jakub@redhat.com>
23334 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
23335 info.nowrite calls with no lhs that can't throw. Return bool
23336 whether gsi_remove has been called or not.
23337 (pass_sprintf_length::handle_gimple_call): Return bool whether
23338 try_substitute_return_value called gsi_remove. Formatting fix.
23339 (pass_sprintf_length::execute): Don't use gsi_remove if
23340 handle_gimple_call returned true.
23343 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
23344 be removed due to side-effects, don't remove following barrier nor
23345 turn the successor edge into fallthru edge.
23347 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23350 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
23351 element-reversing loads and stores as not swappable.
23353 2017-01-12 Nathan Sidwell <nathan@acm.org>
23354 Nicolai Stange <nicstange@gmail.com>
23356 * combine.c (try_combine): Don't ignore result of overlap checking
23357 loop. Combine overlap & asm check into single loop.
23359 2017-01-12 Richard Biener <rguenther@suse.de>
23361 * tree-pretty-print.c (dump_generic_node): Provide -gimple
23362 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
23364 2017-01-12 Richard Biener <rguenther@suse.de>
23366 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
23367 and TS_TARGET_OPTION directly derive from TS_BASE.
23368 * tree-core.h (tree_optimization_option): Derive from tree_base.
23369 (tree_target_option): Likewise.
23371 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
23373 * config/i386/i386.c (memory_address_length): Increase len
23374 only when rip_relative_addr_p returns false.
23376 2017-01-11 Julia Koval <julia.koval@intel.com>
23378 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
23379 (OPTION_MASK_ISA_SGX_SET): New.
23380 (ix86_handle_option): Handle OPT_msgx.
23381 * config.gcc: Added sgxintrin.h.
23382 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
23383 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
23384 * config/i386/i386.c (ix86_target_string): Add -msgx.
23386 (ix86_option_override_internal): Handle new options.
23387 (ix86_valid_target_attribute_inner_p): Add sgx.
23388 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
23389 * config/i386/i386.opt: Add msgx.
23390 * config/i386/sgxintrin.h: New file.
23391 * config/i386/x86intrin.h: Add sgxintrin.h.
23393 2017-01-11 Jakub Jelinek <jakub@redhat.com>
23396 * fold-const.c (maybe_nonzero_address): Return 1 for function
23398 (tree_single_nonzero_warnv_p): Don't handle function local objects
23402 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
23405 2017-01-11 David Malcolm <dmalcolm@redhat.com>
23408 * opts.c: Include "spellcheck.h"
23409 (struct string_fragment): New struct.
23410 (struct edit_distance_traits<const string_fragment &>): New
23412 (get_closest_sanitizer_option): New function.
23413 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
23415 2017-01-11 Jakub Jelinek <jakub@redhat.com>
23417 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
23419 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
23420 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
23421 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
23422 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
23423 for initial die_offset if dwarf_split_debug_info.
23424 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
23425 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
23427 (output_skeleton_debug_sections): Formatting fix. Use
23428 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
23429 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
23431 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
23433 * config/arm/cortex-a53.md: Add bypasses for
23434 cortex_a53_r2f_cvt.
23435 (cortex_a53_r2f): Only use for transfers.
23436 (cortex_a53_f2r): Likewise.
23437 (cortex_a53_r2f_cvt): Add reservation for conversions.
23438 (cortex_a53_f2r_cvt): Likewise.
23440 2017-01-11 Tamar Christina <tamar.christina@arm.com>
23442 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
23443 to all inlined functions, change static to extern.
23445 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
23448 * config/arm/arm.c (legitimize_pic_address): Handle reference to
23450 (arm_assemble_integer): Likewise.
23452 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
23454 * config.gcc: Use new awk script to check CPU, FPU and architecture
23455 parameters for --with-... options.
23456 * config/arm/parsecpu.awk: New file
23457 * config/arm/arm-cpus.in: New file.
23458 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
23460 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
23462 * config/arm/t-arm: Update dependency rules.
23463 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
23464 of processing .def files.
23465 * config/arm/genopt.sh: Deleted.
23466 * config/arm/gentune.sh: Deleted.
23467 * config/arm/arm-cores.def: Deleted.
23468 * config/arm/arm-arches.def: Deleted.
23469 * config/arm/arm-fpus.def: Deleted.
23470 * config/arm/arm-tune.md: Regenerated.
23471 * config/arm/arm-tables.opt: Regenerated.
23472 * config/arm/arm-cpu.h: New generated file.
23473 * config/arm/arm-cpu-data.h: New generated file.
23474 * config/arm/arm-cpu-cdata.h: New generated file.
23476 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
23479 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
23481 (input_varpool_node): Unpack dynamically_initialized bit.
23483 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
23485 PR rtl-optimization/79032
23486 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
23487 the alignment of the adjusted memory reference against that of MODE,
23488 instead of the alignment of the original memory reference.
