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 * target.def (hard_regno_call_part_clobbered): New hook.
6 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
7 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
8 * doc/tm.texi: Regenerate.
9 * hooks.h (hook_bool_uint_mode_false): Declare.
10 * hooks.c (hook_bool_uint_mode_false): New function.
11 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
12 * cselib.c (cselib_process_insn): Use
13 targetm.hard_regno_call_part_clobbered instead of
14 HARD_REGNO_CALL_PART_CLOBBERED.
15 * ira-conflicts.c (ira_build_conflicts): Likewise.
16 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17 * lra-constraints.c (need_for_call_save_p): Likewise.
18 * lra-lives.c: Include target.h.
19 (check_pseudos_live_through_calls): Use
20 targetm.hard_regno_call_part_clobbered instead of
21 HARD_REGNO_CALL_PART_CLOBBERED.
22 * regcprop.c: Include target.h.
23 (copyprop_hardreg_forward_1): Use
24 targetm.hard_regno_call_part_clobbered instead of
25 HARD_REGNO_CALL_PART_CLOBBERED.
26 * reginfo.c (choose_hard_reg_mode): Likewise.
27 * regrename.c (check_new_reg_p): Likewise.
28 * reload.c (find_equiv_reg): Likewise.
29 * reload1.c (emit_reload_insns): Likewise.
30 * sched-deps.c (deps_analyze_insn): Likewise.
31 * sel-sched.c (init_regs_for_mode): Likewise.
32 (mark_unavailable_hard_regs): Likewise.
33 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
34 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
35 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
37 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
38 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
39 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
41 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
43 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
44 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
45 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
47 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
48 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
49 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
51 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
52 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
54 * config/powerpcspe/powerpcspe.c
55 (rs6000_hard_regno_call_part_clobbered): New function.
56 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
57 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
58 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
60 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
61 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
62 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
64 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
65 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
66 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
68 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
69 Alan Hayward <alan.hayward@arm.com>
70 David Sherwood <david.sherwood@arm.com>
72 * rtl.h (subreg_memory_offset): Declare.
73 * emit-rtl.c (subreg_memory_offset): New function.
74 * expmed.c (store_bit_field_1): Use it.
75 * expr.c (undefined_operand_subword_p): Likewise.
76 * simplify-rtx.c (simplify_subreg): Likewise.
78 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
80 PR rtl-optimization/57448
83 * optabs.c (expand_atomic_load): Place compiler memory barriers if
84 using atomic_load pattern.
85 (expand_atomic_store): Likewise.
87 2017-09-04 Jakub Jelinek <jakub@redhat.com>
90 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
91 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
92 handling. Use replace_call_with_value with NULL instead of
93 gsi_replace, unlink_stmt_vdef and release_defs.
95 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
98 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
100 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
103 * rtl.h (emit_library_call_value_1): Declare.
104 (emit_library_call): Replace declaration with a series of overloads.
105 Remove the parameter count argument.
106 (emit_library_call_value): Likewise.
107 * calls.c (emit_library_call_value_1): Make global. Replace varargs
108 with an "rtx_mode_t *".
109 (emit_library_call_value): Delete.
110 (emit_library_call): Likewise.
111 * asan.c (asan_emit_stack_protection): Update calls accordingly.
112 (asan_emit_allocas_unpoison): Likewise.
113 * builtins.c (expand_builtin_powi): Likewise.
114 (expand_asan_emit_allocas_unpoison): Likewise.
115 * cfgexpand.c (expand_main_function): Likewise.
116 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
117 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
118 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
119 * config/arm/arm.c (arm_trampoline_init): Likewise.
120 (arm_call_tls_get_addr): Likewise.
121 (arm_expand_divmod_libfunc): Likewise.
122 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
123 (smulsi3_highpart): Likewise.
124 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
125 (c6x_expand_compare): Likewise.
126 (c6x_expand_movmem): Likewise.
127 * config/frv/frv.c (frv_trampoline_init): Likewise.
128 * config/i386/i386.c (ix86_trampoline_init): Likewise.
129 (ix86_expand_divmod_libfunc): Likewise.
130 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
131 (ia64_expand_compare): Likewise.
132 (ia64_profile_hook): Likewise.
133 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
134 (nonlocal_goto): Likewise.
135 (restore_stack_nonlocal): Likewise.
136 * config/m32r/m32r.c (block_move_call): Likewise.
137 (m32r_trampoline_init): Likewise.
138 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
139 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
140 (m68k_call_m68k_read_tp): Likewise.
141 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
142 (microblaze_expand_divide): Likewise.
143 * config/mips/mips.h (mips_args): Likewise.
144 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
145 (MIPS_ICACHE_SYNC): Likewise.
146 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
147 (nios2_trampoline_init): Likewise.
148 * config/pa/pa.c (hppa_tls_call): Likewise.
149 (pa_trampoline_init): Likewise.
150 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
151 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
152 (expand_strn_compare): Likewise.
153 (rs6000_generate_compare): Likewise.
154 (rs6000_expand_float128_convert): Likewise.
155 (output_profile_hook): Likewise.
156 (rs6000_trampoline_init): Likewise.
157 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
158 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
159 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
160 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
161 (rs6000_generate_compare): Likewise.
162 (rs6000_expand_float128_convert): Likewise.
163 (output_profile_hook): Likewise.
164 (rs6000_trampoline_init): Likewise.
165 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
166 * config/sh/sh.c (sh_trampoline_init): Likewise.
167 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
168 (sparc_emit_float_lib_cmp): Likewise.
169 (sparc32_initialize_trampoline): Likewise.
170 (sparc64_initialize_trampoline): Likewise.
171 (sparc_profile_hook): Likewise.
172 * config/spu/spu.c (ea_load_store): Likewise.
173 * config/spu/spu.md (floatunssidf2): Likewise.
174 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
175 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
176 * config/visium/visium.c (expand_block_move_4): Likewise.
177 (expand_block_move_2): Likewise.
178 (expand_block_move_1): Likewise.
179 (expand_block_set_4): Likewise.
180 (expand_block_set_2): Likewise.
181 (expand_block_set_1): Likewise.
182 (visium_trampoline_init): Likewise.
183 (visium_profile_hook): Likewise.
184 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
185 (xtensa_setup_frame_addresses): Likewise.
186 (xtensa_trampoline_init): Likewise.
187 * except.c (sjlj_emit_function_enter): Likewise.
188 (sjlj_emit_function_exit): Likewise.
189 * explow.c (allocate_dynamic_stack_space): Likewise.
190 (probe_stack_range): Likewise.
191 * expr.c (convert_mode_scalar): Likewise.
192 * optabs.c (expand_binop): Likewise.
193 (expand_twoval_binop_libfunc): Likewise.
194 (expand_unop): Likewise.
195 (prepare_cmp_insn): Likewise.
196 (prepare_float_lib_cmp): Likewise.
197 (expand_float): Likewise.
198 (expand_fix): Likewise.
199 (expand_fixed_convert): Likewise.
200 (maybe_emit_sync_lock_test_and_set): Likewise.
201 (expand_atomic_compare_and_swap): Likewise.
202 (expand_mem_thread_fence): Likewise.
203 (expand_atomic_fetch_op): Likewise.
205 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
207 * doc/generic.texi (OpenACC): Adjust URL.
208 * doc/invoke.texi (C Dialect Options): Ditto.
210 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
212 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
213 predicate for operand 1. Add (m,<S>) constraint.
214 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
215 Prevent memory operand 1 with register operand 2.
217 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
219 PR rtl-optimization/82024
220 * combine.c (try_combine): If the combination result is a PARALLEL,
221 and we only need to retain the SET in there that would be placed
222 at I2, check that we can place that at I3 instead, before doing so.
224 2017-09-01 Jakub Jelinek <jakub@redhat.com>
227 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
229 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
230 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
233 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
234 Jeff Law <law@redhat.com>
236 * varasm.c (bss_initializer_p): Do not put constants into .bss
237 (categorize_decl_for_section): Handle bss_initializer_p returning
238 false when DECL_INITIAL is NULL.
240 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
243 * config/s390/s390.c (s390_can_inline_p): New function.
245 2017-09-01 Jeff Law <law@redhat.com>
247 PR tree-optimization/82052
248 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
249 Always initialize the returned slot after a hash table miss
252 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
254 * config/s390/s390.md (mem_signal_fence): Remove.
255 * doc/md.texi (mem_signal_fence): Remove.
256 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
258 * target-insns.def (mem_signal_fence): Remove.
260 2017-09-01 Jakub Jelinek <jakub@redhat.com>
263 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
266 * asan.c (create_odr_indicator): Strip name encoding from assembler
267 name before appending it after __odr_asan_.
269 2017-09-01 Martin Liska <mliska@suse.cz>
271 PR tree-optimization/82059
272 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
273 frequency only when an edge is redirected.
275 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
277 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
278 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
279 (arc_conditional_register_usage): Remove ARC600 lp_count
281 (arc_file_start): Emit Tag_ARC_CPU_variation.
282 (arc_can_use_doloop_p): New conditions to use ZOLs.
283 (hwloop_fail): New function.
284 (hwloop_optimize): Likewise.
285 (hwloop_pattern_reg): Likewise.
286 (arc_doloop_hooks): New struct, to be used with reorg_loops.
287 (arc_reorg_loops): New function, calls reorg_loops.
288 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
289 (arc600_corereg_hazard): Remove ZOL checking, case handled by
291 (arc_loop_hazard): Remove function, functionality moved into
293 (arc_hazard): Remove arc_loop_hazard call.
294 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
295 into hwloop_optimize.
296 (arc_label_align): Remove ZOL handling.
297 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
298 * config/arc/arc.md (doloop_begin): Remove pattern.
299 (doloop_begin_i): Likewise.
300 (doloop_end_i): Likewise.
301 (doloop_fallback): Likewise.
302 (doloop_fallback_m): Likewise.
303 (doloop_end): Reimplement expand.
304 (arc_lp): New pattern for LP instruction.
305 (loop_end): New pattern.
306 (loop_fail): Likewise.
307 (decrement_and_branch_until_zero): Likewise.
308 * config/arc/arc.opt (mlpc-width): New option.
309 * doc/invoke.texi (mlpc-width): Document option.
311 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
313 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
314 (arc_ccfsm_advance): Fix checking for delay slots.
315 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
317 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
319 * config/arc/arc.md (movqi_insn): Add stores to save constant long
321 (movhi_insn): Update store instruction constraint which are saving
322 6-bit short immediates.
323 (movsi_insn): Consider also short scaled load operations.
324 (zero_extendhisi2_i): Use Usd constraint instead of T.
325 (extendhisi2_i): Add q constraint.
326 (arc_clzsi2): Add type and length attributes.
327 (arc_ctzsi2): Likewise.
328 * config/arc/constraints.md (Usc): Update constraint, the
329 assembler can parse two relocations for a single instruction.
331 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
333 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
334 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
336 2017-08-31 Olivier Hainque <hainque@adacore.com>
338 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
339 match as powerpc-wrs-vxworks*.
341 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
343 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
344 register constraint for by-element operand.
345 (aarch64_mls_elt_merge<mode>): Likewise.
347 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
349 * config/arc/arc.c (arc_can_follow_jump): Check for short
352 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
354 * config.gcc: Use g.opt for arc.
355 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
356 functionality moved to ...
357 (legitimate_scaled_address_p): New function, ...here.
358 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
359 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
360 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
362 (arc_override_options): Handle G option.
363 (arc_output_pic_addr_const): Correct function definition.
364 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
365 (arc_decl_anon_ns_mem_p): Delete.
366 (arc_in_small_data_p): Overhaul this function to take into
367 consideration the value given via G option.
368 (arc_rewrite_small_data_1): Renamed and corrected old
369 arc_rewrite_small_data function.
370 (arc_rewrite_small_data): New function.
371 (small_data_pattern): Don't use pic_offset_table_rtx.
372 * config/arc/arc.h (CC1_SPEC): Recognize G option.
373 * config/arc/simdext.md (movmisalignv2hi): Use
374 prepare_move_operands function.
376 (movmisalign*): Likewise.
377 * doc/invoke.texi (ARC options): Document -G option.
379 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
381 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
383 * config/arc/arc.c (arc_print_operand): Output scalled address for
384 sdata whenever is possible.
385 (arc_in_small_data_p): Allow sdata for 64bit datum when double
386 load/stores are available.
387 (compact_sda_memory_operand): Check for the alignment required by
388 code density instructions.
389 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
391 * config/arc/constraints.md (Usd): Update constraint.
392 (Us0): New constraint.
393 (Usc): Update constraint.
395 2017-08-31 Richard Biener <rguenther@suse.de>
398 * dwarf2out.c (dwarf2out_early_global_decl): Process each
401 2017-08-31 Tamar Christina <tamar.christina@arm.com>
403 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
404 Resize type_signature.
406 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
407 Alan Hayward <alan.hayward@arm.com>
408 David Sherwood <david.sherwood@arm.com>
410 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
411 subregs whose inner modes can be stored in GPRs.
412 (aarch64_classify_index): Likewise.
414 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
415 Alan Hayward <alan.hayward@arm.com>
416 David Sherwood <david.sherwood@arm.com>
418 * config/aarch64/iterators.md (V_cmp_result): Rename to...
419 (V_INT_EQUIV): ...this.
420 (v_cmp_result): Rename to...
421 (v_int_equiv): ...this.
422 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
423 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
424 (copysign<mode>3): Likewise.
425 (aarch64_simd_bsl<mode>_internal): Likewise.
426 (aarch64_simd_bsl<mode>): Likewise.
427 (vec_cmp<mode><mode>): Likewise.
428 (vcond<mode><mode>): Likewise.
429 (vcond<v_cmp_mixed><mode>): Likewise.
430 (vcondu<mode><v_cmp_mixed>): Likewise.
431 (aarch64_cm<optab><mode>): Likewise.
432 (aarch64_cmtst<mode>): Likewise.
433 (aarch64_fac<optab><mode>): Likewise.
434 (vec_perm_const<mode>): Likewise.
435 (vcond_mask_<mode><v_cmp_result>): Rename to...
436 (vcond_mask_<mode><v_int_equiv>): ...this.
437 (vec_cmp<mode><v_cmp_result>): Rename to...
438 (vec_cmp<mode><v_int_equiv>): ...this.
440 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
441 Alan Hayward <alan.hayward@arm.com>
442 David Sherwood <david.sherwood@arm.com>
444 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
446 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
447 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
448 (UNSPEC_LD4_DREG): New unspecs.
449 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
450 (aarch64_ld2<mode>_dreg_be): Replace with...
451 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
453 (aarch64_ld3<mode>_dreg_le)
454 (aarch64_ld3<mode>_dreg_be): Replace with...
455 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
457 (aarch64_ld4<mode>_dreg_le)
458 (aarch64_ld4<mode>_dreg_be): Replace with...
459 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
462 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
464 PR tree-optimization/81987
465 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
466 insert an initializer in a location not dominated by the stride
469 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
471 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
472 on the entire header of the finally block in the fallthru case.
474 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
476 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
478 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
480 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
481 rs6000_emit_move_from_cr and call renamed function.
482 (rs6000_emit_prologue): Call renamed functions.
483 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
484 movesi_from_cr, remove volatile CRs.
486 2017-08-30 Jon Beniston <jon@beniston.com>
487 Richard Biener <rguenther@suse.de>
489 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
490 of VECTOR_MODE_P check.
491 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
492 element vector types.
494 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
496 * df.h (df_read_modify_subreg_p): Remove in favor of...
497 * rtl.h (read_modify_subreg_p): ...this new function. Take a
498 const_rtx instead of an rtx.
499 * cprop.c (local_cprop_find_used_regs): Update accordingly.
500 * df-problems.c (df_word_lr_mark_ref): Likewise.
501 * ira-lives.c (mark_pseudo_reg_live): Likewise.
502 (mark_pseudo_reg_dead): Likewise.
503 (mark_ref_dead): Likewise.
504 * reginfo.c (init_subregs_of_mode): Likewise.
505 * sched-deps.c (sched_analyze_1): Likewise.
506 * df-scan.c (df_def_record_1): Likewise.
507 (df_uses_record): Likewise.
508 (df_read_modify_subreg_p): Remove in favor of...
509 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
510 const_rtx instead of an rtx.
512 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
513 Alan Hayward <alan.hayward@arm.com>
514 David Sherwood <david.sherwood@arm.com>
516 * rtl.h (partial_subreg_p): New function.
517 * caller-save.c (save_call_clobbered_regs): Use it.
518 * calls.c (expand_call): Likewise.
519 * combine.c (combinable_i3pat): Likewise.
520 (simplify_set): Likewise.
521 (make_extraction): Likewise.
522 (make_compound_operation_int): Likewise.
523 (gen_lowpart_or_truncate): Likewise.
524 (force_to_mode): Likewise.
525 (make_field_assignment): Likewise.
526 (reg_truncated_to_mode): Likewise.
527 (record_truncated_value): Likewise.
528 (move_deaths): Likewise.
529 * cse.c (record_jump_cond): Likewise.
530 (cse_insn): Likewise.
531 * cselib.c (cselib_lookup_1): Likewise.
532 * expmed.c (extract_bit_field_using_extv): Likewise.
533 * function.c (assign_parm_setup_reg): Likewise.
534 * ifcvt.c (noce_convert_multiple_sets): Likewise.
535 * ira-build.c (create_insn_allocnos): Likewise.
536 * lra-coalesce.c (merge_pseudos): Likewise.
537 * lra-constraints.c (match_reload): Likewise.
538 (simplify_operand_subreg): Likewise.
539 (curr_insn_transform): Likewise.
540 * lra-lives.c (process_bb_lives): Likewise.
541 * lra.c (new_insn_reg): Likewise.
542 (lra_substitute_pseudo): Likewise.
543 * regcprop.c (mode_change_ok): Likewise.
544 (maybe_mode_change): Likewise.
545 (copyprop_hardreg_forward_1): Likewise.
546 * reload.c (push_reload): Likewise.
547 (find_reloads): Likewise.
548 (find_reloads_subreg_address): Likewise.
549 * reload1.c (alter_reg): Likewise.
550 (eliminate_regs_1): Likewise.
551 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
553 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
555 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
556 back to if statements, including unpack.
558 2017-08-30 Martin Liska <mliska@suse.cz>
561 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
563 (func_checker::compare_asm_inputs_outputs): ... this function.
564 (func_checker::compare_gimple_asm): Use the function to compare
566 * ipa-icf-gimple.h: Rename the function.
568 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
569 Alan Hayward <alan.hayward@arm.com>
570 David Sherwood <david.sherwood@arm.com>
572 * coretypes.h (complex_mode): New type.
573 * gdbhooks.py (build_pretty_printer): Handle it.
574 * machmode.h (complex_mode): New class.
575 (complex_mode::includes_p): New function.
576 (is_complex_int_mode): Likewise.
577 (is_complex_float_mode): Likewise.
578 * genmodes.c (get_mode_class): Handle complex mode classes.
579 * function.c (expand_function_end): Use is_complex_int_mode.
581 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
582 Alan Hayward <alan.hayward@arm.com>
583 David Sherwood <david.sherwood@arm.com>
585 * coretypes.h (scalar_mode_pod): New typedef.
586 * gdbhooks.py (build_pretty_printer): Handle it.
587 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
588 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
589 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
590 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
594 Alan Hayward <alan.hayward@arm.com>
595 David Sherwood <david.sherwood@arm.com>
597 * machmode.h (mode_for_vector): Take a scalar_mode instead
599 * stor-layout.c (mode_for_vector): Likewise.
600 * explow.c (promote_mode): Use as_a <scalar_mode>.
601 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
603 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
604 Alan Hayward <alan.hayward@arm.com>
605 David Sherwood <david.sherwood@arm.com>
607 * target.def (preferred_simd_mode): Take a scalar_mode
608 instead of a machine_mode.
609 * targhooks.h (default_preferred_simd_mode): Likewise.
610 * targhooks.c (default_preferred_simd_mode): Likewise.
611 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
612 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
613 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
614 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
615 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
616 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
617 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
618 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
620 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
621 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
622 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
623 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
624 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
625 * doc/tm.texi: Regenerate.
626 * optabs-query.c (can_vec_mask_load_store_p): Return false for
629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
630 Alan Hayward <alan.hayward@arm.com>
631 David Sherwood <david.sherwood@arm.com>
633 * target.def (scalar_mode_supported_p): Take a scalar_mode
634 instead of a machine_mode.
635 * targhooks.h (default_scalar_mode_supported_p): Likewise.
636 * targhooks.c (default_scalar_mode_supported_p): Likewise.
637 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
638 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
639 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
640 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
641 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
642 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
643 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
644 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
645 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
646 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
647 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
648 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
650 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
651 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
652 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
653 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
654 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
656 * doc/tm.texi: Regenerate.
658 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
659 Alan Hayward <alan.hayward@arm.com>
660 David Sherwood <david.sherwood@arm.com>
662 * coretypes.h (opt_scalar_mode): New typedef.
663 * gdbhooks.py (build_pretty_printers): Handle it.
664 * machmode.h (mode_iterator::get_2xwider): Add overload for
666 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
668 * expr.c (convert_mode_scalar): Likewise.
669 * omp-low.c (omp_clause_aligned_alignment): Likewise.
670 * optabs.c (expand_float): Likewise.
671 (expand_fix): Likewise.
672 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
674 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
675 Alan Hayward <alan.hayward@arm.com>
676 David Sherwood <david.sherwood@arm.com>
678 * optabs.c (expand_float): Explicitly check for scalars before
679 using a branching expansion.
680 (expand_fix): Likewise.
682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
683 Alan Hayward <alan.hayward@arm.com>
684 David Sherwood <david.sherwood@arm.com>
686 * expr.c (convert_mode): Split scalar handling out into...
687 (convert_mode_scalar): ...this new function. Treat the modes
690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
691 Alan Hayward <alan.hayward@arm.com>
692 David Sherwood <david.sherwood@arm.com>
694 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
696 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
699 Alan Hayward <alan.hayward@arm.com>
700 David Sherwood <david.sherwood@arm.com>
702 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
703 rather than a machine_mode.
704 (fixed_from_string): Likewise.
705 (fixed_convert): Likewise.
706 (fixed_convert_from_int): Likewise.
707 (fixed_convert_from_real): Likewise.
708 (real_convert_from_fixed): Likewise.
709 * fixed-value.c (fixed_from_double_int): Likewise.
710 (fixed_from_string): Likewise.
711 (fixed_convert): Likewise.
712 (fixed_convert_from_int): Likewise.
713 (fixed_convert_from_real): Likewise.
714 (real_convert_from_fixed): Likewise.
715 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
718 Alan Hayward <alan.hayward@arm.com>
719 David Sherwood <david.sherwood@arm.com>
721 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
722 of separate mode class checks. Do not allow vector modes here.
723 (immed_wide_int_const): Use as_a <scalar_mode>.
724 * explow.c (trunc_int_for_mode): Likewise.
725 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
726 (wi::shwi): Likewise.
727 (wi::min_value): Likewise.
728 (wi::max_value): Likewise.
729 * dwarf2out.c (loc_descriptor): Likewise.
730 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
733 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
734 Alan Hayward <alan.hayward@arm.com>
735 David Sherwood <david.sherwood@arm.com>
737 * tree.h (SCALAR_TYPE_MODE): New macro.
738 * expr.c (expand_expr_addr_expr_1): Use it.
739 (expand_expr_real_2): Likewise.
740 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
741 (fold_convert_const_fixed_from_int): Likewise.
742 (fold_convert_const_fixed_from_real): Likewise.
743 (native_encode_fixed): Likewise
744 (native_encode_complex): Likewise
745 (native_encode_vector): Likewise.
746 (native_interpret_fixed): Likewise.
747 (native_interpret_real): Likewise.
748 (native_interpret_complex): Likewise.
749 (native_interpret_vector): Likewise.
750 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
751 (simd_clone_adjust_argument_types): Likewise.
752 (simd_clone_init_simd_arrays): Likewise.
753 (simd_clone_adjust): Likewise.
754 * stor-layout.c (layout_type): Likewise.
755 * tree.c (build_minus_one_cst): Likewise.
756 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
757 * tree-inline.c (estimate_move_cost): Likewise.
758 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
759 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
760 (vectorizable_reduction): Likewise.
761 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
762 (vect_recog_mixed_size_cond_pattern): Likewise.
763 (check_bool_pattern): Likewise.
764 (adjust_bool_pattern): Likewise.
765 (search_type_for_mask_1): Likewise.
766 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
767 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
768 (vectorizable_load): Likewise.
769 (vectorizable_store): Likewise.
770 * ubsan.c (ubsan_encode_value): Likewise.
771 * varasm.c (output_constant): Likewise.
773 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
774 Alan Hayward <alan.hayward@arm.com>
775 David Sherwood <david.sherwood@arm.com>
777 * coretypes.h (scalar_mode): New class.
778 * machmode.h (scalar_mode): Likewise.
779 (scalar_mode::includes_p): New function.
780 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
781 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
782 * genmodes.c (get_mode_class): Handle remaining scalar modes.
783 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
784 * expmed.c (store_bit_field_1): Likewise.
785 (extract_bit_field_1): Likewise.
786 * expr.c (write_complex_part): Likewise.
787 (read_complex_part): Likewise.
788 (emit_move_complex_push): Likewise.
789 (expand_expr_real_2): Likewise.
790 * function.c (assign_parm_setup_reg): Likewise.
791 (assign_parms_unsplit_complex): Likewise.
792 * optabs.c (expand_binop): Likewise.
793 * rtlanal.c (subreg_get_info): Likewise.
794 * simplify-rtx.c (simplify_immed_subreg): Likewise.
795 * varasm.c (output_constant_pool_2): Likewise.
797 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
798 Alan Hayward <alan.hayward@arm.com>
799 David Sherwood <david.sherwood@arm.com>
801 * expmed.c (extract_high_half): Use scalar_int_mode and remove
803 (expmed_mult_highpart_optab): Likewise.
804 (expmed_mult_highpart): Likewise.
806 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
807 Alan Hayward <alan.hayward@arm.com>
808 David Sherwood <david.sherwood@arm.com>
810 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
811 instead of a machine_mode.
812 (builtin_memset_read_str): Likewise.
813 * builtins.c (c_readstr): Likewise.
814 (builtin_memcpy_read_str): Likewise.
815 (builtin_strncpy_read_str): Likewise.
816 (builtin_memset_read_str): Likewise.
817 (builtin_memset_gen_str): Likewise.
818 (expand_builtin_signbit): Use scalar_int_mode for local variables.
819 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
820 instead of a machine_mode.
821 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
823 (make_extraction): Likewise.
824 (try_widen_shift_mode): Take and return scalar_int_modes instead
826 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
827 a scalar_int_mode instead of a machine_mode.
828 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
829 (avr_addr_space_pointer_mode): Likewise.
830 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
831 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
832 (msp430_unwind_word_mode): Likewise.
833 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
834 (spu_addr_space_pointer_mode): Likewise.
835 (spu_addr_space_address_mode): Likewise.
836 (spu_libgcc_cmp_return_mode): Likewise.
837 (spu_libgcc_shift_count_mode): Likewise.
838 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
839 (rl78_addr_space_pointer_mode): Likewise.
840 (fl78_unwind_word_mode): Likewise.
841 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
843 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
844 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
845 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
846 (mips_valid_pointer_mode): Likewise.
847 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
848 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
849 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
851 (ft32_addr_space_address_mode): Likewise.
852 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
853 scalar_int_mode instead of a machine_mode.
854 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
856 (m32c_addr_space_address_mode): Likewise.
857 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
858 (rs6000_eh_return_filter_mode): Likewise.
859 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
860 (rs6000_eh_return_filter_mode): Likewise.
861 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
862 (s390_libgcc_shift_count_mode): Likewise.
863 (s390_unwind_word_mode): Likewise.
864 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
866 * target.def (mode_rep_extended): Likewise.
867 (valid_pointer_mode): Likewise.
868 (addr_space.valid_pointer_mode): Likewise.
869 (eh_return_filter_mode): Return a scalar_int_mode rather than
871 (libgcc_cmp_return_mode): Likewise.
872 (libgcc_shift_count_mode): Likewise.
873 (unwind_word_mode): Likewise.
874 (addr_space.pointer_mode): Likewise.
875 (addr_space.address_mode): Likewise.
876 * doc/tm.texi: Regenerate.
877 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
879 (do_jump): Use scalar_int_mode for local variables.
880 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
881 rather than a machine_mode.
882 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
883 (scompare_loc_descriptor_wide): Likewise.
884 (scompare_loc_descriptor_narrow): Likewise.
885 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
887 * except.c (sjlj_emit_dispatch_table): Likewise.
888 (expand_builtin_eh_copy_values): Likewise.
889 * explow.c (convert_memory_address_addr_space_1): Likewise.
890 Take a scalar_int_mode rather than a machine_mode.
891 (convert_memory_address_addr_space): Take a scalar_int_mode rather
893 (memory_address_addr_space): Use scalar_int_mode for local variables.
894 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
895 rather than a machine_mode.
896 * expmed.c (mask_rtx): Likewise.
897 (init_expmed_one_conv): Likewise.
898 (expand_mult_highpart_adjust): Likewise.
899 (extract_high_half): Likewise.
900 (expmed_mult_highpart_optab): Likewise.
901 (expmed_mult_highpart): Likewise.
902 (expand_smod_pow2): Likewise.
903 (expand_sdiv_pow2): Likewise.
904 (emit_store_flag_int): Likewise.
905 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
907 (extract_low_bits): Likewise.
908 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
910 * expr.c (pieces_addr::adjust): Likewise.
911 (can_store_by_pieces): Likewise.
912 (store_by_pieces): Likewise.
913 (clear_by_pieces_1): Likewise.
914 (expand_expr_addr_expr_1): Likewise.
915 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
916 (expand_expr_real_1): Likewise.
917 (try_casesi): Likewise.
918 * final.c (shorten_branches): Likewise.
919 * fold-const.c (fold_convert_const_int_from_fixed): Change the
920 type of "mode" to machine_mode.
921 * internal-fn.c (expand_arith_overflow_result_store): Take a
922 scalar_int_mode rather than a machine_mode.
923 (expand_mul_overflow): Use scalar_int_mode for local variables.
924 * loop-doloop.c (doloop_modify): Likewise.
925 (doloop_optimize): Likewise.
926 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
928 (expand_doubleword_shift_condmove): Likewise.
929 (expand_doubleword_shift): Likewise.
930 (expand_doubleword_clz): Likewise.
931 (expand_doubleword_popcount): Likewise.
932 (expand_doubleword_parity): Likewise.
933 (expand_absneg_bit): Use scalar_int_mode for local variables.
934 (prepare_float_lib_cmp): Likewise.
935 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
936 rather than a machine_mode.
937 (convert_memory_address_addr_space): Likewise.
938 (get_mode_bounds): Likewise.
939 (get_address_mode): Return a scalar_int_mode rather than a
941 * rtlanal.c (get_address_mode): Likewise.
942 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
944 * targhooks.c (default_mode_rep_extended): Likewise.
945 (default_valid_pointer_mode): Likewise.
946 (default_addr_space_valid_pointer_mode): Likewise.
947 (default_eh_return_filter_mode): Return a scalar_int_mode rather
949 (default_libgcc_cmp_return_mode): Likewise.
950 (default_libgcc_shift_count_mode): Likewise.
951 (default_unwind_word_mode): Likewise.
952 (default_addr_space_pointer_mode): Likewise.
953 (default_addr_space_address_mode): Likewise.
954 * targhooks.h (default_eh_return_filter_mode): Likewise.
955 (default_libgcc_cmp_return_mode): Likewise.
956 (default_libgcc_shift_count_mode): Likewise.
957 (default_unwind_word_mode): Likewise.
958 (default_addr_space_pointer_mode): Likewise.
959 (default_addr_space_address_mode): Likewise.
960 (default_mode_rep_extended): Take a scalar_int_mode rather than
962 (default_valid_pointer_mode): Likewise.
963 (default_addr_space_valid_pointer_mode): Likewise.
964 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
966 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
967 rather than a machine_mode.
968 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
970 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
971 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
974 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
975 Alan Hayward <alan.hayward@arm.com>
976 David Sherwood <david.sherwood@arm.com>
978 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
979 the mode argument to scalar_int_mode.
980 (do_jump_by_parts_zero_rtx): Likewise.
981 (do_jump_by_parts_equality_rtx): Likewise.
982 (do_jump_by_parts_greater): Take a mode argument.
983 (do_jump_by_parts_equality): Likewise.
984 (do_jump_1): Update calls accordingly.
986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
987 Alan Hayward <alan.hayward@arm.com>
988 David Sherwood <david.sherwood@arm.com>
990 * is-a.h (safe_dyn_cast): New function.
991 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
992 (jump_table_for_label): Likewise.
993 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
994 instead of an rtx_insn *.
995 (shorten_branches): Use dyn_cast instead of LABEL_P and
996 JUMP_TABLE_DATA_P. Use jump_table_for_label and
997 rtx_jump_table_data::get_data_mode.
998 (final_scan_insn): Likewise.
1000 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1001 Alan Hayward <alan.hayward@arm.com>
1002 David Sherwood <david.sherwood@arm.com>
1004 * combine.c (try_combine): Use is_a <scalar_int_mode> when
1005 trying to combine a full-register integer set with a subreg
1008 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1009 Alan Hayward <alan.hayward@arm.com>
1010 David Sherwood <david.sherwood@arm.com>
1012 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
1013 that is always either address_mode or pointer_mode.
1015 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1016 Alan Hayward <alan.hayward@arm.com>
1017 David Sherwood <david.sherwood@arm.com>
1019 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
1020 when the two are known to be equal.
1022 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1023 Alan Hayward <alan.hayward@arm.com>
1024 David Sherwood <david.sherwood@arm.com>
1026 * simplify-rtx.c (simplify_const_unary_operation): Use
1027 is_a <scalar_int_mode> instead of checking for a nonzero
1028 precision. Forcibly convert op_mode to a scalar_int_mode
1029 in that case. More clearly differentiate the operand and
1030 result modes and use the former when deciding what the value
1031 of a count-bits operation should be. Use is_int_mode instead
1032 of checking for a MODE_INT. Remove redundant check for whether
1033 this mode has a zero precision.
1035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1036 Alan Hayward <alan.hayward@arm.com>
1037 David Sherwood <david.sherwood@arm.com>
1039 * optabs.c (widen_leading): Change the type of the mode argument
1040 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
1041 (widen_bswap): Likewise.
1042 (expand_parity): Likewise.
1043 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
1044 (expand_ffs): Likewise.
1045 (epand_unop): Check for scalar integer modes before calling the
1048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1049 Alan Hayward <alan.hayward@arm.com>
1050 David Sherwood <david.sherwood@arm.com>
1052 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
1054 (expand_expr_real_1): Update call accordingly.
1056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1057 Alan Hayward <alan.hayward@arm.com>
1058 David Sherwood <david.sherwood@arm.com>
1060 * expmed.c (store_bit_field_using_insv): Add op0_mode and
1061 value_mode arguments. Use scalar_int_mode internally.
1062 (store_bit_field_1): Rename the new integer mode from imode
1063 to op0_mode and use it instead of GET_MODE (op0). Update calls
1064 to store_split_bit_field, store_bit_field_using_insv and
1065 store_fixed_bit_field.
1066 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
1067 Use scalar_int_mode internally. Use a bit count rather than a mode
1068 when calculating the largest bit size for get_best_mode.
1069 Update calls to store_split_bit_field and store_fixed_bit_field_1.
1070 (store_fixed_bit_field_1): Add mode and value_mode arguments.
1071 Remove assertion that OP0 has a scalar integer mode.
1072 (store_split_bit_field): Add op0_mode and value_mode arguments.
1073 Update calls to extract_fixed_bit_field.
1074 (extract_bit_field_using_extv): Add an op0_mode argument.
1075 Use scalar_int_mode internally.
1076 (extract_bit_field_1): Rename the new integer mode from imode to
1077 op0_mode and use it instead of GET_MODE (op0). Update calls to
1078 extract_split_bit_field, extract_bit_field_using_extv and
1079 extract_fixed_bit_field.
1080 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
1081 to extract_split_bit_field and extract_fixed_bit_field_1.
1082 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
1083 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
1085 (extract_split_bit_field): Add an op0_mode argument. Update call
1086 to extract_fixed_bit_field.
1088 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1089 Alan Hayward <alan.hayward@arm.com>
1090 David Sherwood <david.sherwood@arm.com>
1092 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
1093 * explow.c (hard_function_value): Likewise.
1094 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
1095 convert_to_mode call outside the loop.
1096 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
1097 for the mode iterator. Require the mode specified by max_pieces
1099 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
1101 (copy_blkmode_to_reg): Likewise.
1102 (set_storage_via_setmem): Likewise.
1103 * optabs.c (prepare_cmp_insn): Likewise.
1104 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1105 * stor-layout.c (finish_bitfield_representative): Likewise.
1107 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1108 Alan Hayward <alan.hayward@arm.com>
1109 David Sherwood <david.sherwood@arm.com>
1111 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
1112 * expr.c (convert_move): Use them.
1113 (convert_modes): Likewise.
1114 (store_expr_with_bounds): Likewise.
1116 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1117 Alan Hayward <alan.hayward@arm.com>
1118 David Sherwood <david.sherwood@arm.com>
1120 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
1121 parameter for the mode of "x". Remove the "known_x", "known_mode"
1122 and "known_ret" arguments. Change the type of the mode argument
1124 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
1125 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
1126 (reg_num_sign_bit_copies_for_combine): Likewise.
1127 * rtlanal.c (nonzero_bits1): Likewise.
1128 (num_sign_bit_copies1): Likewise.
1129 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
1130 (reg_num_sign_bit_copies_general): Likewise.
1131 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
1132 (reg_nonzero_bits_general): Likewise.
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 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
1139 than in subroutines. Return 1 for non-integer modes.
1140 (cached_num_sign_bit_copies): Change the type of the mode parameter
1142 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
1143 classes. Handle CONST_INT_P first and then check whether X also
1144 has a scalar integer mode. Check the same thing for inner registers
1145 of a SUBREG and for values that are being extended or truncated.
1147 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1148 Alan Hayward <alan.hayward@arm.com>
1149 David Sherwood <david.sherwood@arm.com>
1151 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
1152 in subroutines. Return the mode mask for non-integer modes.
1153 (cached_nonzero_bits): Change the type of the mode parameter
1155 (nonzero_bits1): Likewise. Remove early exit for other mode
1156 classes. Handle CONST_INT_P first and then check whether X
1157 also has a scalar integer mode.
1159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1160 Alan Hayward <alan.hayward@arm.com>
1161 David Sherwood <david.sherwood@arm.com>
1163 * expr.c (widest_int_mode_for_size): Make the comment match the code.
1164 Return a scalar_int_mode and assert that the size is greater than
1166 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
1167 (op_by_pieces_d::op_by_pieces_d): Likewise.
1168 (op_by_pieces_d::run): Likewise.
1169 (can_store_by_pieces): Likewise.
1171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1172 Alan Hayward <alan.hayward@arm.com>
1173 David Sherwood <david.sherwood@arm.com>
1175 * combine.c (extract_left_shift): Add a mode argument and update
1177 (make_compound_operation_int): Change the type of the mode parameter
1178 to scalar_int_mode and update the call to extract_left_shift.
1180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1181 Alan Hayward <alan.hayward@arm.com>
1182 David Sherwood <david.sherwood@arm.com>
1184 * combine.c (simplify_and_const_int): Change the type of the mode
1185 parameter to scalar_int_mode.
1186 (simplify_and_const_int_1): Likewise. Update recursive call.
1188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1189 Alan Hayward <alan.hayward@arm.com>
1190 David Sherwood <david.sherwood@arm.com>
1192 * combine.c (simplify_compare_const): Check that the mode is a
1193 scalar_int_mode (rather than VOIDmode) before testing its
1195 (simplify_comparison): Move COMPARISON_P handling out of the
1196 loop and restrict the latter part of the loop to scalar_int_modes.
1197 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
1198 and when considering SUBREG_REGs. Use is_int_mode instead of
1199 checking GET_MODE_CLASS against MODE_INT.
1201 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1202 Alan Hayward <alan.hayward@arm.com>
1203 David Sherwood <david.sherwood@arm.com>
1205 * combine.c (try_widen_shift_mode): Move check for equal modes to...
1206 (simplify_shift_const_1): ...here. Use scalar_int_mode for
1207 shift_unit_mode and for modes involved in scalar shifts.
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 * combine.c (force_int_to_mode): New function, split out from...
1214 (force_to_mode): ...here. Keep xmode up-to-date and use it
1215 instead of GET_MODE (x).
1217 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1218 Alan Hayward <alan.hayward@arm.com>
1219 David Sherwood <david.sherwood@arm.com>
1221 * optabs-query.h (extraction_insn::struct_mode): Change type to
1222 opt_scalar_int_mode and update comment.
1223 (extraction_insn::field_mode): Change type to scalar_int_mode.
1224 (extraction_insn::pos_mode): Likewise.
1225 * combine.c (make_extraction): Update accordingly.
1226 * optabs-query.c (get_traditional_extraction_insn): Likewise.
1227 (get_optab_extraction_insn): Likewise.
1228 * recog.c (simplify_while_replacing): Likewise.
1229 * expmed.c (narrow_bit_field_mem): Change the type of the mode
1230 parameter to opt_scalar_int_mode.
1232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1233 Alan Hayward <alan.hayward@arm.com>
1234 David Sherwood <david.sherwood@arm.com>
1236 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
1237 to a scalar_int_mode instead of a machine_mode.
1238 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
1239 (get_best_mode): Return a boolean and use a pointer argument to store
1240 the selected mode. Replace the limit mode parameter with a bit limit.
1241 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
1242 for the values returned by bit_field_mode_iterator::next_mode.
1243 (store_bit_field): Update call to get_best_mode.
1244 (store_fixed_bit_field): Likewise.
1245 (extract_fixed_bit_field): Likewise.
1246 * expr.c (optimize_bitfield_assignment_op): Likewise.
1247 * fold-const.c (optimize_bit_field_compare): Likewise.
1248 (fold_truth_andor_1): Likewise.
1249 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
1250 Update for new type of m_mode.
1251 (get_best_mode): As above.
1253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1254 Alan Hayward <alan.hayward@arm.com>
1255 David Sherwood <david.sherwood@arm.com>
1257 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
1258 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
1259 (store_bit_field): Check is_a <scalar_int_mode> before calling
1260 strict_volatile_bitfield_p.
1261 (extract_bit_field): Likewise.
1263 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1264 Alan Hayward <alan.hayward@arm.com>
1265 David Sherwood <david.sherwood@arm.com>
1267 * target.def (cstore_mode): Return a scalar_int_mode.
1268 * doc/tm.texi: Regenerate.
1269 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
1270 * targhooks.h (default_cstore_mode): Likewise.
1271 * targhooks.c (default_cstore_mode): Likewise, using a forced
1273 * expmed.c (emit_cstore): Expect the target of the cstore to be
1276 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1277 Alan Hayward <alan.hayward@arm.com>
1278 David Sherwood <david.sherwood@arm.com>
1280 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
1282 (niter_desc): Likewise mode.
1283 (iv_analyze): Add a mode parameter.
1285 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
1286 and change its type to scalar_int_mode.
1287 * loop-iv.c: Update commentary at head of file.
1288 (iv_constant): Pass the mode paraeter before the rtx it describes
1289 and change its type to scalar_int_mode. Remove VOIDmode handling.
1290 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
1291 (iv_extend): Likewise.
1292 (shorten_into_mode): Likewise.
1293 (iv_add): Use scalar_int_mode.
1294 (iv_mult): Likewise.
1295 (iv_shift): Likewise.
1296 (canonicalize_iv_subregs): Likewise.
1297 (get_biv_step_1): Pass the outer_mode parameter before the rtx
1298 it describes and change its mode to scalar_int_mode. Also change
1299 the type of the returned inner_mode to scalar_int_mode.
1300 (get_biv_step): Likewise, turning outer_mode from a pointer
1301 into a direct parameter. Update call to get_biv_step_1.
1302 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
1303 iv_constant and get_biv_step.
1304 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
1305 and change its type to scalar_int_mode. Don't initialise iv->mode
1306 to VOIDmode and remove later checks for its still being VOIDmode.
1307 Update calls to iv_analyze_op and iv_analyze_expr. Check
1308 is_a <scalar_int_mode> when changing the mode under consideration.
1309 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
1310 Update call to iv_analyze_expr.
1311 (iv_analyze_op): Add a mode parameter. Reject subregs whose
1312 inner register is not also a scalar_int_mode. Update call to
1314 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
1315 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
1316 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
1317 separate mode class checks. Update calls to iv_analyze. Remove
1318 fix-up of VOIDmodes after iv_analyze_biv.
1319 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
1320 don't have a scalar_int_mode. Update call to biv_p.
1322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1323 Alan Hayward <alan.hayward@arm.com>
1324 David Sherwood <david.sherwood@arm.com>
1326 * cfgexpand.c (convert_debug_memory_address): Use
1327 as_a <scalar_int_mode>.
1328 * combine.c (expand_compound_operation): Likewise.
1329 (make_extraction): Likewise.
1330 (change_zero_ext): Likewise.
1331 (simplify_comparison): Likewise.
1332 * cse.c (cse_insn): Likewise.
1333 * dwarf2out.c (minmax_loc_descriptor): Likewise.
1334 (mem_loc_descriptor): Likewise.
1335 (loc_descriptor): Likewise.
1336 * expmed.c (init_expmed_one_mode): Likewise.
1337 (synth_mult): Likewise.
1338 (emit_store_flag_1): Likewise.
1339 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
1340 of a comparison with size.
1341 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
1342 (reduce_to_bit_field_precision): Likewise.
1343 * function.c (expand_function_end): Likewise.
1344 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
1345 * loop-doloop.c (doloop_modify): Likewise.
1346 * optabs.c (expand_binop): Likewise.
1347 (expand_unop): Likewise.
1348 (expand_copysign_absneg): Likewise.
1349 (prepare_cmp_insn): Likewise.
1350 (maybe_legitimize_operand): Likewise.
1351 * recog.c (const_scalar_int_operand): Likewise.
1352 * rtlanal.c (get_address_mode): Likewise.
1353 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1354 (simplify_cond_clz_ctz): Likewise.
1355 * tree-nested.c (get_nl_goto_field): Likewise.
1356 * tree.c (build_vector_type_for_mode): Likewise.
1357 * var-tracking.c (use_narrower_mode): Likewise.
1359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1360 Alan Hayward <alan.hayward@arm.com>
1361 David Sherwood <david.sherwood@arm.com>
1363 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
1364 * builtins.c (expand_builtin_signbit): Use it.
1365 * cfgexpand.c (expand_debug_expr): Likewise.
1366 * dojump.c (do_jump): Likewise.
1367 (do_compare_and_jump): Likewise.
1368 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
1369 * expmed.c (make_tree): Likewise.
1370 * expr.c (expand_expr_real_2): Likewise.
1371 (expand_expr_real_1): Likewise.
1372 (try_casesi): Likewise.
1373 * fold-const-call.c (fold_const_call_ss): Likewise.
1374 * fold-const.c (unextend): Likewise.
1375 (extract_muldiv_1): Likewise.
1376 (fold_single_bit_test): Likewise.
1377 (native_encode_int): Likewise.
1378 (native_encode_string): Likewise.
1379 (native_interpret_int): Likewise.
1380 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
1381 * internal-fn.c (expand_addsub_overflow): Likewise.
1382 (expand_neg_overflow): Likewise.
1383 (expand_mul_overflow): Likewise.
1384 (expand_arith_overflow): Likewise.
1385 * match.pd: Likewise.
1386 * stor-layout.c (layout_type): Likewise.
1387 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
1388 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
1389 * tree-ssanames.c (get_range_info): Likewise.
1390 * tree-switch-conversion.c (array_value_type) Likewise.
1391 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
1392 (vect_recog_divmod_pattern): Likewise.
1393 (vect_recog_mixed_size_cond_pattern): Likewise.
1394 * tree-vrp.c (extract_range_basic): Likewise.
1395 (simplify_float_conversion_using_ranges): Likewise.
1396 * tree.c (int_fits_type_p): Likewise.
1397 * ubsan.c (instrument_bool_enum_load): Likewise.
1398 * varasm.c (mergeable_string_section): Likewise.
1399 (narrowing_initializer_constant_valid_p): Likewise.
1400 (output_constant): Likewise.
1402 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1403 Alan Hayward <alan.hayward@arm.com>
1404 David Sherwood <david.sherwood@arm.com>
1406 * machmode.h (NARROWEST_INT_MODE): New macro.
1407 * expr.c (alignment_for_piecewise_move): Use it instead of
1408 GET_CLASS_NARROWEST_MODE (MODE_INT).
1409 (push_block): Likewise.
1410 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1412 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1414 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1415 Alan Hayward <alan.hayward@arm.com>
1416 David Sherwood <david.sherwood@arm.com>
1418 * postreload.c (move2add_valid_value_p): Change the type of the
1419 mode parameter to scalar_int_mode.
1420 (move2add_use_add2_insn): Add a mode parameter and use it instead
1422 (move2add_use_add3_insn): Likewise.