23490 2017-01-11 Martin Jambor <mjambor@suse.cz>
23492 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
23494 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
23495 decorated functions.
23497 2017-01-11 Richard Biener <rguenther@suse.de>
23499 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
23500 set range/nonnull info for PHI results. Do not set it on
23501 stmts marked for removal.
23503 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
23505 * expr.c (store_field): In the bitfield case, fetch the return value
23506 from the registers before applying a single big-endian adjustment.
23507 Always do a final load for a BLKmode value not larger than a word.
23509 2017-01-10 David Malcolm <dmalcolm@redhat.com>
23512 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
23513 that we correctly handle column numbers greater than
23514 LINE_MAP_MAX_COLUMN_NUMBER.
23516 2017-01-10 Martin Sebor <msebor@redhat.com>
23518 PR middle-end/78245
23519 * gimple-ssa-sprintf.c (get_destination_size): Call
23520 {init,fini}object_sizes.
23521 * tree-object-size.c (addr_object_size): Adjust.
23522 (pass_through_call): Adjust.
23523 (pass_object_sizes::execute): Adjust.
23524 * tree-object-size.h (fini_object_sizes): Declare.
23526 2017-01-10 Martin Sebor <msebor@redhat.com>
23528 PR tree-optimization/78775
23529 * builtins.c (get_size_range): Move...
23530 * calls.c: ...to here.
23531 (alloc_max_size): Accept zero argument.
23532 (operand_signed_p): Remove.
23533 (maybe_warn_alloc_args_overflow): Call get_size_range.
23534 * calls.h (get_size_range): Declare.
23536 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
23538 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
23539 from TI's devices.csv file as of September 2016.
23540 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
23542 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
23544 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
23545 * doc/invoke.texi: Likewise.
23546 * doc/md.texi: Likewise.
23547 * doc/objc.texi: Likewise.
23549 2017-01-10 Joshua Conner <joshconner@google.com>
23551 * config/arm/fuchsia-elf.h: New file.
23552 * config/fuchsia.h: New file.
23553 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
23554 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
23556 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
23558 2016-01-10 Richard Biener <rguenther@suse.de>
23560 PR tree-optimization/79034
23561 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
23562 Propagate out degenerate PHIs in the joiner.
23564 2017-01-10 Martin Liska <mliska@suse.cz>
23566 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
23567 (sort_congruence_classes_by_decl_uid): Likewise.
23568 (sort_congruence_class_groups_by_decl_uid): Likewise.
23569 (sem_item_optimizer::merge_classes): Sort class, groups in these
23570 classes and members in the groups by DECL_UID of declarations.
23571 This would make merge operations stable.
23573 2017-01-10 Martin Liska <mliska@suse.cz>
23575 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
23576 usage of m_classes_vec.
23577 (sem_item_optimizer::~sem_item_optimizer): Likewise.
23578 (sem_item_optimizer::get_group_by_hash): Likewise.
23579 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
23580 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
23581 (sem_item_optimizer::verify_classes): Likewise.
23582 (sem_item_optimizer::process_cong_reduction): Likewise.
23583 (sem_item_optimizer::dump_cong_classes): Likewise.
23584 (sem_item_optimizer::merge_classes): Likewise.
23585 * ipa-icf.h (congruence_class_hash): Rename from
23586 congruence_class_group_hash. Remove declaration of m_classes_vec.
23588 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
23590 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
23591 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
23592 * config.gcc: Add avx512vpopcntdqintrin.h.
23593 * config/i386/avx512vpopcntdqintrin.h: New.
23594 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
23595 * config/i386/i386-builtin-types.def: Add new types.
23596 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
23597 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
23598 __builtin_ia32_vpopcountq_v8di_mask): New.
23599 * config/i386/i386-c.c (ix86_target_macros_internal): Define
23600 __AVX512VPOPCNTDQ__.
23601 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
23602 (PTA_AVX512VPOPCNTDQ): Define.
23603 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
23604 TARGET_AVX512VPOPCNTDQ_P): Define.
23605 * config/i386/i386.opt: Add mavx512vpopcntdq.
23606 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
23607 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
23609 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23611 PR middle-end/77484
23612 * predict.def (PRED_CALL): Set to 67.
23614 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
23616 * expr.c (store_field): In the bitfield case, if the value comes from
23617 a function call and is of an aggregate type returned in registers, do
23618 not modify the field mode; extract the value in all cases if the mode
23619 is BLKmode and the size is not larger than a word.
23621 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
23624 * config/i386/cpuid.h: Fix undefined behavior.
23626 2017-01-04 Jeff Law <law@redhat.com>
23628 PR tree-optimization/79007
23629 PR tree-optimization/67955
23630 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
23631 conservative for pt.null when flag_non_call_exceptions is on.
23633 2017-01-09 Jakub Jelinek <jakub@redhat.com>
23635 PR translation/79019
23636 PR translation/79020
23637 * params.def (PARAM_INLINE_MIN_SPEEDUP,
23638 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
23639 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
23641 * config/avr/avr.opt (maccumulate-args): Likewise.
23642 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
23643 * common.opt (freport-bug): Likewise.
23644 * cif-code.def (CIF_FINAL_ERROR): Likewise.
23645 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
23646 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
23647 translatable string.
23648 * config/i386/i386.c (function_value_32): Likewise.
23649 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
23650 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
23652 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
23653 * common/config/msp430/msp430-common.c (msp430_handle_option):
23655 * symtab.c (symtab_node::verify_base): Likewise.
23656 * opts.c (set_debug_level): Likewise.
23657 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
23658 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
23659 missing whitespace to translatable strings.
23660 * config/avr/avr.md (bswapsi2): Fix typo in comment.
23661 * config/sh/superh.h: Likewise.
23662 * config/i386/xopintrin.h: Likewise.
23663 * config/i386/znver1.md: Likewise.
23664 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
23665 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
23666 * double-int.h (struct double_int): Likewise.
23667 * double-int.c (div_and_round_double): Likewise.
23668 * wide-int.cc: Likewise.
23669 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
23670 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
23671 * cfgcleanup.c (crossjumps_occured): Renamed to ...
23672 (crossjumps_occurred): ... this.
23673 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
23676 PR tree-optimization/78899
23677 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
23678 returning bool return struct loop *, NULL for failure and the new
23680 (versionable_outer_loop_p): Don't version outer loop if it has
23681 dont_vectorized bit set.
23682 (tree_if_conversion): When versioning outer loop, ensure
23683 tree_if_conversion is performed also on the inner loop of the
23684 non-vectorizable outer loop copy.
23685 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
23686 LOOP_VECTORIZED in inner loop of the scalar outer loop and
23687 prevent vectorization of it.
23688 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
23689 the outer loop vectorization of the non-scalar version is attempted
23690 before vectorization of the inner loop in scalar version. If
23691 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
23692 vectorization of its inner loop.
23693 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
23694 has 2 inner loops, rename also on edges from bb whose single pred
23695 is outer_loop->header. Fix typo in function comment.
23697 2017-01-09 Martin Sebor <msebor@redhat.com>
23700 * asan.c (asan_emit_stack_protection): Increase local buffer size
23701 to avoid snprintf truncation warning.
23703 2017-01-09 Andrew Pinski <apinski@cavium.com>
23705 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
23706 to reference thunderx2t99 for the tuning structure
23707 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
23709 (thunderx2t99_extra_costs): This.
23710 * config/aarch64/aarch64-tune.md: Regenerate.
23711 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
23712 (vulcan_addrcost_table): This.
23713 (vulcan_regmove_cost): Rename to ...
23714 (thunderx2t99_regmove_cost): This.
23715 (vulcan_vector_cost): Rename to ...
23716 (thunderx2t99_vector_cost): this.
23717 (vulcan_branch_cost): Rename to ...
23718 (thunderx2t99_branch_cost): This.
23719 (vulcan_tunings): Rename to ...
23720 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
23721 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
23723 2017-01-09 Martin Jambor <mjambor@suse.cz>
23727 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
23728 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
23729 (propagate_vr_accross_jump_function): Use the above function for all
23730 value range computations for pass-through jump functions and type
23731 converasion from explicit value range values.
23732 (ipcp_propagate_stage): Do not attempt to deduce types of formal
23733 parameters from TYPE_ARG_TYPES.
23734 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
23735 (ipa_write_node_info): Stream type of the actual argument.
23736 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
23738 2017-01-09 Martin Liska <mliska@suse.cz>
23741 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
23742 (lookup_compiler): Do not show error message with have_E.
23744 2017-01-09 Jakub Jelinek <jakub@redhat.com>
23746 PR tree-optimization/78938
23747 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
23748 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
23749 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
23750 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
23753 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23755 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
23758 2017-01-09 Richard Biener <rguenther@suse.de>
23760 PR tree-optimization/78997
23761 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
23762 name condition properly.
23764 2017-01-09 Richard Biener <rguenther@suse.de>
23767 * dwarf2out.c (is_cxx): New overload with context.
23768 (is_naming_typedef_decl): Use it.
23770 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23772 * invoke.texi (Option Summary): Correct spacing in option lists
23773 and add line breaks to fix over-long lines.
23775 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23777 PR middle-end/17660
23779 * extend.texi (Common Variable Attributes): Add xref to GCC
23780 Internals manual to explain mode attribute keywords.
23782 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23785 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
23786 and Preprocessor Options.
23787 (Options for Linking): Document -pthread here....
23788 (RS/6000 and PowerPC Options): ...not here.
23789 (Solaris 2 Options): ...or here.
23790 * doc/cppopts.texi: Document -pthread.
23792 2017-01-08 Martin Sebor <msebor@redhat.com>
23794 PR middle-end/77708
23795 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
23796 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
23797 New member functions.
23798 (format_directive): Used them.