1423 (reload_cse_move2add): Update accordingly.
1425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1426 Alan Hayward <alan.hayward@arm.com>
1427 David Sherwood <david.sherwood@arm.com>
1429 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
1431 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
1432 * optabs.c (expand_unop): Likewise.
1434 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1435 Alan Hayward <alan.hayward@arm.com>
1436 David Sherwood <david.sherwood@arm.com>
1438 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
1439 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
1440 (popcount_loc_descriptor): Likewise.
1441 (bswap_loc_descriptor): Likewise.
1442 (rotate_loc_descriptor): Likewise.
1443 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
1444 calling the functions above.
1446 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1447 Alan Hayward <alan.hayward@arm.com>
1448 David Sherwood <david.sherwood@arm.com>
1450 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
1452 (try_combine): Likewise.
1453 (simplify_if_then_else): Likewise.
1454 * cse.c (cse_insn): Likewise.
1455 * dwarf2out.c (mem_loc_descriptor): Likewise.
1456 * emit-rtl.c (gen_lowpart_common): Likewise.
1457 * simplify-rtx.c (simplify_truncation): Likewise.
1458 (simplify_binary_operation_1): Likewise.
1459 (simplify_const_relational_operation): Likewise.
1460 (simplify_ternary_operation): Likewise.
1461 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
1463 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1464 Alan Hayward <alan.hayward@arm.com>
1465 David Sherwood <david.sherwood@arm.com>
1467 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
1468 * reload.c (push_reload): Likewise.
1469 (find_reloads): Likewise.
1471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1472 Alan Hayward <alan.hayward@arm.com>
1473 David Sherwood <david.sherwood@arm.com>
1475 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
1476 (make_compound_operation_int): Likewise.
1477 (change_zero_ext): Likewise.
1478 * expr.c (convert_move): Likewise.
1479 (convert_modes): Likewise.
1480 * fwprop.c (forward_propagate_subreg): Likewise.
1481 * loop-iv.c (get_biv_step_1): Likewise.
1482 * optabs.c (widen_operand): Likewise.
1483 * postreload.c (move2add_valid_value_p): Likewise.
1484 * recog.c (simplify_while_replacing): Likewise.
1485 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1486 (simplify_binary_operation_1): Likewise. Remove redundant
1487 mode equality check.
1489 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1490 Alan Hayward <alan.hayward@arm.com>
1491 David Sherwood <david.sherwood@arm.com>
1493 * combine.c (combine_simplify_rtx): Add checks for
1494 is_a <scalar_int_mode>.
1495 (simplify_if_then_else): Likewise.
1496 (make_field_assignment): Likewise.
1497 (simplify_comparison): Likewise.
1498 * ifcvt.c (noce_try_bitop): Likewise.
1499 * loop-invariant.c (canonicalize_address_mult): Likewise.
1500 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1502 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1503 Alan Hayward <alan.hayward@arm.com>
1504 David Sherwood <david.sherwood@arm.com>
1506 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
1507 is_a <scalar_int_mode> instead of != BLKmode.
1509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1510 Alan Hayward <alan.hayward@arm.com>
1511 David Sherwood <david.sherwood@arm.com>
1513 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
1514 instead of != VOIDmode.
1515 * combine.c (if_then_else_cond): Likewise.
1516 (change_zero_ext): Likewise.
1517 * dwarf2out.c (mem_loc_descriptor): Likewise.
1518 (loc_descriptor): Likewise.
1519 * rtlanal.c (canonicalize_condition): Likewise.
1520 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
1522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1523 Alan Hayward <alan.hayward@arm.com>
1524 David Sherwood <david.sherwood@arm.com>
1526 * simplify-rtx.c (simplify_binary_operation_1): Use
1527 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
1529 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1530 Alan Hayward <alan.hayward@arm.com>
1531 David Sherwood <david.sherwood@arm.com>
1533 * wide-int.h (int_traits<unsigned char>) New class.
1534 (int_traits<unsigned short>) Likewise.
1535 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
1536 Use GET_MODE_UNIT_PRECISION and remove redundant test for
1538 * combine.c (set_nonzero_bits_and_sign_copies): Use
1539 is_a <scalar_int_mode>.
1540 (find_split_point): Likewise.
1541 (combine_simplify_rtx): Likewise.
1542 (simplify_logical): Likewise.
1543 (expand_compound_operation): Likewise.
1544 (expand_field_assignment): Likewise.
1545 (make_compound_operation): Likewise.
1546 (extended_count): Likewise.
1547 (change_zero_ext): Likewise.
1548 (simplify_comparison): Likewise.
1549 * dwarf2out.c (scompare_loc_descriptor): Likewise.
1550 (ucompare_loc_descriptor): Likewise.
1551 (minmax_loc_descriptor): Likewise.
1552 (mem_loc_descriptor): Likewise.
1553 (loc_descriptor): Likewise.
1554 * expmed.c (init_expmed_one_mode): Likewise.
1555 * lra-constraints.c (lra_constraint_offset): Likewise.
1556 * optabs.c (prepare_libcall_arg): Likewise.
1557 * postreload.c (move2add_note_store): Likewise.
1558 * reload.c (operands_match_p): Likewise.
1559 * rtl.h (load_extend_op): Likewise.
1560 * rtlhooks.c (gen_lowpart_general): Likewise.
1561 * simplify-rtx.c (simplify_truncation): Likewise.
1562 (simplify_unary_operation_1): Likewise.
1563 (simplify_binary_operation_1): Likewise.
1564 (simplify_const_binary_operation): Likewise.
1565 (simplify_const_relational_operation): Likewise.
1566 (simplify_subreg): Likewise.
1567 * stor-layout.c (bitwise_mode_for_mode): Likewise.
1568 * var-tracking.c (adjust_mems): Likewise.
1569 (prepare_call_arguments): Likewise.
1571 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1572 Alan Hayward <alan.hayward@arm.com>
1573 David Sherwood <david.sherwood@arm.com>
1575 * machmode.h (is_int_mode): New fuction.
1576 * combine.c (find_split_point): Use it.
1577 (combine_simplify_rtx): Likewise.
1578 (simplify_if_then_else): Likewise.
1579 (simplify_set): Likewise.
1580 (simplify_shift_const_1): Likewise.
1581 (simplify_comparison): Likewise.
1582 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
1583 * cse.c (notreg_cost): Likewise.
1584 (cse_insn): Likewise.
1585 * cselib.c (cselib_lookup_1): Likewise.
1586 * dojump.c (do_jump_1): Likewise.
1587 (do_compare_rtx_and_jump): Likewise.
1588 * dse.c (get_call_args): Likewise.
1589 * dwarf2out.c (rtl_for_decl_init): Likewise.
1590 (native_encode_initializer): Likewise.
1591 * expmed.c (emit_store_flag_1): Likewise.
1592 (emit_store_flag): Likewise.
1593 * expr.c (convert_modes): Likewise.
1594 (store_field): Likewise.
1595 (expand_expr_real_1): Likewise.
1596 * fold-const.c (fold_read_from_constant_string): Likewise.
1597 * gimple-ssa-sprintf.c (get_format_string): Likewise.
1598 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
1599 * optabs.c (expand_binop): Likewise.
1600 (expand_unop): Likewise.
1601 (expand_abs_nojump): Likewise.
1602 (expand_one_cmpl_abs_nojump): Likewise.
1603 * simplify-rtx.c (mode_signbit_p): Likewise.
1604 (val_signbit_p): Likewise.
1605 (val_signbit_known_set_p): Likewise.
1606 (val_signbit_known_clear_p): Likewise.
1607 (simplify_relational_operation_1): Likewise.
1608 * tree.c (vector_type_mode): Likewise.
1610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1611 Alan Hayward <alan.hayward@arm.com>
1612 David Sherwood <david.sherwood@arm.com>
1614 * machmode.h (smallest_mode_for_size): Fix formatting.
1615 (smallest_int_mode_for_size): New function.
1616 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
1617 instead of smallest_mode_for_size.
1618 * combine.c (make_extraction): Likewise.
1619 * config/arc/arc.c (arc_expand_movmem): Likewise.
1620 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
1621 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
1622 * config/s390/s390.c (s390_expand_insv): Likewise.
1623 * config/sparc/sparc.c (assign_int_registers): Likewise.
1624 * config/spu/spu.c (spu_function_value): Likewise.
1625 (spu_function_arg): Likewise.
1626 * coverage.c (get_gcov_type): Likewise.
1627 (get_gcov_unsigned_t): Likewise.
1628 * dse.c (find_shift_sequence): Likewise.
1629 * expmed.c (store_bit_field_1): Likewise.
1630 * expr.c (convert_move): Likewise.
1631 (store_field): Likewise.
1632 * internal-fn.c (expand_arith_overflow): Likewise.
1633 * optabs-query.c (get_best_extraction_insn): Likewise.
1634 * optabs.c (expand_twoval_binop_libfunc): Likewise.
1635 * stor-layout.c (layout_type): Likewise.
1636 (initialize_sizetypes): Likewise.
1637 * targhooks.c (default_get_mask_mode): Likewise.
1638 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
1640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1641 Alan Hayward <alan.hayward@arm.com>
1642 David Sherwood <david.sherwood@arm.com>
1644 * machmode.h (opt_mode::else_blk): New function.
1645 (int_mode_for_mode): Declare.
1646 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
1647 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
1649 * cfgexpand.c (expand_debug_expr): Likewise.
1650 * combine.c (gen_lowpart_or_truncate): Likewise.
1651 (gen_lowpart_for_combine): Likewise.
1652 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
1653 * config/avr/avr.c (avr_to_int_mode): Likewise.
1654 (avr_out_plus_1): Likewise.
1655 (avr_out_plus): Likewise.
1656 (avr_out_round): Likewise.
1657 * config/i386/i386.c (ix86_split_to_parts): Likewise.
1658 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
1659 (s390_expand_vcond): Likewise.
1660 * config/spu/spu.c (spu_split_immediate): Likewise.
1661 (spu_expand_mov): Likewise.
1662 * dse.c (get_stored_val): Likewise.
1663 * expmed.c (store_bit_field_1): Likewise.
1664 (convert_extracted_bit_field): Use int_mode_for_mode instead of
1666 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
1667 (extract_low_bits): Likewise.
1668 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
1669 handling rather than repeating the check.
1670 (emit_group_store): Likewise.
1671 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
1672 * optabs.c (expand_absneg_bit): Likewise.
1673 (expand_copysign_absneg): Likewise.
1674 (expand_copysign_bit): Likewise.
1675 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
1676 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
1677 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
1678 * var-tracking.c (prepare_call_arguments): Likewise.
1679 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
1680 int_mode_for_mode instead of mode_for_size.
1681 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
1683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1684 Alan Hayward <alan.hayward@arm.com>
1685 David Sherwood <david.sherwood@arm.com>
1687 * machmode.h (int_mode_for_size): New function.
1688 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
1689 instead of mode_for_size.
1690 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
1692 * combine.c (expand_field_assignment): Use int_mode_for_size
1693 instead of mode_for_size.
1694 (make_extraction): Likewise.
1695 (simplify_shift_const_1): Likewise.
1696 (simplify_comparison): Likewise.
1697 * dojump.c (do_jump): Likewise.
1698 * dwarf2out.c (mem_loc_descriptor): Likewise.
1699 * emit-rtl.c (init_derived_machine_modes): Likewise.
1700 * expmed.c (flip_storage_order): Likewise.
1701 (convert_extracted_bit_field): Likewise.
1702 * expr.c (copy_blkmode_from_reg): Likewise.
1703 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
1704 * internal-fn.c (expand_mul_overflow): Likewise.
1705 * lower-subreg.c (simple_move): Likewise.
1706 * optabs-libfuncs.c (init_optabs): Likewise.
1707 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1708 * tree.c (vector_type_mode): Likewise.
1709 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
1710 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
1711 * tree-vect-generic.c (expand_vector_parallel): Likewise.
1712 * tree-vect-stmts.c (vectorizable_load): Likewise.
1713 (vectorizable_store): Likewise.
1715 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1716 Alan Hayward <alan.hayward@arm.com>
1717 David Sherwood <david.sherwood@arm.com>
1719 * coretypes.h (pod_mode): New type.
1720 (scalar_int_mode_pod): New typedef.
1721 * machmode.h (pod_mode): New class.
1722 (int_n_data_t::m): Change type to scalar_int_mode_pod.
1723 * genmodes.c (emit_mode_int_n): Update accordingly.
1724 * lower-subreg.h (target_lower_subreg): Change type to
1725 scalar_int_mode_pod.
1726 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
1727 scalar_int_mode_pod.
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 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
1734 machine_mode to scalar_int_mode.
1735 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
1736 (rs6000_option_override_internal): Remove cast to int.
1737 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
1738 machine_mode to scalar_int_mode.
1739 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
1740 (rs6000_option_override_internal): Remove cast to int.
1741 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
1742 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
1744 * config/s390/s390.c (s390_expand_builtin): Likewise.
1745 * coretypes.h (scalar_int_mode): New type.
1746 (opt_scalar_int_mode): New typedef.
1747 * machmode.h (scalar_int_mode): New class.
1748 (scalar_int_mode::includes_p): New function.
1749 (byte_mode): Change type to scalar_int_mode.
1750 (word_mode): Likewise.
1751 (ptr_mode): Likewise.
1752 * emit-rtl.c (byte_mode): Likewise.
1753 (word_mode): Likewise.
1754 (ptr_mode): Likewise.
1755 (init_derived_machine_modes): Update accordingly.
1756 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
1757 and MODE_PARTIAL_INT.
1758 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
1759 opt_scalar_int_mode.
1761 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1762 Alan Hayward <alan.hayward@arm.com>
1763 David Sherwood <david.sherwood@arm.com>
1765 * target.def (libgcc_floating_mode_supported_p): Take a
1767 * doc/tm.texi: Regenerate.
1768 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
1770 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
1771 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
1774 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1775 Alan Hayward <alan.hayward@arm.com>
1776 David Sherwood <david.sherwood@arm.com>
1778 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
1779 * doc/tm.texi: Regenerate.
1780 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
1781 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
1782 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
1783 * targhooks.h (default_floatn_mode): Likewise.
1784 * targhooks.c (default_floatn_mode): Likewise.
1785 * tree.c (build_common_tree_nodes): Update accordingly.
1787 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1788 Alan Hayward <alan.hayward@arm.com>
1789 David Sherwood <david.sherwood@arm.com>
1791 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
1792 (mode_iterator::iterate_p): Likewise.
1793 (mode_iterator::get_wider): Likewise.
1794 * expr.c (init_expr_target): Use opt_scalar_float_mode.
1796 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1797 Alan Hayward <alan.hayward@arm.com>
1798 David Sherwood <david.sherwood@arm.com>
1800 * coretypes.h (opt_scalar_float_mode): New typedef.
1801 * machmode.h (float_mode_for_size): New function.
1802 * emit-rtl.c (double_mode): Delete.
1803 (init_emit_once): Use float_mode_for_size.
1804 * stor-layout.c (layout_type): Likewise.
1805 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
1807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1808 Alan Hayward <alan.hayward@arm.com>
1809 David Sherwood <david.sherwood@arm.com>
1811 * output.h (assemble_real): Take a scalar_float_mode.
1812 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
1813 * config/arm/arm.md (consttable_4): Likewise.
1814 (consttable_8): Likewise.
1815 (consttable_16): Likewise.
1816 * config/mips/mips.md (consttable_float): Likewise.
1817 * config/s390/s390.c (s390_output_pool_entry): Likewise.
1818 * varasm.c (assemble_real): Take a scalar_float_mode.
1819 (output_constant_pool_2): Update accordingly.
1820 (output_constant): Likewise.
1822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1823 Alan Hayward <alan.hayward@arm.com>
1824 David Sherwood <david.sherwood@arm.com>
1826 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
1827 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
1828 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
1829 (native_encode_real): Likewise.
1830 (native_interpret_real): Likewise.
1831 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
1832 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1834 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1835 Alan Hayward <alan.hayward@arm.com>
1836 David Sherwood <david.sherwood@arm.com>
1838 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
1839 <scalar_float_mode>. Simplify.
1840 (gen_extend_conv_libfunc): Likewise.
1842 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1843 Alan Hayward <alan.hayward@arm.com>
1844 David Sherwood <david.sherwood@arm.com>
1846 * coretypes.h (scalar_float_mode): New type.
1847 * machmode.h (mode_traits::from_int): Use machine_mode if
1848 USE_ENUM_MODES is defined.
1849 (is_a): New function.
1851 (dyn_cast): Likewise.
1852 (scalar_float_mode): New class.
1853 (scalar_float_mode::includes_p): New function.
1854 (is_float_mode): Likewise.
1855 * gdbhooks.py (MachineModePrinter): New class.
1856 (build_pretty_printer): Use it for scalar_float_mode.
1857 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
1858 (format_helper::format_helper): Turn into a template.
1859 * genmodes.c (get_mode_class): New function.
1860 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
1861 or machine_mode if none.
1862 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
1863 as_a <scalar_float_mode>.
1864 * dwarf2out.c (mem_loc_descriptor): Likewise.
1865 (insert_float): Likewise.
1866 (add_const_value_attribute): Likewise.
1867 * simplify-rtx.c (simplify_immed_subreg): Likewise.
1868 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
1869 (expand_unop): Update accordingly.
1870 (expand_abs_nojump): Likewise.
1871 (expand_copysign_absneg): Take a scalar_float_mode.
1872 (expand_copysign_bit): Likewise.
1873 (expand_copysign): Update accordingly.
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 * coretypes.h (opt_mode): New class.
1880 * machmode.h (opt_mode): Likewise.
1881 (opt_mode::else_void): New function.
1882 (opt_mode::require): Likewise.
1883 (opt_mode::exists): Likewise.
1884 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
1885 (GET_MODE_2XWIDER_MODE): Likewise.
1886 (mode_iterator::get_wider): Update accordingly.
1887 (mode_iterator::get_2xwider): Likewise.
1888 (mode_iterator::get_known_wider): Likewise, turning into a template.
1889 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
1890 forcing a wider mode to exist.
1891 * config/cr16/cr16.h (LONG_REG_P): Likewise.
1892 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1893 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
1894 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
1895 * lower-subreg.c (init_lower_subreg): Likewise.
1896 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
1897 on the final iteration.
1898 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
1899 a wider mode exists before asking for a move pattern.
1900 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
1901 forcing a wider mode to exist.
1902 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
1903 returning false if no such mode exists.
1904 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
1905 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
1906 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
1907 Avoid checking for a MODE_INT if we already know the mode is not a
1909 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
1910 forcing a wider mode to exist.
1911 (expmed_mult_highpart_optab): Likewise.
1912 (expmed_mult_highpart): Likewise.
1913 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
1915 * lto-streamer-in.c (lto_input_mode_table): Likewise.
1916 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
1917 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1918 * internal-fn.c (expand_mul_overflow): Update use of
1919 GET_MODE_2XWIDER_MODE.
1920 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1921 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
1922 GET_MODE_WIDER_MODE.
1923 (convert_plusminus_to_widen): Likewise.
1924 * tree-switch-conversion.c (array_value_type): Likewise.
1925 * var-tracking.c (emit_note_insn_var_location): Likewise.
1926 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1927 Return false inside rather than outside the loop if no wider mode
1929 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
1930 and GET_MODE_2XWIDER_MODE
1931 (can_compare_p): Use else_void.
1932 * gdbhooks.py (OptMachineModePrinter): New class.
1933 (build_pretty_printer): Use it for opt_mode.
1935 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1936 Alan Hayward <alan.hayward@arm.com>
1937 David Sherwood <david.sherwood@arm.com>
1939 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
1940 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
1942 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1943 Alan Hayward <alan.hayward@arm.com>
1944 David Sherwood <david.sherwood@arm.com>
1946 * machmode.h (mode_traits): New structure.
1947 (get_narrowest_mode): New function.
1948 (mode_iterator::start): Likewise.
1949 (mode_iterator::iterate_p): Likewise.
1950 (mode_iterator::get_wider): Likewise.
1951 (mode_iterator::get_known_wider): Likewise.
1952 (mode_iterator::get_2xwider): Likewise.
1953 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
1954 (FOR_EACH_MODE): Likewise.
1955 (FOR_EACH_MODE_FROM): Likewise.
1956 (FOR_EACH_MODE_UNTIL): Likewise.
1957 (FOR_EACH_WIDER_MODE): Likewise.
1958 (FOR_EACH_2XWIDER_MODE): Likewise.
1959 * builtins.c (expand_builtin_strlen): Use new mode iterators.
1960 * combine.c (simplify_comparison): Likewise
1961 * config/i386/i386.c (type_natural_mode): Likewise.
1962 * cse.c (cse_insn): Likewise.
1963 * dse.c (find_shift_sequence): Likewise.
1964 * emit-rtl.c (init_derived_machine_modes): Likewise.
1965 (init_emit_once): Likewise.
1966 * explow.c (hard_function_value): Likewise.
1967 * expmed.c (extract_fixed_bit_field_1): Likewise.
1968 (extract_bit_field_1): Likewise.
1969 (expand_divmod): Likewise.
1970 (emit_store_flag_1): Likewise.
1971 * expr.c (init_expr_target): Likewise.
1972 (convert_move): Likewise.
1973 (alignment_for_piecewise_move): Likewise.
1974 (widest_int_mode_for_size): Likewise.
1975 (emit_block_move_via_movmem): Likewise.
1976 (copy_blkmode_to_reg): Likewise.
1977 (set_storage_via_setmem): Likewise.
1978 (compress_float_constant): Likewise.
1979 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1980 * optabs-query.c (get_best_extraction_insn): Likewise.
1981 * optabs.c (expand_binop): Likewise.
1982 (expand_twoval_unop): Likewise.
1983 (expand_twoval_binop): Likewise.
1984 (widen_leading): Likewise.
1985 (widen_bswap): Likewise.
1986 (expand_parity): Likewise.
1987 (expand_unop): Likewise.
1988 (prepare_cmp_insn): Likewise.
1989 (prepare_float_lib_cmp): Likewise.
1990 (expand_float): Likewise.
1991 (expand_fix): Likewise.
1992 (expand_sfix_optab): Likewise.
1993 * postreload.c (move2add_use_add2_insn): Likewise.
1994 * reg-stack.c (reg_to_stack): Likewise.
1995 * reginfo.c (choose_hard_reg_mode): Likewise.
1996 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1997 * stor-layout.c (mode_for_size): Likewise.
1998 (smallest_mode_for_size): Likewise.
1999 (mode_for_vector): Likewise.
2000 (finish_bitfield_representative): Likewise.
2001 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
2002 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
2003 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2004 * var-tracking.c (prepare_call_arguments): Likewise.
2006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2007 Alan Hayward <alan.hayward@arm.com>
2008 David Sherwood <david.sherwood@arm.com>
2010 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
2011 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
2012 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
2013 * machmode.h (mode_size): Move earlier in file.
2014 (mode_precision): Likewise.
2015 (mode_inner): Likewise.
2016 (mode_nunits): Likewise.
2017 (mode_unit_size): Likewise.
2018 (unit_unit_precision): Likewise.
2019 (mode_wider): Likewise.
2020 (mode_2xwider): Likewise.
2021 (machine_mode): New class.
2022 (mode_to_bytes): New function.
2023 (mode_to_bits): Likewise.
2024 (mode_to_precision): Likewise.
2025 (mode_to_inner): Likewise.
2026 (mode_to_unit_size): Likewise.
2027 (mode_to_unit_precision): Likewise.
2028 (mode_to_nunits): Likewise.
2029 (GET_MODE_SIZE): Use mode_to_bytes.
2030 (GET_MODE_BITSIZE): Use mode_to_bits.
2031 (GET_MODE_PRECISION): Use mode_to_precision.
2032 (GET_MODE_INNER): Use mode_to_inner.
2033 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
2034 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
2035 (GET_MODE_NUNITS): Use mode_to_nunits.
2036 * system.h (ALWAYS_INLINE): New macro.
2037 * config/powerpcspe/powerpcspe-c.c
2038 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
2039 int for arg1_mode and arg2_mode.
2041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2042 Alan Hayward <alan.hayward@arm.com>
2043 David Sherwood <david.sherwood@arm.com>
2045 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2046 Prefix mode names with E_ in case statements.
2047 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2048 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
2049 (aarch64_split_simd_move): Likewise.
2050 (aarch64_gen_storewb_pair): Likewise.
2051 (aarch64_gen_loadwb_pair): Likewise.
2052 (aarch64_gen_store_pair): Likewise.
2053 (aarch64_gen_load_pair): Likewise.
2054 (aarch64_get_condition_code_1): Likewise.
2055 (aarch64_constant_pool_reload_icode): Likewise.
2056 (get_rsqrte_type): Likewise.
2057 (get_rsqrts_type): Likewise.
2058 (get_recpe_type): Likewise.
2059 (get_recps_type): Likewise.
2060 (aarch64_gimplify_va_arg_expr): Likewise.
2061 (aarch64_simd_container_mode): Likewise.
2062 (aarch64_emit_load_exclusive): Likewise.
2063 (aarch64_emit_store_exclusive): Likewise.
2064 (aarch64_expand_compare_and_swap): Likewise.
2065 (aarch64_gen_atomic_cas): Likewise.
2066 (aarch64_emit_bic): Likewise.
2067 (aarch64_emit_atomic_swap): Likewise.
2068 (aarch64_emit_atomic_load_op): Likewise.
2069 (aarch64_evpc_trn): Likewise.
2070 (aarch64_evpc_uzp): Likewise.
2071 (aarch64_evpc_zip): Likewise.
2072 (aarch64_evpc_ext): Likewise.
2073 (aarch64_evpc_rev): Likewise.
2074 (aarch64_evpc_dup): Likewise.
2075 (aarch64_gen_ccmp_first): Likewise.
2076 (aarch64_gen_ccmp_next): Likewise.
2077 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
2078 (alpha_emit_xfloating_libcall): Likewise.
2079 (emit_insxl): Likewise.
2080 (alpha_arg_type): Likewise.
2081 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
2082 (arc_preferred_simd_mode): Likewise.
2083 (arc_secondary_reload): Likewise.
2084 (get_arc_condition_code): Likewise.
2085 (arc_print_operand): Likewise.
2086 (arc_legitimate_constant_p): Likewise.
2087 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2088 * config/arc/arc.md (casesi_load): Likewise.
2089 (casesi_compact_jump): Likewise.
2090 * config/arc/predicates.md (proper_comparison_operator): Likewise.
2091 (cc_use_register): Likewise.
2092 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2093 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2094 (arm_init_iwmmxt_builtins): Likewise.
2095 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
2096 (neon_expand_vector_init): Likewise.
2097 (arm_attr_length_move_neon): Likewise.
2098 (maybe_get_arm_condition_code): Likewise.
2099 (arm_emit_vector_const): Likewise.
2100 (arm_preferred_simd_mode): Likewise.
2101 (arm_output_iwmmxt_tinsr): Likewise.
2102 (thumb1_output_casesi): Likewise.
2103 (thumb2_output_casesi): Likewise.
2104 (arm_emit_load_exclusive): Likewise.
2105 (arm_emit_store_exclusive): Likewise.
2106 (arm_expand_compare_and_swap): Likewise.
2107 (arm_evpc_neon_vuzp): Likewise.
2108 (arm_evpc_neon_vzip): Likewise.
2109 (arm_evpc_neon_vrev): Likewise.
2110 (arm_evpc_neon_vtrn): Likewise.
2111 (arm_evpc_neon_vext): Likewise.
2112 (arm_validize_comparison): Likewise.
2113 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
2114 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
2115 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
2116 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
2117 (c6x_preferred_simd_mode): Likewise.
2118 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
2119 (epiphany_rtx_costs): Likewise.
2120 * config/epiphany/predicates.md (proper_comparison_operator):
2122 * config/frv/frv.c (condexec_memory_operand): Likewise.
2123 (frv_emit_move): Likewise.
2124 (output_move_single): Likewise.
2125 (output_condmove_single): Likewise.
2126 (frv_hard_regno_mode_ok): Likewise.
2127 (frv_matching_accg_mode): Likewise.
2128 * config/h8300/h8300.c (split_adds_subs): Likewise.
2129 (h8300_rtx_costs): Likewise.
2130 (h8300_print_operand): Likewise.
2131 (compute_mov_length): Likewise.
2132 (output_logical_op): Likewise.
2133 (compute_logical_op_length): Likewise.
2134 (compute_logical_op_cc): Likewise.
2135 (h8300_shift_needs_scratch_p): Likewise.
2136 (output_a_shift): Likewise.
2137 (compute_a_shift_length): Likewise.
2138 (compute_a_shift_cc): Likewise.
2139 (expand_a_rotate): Likewise.
2140 (output_a_rotate): Likewise.
2141 * config/i386/i386.c (classify_argument): Likewise.
2142 (function_arg_advance_32): Likewise.
2143 (function_arg_32): Likewise.
2144 (function_arg_64): Likewise.
2145 (function_value_64): Likewise.
2146 (ix86_gimplify_va_arg): Likewise.
2147 (ix86_legitimate_constant_p): Likewise.
2148 (put_condition_code): Likewise.
2149 (split_double_mode): Likewise.
2150 (ix86_avx256_split_vector_move_misalign): Likewise.
2151 (ix86_expand_vector_logical_operator): Likewise.
2152 (ix86_split_idivmod): Likewise.
2153 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
2154 (ix86_build_const_vector): Likewise.
2155 (ix86_build_signbit_mask): Likewise.
2156 (ix86_match_ccmode): Likewise.
2157 (ix86_cc_modes_compatible): Likewise.
2158 (ix86_expand_branch): Likewise.
2159 (ix86_expand_sse_cmp): Likewise.
2160 (ix86_expand_sse_movcc): Likewise.
2161 (ix86_expand_int_sse_cmp): Likewise.
2162 (ix86_expand_vec_perm_vpermi2): Likewise.
2163 (ix86_expand_vec_perm): Likewise.
2164 (ix86_expand_sse_unpack): Likewise.
2165 (ix86_expand_int_addcc): Likewise.
2166 (ix86_split_to_parts): Likewise.
2167 (ix86_vectorize_builtin_gather): Likewise.
2168 (ix86_vectorize_builtin_scatter): Likewise.
2169 (avx_vpermilp_parallel): Likewise.
2170 (inline_memory_move_cost): Likewise.
2171 (ix86_tieable_integer_mode_p): Likewise.
2172 (x86_maybe_negate_const_int): Likewise.
2173 (ix86_expand_vector_init_duplicate): Likewise.
2174 (ix86_expand_vector_init_one_nonzero): Likewise.
2175 (ix86_expand_vector_init_one_var): Likewise.
2176 (ix86_expand_vector_init_concat): Likewise.
2177 (ix86_expand_vector_init_interleave): Likewise.
2178 (ix86_expand_vector_init_general): Likewise.
2179 (ix86_expand_vector_set): Likewise.
2180 (ix86_expand_vector_extract): Likewise.
2181 (emit_reduc_half): Likewise.
2182 (ix86_emit_i387_round): Likewise.
2183 (ix86_mangle_type): Likewise.
2184 (ix86_expand_round_sse4): Likewise.
2185 (expand_vec_perm_blend): Likewise.
2186 (canonicalize_vector_int_perm): Likewise.
2187 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
2188 (expand_vec_perm_1): Likewise.
2189 (expand_vec_perm_interleave3): Likewise.
2190 (expand_vec_perm_even_odd_pack): Likewise.
2191 (expand_vec_perm_even_odd_1): Likewise.
2192 (expand_vec_perm_broadcast_1): Likewise.
2193 (ix86_vectorize_vec_perm_const_ok): Likewise.
2194 (ix86_expand_vecop_qihi): Likewise.
2195 (ix86_expand_mul_widen_hilo): Likewise.
2196 (ix86_expand_sse2_abs): Likewise.
2197 (ix86_expand_pextr): Likewise.
2198 (ix86_expand_pinsr): Likewise.
2199 (ix86_preferred_simd_mode): Likewise.
2200 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
2201 * config/i386/sse.md (*andnot<mode>3): Likewise.
2202 (<mask_codefor><code><mode>3<mask_name>): Likewise.
2203 (*<code><mode>3): Likewise.
2204 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
2205 (ia64_expand_atomic_op): Likewise.
2206 (ia64_arg_type): Likewise.
2207 (ia64_mode_to_int): Likewise.
2208 (ia64_scalar_mode_supported_p): Likewise.
2209 (ia64_vector_mode_supported_p): Likewise.
2210 (expand_vec_perm_broadcast): Likewise.
2211 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
2212 (iq2000_function_arg_advance): Likewise.
2213 (iq2000_function_arg): Likewise.
2214 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
2215 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
2216 (m68k_libcall_value): Likewise.
2217 (m68k_function_value): Likewise.
2218 (sched_attr_op_type): Likewise.
2219 * config/mcore/mcore.c (mcore_output_move): Likewise.
2220 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
2222 (microblaze_function_arg): Likewise.
2223 * config/mips/mips.c (mips16_build_call_stub): Likewise.
2224 (mips_print_operand): Likewise.
2225 (mips_mode_ok_for_mov_fmt_p): Likewise.
2226 (mips_vector_mode_supported_p): Likewise.
2227 (mips_preferred_simd_mode): Likewise.
2228 (mips_expand_vpc_loongson_even_odd): Likewise.
2229 (mips_expand_vec_unpack): Likewise.
2230 (mips_expand_vi_broadcast): Likewise.
2231 (mips_expand_vector_init): Likewise.
2232 (mips_expand_vec_reduc): Likewise.
2233 (mips_expand_msa_cmp): Likewise.
2234 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
2235 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
2236 (cc_flags_for_mode): Likewise.
2237 * config/msp430/msp430.c (msp430_print_operand): Likewise.
2238 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
2239 (nds32_output_casesi_pc_relative): Likewise.
2240 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2241 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
2242 (nvptx_gen_unpack): Likewise.
2243 (nvptx_gen_pack): Likewise.
2244 (nvptx_gen_shuffle): Likewise.
2245 (nvptx_gen_wcast): Likewise.
2246 (nvptx_preferred_simd_mode): Likewise.
2247 * config/pa/pa.c (pa_secondary_reload): Likewise.
2248 * config/pa/predicates.md (base14_operand): Likewise.
2249 * config/powerpcspe/powerpcspe-c.c
2250 (altivec_resolve_overloaded_builtin): Likewise.
2251 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
2253 (rs6000_preferred_simd_mode): Likewise.
2254 (output_vec_const_move): Likewise.
2255 (rs6000_expand_vector_extract): Likewise.
2256 (rs6000_split_vec_extract_var): Likewise.
2257 (reg_offset_addressing_ok_p): Likewise.
2258 (rs6000_legitimate_offset_address_p): Likewise.
2259 (rs6000_legitimize_address): Likewise.
2260 (rs6000_emit_set_const): Likewise.
2261 (rs6000_const_vec): Likewise.
2262 (rs6000_emit_move): Likewise.
2263 (spe_build_register_parallel): Likewise.
2264 (rs6000_darwin64_record_arg_recurse): Likewise.
2265 (swap_selector_for_mode): Likewise.
2266 (spe_init_builtins): Likewise.
2267 (paired_init_builtins): Likewise.
2268 (altivec_init_builtins): Likewise.
2269 (do_load_for_compare): Likewise.
2270 (rs6000_generate_compare): Likewise.
2271 (rs6000_expand_float128_convert): Likewise.
2272 (emit_load_locked): Likewise.
2273 (emit_store_conditional): Likewise.
2274 (rs6000_output_function_epilogue): Likewise.
2275 (rs6000_handle_altivec_attribute): Likewise.
2276 (rs6000_function_value): Likewise.
2277 (emit_fusion_gpr_load): Likewise.
2278 (emit_fusion_p9_load): Likewise.
2279 (emit_fusion_p9_store): Likewise.
2280 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
2281 (fusion_gpr_mem_load): Likewise.
2282 (fusion_addis_mem_combo_load): Likewise.
2283 (fusion_addis_mem_combo_store): Likewise.
2284 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
2285 (fusion_gpr_mem_load): Likewise.
2286 (fusion_addis_mem_combo_load): Likewise.
2287 (fusion_addis_mem_combo_store): Likewise.
2288 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2290 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
2291 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
2292 (rs6000_preferred_simd_mode): Likewise.
2293 (output_vec_const_move): Likewise.
2294 (rs6000_expand_vector_extract): Likewise.
2295 (rs6000_split_vec_extract_var): Likewise.
2296 (reg_offset_addressing_ok_p): Likewise.
2297 (rs6000_legitimate_offset_address_p): Likewise.
2298 (rs6000_legitimize_address): Likewise.
2299 (rs6000_emit_set_const): Likewise.
2300 (rs6000_const_vec): Likewise.
2301 (rs6000_emit_move): Likewise.
2302 (rs6000_darwin64_record_arg_recurse): Likewise.
2303 (swap_selector_for_mode): Likewise.
2304 (paired_init_builtins): Likewise.
2305 (altivec_init_builtins): Likewise.
2306 (rs6000_expand_float128_convert): Likewise.
2307 (emit_load_locked): Likewise.
2308 (emit_store_conditional): Likewise.
2309 (rs6000_output_function_epilogue): Likewise.
2310 (rs6000_handle_altivec_attribute): Likewise.
2311 (rs6000_function_value): Likewise.
2312 (emit_fusion_gpr_load): Likewise.
2313 (emit_fusion_p9_load): Likewise.
2314 (emit_fusion_p9_store): Likewise.
2315 * config/rx/rx.c (rx_gen_move_template): Likewise.
2316 (flags_from_mode): Likewise.
2317 * config/s390/predicates.md (s390_alc_comparison): Likewise.
2318 (s390_slb_comparison): Likewise.
2319 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
2320 (s390_vector_mode_supported_p): Likewise.
2321 (s390_cc_modes_compatible): Likewise.
2322 (s390_match_ccmode_set): Likewise.
2323 (s390_canonicalize_comparison): Likewise.
2324 (s390_emit_compare_and_swap): Likewise.
2325 (s390_branch_condition_mask): Likewise.
2326 (s390_rtx_costs): Likewise.
2327 (s390_secondary_reload): Likewise.
2328 (__SECONDARY_RELOAD_CASE): Likewise.
2329 (s390_expand_cs): Likewise.
2330 (s390_preferred_simd_mode): Likewise.
2331 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
2332 * config/sh/sh.c (sh_print_operand): Likewise.
2333 (dump_table): Likewise.
2334 (sh_secondary_reload): Likewise.
2335 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2336 * config/sh/sh.md (casesi_worker_1): Likewise.
2337 (casesi_worker_2): Likewise.
2338 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
2339 (fcc_comparison_operator): Likewise.
2340 * config/sparc/sparc.c (sparc_expand_move): Likewise.
2341 (emit_soft_tfmode_cvt): Likewise.
2342 (sparc_preferred_simd_mode): Likewise.
2343 (output_cbranch): Likewise.
2344 (sparc_print_operand): Likewise.
2345 (sparc_expand_vec_perm_bmask): Likewise.
2346 (vector_init_bshuffle): Likewise.
2347 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
2348 (spu_vector_mode_supported_p): Likewise.
2349 (spu_expand_insv): Likewise.
2350 (spu_emit_branch_or_set): Likewise.
2351 (spu_handle_vector_attribute): Likewise.
2352 (spu_builtin_splats): Likewise.
2353 (spu_builtin_extract): Likewise.
2354 (spu_builtin_promote): Likewise.
2355 (spu_expand_sign_extend): Likewise.
2356 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
2357 (tilegx_simd_int): Likewise.
2358 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
2359 (tilepro_simd_int): Likewise.
2360 * config/v850/v850.c (const_double_split): Likewise.
2361 (v850_print_operand): Likewise.
2362 (ep_memory_offset): Likewise.
2363 * config/vax/vax.c (vax_rtx_costs): Likewise.
2364 (vax_output_int_move): Likewise.
2365 (vax_output_int_add): Likewise.
2366 (vax_output_int_subtract): Likewise.
2367 * config/visium/predicates.md (visium_branch_operator): Likewise.
2368 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
2369 (visium_print_operand_address): Likewise.
2370 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2371 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
2372 (xtensa_expand_conditional_branch): Likewise.
2373 (xtensa_copy_incoming_a7): Likewise.
2374 (xtensa_output_literal): Likewise.
2375 * dfp.c (decimal_real_maxval): Likewise.
2376 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
2378 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2379 Alan Hayward <alan.hayward@arm.com>
2380 David Sherwood <david.sherwood@arm.com>
2382 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
2383 (mode_nunits_inline): Likewise.
2384 (mode_inner_inline): Likewise.
2385 (mode_unit_size_inline): Likewise.
2386 (mode_unit_precision_inline): Likewise.
2387 (emit_insn_modes_h): Likewise. Also emit a #define of the
2389 (emit_mode_wider): Add an E_ prefix to mode names.
2390 (emit_mode_complex): Likewise.
2391 (emit_mode_inner): Likewise.
2392 (emit_mode_adjustments): Likewise.
2393 (emit_mode_int_n): Likewise.
2394 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
2395 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
2396 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
2397 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
2398 (CRC32_BUILTIN, ENTRY): Likewise.
2399 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
2400 (aarch64_pop_regs): Likewise.
2401 (aarch64_process_components): Likewise.
2402 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
2403 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
2404 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
2405 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
2406 * config/arm/arm.c (arm_init_libfuncs): Likewise.
2407 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
2409 * config/i386/i386-builtin.def (pcmpestr): Likewise.
2410 (pcmpistr): Likewise.
2411 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
2412 * config/mmix/mmix.c (mmix_output_condition): Likewise.
2413 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
2415 * config/rl78/rl78.c (mduc_regs): Likewise.
2416 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2417 (htm_expand_builtin): Likewise.
2418 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
2419 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
2420 * config/xtensa/xtensa.c (print_operand): Likewise.
2421 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
2422 (NUM_MODE_VECTOR_INT): Likewise.
2423 * genoutput.c (null_operand): Likewise.
2424 (output_operand_data): Likewise.
2425 * genrecog.c (print_parameter_value): Likewise.
2426 * lra.c (debug_operand_data): Likewise.
2428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2430 * dwarf2out.c (scompare_loc_descriptor_wide)
2431 (scompare_loc_descriptor_narrow): New functions, split out from...
2432 (scompare_loc_descriptor): ...here.
2433 * expmed.c (emit_store_flag_int): New function, split out from...
2434 (emit_store_flag): ...here.
2436 2017-08-30 Richard Biener <rguenther@suse.de>
2438 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
2439 (dwarf2out_early_finish): Move setting of AT_pubnames from
2440 early debug output to early finish.
2442 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
2444 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
2445 and -mdata-region to the assembler.
2447 2017-08-30 Richard Biener <rguenther@suse.de>
2449 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
2451 (gen_subprogram_die): Add DW_AT_object_pointer only early.
2452 (dwarf2out_early_global_decl): Only generate a DIE for the
2453 abstract origin if it doesn't already exist or is a declaration DIE.
2454 (resolve_addr): Do not add the linkage name twice when
2455 generating a stub DIE for the DW_TAG_GNU_call_site target.
2457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2459 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2460 Use machine_mode rather than int for arg1_mode.
2462 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
2465 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
2466 that the second argument of the built-in functions to unpack
2467 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
2468 switch statement instead a lot of if statements.
2469 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
2470 Allow 64-bit values to be in Altivec registers as well as
2471 traditional floating point registers.
2472 (pack<mode>, FMOVE128_VSX iterator): Likewise.
2474 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
2476 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
2477 MAX_REGS_PER_ADDRESS == 1.
2479 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
2481 * config/i386/i386.opt (flag_fentry): Do not init to -1.
2482 * config/i386/i386.c (ix86_option_override_internal): Simplify
2483 setting of opts->x_flag_entry.
2485 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2486 Jakub Jelinek <jakub@redhat.com>
2487 Richard Biener <rguenther@suse.de>
2489 PR tree-optimization/81503
2490 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
2491 folded constant fits in the target type; reorder tests for clarity.
2493 2017-08-29 Martin Liska <mliska@suse.cz>
2495 * passes.def: Include pass_lower_switch.
2496 * stmt.c (dump_case_nodes): Remove and move to
2497 tree-switch-conversion.
2498 (case_values_threshold): Likewise.
2499 (expand_switch_as_decision_tree_p): Likewise.
2500 (emit_case_decision_tree): Likewise.
2501 (expand_case): Likewise.
2502 (balance_case_nodes): Likewise.
2503 (node_has_low_bound): Likewise.
2504 (node_has_high_bound): Likewise.
2505 (node_is_bounded): Likewise.
2506 (emit_case_nodes): Likewise.
2507 (struct simple_case_node): New struct.
2508 (add_case_node): Remove.
2509 (emit_case_dispatch_table): Use vector instead of case_list.
2510 (reset_out_edges_aux): Remove.
2511 (compute_cases_per_edge): Likewise.
2512 (expand_case): Build list of simple_case_node.
2513 (expand_sjlj_dispatch_table): Use it.
2514 * tree-switch-conversion.c (struct case_node): Moved from
2515 stmt.c and adjusted.
2516 (emit_case_nodes): Likewise.
2517 (node_has_low_bound): Likewise.
2518 (node_has_high_bound): Likewise.
2519 (node_is_bounded): Likewise.
2520 (case_values_threshold): Likewise.
2521 (reset_out_edges_aux): Likewise.
2522 (compute_cases_per_edge): Likewise.
2523 (add_case_node): Likewise.
2524 (dump_case_nodes): Likewise.
2525 (balance_case_nodes): Likewise.
2526 (expand_switch_as_decision_tree_p): Likewise.
2527 (emit_jump): Likewise.
2528 (emit_case_decision_tree): Likewise.
2529 (try_switch_expansion): Likewise.
2530 (do_jump_if_equal): Likewise.
2531 (emit_cmp_and_jump_insns): Likewise.
2532 (fix_phi_operands_for_edge): New function.
2533 (record_phi_operand_mapping): Likewise.
2534 (class pass_lower_switch): New pass.
2535 (pass_lower_switch::execute): New function.
2536 (make_pass_lower_switch): Likewise.
2537 (conditional_probability):
2538 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
2539 * tree-pass.h: Add make_pass_lower_switch.
2541 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
2544 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
2547 2017-08-29 Richard Biener <rguenther@suse.de>
2549 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
2550 we do not add a DW_AT_inline attribute twice.
2551 (gen_subprogram_die): Remove code setting DW_AT_inline on
2552 DECL_ABSTRACT_P nodes.
2554 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
2556 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
2557 calls to internal functions.
2558 (gimplify_modify_expr): Likewise.
2559 * tree-call-cdce.c (use_internal_fn): Likewise.
2560 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
2561 (convert_to_divmod): Set the nothrow flag.
2562 * tree-if-conv.c (predicate_mem_writes): Likewise.
2563 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
2564 (vectorizable_call): Likewise.
2565 (vectorizable_store): Likewise.
2566 (vectorizable_load): Likewise.
2567 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
2568 (vect_recog_mask_conversion_pattern): Likewise.
2570 2017-08-29 Martin Liska <mliska@suse.cz>
2573 * gcc.c (driver_handle_option): Add new argument.
2574 * opts-common.c (handle_option): Pass
2575 target_option_override_hook.
2576 * opts-global.c (lang_handle_option): Add new option.
2577 (set_default_handlers): Add new argument.
2578 (decode_options): Likewise.
2579 * opts.c (target_handle_option): Likewise.
2580 (common_handle_option): Call target_option_override_hook.
2581 * opts.h (struct cl_option_handler_func): Add hook for
2582 target option override.
2583 (struct cl_option_handlers): Likewise.
2584 (set_default_handlers): Add new argument.
2585 (decode_options): Likewise.
2586 (common_handle_option): Likewise.
2587 (target_handle_option): Likewise.
2588 * toplev.c (toplev::main): Pass targetm.target_option.override
2591 2017-08-29 Richard Biener <rguenther@suse.de>
2592 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2594 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
2595 life to the active subtree.
2597 2017-08-28 Jeff Law <law@redhat.com>
2599 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
2600 derive_equivalences.
2601 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
2603 (edge_info::derive_equivalences): New private member function
2605 * tree-ssa-dom.c (class edge_info): Changed from a struct
2606 to a class. Add ctor/dtor, methods and data members.
2607 (edge_info::edge_info): Renamed from allocate_edge_info.
2608 Initialize additional members.
2609 (edge_info::~edge_info): New.
2610 (free_dom_edge_info): Delete the edge info.
2611 (record_edge_info): Use new class & associated member functions.
2612 Tighten forms for testing for edge equivalences.
2613 (record_temporary_equivalences): Iterate over the simple
2614 equivalences rather than assuming there's only one per edge.
2615 (cprop_into_successor_phis): Iterate over the simple
2616 equivalences rather than assuming there's only one per edge.
2617 (optimize_stmt): Use operand_equal_p rather than pointer
2618 equality for mini-DSE code.
2620 2017-08-28 Nathan Sidwell <nathan@acm.org>
2622 * gcc.c (execute): Fold SIGPIPE handling into switch
2623 statement. Adjust internal error message.