23800 (pass_sprintf_length::handle_gimple_call): Same.
23801 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
23802 to avoid truncation for any argument.
23803 (extract_affine_mul): Same.
23804 * tree.c (get_file_function_name): Same.
23806 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23808 PR middle-end/77484
23809 * predict.def (PRED_INDIR_CALL): Set to 86.
23811 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23813 PR preprocessor/54124
23814 * doc/cppopts.texi: Reformat -d subtable to list the full name
23815 of the options. Add cross-reference to the docs for the general
23816 compiler -d options.
23817 * doc/invoke.texi (Developer Options): Add cross-reference to the
23818 preprocessor-specific -d option documentation.
23820 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23822 PR preprocessor/13498
23823 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
23824 redudant material, and reflect new command-line options.
23825 (System Headers): Likewise.
23827 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23829 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
23830 -isystem, and -idirafter. Copy-edit.
23831 * doc/cppopts.texi: Copy-edit. Remove contradiction about
23832 default for -ftrack-macro-expansion. Delete obsolete and
23833 badly-formatted implementation details about -fdebug-cpp output.
23834 * doc/cppwarnopts.texi: Copy-edit.
23836 2017-01-07 David Malcolm <dmalcolm@redhat.com>
23839 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
23840 that the transition from a max line width >= 1<<10 to narrower
23841 lines works correctly.
23843 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
23845 * doc/options.texi (PerFunction): New.
23846 * opt-functions.awk (switch_flags): Map both Optimization and
23847 PerFunction to CL_OPTIMIZATION.
23848 * opth-gen.awk: Test for PerFunction flag along with
23850 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
23851 it only when the latter is present. Skip those that don't in
23852 the hash function generator.
23853 * common.opt (fvar-tracking): Mark as PerFunction instead of
23855 (fvar-tracking-assignments): Likewise.
23856 (fvar-tracking-assignments-toggle): Likewise.
23857 (fvar-tracking-uninit): Likewise.
23859 2017-01-07 Jakub Jelinek <jakub@redhat.com>
23861 PR translation/79018
23862 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
23865 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
23868 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
23869 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
23870 TARGET_LONG_JUMP_TABLE_OFFSETS.
23871 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
23872 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
23873 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
23874 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
23875 * config/m68k/m68k.md (tablejump expander): Likewise.
23876 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
23877 TARGET_LONG_JUMP_TABLE_OFFSETS.
23878 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
23879 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
23881 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
23882 David Holsgrove <david.holsgrove@xilinx.com>
23884 * common/config/microblaze/microblaze-common.c
23885 (TARGET_EXCEPT_UNWIND_INFO): Remove.
23886 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
23888 * config/microblaze/microblaze.c (microblaze_must_save_register)
23889 (microblaze_expand_epilogue, microblaze_return_addr): Handle
23891 (microblaze_eh_return): New function.
23892 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
23893 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
23894 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
23895 * config/microblaze/microblaze.md (eh_return): New pattern.
23897 2017-01-06 Jakub Jelinek <jakub@redhat.com>
23899 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
23900 GCC_DIAGNOSTIC_STRINGIFY): Define.
23902 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
23904 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23906 * config/arm/arm.md (<mcrr>): New.
23908 * config/arm/arm.c (arm_arch5te): New.
23909 (arm_option_override): Set arm_arch5te.
23910 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
23912 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
23913 (arm_mcrr_qualifiers): ... this. New.
23914 (MRRC_QUALIFIERS): Define to...
23915 (arm_mrrc_qualifiers): ... this. New.
23916 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
23918 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
23919 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
23920 (MRRCI, mrrc, MRRC): New.
23921 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
23922 VUNSPEC_MRRC2): New.
23924 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23926 * config/arm/arm.md (<mcr>): New.
23928 * config/arm/arm.c (arm_coproc_builtin_available): Add
23929 support for mcr, mrc, mcr2 and mrc2.
23930 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
23931 (arm_mcr_qualifiers): ... this. New.
23932 (MRC_QUALIFIERS): Define to ...
23933 (arm_mrc_qualifiers): ... this. New.
23934 (MCR_QUALIFIERS): Define to ...
23935 (arm_mcr_qualifiers): ... this. New.
23936 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
23938 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
23939 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
23940 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
23941 VUNSPEC_MRC2): New.
23943 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23945 * config/arm/arm.md (*ldc): New.
23949 * config/arm/arm.c (arm_coproc_builtin_available): Add
23950 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
23951 (arm_coproc_ldc_stc_legitimate_address): New.
23952 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
23953 'qualifier_const_pointer'.
23954 (LDC_QUALIFIERS): Define to...
23955 (arm_ldc_qualifiers): ... this. New.
23956 (STC_QUALIFIERS): Define to...
23957 (arm_stc_qualifiers): ... this. New.