2625 2017-08-28 Richard Biener <rguenther@suse.de>
2628 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
2629 Do nothing for removed DIEs.
2631 2017-08-28 Richard Biener <rguenther@suse.de>
2633 PR tree-optimization/81977
2634 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
2637 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
2640 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
2641 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
2642 using targetm.gen_mem_thread_fence.
2644 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
2647 * config/i386/i386.md (*<btsc><mode>): Change operand 2
2648 predicate to register_operand. Reorder operands.
2649 (*btr<mode>): Ditto.
2650 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
2651 (*btr<mode>_mask): Ditto.
2653 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
2655 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
2656 * config/rs6000/xmmintrin.h: New file.
2657 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
2659 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2662 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
2663 parameter and_insn and return it.
2664 (recombine_lvx_pattern): Insert a copy to ensure availability of
2665 the base register of the copied masking operation at the point of
2666 the instruction replacement.
2667 (recombine_stvx_pattern): Likewise.
2669 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
2671 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
2672 undocumented switches.
2673 (-mpower9-dform-vector): Likewise.
2674 (-mpower9-dform): Likewise.
2675 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
2676 comments to delete references to -mpower9-dform* switches.
2677 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
2678 Delete reference to -mpower9-dform* switches, test for
2679 -mpower9-vector instead.
2680 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
2681 (OTHER_P9_VECTOR_MASKS): Likewise.
2682 (POWERPC_MASKS): Likewise.
2683 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
2684 tests against -mpower9-dform* to -mpower9-vector. Delete code
2685 that checked for -mpower9-dform* consistancy with other options.
2686 Add test for -mpower9-misc to enable other power9 switches.
2687 (rs6000_init_hard_regno_mode_ok): Likewise.
2688 (rs6000_option_override_internal): Likewise.
2689 (rs6000_emit_prologue): Likewise.
2690 (rs6000_emit_epilogue): Likewise.
2691 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
2692 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
2693 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
2695 (emit_fusion_p9_store): Likewise.
2696 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
2697 resetting these macros if the assembler does not support ISA 3.0
2699 (TARGET_P9_DFORM_VECTOR): Likewise.
2700 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
2701 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
2703 2017-08-25 Alan Modra <amodra@gmail.com>
2705 PR rtl-optimization/81747
2706 * cse.c (cse_extended_basic_block): Don't attempt to record
2707 equivalences for degenerate conditional branches that branch
2708 to their fall-through.
2710 2017-08-24 Martin Sebor <msebor@redhat.com>
2713 * gimple-fold.c (size_must_be_zero_p): New function.
2714 (gimple_fold_builtin_memory_op): Call it.
2716 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
2718 * config/rs6000/mm_malloc.h: New file.
2720 2017-08-24 Bin Cheng <bin.cheng@arm.com>
2722 PR tree-optimization/81913
2723 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
2724 analysis when either IVs in condition can wrap.
2726 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
2728 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
2729 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
2731 2017-08-24 Richard Biener <rguenther@suse.de>
2734 * targhooks.c (default_target_can_inline_p): Properly
2735 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
2736 is present and always compare.
2737 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
2738 infer -mfpmath=sse from TARGET_SSE_P.
2739 (ix86_can_inline_p): Properly use target_option_default_node when
2740 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
2742 2017-08-24 Richard Biener <rguenther@suse.de>
2745 * dwarf2out.c (output_die): Handle flag_generate_offload like
2747 (output_comp_unit): Likewise.
2748 (gen_array_type_die): Likewise.
2749 (dwarf2out_early_finish): Likewise.
2750 (note_variable_value_in_expr): Likewise.
2751 (dwarf2out_finish): Likewise. Adjust assert.
2752 * cgraphunit.c (symbol_table::compile): Move setting of
2753 flag_generate_offload earlier ...
2754 (symbol_table::finalize_compilation_unit): ... here, before
2755 early debug finalization.
2757 2017-08-24 Richard Biener <rguenther@suse.de>
2759 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
2760 and ipa-fnsummary.h.
2761 (ix86_can_inline_p): When ix86_fpmath flags do not match
2762 check whether the callee uses FP math at all.
2764 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
2767 * tree-ssanames.c (get_nonzero_bits): Use element_precision
2768 instead of TYPE_PRECISION.
2770 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
2771 Alan Hayward <alan.hayward@arm.com>
2772 David Sherwood <david.sherwood@arm.com>
2774 * combine.c (make_extraction): Use subreg_offset_from_lsb.
2776 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
2778 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
2780 (ix86_frame::stack_realign_allocate): New field.
2781 (struct machine_frame_state): Modify comments.
2782 (machine_frame_state::sp_realigned_fp_end): New field.
2783 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
2785 (sp_valid_at): Add assertion to assure no attempt to access invalid
2786 offset of a realigned stack.
2787 (fp_valid_at): Likewise.
2788 (choose_baseaddr): Modify comments.
2789 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
2790 ix86_expand_prologue.
2791 (ix86_expand_prologue): Modify stack realignment and allocation.
2792 (ix86_expand_epilogue): Modify comments.
2793 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
2794 avx2_runtime, avx512f, and avx512f_runtime.
2796 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
2798 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
2799 (mstackrealign): Do not init to -1.
2800 * config/i386/i386.c (ix86_option_override_internal):
2801 Check opts_set, not opts when setting default value of
2802 opts->x_ix86_force_align_arg_pointer.
2804 2017-08-23 Richard Biener <rguenther@suse.de>
2806 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
2807 lang_hooks.decl_printable_name.
2808 * print-rtl-function.c (print_rtx_function): Likewise.
2809 * tree-pretty-print.c (dump_function_header): Likewise.
2811 2017-08-23 Richard Biener <rguenther@suse.de>
2814 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
2815 -g0 at compile-time.
2817 2017-08-23 Tamar Christina <tamar.christina@arm.com>
2820 * doc/sourcebuild.texi (Other hardware attributes):
2823 2017-08-23 Tamar Christina <tamar.christina@arm.com>
2826 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
2827 Fix single-use check.
2829 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
2831 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
2833 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
2835 * doc/install.texi: Modify to add more details on running selected
2838 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
2840 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
2841 is combined with -mabi=ms.
2842 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
2845 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2847 PR tree-optimization/81488
2848 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
2849 and cached_basis fields.
2850 (MAX_SPREAD): New constant.
2851 (alloc_cand_and_find_basis): Initialize new fields.
2852 (clear_visited): New function.
2853 (create_phi_basis_1): Rename from create_phi_basis, set visited
2854 and cached_basis fields.
2855 (create_phi_basis): New wrapper function.
2856 (phi_add_costs_1): Rename from phi_add_costs, add spread
2857 parameter, set visited field, short-circuit when limits reached.
2858 (phi_add_costs): New wrapper function.
2859 (record_phi_increments_1): Rename from record_phi_increments, set
2861 (record_phi_increments): New wrapper function.
2862 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
2863 (phi_incr_cost): New wrapper function.
2864 (all_phi_incrs_profitable_1): Rename from
2865 all_phi_incrs_profitable, set visited field.
2866 (all_phi_incrs_profitable): New wrapper function.
2868 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
2869 Alan Hayward <alan.hayward@arm.com>
2870 David Sherwood <david.sherwood@arm.com>
2872 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
2873 that takes the outer and inner modes.
2874 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
2875 comparison as the canonical test for a paradoxical subreg.
2876 * combine.c (simplify_set): Use paradoxical_subreg_p.
2877 (make_extraction): Likewise.
2878 (force_to_mode): Likewise.
2879 (rtx_equal_for_field_assignment_p): Likewise.
2880 (gen_lowpart_for_combine): Likewise.
2881 (simplify_comparison): Likewise.
2882 * cse.c (equiv_constant): Likewise.
2883 * expmed.c (store_bit_field_1): Likewise.
2884 * final.c (alter_subreg): Likewise.
2885 * fwprop.c (propagate_rtx): Likewise.
2886 (forward_propagate_subreg): Likewise.
2887 * ira-conflicts.c (ira_build_conflicts): Likewise.
2888 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
2889 * lra-constraints.c (curr_insn_transform): Likewise.
2890 (split_reg): Likewise.
2891 * lra-eliminations.c (move_plus_up): Likewise.
2892 (lra_eliminate_regs_1): Likewise.
2893 * recog.c (general_operand): Likewise.
2894 * ree.c (combine_reaching_defs): Likewise.
2895 * reload.c (push_reload): Likewise.
2896 (find_reloads): Likewise.
2897 * reload1.c (elimination_effects): Likewise.
2898 (compute_reload_subreg_offset): Likewise.
2899 (choose_reload_regs): Likewise.
2900 * rtlanal.c (subreg_lsb_1): Likewise.
2901 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2902 (simplify_subreg): Likewise.
2903 * var-tracking.c (track_loc_p): Likewise.
2904 * emit-rtl.c (byte_lowpart_offset): Likewise.
2905 (paradoxical_subreg_p): Delete out-of-line definition.
2907 2017-08-22 Jeff Law <law@redhat.com>
2909 PR tree-optimization/81741
2910 PR tree-optimization/71947
2911 * tree-ssa-dom.c: Include tree-inline.h.
2912 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
2913 equivalences if one is more expensive to compute than the other.
2914 * tree-ssa-scopedtables.h (class const_or_copies): Make
2915 record_const_or_copy_raw method private.
2916 (class avail_exprs_stack): New method simplify_binary_operation.
2917 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
2918 avail_exprs_stack::simplify_binary_operation as needed.
2919 (avail_exprs_stack::simplify_binary_operation): New function.
2921 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
2923 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
2924 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
2925 (DOT_SYMBOLS): Likewise.
2926 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
2927 (RELOCATABLE_NEEDS_FIXUP): Likewise.
2928 (RS6000_ABI_NAME): Likewise.
2929 (TARGET_CMODEL): Likewise.
2930 (TOC_SECTION_ASM_OP): Likewise.
2931 (SET_CMODEL): New macro.
2932 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
2934 2017-08-22 Richard Biener <rguenther@suse.de>
2936 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
2937 to allow for free-lang-data replacements similar to verify_type_variant.
2939 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
2940 Alan Hayward <alan.hayward@arm.com>
2941 David Sherwood <david.sherwood@arm.com>
2943 * config/aarch64/aarch64.md (casesi): Use DImode rather than
2944 VOIDmode for the LABEL_REF.
2946 2017-08-22 Richard Biener <rguenther@suse.de>
2948 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
2950 2017-08-22 Richard Biener <rguenther@suse.de>
2952 * common.opt (feliminate-dwarf2-dups): Ignore.
2953 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
2954 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
2955 same_die_p_wrap, compute_section_prefix,
2956 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
2957 (comdat_symbol_id, comdat_symbol_number): Likewise.
2958 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
2960 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
2961 (output_die): Mark unreachable path unreachable.
2962 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
2963 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
2964 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
2965 (dwarf2out_early_finish): Likewise.
2967 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
2969 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
2971 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
2974 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
2975 not VAR_P. Filter attribute warnings with OPT_Wattributes.
2976 (avr_attribute_table) <io, io_low, address>: Initialize
2977 .decl_required with true.
2979 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
2981 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
2982 undocumented debugging options.
2983 (-mvsx-scalar-double): Likewise.
2984 (-mallow-df-permute): Likewise.
2985 (-mvectorize-builtins): Likewise.
2986 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2987 (rs6000_builtin_vectorized_function): Likewise.
2988 (rs6000_builtin_md_vectorized_function): Likewise.
2989 (rs6000_opt_vars): Likewise.
2991 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
2994 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
2995 (*btrq_imm): Rename from *btrq.
2996 (*btcq_imm): Rename from *btcq.
2997 (btsc): New code attribute.
2998 (*<btsc><mode>): New insn pattern.
2999 (*btr<mode>): Ditto.
3000 (*<btsc><mode>_mask): New insn_and_split pattern.
3001 (*btr<mode>_mask): Ditto.
3003 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3004 Alan Hayward <alan.hayward@arm.com>
3005 David Sherwood <david.sherwood@arm.com>
3007 * function.c (pad_below): Simplify padding calculation.
3009 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3010 Alan Hayward <alan.hayward@arm.com>
3011 David Sherwood <david.sherwood@arm.com>
3013 * target.def (function_prologue): Remove frame size argument.
3014 (function_epilogue): Likewise.
3015 * doc/tm.texi: Regenerate.
3016 * final.c (final_start_function): Update call to function_prologue.
3017 (final_end_function): Update call to function_epilogue.
3018 (default_function_pro_epilogue): Remove frame size argument.
3019 * output.h (default_function_pro_epilogue): Likewise.
3020 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
3021 (arm_output_function_prologue): Likewise.
3022 * config/frv/frv.c (frv_function_prologue): Likewise.
3023 (frv_function_epilogue): Likewise.
3024 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
3025 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
3026 (ia64_output_function_epilogue): Likewise.
3027 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
3028 (m32r_output_function_epilogue): Likewise.
3029 * config/microblaze/microblaze.c (microblaze_function_prologue)
3030 (microblaze_function_epilogue): Likewise.
3031 * config/mips/mips.c (mips_output_function_prologue): Likewise.
3032 (mips_output_function_epilogue): Likewise.
3033 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
3034 (mmix_target_asm_function_epilogue): Likewise.
3035 * config/msp430/msp430.c (msp430_start_function): Likewise.
3036 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
3037 (nds32_asm_function_epilogue): Likewise.
3038 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
3039 * config/pa/pa.c (pa_output_function_prologue): Likewise.
3040 (pa_output_function_epilogue): Likewise.
3041 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
3042 (rs6000_output_function_epilogue): Likewise.
3043 * config/rl78/rl78.c (rl78_start_function): Likewise.
3044 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
3045 (rs6000_output_function_epilogue): Likewise.
3046 * config/rx/rx.c (rx_output_function_prologue): Likewise.
3047 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
3048 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
3049 (sparc_asm_function_epilogue): Likewise.
3051 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3053 * tree.h (type_has_mode_precision_p): New function.
3054 * convert.c (convert_to_integer_1): Use it.
3055 * expr.c (expand_expr_real_2): Likewise.
3056 (expand_expr_real_1): Likewise.
3057 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
3058 * match.pd: Likewise.
3059 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
3060 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
3061 * tree-tailcall.c (process_assignment): Likewise.
3062 * tree-vect-loop.c (vectorizable_reduction): Likewise.
3063 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
3064 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
3065 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3066 (vectorizable_assignment): Likewise.
3067 (vectorizable_shift): Likewise.
3068 (vectorizable_operation): Likewise.
3069 * tree-vrp.c (register_edge_assert_for_2): Likewise.
3071 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
3073 * match.pd: Add pow (C, x) simplification.
3075 2017-08-21 Richard Biener <rguenther@suse.de>
3077 PR tree-optimization/81900
3078 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
3079 for blocks with abnormal predecessors.
3080 (compute_antic): Do not set visited flag prematurely.
3082 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
3085 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
3087 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3089 * stor-layout.h (vector_type_mode): Move to...
3090 * tree.h (vector_type_mode): ...here.
3091 * stor-layout.c (vector_type_mode): Move to...
3092 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
3094 2017-08-21 Richard Biener <rguenther@suse.de>
3096 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
3097 register_external_die hooks.
3098 (debug_false_tree_charstarstar_uhwistar): Declare.
3099 (debug_nothing_tree_charstar_uhwi): Likewise.
3100 * debug.c (do_nothing_debug_hooks): Adjust.
3101 (debug_false_tree_charstarstar_uhwistar): New do nothing.
3102 (debug_nothing_tree_charstar_uhwi): Likewise.
3103 * dbxout.c (dbx_debug_hooks): Adjust.
3104 (xcoff_debug_hooks): Likewise.
3105 * sdbout.c (sdb_debug_hooks): Likewise.
3106 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3107 * dwarf2out.c (macinfo_label_base): New global.
3108 (dwarf2out_register_external_die): New function for the
3109 register_external_die hook.
3110 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
3111 (dwarf2_debug_hooks): Use them.
3112 (dwarf2_lineno_debug_hooks): Adjust.
3113 (struct die_struct): Add with_offset flag.
3114 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
3115 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
3116 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
3117 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
3118 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
3119 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
3120 defining section names for the early LTO debug variants.
3121 (reset_indirect_string): New helper.
3122 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
3123 (print_dw_val): Add support for offsetted symbol references.
3124 (get_ultimate_context): Split out from is_cxx.
3125 (is_cxx): Use get_ultimate_context.
3126 (is_fortran): Add decl overload.
3127 (compute_comp_unit_symbol): Split out worker from
3128 compute_section_prefix.
3129 (compute_section_prefix): Call compute_comp_unit_symbol and
3130 set comdat_type_p here.
3131 (output_die): Skip DIE symbol output for the LTO added one.
3132 Handle DIE symbol references with offset.
3133 (output_comp_unit): Guard section name mangling properly.
3134 For LTO debug sections emit a symbol at the section beginning
3135 which we use to refer to its DIEs.
3136 (add_abstract_origin_attribute): For DIEs registered via
3137 dwarf2out_register_external_die directly refer to the early
3138 DIE rather than indirectly through the shadow one we created.
3139 Remove obsolete call to dwarf2out_abstract_function for
3140 non-function/block origins.
3141 (gen_array_type_die): When generating early LTO debug do
3142 not emit DW_AT_string_length.
3143 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
3144 late when in LTO. As suggested place a gcc_unreachable for
3145 the DECL_ABSTRACT_P case.
3146 (gen_subprogram_die): Avoid another specification DIE
3147 for early built declarations/definitions for the late LTO case.
3148 (gen_variable_die): Add type references for late duplicated VLA dies
3150 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
3151 we have the abstract instance already.
3152 (process_scope_var): Adjust decl DIE contexts in LTO which
3153 first puts them in limbo.
3154 (gen_decl_die): Do not generate type DIEs late apart from
3155 types for VLAs or for decls we do not yet have a DIE. Do not
3156 call dwarf2out_abstract_function late.
3157 (dwarf2out_early_global_decl): Make sure to create DIEs
3158 for abstract instances of a decl first.
3159 (dwarf2out_late_global_decl): Adjust comment.
3160 (output_macinfo_op): With multiple macro sections use
3161 macinfo_label_base to distinguish labels.
3162 (output_macinfo): Likewise. Update macinfo_label_base.
3163 Pass in the line info label.
3164 (note_variable_value_in_expr): When generating LTO resolve
3165 all variable values here by generating DIEs as needed.
3166 (init_sections_and_labels): Add early LTO debug flag parameter
3167 and generate different sections and names if set. Add generation
3168 counter for the labels so we can have multiple of them.
3169 (reset_dies): Helper to allow DIEs to be output multiple times.
3170 (dwarf2out_finish): When outputting DIEs to the fat part of an
3171 LTO object first reset DIEs.
3172 (dwarf2out_early_finish): Output early DIEs when generating LTO.
3173 (modified_type_die): Check for decl_ultimate_origin being self
3175 (gen_type_die_with_usage): Likewise.
3176 (gen_typedef_die): Allow decl_ultimate_origin being self.
3177 (set_decl_abstract_flags): Remove.
3178 (set_block_abstract_flags): Likewise.
3179 (dwarf2out_abstract_function): Treat the early generated DIEs
3180 as the abstract copy and only add DW_AT_inline and
3181 DW_AT_artificial here and call set_decl_origin_self.
3182 If the DIE has an abstract origin don't do anything.
3183 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
3184 if we have none yet (Go fails to build one, PR78628).
3185 (variably_modified_type_p): Prevent endless recursion for Ada
3186 cyclic pointer types.
3187 * lto-streamer-in.c: Include debug.h.
3188 (dref_queue): New global.
3189 (lto_read_tree_1): Stream in DIE references.
3190 (lto_input_tree): Register DIE references.
3191 (input_function): Stream DECL_DEBUG_ARGS.
3192 * lto-streamer-out.c: Include debug.h.
3193 (lto_write_tree_1): Output DIE references.
3194 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
3195 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
3196 (output_function): Stream DECL_DEBUG_ARGS.
3197 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
3198 Stream DECL_ABSTRACT_ORIGIN.
3199 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
3200 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
3201 DECL_CONTEXT for file-scope decls.
3202 * lto-streamer.h (struct dref_entry): Declare.
3203 (dref_queue): Likewise.
3204 * cfgexpand.c (pass_expand::execute): Do not call the
3205 outlining_inline_function hook here.
3206 * lto-wrapper.c (debug_obj): New global.
3207 (tool_cleanup): Unlink it if required.
3208 (debug_objcopy): New function.
3209 (run_gcc): Handle early debug sections in the IL files by
3210 extracting them to separate files, partially linkin them and
3211 feeding the result back as result to the linker.
3212 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
3213 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
3214 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
3215 sections into a separate segment.
3216 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
3218 (darwin_asm_dwarf_section): Likewise.
3219 (darwin_asm_output_dwarf_offset): Likewise.
3220 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
3222 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3223 Alan Hayward <alan.hayward@arm.com>
3224 David Sherwood <david.sherwood@arm.com>
3226 * read-md.h (md_reader::record_potential_iterator_use): Replace
3227 pointer argument with an rtx and an index.
3228 * read-rtl.c (iterator_group::apply_iterator): Likewise.
3229 (apply_mode_iterator): Likewise.
3230 (apply_code_iterator): Likewise.
3231 (apply_int_iterator): Likewise.
3232 (apply_subst_iterator): Likewise.
3233 (record_iterator_use): Likewise.
3234 (record_attribute_use): Likewise.
3235 (md_reader::record_potential_iterator_use): Likewise. Update calls
3236 to record_iterator_use and apply_iterator.
3237 (iterator_use): Replace ptr with x and index.
3238 (attribute_use): Likewise.
3239 (apply_attribute_uses): Update calls to apply_iterator.
3240 (apply_iterators): Likewise. Update initialization of iterator_use.
3241 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
3242 and record_potential_iterator_use.
3243 (rtx_reader::read_rtx_operand): Likewise.
3245 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3246 Alan Hayward <alan.hayward@arm.com>
3247 David Sherwood <david.sherwood@arm.com>
3249 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
3252 2017-08-21 Richard Biener <rguenther@suse.de>
3255 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
3256 at struct end conservatively when comparing common bases.
3258 2017-08-21 Richard Biener <rguenther@suse.de>
3260 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
3261 (mem_ref_in_stmt): Remove.
3262 (determine_max_movement): Use ref index to get at the reference.
3263 (invariantness_dom_walker::before_dom_children): Deal with
3264 lim data already initialized.
3265 (gather_mem_refs_stmt): Initialize lim data and record ref index.
3267 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
3269 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
3270 (TARGET_ISA_ROUND): Ditto.
3271 (TARGET_ROUND): Ditto.
3272 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
3273 * config/i386/i386.md: Ditto.
3274 * config/i386/sse.md: Ditto.
3275 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
3276 with OPTION_MASK_ISA_SSE4_1.
3278 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
3281 * doc/extend.texi (x86 Built-in Functions): Correct the name of
3282 __builtin_ia32_lzcnt_u16.
3284 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
3287 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
3288 (rs6000_set_current_function): Rewrite function to use it.
3290 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
3293 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
3294 and TYPE_WARN_IF_NOT_ALIGN.
3295 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
3296 (handle_warn_if_not_align): New.
3297 (place_union_field): Call handle_warn_if_not_align.
3298 (place_field): Call handle_warn_if_not_align.
3299 Copy TYPE_WARN_IF_NOT_ALIGN.
3300 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
3301 (layout_type): Likewise.
3302 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
3304 (tree_decl_common): Add warn_if_not_align.
3305 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
3306 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
3307 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
3308 (DECL_WARN_IF_NOT_ALIGN): Likewise.
3309 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
3310 * doc/extend.texi: Document warn_if_not_aligned attribute.
3311 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
3313 2017-08-17 Martin Liska <mliska@suse.cz>
3316 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
3317 (get_data_dependence): Use it as pointer type.
3318 (distribute_loop): Likewise.
3320 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3322 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
3323 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
3324 (p8_vmrgow_<mode>_direct): New define_insn.
3325 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
3326 handle endianness for vmrgew and vmrgow permute patterns.
3328 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
3330 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
3331 * config/rs6000/rs6000-cpus.def: Remove comment.
3332 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
3333 (POWERPC_MASKS): Likewise.
3334 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
3335 use of TARGET_VSX_TIMODE.
3336 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
3337 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
3338 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
3339 (rs6000_option_override_internal): Remove dead code.
3340 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
3341 (rs6000_legitimize_reload_address): Likewise.
3342 (rs6000_legitimate_address_p): Likewise.
3343 (rs6000_opt_masks): Delete "vsx-timode".
3344 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
3345 from function comment.
3346 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
3347 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
3348 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
3350 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
3351 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
3352 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
3355 (define_peephole2): Likewise.
3357 2017-08-17 Martin Sebor <msebor@redhat.com>
3360 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
3361 past the end of an array.
3362 (test_pp_format): Add test cases.
3364 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
3366 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
3367 missing ECF_NOTHROW flags.
3369 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
3372 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
3373 operands residing in integer registers.
3374 (*vsx_le_perm_load_<mode>): Likewise.
3375 (*vsx_le_perm_store_<mode>): Likewise.
3376 (define_peephole2): Add peepholes to optimize the above.
3378 2017-08-17 Marek Polacek <polacek@redhat.com>
3381 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
3382 to mimic what shorten_compare did. Change the return type to bool.
3383 (fold_cond_expr_with_comparison): Update call to
3384 operand_equal_for_comparison_p.
3385 (fold_ternary_loc): Likewise.
3387 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
3389 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
3391 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
3393 2017-08-17 Richard Biener <rguenther@suse.de>
3395 * tree-ssa-structalias.c (solve_graph): When propagating
3396 to successors update the graphs succ edges and avoid duplicate work.
3398 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
3401 * config/i386/i386.c (ix86_option_override_internal): Save target
3402 specific options after ix86_stack_protector_guard_reg was changed.
3404 2017-08-17 Richard Biener <rguenther@suse.de>
3406 PR tree-optimization/81827
3407 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
3408 (new_var_info): Initialize it conservatively.
3409 (get_call_vi): Mark register vars.
3410 (new_scalar_tmp_constraint_exp): Likewise.
3411 (handle_rhs_call): Likewise.
3412 (handle_const_call): Likewise.
3413 (create_function_info_for): Likewise.
3414 (solve_constraints): Sort varinfos to separate register from
3415 non-register vars to pack points-to solution bitmaps during
3418 2017-08-17 Marek Polacek <polacek@redhat.com>
3420 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
3422 2017-08-17 Richard Biener <rguenther@suse.de>
3424 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
3425 to true when overflow is undefined and we saturated the result.
3427 2017-08-17 Alan Modra <amodra@gmail.com>
3430 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
3431 Don't use store multiple if only one reg needs saving.
3432 (interesting_frame_related_regno): New function.
3433 (rs6000_frame_related): Don't emit frame info for regs that
3435 (rs6000_emit_epilogue): Likewise.
3437 2017-08-16 Nathan Sidwell <nathan@acm.org>
3439 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
3440 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
3441 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
3442 * tree.c (free_lang_data_in_type): Use else-if chain. Always
3443 clear TYPE_LANG_1. Remove obsolete member-function stripping.
3444 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
3445 (verify_type): Adjust for TYPE_BINFO move.
3446 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
3447 process TYPE_BINFO directly.
3448 (hash_tree): Likewise.
3449 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
3451 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
3454 2017-08-16 David Malcolm <dmalcolm@redhat.com>
3456 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
3458 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
3461 * config/i386/i386.md (*anddi_1_btr): Change predicates of
3462 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
3463 Add ix86_binary_operator_ok to insn constraint.
3464 (*iordi_1_bts): Ditto.
3465 (*xordi_1_btc): Ditto.
3466 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
3467 Update corresponding peephole2 pattern.
3471 2017-08-16 Bin Cheng <bin.cheng@arm.com>
3473 PR tree-optimization/81832
3474 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
3475 copy loop header which has IFN_LOOP_DIST_ALIAS call.
3477 2017-08-16 Marek Polacek <polacek@redhat.com>
3480 * fold-const.c (fold_indirect_ref_1): Restore original behavior
3481 regarding size_zero_node.
3483 2017-08-16 Martin Liska <mliska@suse.cz>
3486 * config.gcc: Respect previously set extra_objs in case
3489 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
3491 PR tree-optimization/81835
3492 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
3493 the phi SSA_NAME. Check that the condition in a COND_EXPR does
3494 not depend on the phi.
3496 2017-08-16 Alan Modra <amodra@gmail.com>
3498 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
3501 2017-08-16 Alan Modra <amodra@gmail.com>
3503 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
3504 (save_reg_p): ..into this. Update all callers.
3505 (first_reg_to_save): Simplify.
3507 2017-08-16 Alan Modra <amodra@gmail.com>
3509 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
3512 2017-08-15 Joseph Myers <joseph@codesourcery.com>
3516 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
3517 constant count if that count is less than 32.
3519 2017-08-15 Nathan Sidwell <nathan@acm.org>
3521 * gcc.c (execute): Emit friendlier message if inferior is killed
3522 by an external cause.
3524 2017-08-15 Richard Biener <rguenther@suse.de>
3526 PR tree-optimization/81790
3527 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
3528 CONSTRUCTORs from simplifying and VN.
3530 2017-08-14 Martin Sebor <msebor@redhat.com>
3532 * builtin-attrs.def: Add comments.
3534 2017-08-14 Martin Sebor <msebor@redhat.com>
3537 * doc/extend.texi (attribute nonstring): Document new attribute.
3539 2017-08-14 Martin Sebor <msebor@redhat.com>
3542 * tree-diagnostic.c (default_tree_printer): Handle %G.
3543 * gimple-pretty-print.h (percent_G_format): Declare new function.
3544 * gimple-pretty-print.c (percent_G_format): Define.
3545 * tree-pretty-print.c (percent_K_format): Add argument.
3547 2017-08-14 Martin Sebor <msebor@redhat.com>
3549 PR translation/79998
3550 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
3551 Remove a stray space.
3553 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
3556 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
3557 (*iordi_1_bts): Ditto.
3558 (*xordi_1_btc): Ditto.
3560 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3563 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
3564 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3566 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
3567 quoted strings, and make more translator-friendly.
3568 (darwin_rs6000_override_options): Likewise.
3569 (rs6000_option_override_internal): Likewise.
3570 (rs6000_return_in_memory): Fix overlong line.
3571 (init_cmulative_args): Use quoted strings, and make more
3572 translator-friendly.
3573 (rs6000_pass_by_reference): Fix overlong line.
3574 (def_builtin): Use quoted strings.
3575 (altivec_expand_predicate_builtin): Use quoted strings, and make
3576 more translator-friendly.
3577 (htm_expand_builtin): Use quoted strings.
3578 (cpu_expand_builtin): Use quoted strings, and make more
3579 translator-friendly.
3580 (altivec_expand_builtin): Likewise.
3581 (paired_expand_predicate_builtin): Likewise.
3582 (rs6000_invalid_builtin): Likewise.
3583 (builtin_function_type): Use quoted strings.
3584 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
3585 more translator-friendly.
3586 (rs6000_trampoline_init): Likewise.
3587 (rs6000_handle_altivec_attribute): Likewise.
3588 (rs6000_inner_target_options): Use quoted strings.
3589 (rs6000_disable_incompatible_switches): Likewise.
3590 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
3591 strings, and make more translator-friendly.
3592 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
3594 2017-08-14 Bin Cheng <bin.cheng@arm.com>
3596 PR tree-optimization/81799
3597 * tree-loop-distribution.c (version_loop_by_alias_check): Force
3598 cond_expr to simple gimple operand.
3600 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
3603 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
3605 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
3609 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
3610 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
3611 TARGET_GASISR_PROLOGUES.
3612 * config/avr/avr.c (avr_option_override): Same.
3613 (avr_pass_pre_proep::execute): Same.
3615 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
3618 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
3619 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
3621 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
3623 * config/i386/i386.md (*load_tp_<mode>): Redefine as
3624 define_insn_and_split. Split to a memory load from 0 in
3625 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
3626 using PTR mode iterator.
3627 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
3628 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
3629 (*add_tp_<mode>): Redefine as define_insn_and_split.
3630 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
3631 address space. Merge with *add_tp_x32 using PTR mode iterator.
3632 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
3633 Split to an add with a memory load from 0 in
3634 DEFAULT_TLS_SEG_REG address space.
3636 2017-08-12 Andrew Pinski <apinski@cavium.com>
3638 * config/aarch64/aarch64-option-extensions.def (rdma):
3639 Fix feature string to what Linux prints out in /proc/cpuinfo.
3641 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
3644 * dwarf2out.c (modified_type_die): For C typedef types that have
3645 an ultimate origin, process the ultimate origin instead of the
3647 (gen_typedef_die): Assert that input DECLs have no ultimate
3649 (gen_type_die_with_usage): For typedef variants that have an
3650 ultimate origin, just call gen_decl_die on the original DECL.
3651 (process_scope_var): Avoid creating DIEs for local typedefs and
3652 concrete static variables.
3654 2017-08-12 Alan Modra <amodra@gmail.com>
3658 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
3659 match gnu-user.h startfile.
3660 (ENDFILE_LINUX_SPEC): Similarly.
3662 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
3665 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
3667 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
3669 2017-08-11 Tamar Christina <tamar.christina@arm.com>
3670 * config/aarch64/aarch64.md (mov<mode>): Change.
3671 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
3672 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
3673 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
3675 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
3677 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
3678 for storage order barriers.
3680 2017-08-11 Martin Liska <mliska@suse.cz>
3683 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
3684 variables of void type.
3686 2017-08-11 Martin Liska <mliska@suse.cz>
3688 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
3689 TARGET_SUPPORTS_ALIASES.
3690 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
3691 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
3692 (optimize_weakref): Likewise.
3693 * symtab.c (symtab_node::noninterposable_alias): Likewise.
3694 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
3695 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
3697 2017-08-11 Martin Liska <mliska@suse.cz>
3700 * config/i386/i386.c (make_resolver_func): Do complete
3701 refactoring of the function.
3703 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
3706 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
3707 * config/i386/i386.c (ix86_stack_protect_guard): Use
3708 ix86_stack_protect_guard_symbol_str to generate varible declaration.
3709 * doc/invoke.texi (x86 Options): Document
3710 -mstack-protector-guard-symbol= option.
3712 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
3714 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
3715 * config/i386/i386.c (ix86_split_stack_guard): New function.
3716 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
3717 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
3718 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
3719 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
3720 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
3721 (split_stack_space_check): Call ix86_split_stack_guard.
3723 2017-08-10 Martin Sebor <msebor@redhat.com>
3725 * print-tree.c (print_node): Print location using the established
3727 Replace spaces with colons.
3728 (debug_raw, debug): Ditto.
3730 2017-08-10 Martin Sebor <msebor@redhat.com>
3733 * pretty-print.c (pp_format): Correct the handling of %s precision.
3735 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
3738 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
3740 (ix86_finalize_stack_frame_flags): This. Also clear
3741 frame_pointer_needed if -fno-omit-frame-pointer is used without
3743 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
3744 with ix86_finalize_stack_frame_flags.
3745 (ix86_expand_epilogue): Likewise.
3746 (ix86_expand_split_stack_prologue): Likewise.
3747 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
3749 2017-08-10 Martin Liska <mliska@suse.cz>
3752 * c-attribs.c (handle_target_attribute):
3753 Report warning for an empty string argument of target attribute.
3755 2017-08-09 Jakub Jelinek <jakub@redhat.com>
3758 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
3760 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
3761 or DECL_NONLOCAL labels.
3762 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
3763 or DECL_NONLOCAL labels here.
3765 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
3767 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
3768 to indicate when early gimple folding has been disabled.
3769 (rs6000_gimple_fold_builtin): Add debug content.
3770 (rs6000_invalid_builtin): Fix whitespace.
3771 (rs6000_expand_builtin): Fix whitespace.
3772 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
3774 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
3777 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
3778 SAVE_MULTIPLE if not all the registers that saves, should be saved.
3780 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
3782 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
3783 (qdf24xx): Likewise.
3784 * config/aarch64/aarch64.md: Include falkor.md.
3785 * config/aarch64/falkor.md: New.
3787 2017-08-09 Marek Polacek <polacek@redhat.com>
3790 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
3791 * diagnostic.c (emit_diagnostic): Add a comment.
3792 (emit_diagnostic_valist): New function.
3794 2017-08-09 Marek Polacek <polacek@redhat.com>
3797 * input.c (make_location): New overload.
3798 * input.h (make_location): Declare.
3800 2017-08-08 Alan Modra <amodra@gmail.com>
3801 H.J. Lu <hongjiu.lu@intel.com>
3804 * gcc.c (NO_PIE_SPEC): Delete.
3805 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
3807 (LINK_PIE_SPEC): ..to here.
3808 (LINK_COMMAND_SPEC): Support -no-pie.
3809 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
3810 chain of crtbegin*.o selection, update for PIE_SPEC changes and
3812 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
3813 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
3814 (ENDFILE_CRTEND_SPEC): Similarly.
3816 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
3819 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
3820 (mstack-protector-guard-offset=): Ditto.
3821 * config/i386/i386.c (ix86_option_override): Handle
3822 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
3824 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
3825 ix86_stack_protect_guard_offset variables.
3826 (TARGET_STACK_PROTECT_GUARD): Always define.
3827 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
3828 and -mstack-protector-guard-offset= options.
3830 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3832 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
3833 boundary case for the last candidate.
3835 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3837 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
3838 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
3840 2017-08-08 Tamar Christina <tamar.christina@arm.com>
3843 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
3844 * config/aarch64/aarch64-builtins.c
3845 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
3846 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
3847 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
3849 2017-08-08 Tamar Christina <tamar.christina@arm.com>
3850 Andrew Pinski <pinskia@gmail.com>
3853 * internal-fn.def (XORSIGN): New.
3854 * optabs.def (xorsign_optab): New.
3855 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
3856 (convert_expand_mult_copysign): New.
3857 (pass_optimize_widening_mul::execute): Call
3858 convert_expand_mult_copysign.
3860 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3862 PR tree-optimization/81354
3863 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
3864 Insert on edges rather than explicitly creating landing pads.
3865 (analyze_candidates_and_replace): Commit edge inserts.
3867 2017-08-08 Richard Biener <rguenther@suse.de>
3870 * tree-ssa-loop-niter.c: Include tree-dfa.h.
3871 (expand_simple_operations): Also look through ADDR_EXPRs with
3872 MEM_REF bases treating them as POINTER_PLUS_EXPR.
3874 2017-08-08 Richard Biener <rguenther@suse.de>
3876 PR tree-optimization/81723
3877 * tree-vect-slp.c (struct bst_traits): New hash traits.
3878 (bst_fail): New global.
3879 (vect_build_slp_tree_2): New worker, split out from ...
3880 (vect_build_slp_tree): ... this now wrapping it with using
3881 bst_fail set to cache SLP tree build fails. Properly handle
3883 (vect_analyze_slp_instance): Allocate and free bst_fail.
3885 2017-08-08 Martin Liska <mliska@suse.cz>
3888 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
3889 LABEL_DECLs that can be from a different function.
3891 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3893 PR tree-optimization/81744
3894 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
3895 loop's number of iterations.
3897 2017-08-08 Martin Liska <mliska@suse.cz>
3899 * asan.c: Include header files.
3900 * attribs.c (build_decl_attribute_variant): New function moved
3902 (build_type_attribute_qual_variant): Likewise.
3903 (cmp_attrib_identifiers): Likewise.
3904 (simple_cst_list_equal): Likewise.
3905 (omp_declare_simd_clauses_equal): Likewise.
3906 (attribute_value_equal): Likewise.
3907 (comp_type_attributes): Likewise.
3908 (build_type_attribute_variant): Likewise.
3909 (lookup_ident_attribute): Likewise.
3910 (remove_attribute): Likewise.
3911 (merge_attributes): Likewise.
3912 (merge_type_attributes): Likewise.
3913 (merge_decl_attributes): Likewise.
3914 (merge_dllimport_decl_attributes): Likewise.
3915 (handle_dll_attribute): Likewise.
3916 (attribute_list_equal): Likewise.
3917 (attribute_list_contained): Likewise.
3918 * attribs.h (lookup_attribute): New function moved from tree.[ch].
3919 (lookup_attribute_by_prefix): Likewise.
3920 * bb-reorder.c: Include header files.
3921 * builtins.c: Likewise.
3922 * calls.c: Likewise.
3923 * cfgexpand.c: Likewise.
3924 * cgraph.c: Likewise.
3925 * cgraphunit.c: Likewise.
3926 * convert.c: Likewise.
3927 * dwarf2out.c: Likewise.
3928 * final.c: Likewise.
3929 * fold-const.c: Likewise.
3930 * function.c: Likewise.
3931 * gimple-expr.c: Likewise.
3932 * gimple-fold.c: Likewise.
3933 * gimple-pretty-print.c: Likewise.
3934 * gimple.c: Likewise.
3935 * gimplify.c: Likewise.
3936 * hsa-common.c: Likewise.
3937 * hsa-gen.c: Likewise.
3938 * internal-fn.c: Likewise.
3939 * ipa-chkp.c: Likewise.
3940 * ipa-cp.c: Likewise.
3941 * ipa-devirt.c: Likewise.
3942 * ipa-fnsummary.c: Likewise.
3943 * ipa-inline.c: Likewise.
3944 * ipa-visibility.c: Likewise.
3946 * lto-cgraph.c: Likewise.
3947 * omp-expand.c: Likewise.
3948 * omp-general.c: Likewise.
3949 * omp-low.c: Likewise.
3950 * omp-offload.c: Likewise.
3951 * omp-simd-clone.c: Likewise.
3952 * opts-global.c: Likewise.
3953 * passes.c: Likewise.
3954 * predict.c: Likewise.
3955 * sancov.c: Likewise.
3956 * sanopt.c: Likewise.
3957 * symtab.c: Likewise.
3958 * toplev.c: Likewise.
3959 * trans-mem.c: Likewise.
3960 * tree-chkp.c: Likewise.
3961 * tree-eh.c: Likewise.
3962 * tree-into-ssa.c: Likewise.
3963 * tree-object-size.c: Likewise.
3964 * tree-parloops.c: Likewise.
3965 * tree-profile.c: Likewise.
3966 * tree-ssa-ccp.c: Likewise.
3967 * tree-ssa-live.c: Likewise.
3968 * tree-ssa-loop.c: Likewise.
3969 * tree-ssa-sccvn.c: Likewise.
3970 * tree-ssa-structalias.c: Likewise.
3971 * tree-ssa.c: Likewise.
3972 * tree-streamer-in.c: Likewise.
3973 * tree-vectorizer.c: Likewise.
3974 * tree-vrp.c: Likewise.
3976 * ubsan.c: Likewise.
3977 * varasm.c: Likewise.
3978 * varpool.c: Likewise.
3979 * tree.c: Remove functions moved to attribs.[ch].
3981 * config/aarch64/aarch64.c: Add attrs.h header file.
3982 * config/alpha/alpha.c: Likewise.
3983 * config/arc/arc.c: Likewise.
3984 * config/arm/arm.c: Likewise.
3985 * config/avr/avr.c: Likewise.
3986 * config/bfin/bfin.c: Likewise.
3987 * config/c6x/c6x.c: Likewise.
3988 * config/cr16/cr16.c: Likewise.
3989 * config/cris/cris.c: Likewise.
3990 * config/darwin.c: Likewise.
3991 * config/epiphany/epiphany.c: Likewise.
3992 * config/fr30/fr30.c: Likewise.
3993 * config/frv/frv.c: Likewise.
3994 * config/ft32/ft32.c: Likewise.
3995 * config/h8300/h8300.c: Likewise.
3996 * config/i386/winnt.c: Likewise.
3997 * config/ia64/ia64.c: Likewise.
3998 * config/iq2000/iq2000.c: Likewise.
3999 * config/lm32/lm32.c: Likewise.
4000 * config/m32c/m32c.c: Likewise.
4001 * config/m32r/m32r.c: Likewise.
4002 * config/m68k/m68k.c: Likewise.
4003 * config/mcore/mcore.c: Likewise.
4004 * config/microblaze/microblaze.c: Likewise.
4005 * config/mips/mips.c: Likewise.
4006 * config/mmix/mmix.c: Likewise.
4007 * config/mn10300/mn10300.c: Likewise.
4008 * config/moxie/moxie.c: Likewise.
4009 * config/msp430/msp430.c: Likewise.
4010 * config/nds32/nds32-isr.c: Likewise.
4011 * config/nds32/nds32.c: Likewise.
4012 * config/nios2/nios2.c: Likewise.
4013 * config/nvptx/nvptx.c: Likewise.
4014 * config/pa/pa.c: Likewise.
4015 * config/pdp11/pdp11.c: Likewise.
4016 * config/powerpcspe/powerpcspe.c: Likewise.
4017 * config/riscv/riscv.c: Likewise.
4018 * config/rl78/rl78.c: Likewise.
4019 * config/rx/rx.c: Likewise.
4020 * config/s390/s390.c: Likewise.
4021 * config/sh/sh.c: Likewise.
4022 * config/sol2.c: Likewise.
4023 * config/sparc/sparc.c: Likewise.
4024 * config/spu/spu.c: Likewise.
4025 * config/stormy16/stormy16.c: Likewise.
4026 * config/tilegx/tilegx.c: Likewise.
4027 * config/tilepro/tilepro.c: Likewise.
4028 * config/v850/v850.c: Likewise.
4029 * config/vax/vax.c: Likewise.
4030 * config/visium/visium.c: Likewise.
4031 * config/xtensa/xtensa.c: Likewise.
4033 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
4036 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
4037 constraints since the -mupper-regs-* switches have been
4039 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
4040 into a vector from a double word element that was extracted from
4041 another vector, and eliminate extra XXPERMDI instructions.
4042 (vsx_concat_<mode>_2): Likewise.
4043 (vsx_concat_<mode>_3): Likewise.
4044 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
4045 concat to allow optimizing inserts from previous extracts.
4047 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
4049 * config/i386/i386.c (ix86_stack_protect_guard): Generate
4050 memory reference to a SSP offset in TLS address space.
4051 (ix86_print_operand) <case '@'>: Remove.
4052 (ix86_print_operand_punct_valid_p): Remove '@' code.
4053 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
4055 (stack_tls_protect_set_<mode>): Remove.
4056 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
4057 (stack_tls_protect_test_<mode>): Remove.
4058 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
4060 2017-08-07 Olivier Hainque <hainque@adacore.com>
4063 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
4065 2017-08-07 Douglas Rupp <rupp@adacore.com>
4067 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
4068 variable was referenced as multidir in command.
4070 2017-08-07 Jakub Jelinek <jakub@redhat.com>
4073 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
4076 2017-08-07 Martin Liska <mliska@suse.cz>
4078 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
4079 * config/rl78/rl78.c: Add include of attribs.h.
4080 * config/sh/sh.c: Likewise.
4081 * config/v850/v850.c: Likewise.
4083 2017-08-07 Tom de Vries <tom@codesourcery.com>
4086 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
4088 2017-08-07 Martin Liska <mliska@suse.cz>
4090 * config/mips/mips.c: Include attribs.h.
4092 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
4095 * doc/invoke.texi: Document change in behvaior for -Ofast for
4098 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
4100 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
4102 (aarch64_pop_regs): Likewise.
4103 (aarch64_gen_load_pair): Likewise.
4104 (aarch64_save_callee_saves): Likewise.
4105 (aarch64_restore_callee_saves): Likewise.
4107 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4109 * config/i386/i386.c: Revert the last change.
4111 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4114 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
4116 (ix86_finalize_stack_frame_flags): This. Also clear
4117 frame_pointer_needed if -fno-omit-frame-pointer is used without
4119 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
4120 with ix86_finalize_stack_frame_flags.
4121 (ix86_expand_epilogue): Likewise.
4122 (ix86_expand_split_stack_prologue): Likewise.
4124 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4127 * config/i386/i386.c (get_builtin_code_for_version): Set priority
4128 to P_AES for Westmere.
4130 2017-08-07 Jonathan Yong <10walls@gmail.com>
4132 * config/i386/mingw.opt (fset-stack-executable): Removed.
4133 * config/i386/cygming.opt (fset-stack-executable): Moved
4135 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
4137 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
4139 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
4141 2017-08-07 Marek Polacek <polacek@redhat.com>
4144 * fold-const.c (fold_indirect_ref_1): Check type_domain.
4146 2017-08-07 Martin Liska <mliska@suse.cz>
4148 * attribs.h (canonicalize_attr_name): New function.
4149 (cmp_attribs): Move from c-format.c and adjusted.
4150 (is_attribute_p): Moved from tree.h.
4151 * tree-inline.c: Add new includes.
4152 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
4153 (private_is_attribute_p): Remove.
4154 (private_lookup_attribute): Likewise.
4155 (private_lookup_attribute_by_prefix): Simplify.
4156 (remove_attribute): Use is_attribute_p.
4157 * tree.h: Remove removed declarations.
4159 2017-08-07 Jakub Jelinek <jakub@redhat.com>
4162 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
4163 instead of computing it in the function. Formatting fix.
4164 (expand_case): Don't rely on default_edge being the first edge,
4165 clear it if removing it, pass default_edge to
4166 emit_case_dispatch_table.
4167 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
4170 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
4172 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
4173 insn in the function, emit NOP after the insn.