23958 * config/arm/arm-protos.h
23959 (arm_coproc_ldc_stc_legitimate_address): New.
23960 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
23961 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
23962 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
23963 stc2, stcl, stc2l): New.
23964 * config/arm/constraints.md (Uz): New.
23965 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
23966 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
23967 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
23968 VUNSPEC_STC2L): New.
23970 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23972 * config/arm/arm.md (<cdp>): New.
23973 * config/arm/arm.c (neon_const_bounds): Rename this ...
23974 (arm_const_bounds): ... this.
23975 (arm_coproc_builtin_available): New.
23976 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
23977 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
23978 (CDP_QUALIFIERS): Define to...
23979 (arm_cdp_qualifiers): ... this. New.
23981 (arm_expand_builtin_args): Add case for 6 arguments.
23982 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
23983 (arm_const_bounds): ... this.
23984 (arm_coproc_builtin_available): New.
23985 * config/arm/arm_acle.h (__arm_cdp): New.
23987 * config/arm/arm_acle_builtins.def (cdp): New.
23989 * config/arm/iterators.md (CDPI,CDP,cdp): New.
23990 * config/arm/neon.md: Rename all 'neon_const_bounds' to
23991 'arm_const_bounds'.
23992 * config/arm/types.md (coproc): New.
23993 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
23994 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
23995 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
23996 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
23998 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24000 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
24001 (UBINOP_QUALIFIERS): New.
24003 (acle_builtin_data): New. Change comment.
24004 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
24005 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
24006 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
24007 arm_acle_builtins.def.
24008 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
24009 (arm_init_acle_builtins): New.
24010 (CRC32_BUILTIN): Remove.
24011 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
24012 crc32cb, crc32ch and crc32cw.
24013 (arm_init_crc32_builtins): Remove.
24014 (arm_init_builtins): Use arm_init_acle_builtins rather
24015 than arm_init_crc32_builtins.
24016 (arm_expand_acle_builtin): New.
24017 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
24018 * config/arm/arm_acle_builtins.def: New.
24020 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24022 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
24023 (arm_builtin_datum): ... this.
24024 (arm_init_neon_builtin): Rename to ...
24025 (arm_init_builtin): ... this. Add a new parameters PREFIX
24026 and USE_SIG_IN_NAME.
24027 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
24028 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
24029 'arm_builtin_datum'.
24030 (arm_init_vfp_builtins): Likewise.
24031 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
24032 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
24033 (arm_expand_neon_args): Rename to ...
24034 (arm_expand_builtin_args): ... this. Rename builtin_arg
24035 enum values and differentiate between ARG_BUILTIN_MEMORY
24036 and ARG_BUILTIN_NEON_MEMORY.
24037 (arm_expand_neon_builtin_1): Rename to ...
24038 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
24039 values, arm_expand_builtin_args and add bool parameter NEON.
24040 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
24041 (arm_expand_vfp_builtin): Likewise.
24042 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
24044 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24046 PR middle-end/77484
24047 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
24048 * predict.c (tree_estimate_probability_bb): Reverse direction of
24049 polymorphic call predictor.
24051 2017-01-06 David Malcolm <dmalcolm@redhat.com>
24053 * passes.c (execute_one_pass): Split out pass-skipping logic into...
24054 (determine_pass_name_match): ...this new function and...
24055 (should_skip_pass_p): ...this new function.
24057 2017-01-06 Nathan Sidwell <nathan@acm.org>
24059 * ipa-visibility.c (function_and_variable_visibility): Reformat
24060 comments and long lines. Remove extrneous if.
24061 * symtab.c (symtab_node::make_decl_local): Fix code format.
24062 (symtab_node::set_section_for_node): Fix comment typo.
24064 2017-01-06 Martin Liska <mliska@suse.cz>
24067 * lra-constraints.c: Rename invariant to lra_invariant.
24068 * predict.c (set_even_probabilities): Initialize e to NULL.
24070 2017-01-05 Martin Sebor <msebor@redhat.com>
24072 PR tree-optimization/78910
24073 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
24074 (format_integer): Correct off-by-one error in the handling
24075 of precision with negative numbers in signed conversions..
24077 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
24079 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
24081 2017-01-05 Jakub Jelinek <jakub@redhat.com>
24083 PR tree-optimization/71016
24084 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
24085 factor_out_conditional_conversion. Formatting fix.
24086 (factor_out_conditional_conversion): Add cond_stmt argument.
24087 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
24088 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
24091 2017-01-05 David Malcolm <dmalcolm@redhat.com>
24093 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
24094 read-rtl-function.o, and selftest-rtl.o.
24095 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
24096 (selftest::aarch64_test_loading_full_dump): New function.
24097 (selftest::aarch64_run_selftests): New function.
24098 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
24099 selftest::aarch64_run_selftests.
24100 * config/i386/i386.c
24101 (selftest::ix86_test_loading_dump_fragment_1): New function.