4175 2017-08-06 Tom de Vries <tom@codesourcery.com>
4177 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
4180 2017-08-06 Tom de Vries <tom@codesourcery.com>
4182 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
4185 2017-08-04 Yury Gribov <tetra2005@gmail.com>
4187 PR tree-optimization/57371
4188 * match.pd: New pattern.
4190 2017-08-04 Marek Polacek <polacek@redhat.com>
4193 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
4194 perform the computation in offset_int.
4196 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
4198 PR tree-optimization/81136
4199 * tree-vectorizer.h: Include tree-hash-traits.h.
4200 (vec_base_alignments): New typedef.
4201 (vec_info): Add a base_alignments field.
4202 (vect_record_base_alignments): Declare.
4203 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
4205 (DR_IS_CONDITIONAL_IN_STMT): New macro.
4206 (create_data_ref): Add an is_conditional_in_stmt argument.
4207 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
4208 the is_conditional_in_stmt field.
4209 (data_ref_loc): Add an is_conditional_in_stmt field.
4210 (get_references_in_stmt): Set the is_conditional_in_stmt field.
4211 (find_data_references_in_stmt): Update call to create_data_ref.
4212 (graphite_find_data_references_in_stmt): Likewise.
4213 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
4214 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
4215 (vect_record_base_alignment): New function.
4216 (vect_record_base_alignments): Likewise.
4217 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
4218 for nested statements even if we fail to compute a misalignment.
4219 Use pooled base alignments for unconditional references.
4220 (vect_find_same_alignment_drs): Compare base addresses instead
4222 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
4223 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
4225 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
4227 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
4228 Add an explicit name for the enum. Use auto_vec for slp_instances
4230 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
4232 (_bb_vec_info): Add a constructor and destructor.
4233 (vinfo_for_stmt): Return NULL for uids of -1 as well.
4234 (destroy_loop_vec_info): Delete.
4235 (vect_destroy_datarefs): Likewise.
4236 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
4237 (vec_info::vec_info): New function.
4238 (vec_info::~vec_info): Likewise.
4239 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
4240 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
4241 destroy_loop_vec_info.
4242 * tree-vect-loop.c (new_loop_vec_info): Replace with...
4243 (_loop_vec_info::_loop_vec_info): ...this.
4244 (destroy_loop_vec_info): Replace with...
4245 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
4246 the stmt_vec_infos. Leave handling of vec_info information to its
4247 destructor. Remove explicit vector releases.
4248 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
4249 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
4250 * tree-vect-slp.c (new_bb_vec_info): Replace with...
4251 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
4252 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
4253 (destroy_bb_vec_info): Replace with...
4254 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
4255 information to its destructor.
4256 (vect_slp_analyze_bb_1): Use new and delete instead of
4257 new_bb_vec_info and destroy_bb_vec_info.
4258 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
4261 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
4263 * tree-data-ref.h (subscript): Add access_fn field.
4264 (data_dependence_relation): Add could_be_independent_p.
4265 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
4266 (same_access_functions): Move to tree-data-ref.c.
4267 * tree-data-ref.c (ref_contains_union_access_p): New function.
4268 (access_fn_component_p): Likewise.
4269 (access_fn_components_comparable_p): Likewise.
4270 (dr_analyze_indices): Add a reference to access_fn_component_p.
4271 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
4273 (constant_access_functions): Likewise.
4274 (add_other_self_distances): Likewise.
4275 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
4276 (initialize_data_dependence_relation): Use XCNEW and remove
4277 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
4278 of access functions that have the same type. Allow the
4279 subsequence to end with different bases in some circumstances.
4280 Record the chosen access functions in SUB_ACCESS_FN.
4281 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
4282 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
4283 (subscript_dependence_tester_1): Likewise dra and drb.
4284 (build_classic_dist_vector): Update calls accordingly.
4285 (subscript_dependence_tester): Likewise.
4286 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
4287 DDR_COULD_BE_INDEPENDENT_P.
4288 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
4289 comp_alias_ddrs instead of may_alias_ddrs.
4290 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
4292 (vect_analyze_data_ref_dependence): Use it if
4293 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
4294 distance vectors if that fails.
4295 (dependence_distance_ge_vf): New function.
4296 (vect_prune_runtime_alias_test_list): Use it. Don't clear
4297 LOOP_VINFO_MAY_ALIAS_DDRS.
4299 2017-08-04 Richard Biener <rguenther@suse.de>
4302 * fold-const.c (fold_binary_loc): Properly restrict
4303 minus_var0 && minus_var1 case when associating undefined overflow
4306 2017-08-04 Tom de Vries <tom@codesourcery.com>
4308 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
4310 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4312 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4313 Don't start diagnostic messages with a capital letter.
4314 * config/rs6000/rs6000.c (rs6000_option_override_internal):
4316 (rs6000_invalid_builtin): Likewise.
4317 (rs6000_trampoline_init): Likewise.
4319 2017-08-03 Jakub Jelinek <jakub@redhat.com>
4322 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
4323 after setting changeable df flags.
4325 2017-08-03 Richard Biener <rguenther@suse.de>
4327 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
4328 up if the use is in USE - X.
4330 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
4332 * toplev.c (dumpfile.h): New include.
4333 (internal_error_reentered): New static function. Use it...
4334 (internal_error_function): ...here to handle reentered internal_error.
4336 2017-08-03 Richard Biener <rguenther@suse.de>
4339 * fold-const.c (split_tree): Add minus_var and minus_con
4340 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
4341 here but always use minus_*.
4342 (associate_trees): Assert we never associate with MINUS_EXPR
4343 and NULL first operand. Do not recurse for PLUS_EXPR operands
4344 when associating as MINUS_EXPR either.
4345 (fold_binary_loc): Track minus_var and minus_con.
4347 2017-08-03 Tom de Vries <tom@codesourcery.com>
4350 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
4351 ACCEL_COMPILER, apply finish_options on
4352 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
4354 2017-08-03 Tom de Vries <tom@codesourcery.com>
4357 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
4358 function_entry_patch_area_size > 0.
4360 2017-08-03 Jakub Jelinek <jakub@redhat.com>
4363 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
4364 instead of 10??LU, perform unit multiplication in wide_int,
4365 don't change alloc_object_size_limit if the limit is larger
4368 PR tree-optimization/81655
4369 PR tree-optimization/81588
4370 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
4371 the case when ranges[i].low and high are 1 for unsigned type with
4375 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
4376 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
4378 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4380 * tree-vrp.h: Add include guard.
4382 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
4385 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
4386 (ud2): New insn pattern.
4387 * config/i386/i386.c (ix86_expand_epilogue):
4388 For naked functions, generate ud2 instead of trap insn.
4390 2017-08-02 Marek Polacek <polacek@redhat.com>
4393 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
4395 2017-08-02 Tom de Vries <tom@codesourcery.com>
4396 Cesar Philippidis <cesar@codesourcery.com>
4398 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
4399 Add missing edge probabilities.
4401 2017-08-02 Tamar Christina <tamar.christina@arm.com>
4403 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
4406 2017-08-02 Jakub Jelinek <jakub@redhat.com>
4409 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
4410 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
4411 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
4412 prologue_seq sequences - if any.
4414 2017-08-02 Richard Biener <rguenther@suse.de>
4416 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
4417 via vectors if supported, integer extracts via punning if supported
4418 or otherwise vector extracts.
4420 2017-08-02 Richard Biener <rguenther@suse.de>
4422 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
4424 (bitmap_insert_into_set): ... this.
4426 2017-08-02 Richard Biener <rguenther@suse.de>
4428 PR tree-optimization/81633
4430 2015-08-17 Alan Hayward <alan.hayward@arm.com>
4432 PR tree-optimization/71752
4433 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
4435 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
4437 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
4438 (machine_function::call_ms2sysv_pad_out): Remove field.
4439 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
4440 (ix86_compute_frame_layout): Likewise.
4442 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
4445 * config/i386/i386.c (ix86_set_func_type): Disallow naked
4446 attribute with interrupt attribute.
4448 2017-08-01 Andrew Pinski <apinski@cavium.com>
4450 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
4451 BIT_INSERT_EXPR's operand 1
4452 to see if the types precision matches.
4454 2017-08-01 Martin Liska <mliska@suse.cz>
4457 * builtins.c (expand_builtin_memcpy_args): Remove.
4458 (expand_builtin_memcpy): Call newly added function
4459 expand_builtin_memory_copy_args.
4460 (expand_builtin_memcpy_with_bounds): Likewise.
4461 (expand_builtin_mempcpy): Remove last argument.
4462 (expand_builtin_mempcpy_with_bounds): Likewise.
4463 (expand_builtin_memory_copy_args): New function created from
4464 expand_builtin_mempcpy_args with small modifications.
4465 (expand_builtin_mempcpy_args): Remove.
4466 (expand_builtin_stpcpy): Remove unused argument.
4467 (expand_builtin): Likewise.
4468 (expand_builtin_with_bounds): Likewise.
4470 2017-08-01 Martin Liska <mliska@suse.cz>
4473 Make mempcpy more optimal (PR middle-end/70140).
4475 2017-08-01 Jakub Jelinek <jakub@redhat.com>
4478 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
4479 __builtin_vec_cmpne verify both arguments are compatible vectors
4480 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
4481 verify arg1_type is a pointer or array type. For __builtin_vec_st,
4482 move computation of aligned to after checking the argument types.
4486 * config/rs6000/vsx.md (vextract_fp_from_shorth,
4487 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
4490 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
4491 Doug Rupp <rupp@adacore.com>
4492 Olivier Hainque <hainque@adacore.com>
4494 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
4495 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
4497 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
4498 for TARGET_OS_CPP_BUILTIN.
4499 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
4500 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
4502 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
4503 passing required abi options to the assembler for EABI configurations.
4504 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
4505 of .text.hot and .text.unlikely sections for kernel modules when
4506 using ARM style exceptions.
4507 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
4508 options. Add EXTRA_CC1_SPEC.
4509 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
4511 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
4513 (ARM_TARGET2_DWARF_FORMAT): Define.
4514 * config/arm/t-vxworks: Adjust multilib control to removal of the
4515 Diab command line options.
4517 2017-08-01 Martin Liska <mliska@suse.cz>
4519 PR gcov-profile/81561
4520 * gcov.c (unblock): Make unblocking safe as we need to preserve
4521 index correspondence of blocks and block_lists.
4523 2017-08-01 Richard Biener <rguenther@suse.de>
4525 PR tree-optimization/81181
4526 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4527 (compute_antic): ... end of iteration here.
4529 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
4531 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
4532 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
4533 (ftree-slp-vectorize): Likewise.
4534 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
4535 can no longer be set independent of flag_tree_loop_vectorize.
4536 * omp-general.c (emp_max_vf): Likewise.
4537 * opts.c (enable_fdo_optimizations): Remove references to
4538 flag_tree_vectorize, these are now implicit.
4539 (common_handle_option): Remove handling for OPT_ftree_vectorize,
4540 and leave it for the options machinery.
4542 2017-08-01 Martin Liska <mliska@suse.cz>
4545 * builtins.c (expand_builtin_memcpy_args): Remove.
4546 (expand_builtin_memcpy): Call newly added function
4547 expand_builtin_memory_copy_args.
4548 (expand_builtin_memcpy_with_bounds): Likewise.
4549 (expand_builtin_mempcpy): Remove last argument.
4550 (expand_builtin_mempcpy_with_bounds): Likewise.
4551 (expand_builtin_memory_copy_args): New function created from
4552 expand_builtin_mempcpy_args with small modifications.
4553 (expand_builtin_mempcpy_args): Remove.
4554 (expand_builtin_stpcpy): Remove unused argument.
4555 (expand_builtin): Likewise.
4556 (expand_builtin_with_bounds): Likewise.
4558 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
4561 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
4562 print "ds:" only for immediates in generic address space.
4564 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
4567 * config/i386/i386.c (ix86_funciton_naked): New prototype.
4568 (ix86_function_ok_for_sibcall): Return false for naked functions.
4570 2017-08-01 Richard Biener <rguenther@suse.de>
4572 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
4573 (compute_antic): Seed worklist with exit block predecessors.
4574 * cfganal.c (dfs_find_deadend): For a cycle return the source
4575 of the edge closing it.
4577 2017-08-01 Tamar Christina <tamar.christina@arm.com>
4579 * config/aarch64/aarch64.c
4580 (aarch64_can_const_movi_rtx_p): Move 0 check.
4582 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4584 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
4585 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
4587 * match.pd: Ditto in address comparison pattern.
4589 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4591 PR tree-optimization/81627
4592 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
4593 closed ssa form for store-store chain.
4595 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4597 PR tree-optimization/81620
4598 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
4599 for store-store chain.
4601 2017-08-01 Jakub Jelinek <jakub@redhat.com>
4603 PR tree-optimization/81588
4604 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
4605 ranges[i].in_p, invert comparison code ccode. For >/>=,
4606 swap rhs1 and rhs2 and comparison code unconditionally,
4607 for </<= don't do that. Don't swap rhs1/rhs2 again if
4608 ranges[i].in_p, instead invert comparison code ccode if
4609 opcode or oe->rank is BIT_IOR_EXPR.
4612 * optabs.def (vec_extract_optab, vec_init_optab): Change from
4613 a direct optab to conversion optab.
4614 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
4615 with GET_MODE_INNER as last argument instead of optab_handler.
4616 * expmed.c (extract_bit_field_1): Likewise. Use vector from
4617 vector extraction if possible and optab is available.
4618 * expr.c (store_constructor): Use convert_optab_handler instead
4619 of optab_handler. Use vector initialization from smaller
4620 vectors if possible and optab is available.
4621 * tree-vect-stmts.c (vectorizable_load): Likewise.
4622 * doc/md.texi (vec_extract, vec_init): Document that the optabs
4624 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
4625 of vec_init from half-sized vectors with the same element mode.
4626 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
4627 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
4628 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
4629 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
4630 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
4631 after mode in gen_vec_extract* calls.
4632 (vec_extract<mode>): Renamed to ...
4633 (vec_extract<mode><ssescalarmodelower>): ... this.
4634 (vec_extract<mode><ssehalfvecmodelower>): New expander.
4635 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
4636 element mode after mode in gen_vec_init* calls.
4637 (VEC_INIT_HALF_MODE): New mode iterator.
4638 (vec_init<mode>): Renamed to ...
4639 (vec_init<mode><ssescalarmodelower>): ... this.
4640 (vec_init<mode><ssehalfvecmodelower>): New expander.
4641 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
4642 (vec_extractv2sfsf): ... this.
4643 (vec_initv2sf): Renamed to ...
4644 (vec_initv2sfsf): ... this.
4645 (vec_extractv2si): Renamed to ...
4646 (vec_extractv2sisi): ... this.
4647 (vec_initv2si): Renamed to ...
4648 (vec_initv2sisi): ... this.
4649 (vec_extractv4hi): Renamed to ...
4650 (vec_extractv4hihi): ... this.
4651 (vec_initv4hi): Renamed to ...
4652 (vec_initv4hihi): ... this.
4653 (vec_extractv8qi): Renamed to ...
4654 (vec_extractv8qiqi): ... this.
4655 (vec_initv8qi): Renamed to ...
4656 (vec_initv8qiqi): ... this.
4657 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
4658 (vec_init<mode>): Renamed to ...
4659 (vec_init<mode><VEC_base_l>): ... this.
4660 (vec_extract<mode>): Renamed to ...
4661 (vec_extract<mode><VEC_base_l>): ... this.
4662 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
4663 (vec_initv2sfsf): ... this.
4664 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
4665 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
4666 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
4667 element mode after mode in gen_vec_init* calls.
4668 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
4669 (vec_init<mode><Vel>): ... this.
4670 (vec_extract<mode>): Renamed to ...
4671 (vec_extract<mode><Vel>): ... this.
4672 * config/aarch64/iterators.md (Vel): New mode attribute.
4673 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
4674 Add element mode after mode in gen_vec_extract* calls.
4675 * config/s390/vector.md (non_vec_l): New mode attribute.
4676 (vec_extract<mode>): Renamed to ...
4677 (vec_extract<mode><non_vec_l>): ... this.
4678 (vec_init<mode>): Renamed to ...
4679 (vec_init<mode><non_vec_l>): ... this.
4680 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
4681 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
4683 * config/arm/iterators.md (V_elem_l): New mode attribute.
4684 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
4685 (vec_extract<mode><V_elem_l>): ... this.
4686 (vec_extractv2di): Renamed to ...
4687 (vec_extractv2didi): ... this.
4688 (vec_init<mode>): Renamed to ...
4689 (vec_init<mode><V_elem_l>): ... this.
4690 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
4691 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
4692 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
4693 Add element mode after gen_vec_extract* calls.
4694 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
4695 (vec_init<mode><unitmode>): ... this.
4696 (vec_extract<mode>): Renamed to ...
4697 (vec_extract<mode><unitmode>): ... this.
4698 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
4699 (vec_init<mode><unitmode>): ... this.
4700 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
4701 (vec_initv2sfsf): ... this.
4702 (vec_extractv2sf): Renamed to ...
4703 (vec_extractv2sfsf): ... this.
4704 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
4705 Add element mode after gen_vec_extract* calls.
4706 * config/mips/mips.md (unitmode): New mode iterator.
4707 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
4708 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
4709 * config/spu/spu.md (inner_l): New mode attribute.
4710 (vec_init<mode>): Renamed to ...
4711 (vec_init<mode><inner_l>): ... this.
4712 (vec_extract<mode>): Renamed to ...
4713 (vec_extract<mode><inner_l>): ... this.
4714 * config/sparc/sparc.md (veltmode): New mode iterator.
4715 (vec_init<VMALL:mode>): Renamed to ...
4716 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
4717 * config/ia64/vect.md (vec_initv2si): Renamed to ...
4718 (vec_initv2sisi): ... this.
4719 (vec_initv2sf): Renamed to ...
4720 (vec_initv2sfsf): ... this.
4721 (vec_extractv2sf): Renamed to ...
4722 (vec_extractv2sfsf): ... this.
4723 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
4724 (vec_init<mode>): Renamed to ...
4725 (vec_init<mode><VEC_base_l>): ... this.
4726 (vec_extract<mode>): Renamed to ...
4727 (vec_extract<mode><VEC_base_l>): ... this.
4728 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
4729 (vec_initv2sfsf): ... this.
4730 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
4731 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
4732 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
4733 gen_vec_init* calls.
4735 2017-08-01 Richard Biener <rguenther@suse.de>
4737 PR tree-optimization/81297
4738 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
4739 TREE_OVERFLOW from INTEGER_CSTs.
4741 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
4743 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
4745 2017-07-31 Carl Love <cel@us.ibm.com>
4747 * config/rs6000/rs6000-c: Add support for built-in functions
4748 vector signed char vec_xl_be (signed long long, signed char *);
4749 vector unsigned char vec_xl_be (signed long long, unsigned char *);
4750 vector signed int vec_xl_be (signed long long, signed int *);
4751 vector unsigned int vec_xl_be (signed long long, unsigned int *);
4752 vector signed long long vec_xl_be (signed long long, signed long long *);
4753 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
4754 vector signed short vec_xl_be (signed long long, signed short *);
4755 vector unsigned short vec_xl_be (signed long long, unsigned short *);
4756 vector double vec_xl_be (signed long long, double *);
4757 vector float vec_xl_be (signed long long, float *);
4758 * config/rs6000/altivec.h (vec_xl_be): Add #define.
4759 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
4760 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
4762 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
4763 (altivec_expand_builtin): Add switch statement to call
4764 altivec_expand_xl_be for each builtin.
4765 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
4766 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
4767 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
4768 __builtin_vsx_le_be_v16qi.
4769 * doc/extend.texi: Update the built-in documentation file for the
4770 new built-in functions.
4772 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
4775 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
4777 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
4779 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4781 * config.gcc: Add z14.
4782 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
4783 CPU model numbers for z13s and z14.
4784 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
4786 * config/s390/s390-opts.h (enum processor_type): Rename
4787 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
4788 * config/s390/s390.c (processor_table): Add field for CPU name to
4789 be passed to Binutils.
4790 (s390_asm_output_machine_for_arch): Use the new field in
4791 processor_table for Binutils.
4792 (s390_expand_builtin): Replace arch12 with z14.
4793 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
4794 (s390_get_sched_attrmask): Likewise.
4795 (s390_get_unit_mask): Likewise.
4796 * config/s390/s390.opt: Add z14 to processor_type enum.
4798 2017-07-31 Martin Jambor <mjambor@suse.cz>
4801 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
4802 regardless of optimization level.
4804 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
4805 Martin Liska <mliska@suse.cz>
4807 * predict.def: Remove old comment and adjust probability.
4808 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
4811 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
4814 * config/i386/i386.c (ix86_function_naked): New function.
4815 (ix86_can_use_return_insn_p): Return false for naked functions.
4816 (ix86_expand_prologue): Skip prologue for naked functions.
4817 (ix86_expand_epilogue): Skip epilogue for naked functions
4818 and emit trap instruction.
4819 (ix86_warn_func_return): New function.
4820 (ix86_attribute_table): Add "naked" attribute specification.
4821 (TARGET_WARN_FUNC_RETURN): Define.
4822 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
4824 2017-07-31 Martin Liska <mliska@suse.cz>
4826 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
4827 (dump_gimple_bb_header): Always dump BB info.
4828 (pp_cfg_jump): Do not append info about BB when dumping a jump.
4830 2017-07-31 Martin Liska <mliska@suse.cz>
4833 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
4834 also with current_function_decl non-null equality.
4836 2017-07-31 Jakub Jelinek <jakub@redhat.com>
4839 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
4840 change type to the element type, instead add eltype variable and
4841 use it where we are interested in the element type.
4843 PR tree-optimization/81603
4844 * ipa-polymorphic-call.c
4845 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
4846 offset arithmetic in offset_int, bail out if the resulting bit offset
4847 doesn't fit into shwi.
4849 2017-07-31 Martin Liska <mliska@suse.cz>
4851 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
4852 (gimplify_save_expr): Fix comment.
4854 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
4857 * config/i386/i386.c (ix86_function_arg): Update arguments for
4859 (ix86_compute_frame_layout): Set the initial stack offset to
4860 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
4861 INCOMING_FRAME_SP_OFFSET.
4862 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
4863 stack before exception handler returns.
4864 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
4865 the 'ERROR_CODE' for exception handler.
4867 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
4869 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
4870 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
4871 (ASM_OUTPUT_REG_POP): Ditto.
4872 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
4873 instead of asm_fprintf to output pure string.
4875 2017-07-29 Jakub Jelinek <jakub@redhat.com>
4877 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
4878 to imported_module_or_decl hook.
4879 (debug_nothing_tree_tree_tree_bool): Remove.
4880 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
4881 * debug.c (do_nothing_debug_hooks): Use
4882 debug_nothing_tree_tree_tree_bool_bool instead of
4883 debug_nothing_tree_tree_tree_bool.
4884 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4885 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
4886 * sdbout.c (sdb_debug_hooks): Likewise.
4887 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
4888 (gen_namespace_die): Add DW_AT_export_symbols attribute if
4890 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
4891 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
4892 attribute, don't add anything.
4894 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4896 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
4897 (fold_build2_stat_loc): Likewise.
4898 (fold_build3_stat_loc): Likewise.
4899 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
4900 (fold_build1_loc): Remove macro.
4901 (fold_build2_loc): Likewise.
4902 (fold_build3_loc): Likewise.
4904 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4906 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
4907 (gimple_build_debug_bind_source_stat): Likewise.
4908 * gimple.h (gimple_build_debug_bind): Remove macro.
4909 (gimple_build_debug_bind_source): Likewise.
4911 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4913 * bitmap.c (bitmap_alloc): Adjust.
4914 (bitmap_gc_alloc): Likewise.
4915 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
4917 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4919 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
4920 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
4921 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
4922 (bitmap_gc_alloc_stat): Likewise.
4923 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
4925 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4927 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
4928 * rtl.h (shallow_copy_rtx): Remove macro.
4930 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4932 * emit-rtl.c (gen_raw_REG): Adjust.
4933 * gengenrtl.c (gendef): Likewise.
4934 * rtl.c (rtx_alloc_stat): Remove _stat from name.
4935 * rtl.h (rtx_alloc): Remove macro.
4937 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4939 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
4940 (build_tree_list_stat): Likewise.
4941 * tree.h (build_tree_list): Remove macro.
4942 (build_tree_list_vec): Likewise.
4944 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4946 * tree.c (make_vector_stat): Remove _stat from name.
4947 (build_vector_stat): Likewise.
4948 * tree.h (make_vector_stat): Remove macro.
4949 (build_vector_stat): Likewise.
4951 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4953 * tree.h (build_var_debug_value): Remove prototype.
4955 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4957 * tree.c (tree_cons_stat): Remove _stat from name.