24102 (selftest::ix86_test_loading_call_insn): New function.
24103 (selftest::ix86_test_loading_full_dump): New function.
24104 (selftest::ix86_test_loading_unspec): New function.
24105 (selftest::ix86_run_selftests): Call the new functions.
24106 * emit-rtl.c (maybe_set_max_label_num): New function.
24107 * emit-rtl.h (maybe_set_max_label_num): New decl.
24108 * function.c (instantiate_decls): Guard call to
24109 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
24110 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
24112 * gensupport.c (gen_reader::gen_reader): Pass "false"
24113 for new "compact" param of rtx_reader.
24114 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
24115 rather than an empty string for NULL strings.
24116 * read-md.c: Potentially include config.h rather than bconfig.h.
24117 Wrap include of errors.h with #ifdef GENERATOR_FILE.
24118 (have_error): New global, copied from errors.c.
24119 (md_reader::read_name): Rename to...
24120 (md_reader::read_name_1): ...this, adding "out_loc" param,
24121 and converting "missing name or number" to returning false, rather
24123 (md_reader::read_name): Reimplement in terms of read_name_1.
24124 (md_reader::read_name_or_nil): New function.
24125 (md_reader::read_string): Handle "(nil)" by returning NULL.
24126 (md_reader::md_reader): Add new param "compact".
24127 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
24128 (md_reader::read_file): New method.
24129 * read-md.h (md_reader::md_reader): Add new param "compact".
24130 (md_reader::read_file): New method.
24131 (md_reader::is_compact): New accessor.
24132 (md_reader::read_name): Convert return type from void to file_location.
24133 (md_reader::read_name_or_nil): New decl.
24134 (md_reader::read_name_1): New decl.
24135 (md_reader::m_compact): New field.
24136 (noop_reader::noop_reader): Pass "false" for new "compact" param
24138 (rtx_reader::rtx_reader): Add new "compact" param.
24139 (rtx_reader::read_rtx_operand): Make virtual and convert return
24140 type from void to rtx.
24141 (rtx_reader::read_until): New decl.
24142 (rtx_reader::handle_any_trailing_information): New virtual function.
24143 (rtx_reader::postprocess): New virtual function.
24144 (rtx_reader::finalize_string): New virtual function.
24145 (rtx_reader::m_in_call_function_usage): New field.
24146 (rtx_reader::m_reuse_rtx_by_id): New field.
24147 * read-rtl-function.c: New file.
24148 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
24149 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
24150 (selftest::verify_three_block_rtl_cfg): New decl.
24151 * read-rtl-function.h: New file.
24152 * read-rtl.c: Potentially include config.h rather than bconfig.h.
24153 For host, include function.h, memmodel.h, and emit-rtl.h.
24154 (one_time_initialization): New function.
24155 (struct compact_insn_name): New struct.
24156 (compact_insn_names): New array.
24157 (find_code): Handle insn codes in compact dumps.
24158 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
24159 (bind_subst_iter_and_attr): Likewise.
24160 (add_condition_to_string): Likewise.
24161 (add_condition_to_rtx): Likewise.
24162 (apply_attribute_uses): Likewise.
24163 (add_current_iterators): Likewise.
24164 (apply_iterators): Likewise.
24165 (initialize_iterators): Guard usage of apply_subst_iterator with
24166 #ifdef GENERATOR_FILE.
24167 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
24168 (md_reader::read_mapping): Likewise.
24169 (add_define_attr_for_define_subst): Likewise.
24170 (add_define_subst_attr): Likewise.
24171 (read_subst_mapping): Likewise.
24172 (check_code_iterator): Likewise.
24173 (rtx_reader::read_rtx): Likewise. Move one-time initialization
24175 (one_time_initialization): New function.
24176 (rtx_reader::read_until): New method.
24177 (read_flags): New function.
24178 (parse_reg_note_name): New function.
24179 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
24180 Handle reuse_rtx ids.
24181 Wrap iterator lookup within #ifdef GENERATOR_FILE.
24182 Add parsing support for RTL dumps, mirroring the special-cases in
24183 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
24184 values, and calling handle_any_trailing_information.
24185 (rtx_reader::read_rtx_operand): Convert return type from void
24186 to rtx, returning return_rtx. Handle case 'e'. Call
24187 finalize_string on XSTR and XTMPL fields.
24188 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
24189 "(nil)" values were omitted. Call the postprocess vfunc on the
24191 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
24192 class ctor. Initialize m_in_call_function_usage. Call
24193 one_time_initialization.
24194 * rtl-tests.c (selftest::test_uncond_jump): Call
24195 set_new_first_and_last_insn.
24196 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
24197 * selftest-rtl.c: New file.
24198 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
24199 (selftest::get_insn_by_uid): New decl.
24200 * selftest-run-tests.c (selftest::run_tests): Call
24201 read_rtl_function_c_tests.
24202 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
24203 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
24206 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
24208 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
24209 operands in a special way. Assert that pos+len <= mode precision.
24211 2017-01-05 Jakub Jelinek <jakub@redhat.com>
24213 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
24214 3 argument Alias with unlimited for the negative form.
24215 (fno-vect-cost-model): Removed.
24217 2017-01-05 Martin Liska <mliska@suse.cz>
24219 * hsa-gen.c (gen_hsa_divmod): New function.
24220 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
24222 2017-01-05 Martin Liska <mliska@suse.cz>
24225 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
24228 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24230 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
24231 small constant length operands.
24233 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24235 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
24236 between loop iterations.
24238 2017-01-05 Martin Liska <mliska@suse.cz>
24241 * gimplify.c (gimplify_decl_expr): Compare to
24242 asan_poisoned_variables instread of checking flags.
24243 (gimplify_target_expr): Likewise.
24244 (gimplify_expr): Likewise.
24245 (gimplify_function_tree): Conditionally initialize
24246 asan_poisoned_variables.
24248 2017-01-04 Jeff Law <law@redhat.com>
24250 PR tree-optimizatin/78812
24251 * rtl.h (contains_mem_rtx_p): Prototype.
24252 * ifcvt.c (containts_mem_rtx_p): Move from here to...
24253 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
24254 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
24255 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
24256 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
24258 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24260 * input.c (assert_char_at_range): Default-initialize actual_range.
24262 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24264 * df-scan.c (df_ref_create_structure): Make regno unsigned,
24265 to match the caller.
24267 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24269 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
24270 insns after final jump in test to emit dummy move.
24272 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24274 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
24275 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
24277 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24279 * multiple_target.c (create_dispatcher_calls): Init e_next.
24280 * tree-ssa-loop-split.c (split_loop): Init border.
24281 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
24284 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
24289 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
24290 (altivec_register_operand): Do not return true if the operand
24291 contains a SUBREG mixing SImode and SFmode.
24292 (vsx_register_operand): Likewise.
24293 (vsx_reg_sfsubreg_ok): New predicate.
24294 (vfloat_operand): Do not return true if the operand contains a
24295 SUBREG mixing SImode and SFmode.
24296 (vint_operand): Likewise.
24297 (vlogical_operand): Likewise.
24298 (gpc_reg_operand): Likewise.
24299 (int_reg_operand): Likewise.
24300 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
24301 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
24302 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
24304 (rs6000_emit_move_si_sf_subreg): New helper function.
24305 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
24306 fixup SUBREGs involving SImode and SFmode.
24307 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
24308 numbers for the new peephole2 optimization.
24309 (peephole2 for SFmode unions): New peephole2 to optimize cases in
24310 the GLIBC math library that do AND/IOR/XOR operations on single
24311 precision floating point.
24312 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
24313 target macros to say whether we need to avoid SUBREGs mixing
24315 (TARGET_ALLOW_SF_SUBREG): Likewise.
24316 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
24317 (UNSPEC_SI_FROM_SF): Likewise.
24318 (iorxor): Change spacing.
24319 (and_ior_xor): New iterator for AND, IOR, and XOR.
24320 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
24321 (movdi_from_sf_zero_ext): Likewise.
24322 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
24323 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
24324 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
24325 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
24326 (fms<mode>4): Likewise.
24327 (fnma<mode>4): Likewise.
24328 (fnms<mode>4): Likewise.
24329 (nfma<mode>4): Likewise.
24330 (nfms<mode>4): Likewise.
24332 2017-01-04 Marek Polacek <polacek@redhat.com>
24335 * doc/invoke.texi: Document -Wpointer-compare.
24337 2017-01-04 Jakub Jelinek <jakub@redhat.com>
24339 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
24342 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
24343 descriptions for -gdwarf-5 and emit them as uleb128 instead of
24346 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
24349 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
24350 documentation of the powerpc_popcntb_ok attribute.
24351 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24352 code to issue warning messages if a requested CPU configuration is
24353 not supported by the binary (assembler and loader) toolchain.
24354 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
24355 made to define a built-in function that has been disabled.
24356 (paired_init_builtins): Add assertion to prevent ICE if attempt is
24357 made to define a built-in function that has been disabled.
24358 (altivec_init_builtins): Add comment explaining why definition
24359 of the DST built-in functions is not preceded by an assertion
24360 check. Add assertions to prevent ICE if attempts are made to
24361 define an altivec predicate or an abs* built-in function that has
24363 (htm_init_builtins): Add comment explaining why definition of the
24364 htm built-in functions is not preceded by an assertion check.
24366 2017-01-04 Jeff Law <law@redhat.com>
24368 PR tree-optimizatin/67955
24369 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
24370 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
24371 the points-to solution does not include pt_null. Use DECL_PT_UID
24374 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
24376 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
24377 Use gen_int_mode instead of gen_lopwart for const_int operands.
24379 2017-01-04 Jakub Jelinek <jakub@redhat.com>
24381 PR tree-optimization/71563
24382 * match.pd: Simplify X << Y into X if Y is known to be 0 or
24383 out of range value - has low bits known to be zero.
24385 2017-01-04 Alan Modra <amodra@gmail.com>
24387 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
24388 * configure: Regenerate.
24389 * config.in: Regenerate.
24391 2017-01-04 Jakub Jelinek <jakub@redhat.com>
24394 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
24395 a substring of the message, but strcmp with the whole message. Ifdef
24396 ENABLE_NLS, translate the message first using dgettext.
24398 2017-01-03 Jeff Law <law@redhat.com>
24400 PR tree-optimizatin/78856
24401 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
24402 (mark_threaded_blocks): Remove code to truncate thread paths that
24403 cross multiple loop headers. Instead invalidate the cached loop
24404 iteration information and handle case of a thread path walking
24405 into an irreducible region.
24407 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
24410 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
24411 assertions. Add support for doing the signbit if the IEEE 128-bit
24412 floating point value is in a GPR.
24413 * config/rs6000/rs6000.md (Fsignbit): Delete.
24414 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
24415 Update the length attribute if the value is in a GPR.
24416 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
24417 the sign or zero extension instruction, since the value is always 0/1.
24418 (signbit<mode>2_dm2): Delete using <Fsignbit>.
24421 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
24422 extracting SImode to a GPR register so that we can generate a
24423 store, limit the vector to be in a traditional Altivec register
24424 for the vextuwrx instruction.
24426 2017-01-03 Ian Lance Taylor <iant@google.com>
24428 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
24430 2017-01-03 Martin Sebor <msebor@redhat.com>
24432 PR tree-optimization/78696
24433 * gimple-ssa-sprintf.c (format_floating): Correct handling of
24434 precision. Use MPFR for %f for greater fidelity. Correct handling
24436 (pass_sprintf_length::compute_format_length): Set width and precision
24437 specified by asrerisk to void_node for vararg functions.
24438 (try_substitute_return_value): Adjust dump output.
24440 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
24442 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
24444 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
24446 * doc/invoke.texi (SPARC options): Document -mlra as the default.
24447 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
24448 -mlra/-mno-lra was passed to the compiler.
24450 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
24452 PR rtl-optimization/65618
24453 * emit-rtl.c (try_split): Move initialization of "before" and
24454 "after" to just before the call to emit_insn_after_setloc.
24456 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
24458 * doc/md.texi (Standard Names): Remove reference to Java frontend.
24460 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
24462 * dwarf2out.c (gen_enumeration_type_die): When
24463 -gno-strict-dwarf, add a DW_AT_encoding attribute.
24465 2017-01-03 Jakub Jelinek <jakub@redhat.com>
24467 PR tree-optimization/78965
24468 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
24469 Change first argument from const call_info & to call_info &. For %n
24470 set info.nowrite to false.
24472 PR middle-end/78901
24473 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
24474 possibly throwing calls.
24476 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
24477 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
24478 and fns handling, rather than in a separate case SSA_NAME.
24480 2017-01-02 Jeff Law <law@redhat.com>
24482 * config/darwin-driver.c (darwin_driver_init): Const-correctness
24483 fixes for first_period and second_period variables.
24485 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
24488 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
24489 (*insvqi_1): New insn pattern.
24490 (*insvqi_1_mem_rex64): Ditto.
24491 (*insvqi_2): Ditto.
24492 (*insvqi_3): Rename from *insvqi.
24494 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
24496 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
24498 * doc/cfg.texi (Edges): Remove reference to Java.
24499 (Maintaining the CFG): Ditto.
24501 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24503 PR middle-end/77674
24504 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
24505 transparent aliases.
24507 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24509 PR middle-end/77484
24510 * predict.def (PRED_CALL): Update hitrate.
24511 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
24512 * predict.c (tree_estimate_probability_bb): Split CALL predictor
24513 into direct/indirect/polymorphic variants.
24515 2017-01-01 Jakub Jelinek <jakub@redhat.com>
24517 Update copyright years.
24519 * gcc.c (process_command): Update copyright notice dates.
24520 * gcov-dump.c (print_version): Ditto.
24521 * gcov.c (print_version): Ditto.
24522 * gcov-tool.c (print_version): Ditto.
24523 * gengtype.c (create_file): Ditto.
24524 * doc/cpp.texi: Bump @copying's copyright year.
24525 * doc/cppinternals.texi: Ditto.
24526 * doc/gcc.texi: Ditto.
24527 * doc/gccint.texi: Ditto.
24528 * doc/gcov.texi: Ditto.
24529 * doc/install.texi: Ditto.
24530 * doc/invoke.texi: Ditto.
24532 Copyright (C) 2017 Free Software Foundation, Inc.
24534 Copying and distribution of this file, with or without modification,
24535 are permitted in any medium without royalty provided the copyright
24536 notice and this notice are preserved.