4958 * tree.h (tree_cons): Remove macro.
4960 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4962 * tree.c (build_vl_exp_stat): Remove _stat from name.
4963 * tree.h (build_vl_exp): Remove macro.
4965 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4967 * tree.c (build_decl_stat): Remove _stat from name.
4968 * tree.h (build_decl): Remove macro.
4970 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4972 * gimple.c (gimple_build_with_ops_stat): Adjust.
4973 (gimple_alloc_stat): Remove _stat from name.
4974 * gimple.h (gimple_alloc): Remove macro.
4976 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4978 * tree.c (make_tree_vec_stat): Remove _stat from name.
4979 (grow_tree_vec_stat): Likewise.
4980 * tree.h (make_tree_vec_stat): Adjust prototype.
4981 (grow_tree_vec_stat): Likewise.
4982 (make_tree_vec): Remove macro.
4983 (grow_tree_vec): Likewise.
4985 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4987 * fold-const.c (fold_build1_stat_loc): Adjust.
4988 (fold_build2_stat_loc): Likewise.
4989 (fold_build3_stat_loc): Likewise.
4990 * tree.c (build0_stat): Remove _stat from name.
4991 (build1_stat): Likewise.
4992 (build2_stat): Likewise.
4993 (build3_stat): Likewise.
4994 (build4_stat): Likewise.
4995 (build5_stat): Likewise.
4996 * tree.h (build1_loc): Remove macro, and rename _stat function
4998 (build2_loc): Likewise.
4999 (build3_loc): Likewise.
5000 (build4_loc): Likewise.
5001 (build5_loc): Likewise.
5003 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5005 * tree.c (make_int_cst_stat): Remove _stat from name.
5006 * tree.h (make_int_cst_stat): Adjust prototype.
5007 (make_int_cst): Remove macro.
5009 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5011 * tree.c (make_tre_binfo_stat): Remove _stat from name.
5012 * tree.h (make_tree_binfo_stat): Adjust prototype.
5013 (make_tree_binfo): Remove.
5015 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5017 * tree.c (copy_node_stat): Rename to copy_node.
5018 (build_distinct_type_copy): Adjust.
5019 * tree.h (copy_node_stat): Adjust prototype.
5020 (copy_node): Remove macro.
5022 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5024 * tree.c (make_node_stat): rename to make_node.
5025 (build_tree_list_stat): Adjust.
5026 (build0_stat): Likewise.
5027 (build2_stat): Likewise.
5028 (build3_stat): Likewise.
5029 (build4_stat): Likewise.
5030 (build5_stat): Likewise.
5031 (build_decl_stat): Likewise.
5032 * tree.h (make_node_stat): Adjust prototype.
5033 (make_node): remove macro.
5035 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
5037 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
5038 (PPC_FEATURE2_SCV): Likewise.
5039 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
5041 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5043 * config/aarch64/aarch64.c
5044 (aarch64_internal_mov_immediate): Add new special pattern.
5045 * config/aarch64/aarch64.md (*movdi_aarch64):
5046 Add reg/32bit const mov case.
5048 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5049 Richard Sandiford <richard.sandiford@linaro.org>
5051 * config/aarch64/aarch64.md (mov<mode>): Generalize.
5052 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
5053 Add integer and movi cases.
5054 (movi-split-hf-df-sf split, fp16): New.
5055 (enabled): Added TARGET_FP_F16INST.
5056 * config/aarch64/iterators.md (GPF_HF): New.
5057 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
5059 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5061 * config/aarch64/aarch64.c
5062 (aarch64_simd_container_mode): Add prototype.
5063 (aarch64_expand_mov_immediate): Add HI support.
5064 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
5065 (aarch64_can_const_movi_rtx_p): New.
5066 (aarch64_preferred_reload_class):
5067 Remove restrictions of using FP registers for certain SIMD operations.
5068 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
5069 (aarch64_valid_floating_const): Add integer move validation.
5070 (aarch64_simd_imm_scalar_p): Remove.
5071 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
5072 (aarch64_legitimate_constant_p): Expand list of supported cases.
5073 * config/aarch64/aarch64-protos.h
5074 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
5075 (aarch64_reinterpret_float_as_int): New.
5076 (aarch64_simd_imm_scalar_p): Remove.
5077 * config/aarch64/constraints.md (Uvi): New.
5078 (Dd): Split into Ds and new Dd.
5079 * config/aarch64/aarch64.md (*movsi_aarch64):
5081 (*movdi_aarch64): Add SIMD mov case.
5083 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5085 * tree-predcom.c: (struct chain): Handle store-store chain in which
5086 stores for elimination only store loop invariant values.
5087 (execute_pred_commoning_chain): Ditto.
5088 (prepare_initializers_chain_store_elim): Ditto.
5089 (prepare_finalizers): Ditto.
5090 (is_inv_store_elimination_chain): New function.
5091 (initialize_root_vars_store_elim_1): New function.
5093 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5095 * tree-predcom.c: Revise general description of the pass.
5096 (enum chain_type): New enum type for store elimination.
5097 (struct chain): New field supporting store elimination.
5098 (struct component): Ditto.
5099 (dump_chain): Dump store-stores chain.
5100 (release_chain): Release resources.
5101 (split_data_refs_to_components): Compute and create component
5102 contains only stores for elimination.
5103 (get_chain_last_ref_at): New function.
5104 (make_invariant_chain): Initialization.
5105 (make_rooted_chain): Specify chain type in parameter and record it.
5106 (add_looparound_copies): Skip for store-stores chain.
5107 (determine_roots_comp): Compute type of chain and pass it to
5109 (initialize_root_vars_store_elim_2): New function.
5110 (finalize_eliminated_stores): New function.
5111 (remove_stmt): Handle store for elimination.
5112 (execute_pred_commoning_chain): Execute predictive commoning on
5114 (determine_unroll_factor): Skip unroll for store-stores chain.
5115 (prepare_initializers_chain_store_elim): New function.
5116 (prepare_initializers_chain): Hanlde store-store chain.
5117 (prepare_finalizers_chain, prepare_finalizers): New function.
5118 (tree_predictive_commoning_loop): Return integer value indicating
5119 if loop is unrolled or lcssa form is corrupted.
5120 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
5122 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5124 * tree-predcom.c (initialize_root): Delete.
5125 (execute_pred_commoning_chain): Initialize root vars and replace
5126 reference of non-combined chain directly, rather than call above
5129 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5131 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
5132 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
5134 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5136 * tree-predcom.c (struct chain): New field init_seq.
5137 (release_chain): Release init_seq.
5138 (prepare_initializers_chain): Record intialization stmts in above
5140 (insert_init_seqs): New function.
5141 (tree_predictive_commoning_loop): Call insert_init_seqs.
5143 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5145 * tree-predcom.c (determine_roots_comp): Skip trivial components.
5147 2017-07-28 Richard Biener <rguenther@suse.de>
5149 * match.pd: Remove superfluous :c.
5150 * genmatch.c (simplify::id): Add member.
5151 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
5153 (current_id): New global.
5154 (dt_node::parent): Move from ...
5155 (dt_operand::parent): ... here. Add for_id member.
5156 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
5157 (decision_tree::find_node): Relax order requirement when
5158 merging DT_TRUE nodes to ones inbetween the current simplify
5159 and the one we try to merge with. Add diagnostic whenever
5160 we need to enforce pattern order by not merging.
5161 (decision_tree::insert): Set current_id.
5162 (decision_tree::print_node): Dump parent node and for_id.
5163 (parser::last_id): Add member.
5164 (parser::push_simplify): Assign unique id.
5165 (parser::parser): Initialize last_id.
5167 2017-07-28 Martin Liska <mliska@suse.cz>
5170 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
5171 gimple_build_debug_bind.
5173 2017-07-28 Richard Biener <rguenther@suse.de>
5175 PR tree-optimization/81502
5176 * match.pd: Add pattern combining BIT_INSERT_EXPR with
5178 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
5180 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
5181 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
5182 for BIT_FIELD_REF args.
5183 * fold-const.c (make_bit_field_ref): Likewise.
5184 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5186 2017-07-28 Jakub Jelinek <jakub@redhat.com>
5189 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
5190 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
5191 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
5192 Or it into SANITIZER_UNDEFINED.
5193 * ubsan.c: Include gimple-fold.h and varasm.h.
5194 (ubsan_expand_ptr_ifn): New function.
5195 (instrument_pointer_overflow): New function.
5196 (maybe_instrument_pointer_overflow): New function.
5197 (instrument_object_size): Formatting fix.
5198 (pass_ubsan::execute): Call instrument_pointer_overflow
5199 and maybe_instrument_pointer_overflow.
5200 * internal-fn.c (expand_UBSAN_PTR): New function.
5201 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
5202 * sanitizer.def (__ubsan_handle_pointer_overflow,
5203 __ubsan_handle_pointer_overflow_abort): New builtins.
5204 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
5205 * internal-fn.def (UBSAN_PTR): New internal function.
5206 * opts.c (sanitizer_opts): Add pointer-overflow.
5207 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
5208 * fold-const.c (build_range_check): Compute pointer range check in
5209 integral type if pointer arithmetics would be needed. Formatting
5212 2017-07-28 Martin Liska <mliska@suse.cz>
5215 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
5216 parameters that are of a variable-length.
5218 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
5220 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
5222 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
5223 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
5224 (CRT_CALL_STATIC_FUNCTION): Likewise.
5225 (ASM_DEFAULT_SPEC): New define.
5226 (ASM_SPEC32): Likewise.
5227 (ASM_SPEC64): Likewise.
5228 (ASM_SPEC_COMMON): Likewise.
5229 (ASM_SPEC): Likewise.
5230 (INVALID_64BIT): Likewise.
5231 (LINK_OS_DEFAULT_SPEC): Likewise.
5232 (LINK_OS_SPEC32): Likewise.
5233 (LINK_OS_SPEC64): Likewise.
5234 (POWERPC_LINUX): Likewise.
5235 (PTRDIFF_TYPE): Likewise.
5236 (RESTORE_FP_PREFIX): Likewise.
5237 (RESTORE_FP_SUFFIX): Likewise.
5238 (SAVE_FP_PREFIX): Likewise.
5239 (SAVE_FP_SUFFIX): Likewise.
5240 (SIZE_TYPE): Likewise.
5241 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
5242 (TARGET_64BIT): Likewise.
5243 (TARGET_64BIT): Likewise.
5244 (TARGET_AIX): Likewise.
5245 (WCHAR_TYPE_SIZE): Likewise.
5246 (WCHAR_TYPE): Undefine.
5247 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
5248 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
5249 (CPP_OS_RTEMS_SPEC): Delete.
5250 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
5251 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
5253 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
5255 2017-07-28 Jakub Jelinek <jakub@redhat.com>
5257 PR tree-optimization/81578
5258 * tree-parloops.c (build_new_reduction): Bail out if
5259 reduction_code isn't one of the standard OpenMP reductions.
5260 Move the details printing after that decision.
5262 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
5264 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
5265 related to reload_in_progress.
5266 (splat_input_operand): Likewise.
5267 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
5269 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
5271 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
5272 (TARGET_INSTANTIATE_DECLS): Likewise.
5273 (legitimate_indexed_address_p): Delete reload_in_progress code.
5274 (rs6000_debug_legitimate_address_p): Likewise.
5275 (rs6000_eliminate_indexed_memrefs): Likewise.
5276 (rs6000_emit_le_vsx_store): Likewise.
5277 (rs6000_emit_move_si_sf_subreg): Likewise.
5278 (rs6000_emit_move): Likewise.
5279 (register_to_reg_type): Likewise.
5280 (rs6000_pre_atomic_barrier): Likewise.
5281 (rs6000_machopic_legitimize_pic_address): Likewise.
5282 (rs6000_allocate_stack_temp): Likewise.
5283 (rs6000_address_for_fpconvert): Likewise.
5284 (rs6000_address_for_altivec): Likewise.
5285 (rs6000_secondary_memory_needed_rtx): Delete function.
5286 (rs6000_check_sdmode): Likewise.
5287 (rs6000_alloc_sdmode_stack_slot): Likewise.
5288 (rs6000_instantiate_decls): Likewise.
5289 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
5290 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
5291 Delete reload_in_progress.
5292 (*vec_reload_and_plus_<mptrsize>): Likewise.
5293 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
5294 (vsx_div_v2di): Likewise.
5295 (vsx_udiv_v2di): Likewise.
5297 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
5299 * config/rs6000/rs6000.opt (mlra): Replace with stub.
5300 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
5301 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
5302 (rs6000_debug_reg_global): Delete print of LRA status.
5303 (rs6000_option_override_internal): Delete dead LRA related code.
5304 (rs6000_lra_p): Delete function.
5305 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
5307 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
5309 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
5310 * config/riscv/rtems.h: New file.
5312 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5313 Sudakshina Das <sudi.das@arm.com>
5315 * config/aarch64/aarch64.md
5316 (define_split for and<mode>3nr_compare): Move
5317 non aarch64_logical_operand to a register.
5318 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
5319 register immediate operand to a register.
5320 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
5322 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
5325 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
5327 2017-07-27 Richard Biener <rguenther@suse.de>
5329 PR tree-optimization/81573
5330 PR tree-optimization/81494
5331 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
5332 multi defuse cycle case.
5334 2017-07-27 Richard Biener <rguenther@suse.de>
5336 PR tree-optimization/81571
5337 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
5340 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
5342 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
5343 earlier and only if MASK_FPU is set. Adjust formatting.
5345 2017-07-27 Martin Liska <mliska@suse.cz>
5347 * opt-functions.awk: Add validation of value of Init.
5348 * optc-gen.awk: Pass new argument.
5350 2017-07-27 Martin Liska <mliska@suse.cz>
5352 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
5353 Fix wrong condition.
5355 2017-07-27 Martin Liska <mliska@suse.cz>
5357 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
5358 BBs and edges seen by autoFDO.
5360 2017-07-27 Richard Biener <rguenther@suse.de>
5362 PR tree-optimization/81502
5363 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
5364 with incompatible but same sized type.
5365 (execute_update_addresses_taken): Likewise.
5367 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
5369 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
5370 flag_tree_loop_vectorize rather than flag_tree_vectorize.
5372 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5375 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
5376 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
5377 Change s_operand to memory_operand.
5379 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
5381 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
5382 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
5383 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
5384 Emit instructions rather than returning an expression. Handle TFmode
5385 and KFmode by casting to TImode.
5386 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
5387 (rs6000_emit_le_vsx_store): Likewise.
5388 * config/rs6000/vsx.md (VSX_TI): New iterator.
5389 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
5390 (*vsx_le_undo_permute_<mode>): Likewise.
5391 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
5392 emit the split sequence.
5393 (*vsx_le_perm_store_<mode>): Likewise.
5395 2017-07-27 Jakub Jelinek <jakub@redhat.com>
5397 PR tree-optimization/81555
5398 PR tree-optimization/81556
5399 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
5400 if true, force CHANGED for the recursive invocation.
5401 (reassociate_bb): Remember original length of ops array, pass
5402 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
5404 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
5405 attributes for noipa attribute. For naked attribute use
5406 lookup_attribute first before lookup_attribute_spec.
5407 * final.c (rest_of_handle_final): Disable IPA RA for functions with
5409 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
5410 for functions with noipa attribute.
5411 (cgraph_externally_visible_p): Return true for functions with noipa
5413 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
5414 for functions with noipa attribute.
5415 * doc/extend.texi: Document noipa function attribute.
5416 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
5417 also for functions with noipa attribute.
5418 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
5420 2017-07-26 Andrew Pinski <apinski@cavium.com>
5422 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
5423 vec_unalign_load_cost and vec_unalign_store_cost.
5425 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5427 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
5428 -mvsx-small-integer option.
5429 (ISA_3_0_MASKS_IEEE): Likewise.
5430 (OTHER_VSX_VECTOR_MASKS): Likewise.
5431 (POWERPC_MASKS): Likewise.
5432 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
5433 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
5434 code, only testing for DImode being allowed in non-VSX floating
5436 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
5437 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
5439 (rs6000_option_override_internal): Delete -mvsx-small-integer.
5440 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
5441 TARGET_P8_VECTOR test.
5442 (rs6000_secondary_reload_simple_move): Likewise.
5443 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
5444 since TARGET_P9_VECTOR was already tested.
5445 (rs6000_opt_masks): Remove -mvsx-small-integer.
5446 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
5447 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
5449 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
5450 test for TARGET_VEXTRACTUB was used, and that uses
5452 (p9 extract splitter): Likewise.
5453 (vsx_extract_<mode>_di_p9): Likewise.
5454 (vsx_extract_<mode>_store_p9): Likewise.
5455 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
5456 for TARGET_P9_VECTOR was used. Delete code that is now dead with
5457 the elimination of TARGET_VSX_SMALL_INTEGER.
5458 (vsx_extract_<mode>_p8): Likewise.
5459 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
5460 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
5461 (vsx_set_<mode>_p9): Likewise.
5462 (vsx_set_v4sf_p9): Likewise.
5463 (vsx_set_v4sf_p9_zero): Likewise.
5464 (vsx_insert_extract_v4sf_p9): Likewise.
5465 (vsx_insert_extract_v4sf_p9_2): Likewise.
5466 * config/rs6000/rs6000.md (sign extend splitter): Change
5467 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
5468 (floatsi<mode>2_lfiwax_mem): Likewise.
5469 (floatunssi<mode>2_lfiwzx_mem): Likewise.
5470 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
5471 since a test for TARGET_P9_VECTOR was used.
5472 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
5473 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
5474 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
5475 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
5476 TARGET_P8_VECTOR test.
5477 (fix_trunc<mode>si2_stfiwx): Likewise.
5478 (fix_trunc<mode>si2_internal): Likewise.
5479 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
5480 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
5482 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
5483 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
5484 TARGET_P8_VECTOR test.
5485 (fixuns_trunc<mode>si2_stfiwx): Likewise.
5486 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
5487 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
5489 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
5490 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
5491 since a test for TARGET_P9_VECTOR was used.
5492 (splitter for loading small constants): Likewise.
5494 2017-07-26 Andrew Pinski <apinski@cavium.com>
5496 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
5499 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
5502 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
5503 (fp_valid_at): Likewise.
5505 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
5507 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
5508 (qdf24xx_addrcost_table): Likewise.
5509 (cortexa57_tunings): Update to use generic_branch_cost.
5510 (cortexa72_tunings): Likewise.
5511 (cortexa73_tunings): Likewise.
5512 (qdf24xx_tunings): Likewise.
5514 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
5516 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
5517 (thunderx2t99_branch_cost): Likewise.
5518 (cortexa35_tunings): Update to use generic_branch_cost.
5519 (cortexa53_tunings): Likewise.
5520 (cortexa57_tunings): Likewise.
5521 (cortexa72_tunings): Likewise.
5522 (cortexa73_tunings): Likewise.
5523 (thunderx2t99_tunings): Likewise.
5525 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
5527 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
5528 (sparc_option_override): Honour MASK_FSMULD.
5529 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
5530 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
5531 * config/sparc/sparc.opt (mfsmuld): New option.
5532 * doc/invoke.texi (mfsmuld): Document option.
5534 2017-07-26 Marek Polacek <polacek@redhat.com>
5537 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
5539 2017-07-26 Richard Biener <rguenther@suse.de>
5541 * gimple-match-head.c (do_valueize): Return OP if valueize
5543 (get_def): New helper to get at the def stmt of a SSA name
5545 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
5546 do_valueize to get at the def stmt.
5547 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
5549 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
5552 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
5554 2017-07-26 Martin Liska <mliska@suse.cz>
5557 * function.c (expand_function_start): Make expansion of
5558 nonlocal_goto_save_area after parm_birth_insn.
5560 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
5562 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
5563 from all CPU target flags enable members.
5565 2017-07-26 Richard Biener <rguenther@suse.de>
5567 * genmatch.c (dt_simplify::gen): Make iterator vars const.
5568 (decision_tree::gen): Make 'type' const.
5569 (write_predicate): Likewise.
5571 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
5573 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
5574 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
5575 (rs6000_option_override_internal): Likewise.
5576 (rs6000_expand_vector_set): Likewise.
5577 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
5578 (TARGET_UPPER_REGS_SF): Likewise.
5579 (TARGET_UPPER_REGS_DI): Likewise.
5580 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
5581 (TARGET_DIRECT_MOVE_64BIT): Likewise.
5582 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
5583 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
5584 (Splitters for DI constants in Altivec registers): Likewise.
5585 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
5586 (vsx_set_v4sf_p9): Likewise.
5587 (vsx_set_v4sf_p9_zero): Likewise.
5588 (vsx_insert_extract_v4sf_p9): Likewise.
5589 (vsx_insert_extract_v4sf_p9_2): Likewise.
5591 2017-07-25 Carl Love <cel@us.ibm.com>
5593 * doc/extend.texi: Update the built-in documentation file for the
5594 existing built-in functions
5595 vector signed char vec_cnttz (vector signed char);
5596 vector unsigned char vec_cnttz (vector unsigned char);
5597 vector signed short vec_cnttz (vector signed short);
5598 vector unsigned short vec_cnttz (vector unsigned short);
5599 vector signed int vec_cnttz (vector signed int);
5600 vector unsigned int vec_cnttz (vector unsigned int);
5601 vector signed long long vec_cnttz (vector signed long long);
5602 vector unsigned long long vec_cnttz (vector unsigned long long);
5604 2017-07-25 Andrew Pinski <apinski@cavium.com>
5606 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
5607 accesses where the use is for the first operand of a BIT_INSERT.
5609 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
5612 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
5613 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
5615 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
5617 * config/i386/gstabs.h: Delete.
5618 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
5620 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
5622 * config/i386/i386.c (ix86_decompose_address): Do not check for
5623 register RTX when looking at index_reg or base_reg.
5624 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
5626 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
5628 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
5629 to update EH info here.
5631 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5633 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
5635 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5637 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
5639 2017-07-25 Torsten Duwe <duwe@suse.de>
5641 * common.opt: Introduce -fpatchable-function-entry
5642 command line option, and its variables function_entry_patch_area_size
5643 and function_entry_patch_area_start.
5644 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
5645 including a two-value parser.
5646 * target.def (print_patchable_function_entry): New target hook.
5647 * targhooks.h (default_print_patchable_function_entry): New function.
5648 * targhooks.c (default_print_patchable_function_entry): Likewise.
5649 * toplev.c (process_options): Switch off IPA-RA if
5650 patchable function entries are being generated.
5651 * varasm.c (assemble_start_function): Look at the
5652 patchable-function-entry command line switch and current
5653 function attributes and maybe generate NOP instructions by
5654 calling the print_patchable_function_entry hook.
5655 * doc/extend.texi: Document patchable_function_entry attribute.
5656 * doc/invoke.texi: Document -fpatchable_function_entry
5657 command line option.
5658 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
5660 * doc/tm.texi: Re-generate.
5662 2017-07-25 Jakub Jelinek <jakub@redhat.com>
5665 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
5666 TARGET_AVX512DQ rather than TARGET_AVX512BW.
5668 2017-07-25 Tamar Christina <tamar.christina@arm.com>
5670 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
5672 2017-07-25 Richard Biener <rguenther@suse.de>
5674 PR tree-optimization/81455
5675 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
5676 not walk in cycles when looking for guards.
5678 2017-07-25 Richard Biener <rguenther@suse.de>
5680 PR tree-optimization/81529
5681 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
5682 when optimizing backedge uses.
5684 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
5686 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
5688 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
5689 to dl_section_ref. On AIX, append an expression to subtract
5690 the size of the section length to dl_section_ref.
5692 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
5694 * configure.ac: If any of the config.* scripts fail, exit 1.
5695 * configure: Regenerate.
5697 2017-07-25 Richard Biener <rguenther@suse.de>
5700 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
5701 of immediate uses, be more verbose on errors.
5703 2017-07-25 Richard Biener <rguenther@suse.de>
5705 PR tree-optimization/81510
5706 * tree-vect-loop.c (vect_is_simple_reduction): When the
5707 reduction stmt is not inside the loop bail out.
5709 2017-07-25 Richard Biener <rguenther@suse.de>
5711 PR tree-optimization/81303
5712 * tree-vect-loop-manip.c (vect_loop_versioning): Build
5713 profitability check against LOOP_VINFO_NITERSM1.
5715 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5717 * domwalk.c (cmp_bb_postorder): Simplify.
5718 (sort_bbs_postorder): New function. Use it...
5719 (dom_walker::walk): ...here to optimize common cases.
5721 2017-07-25 Martin Liska <mliska@suse.cz>
5724 * ipa-visibility.c (function_and_variable_visibility): Make the
5725 redirection just on target that supports aliasing.
5726 Fix GNU coding style.
5728 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
5731 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
5733 * config/arm/rtems.h: Likewise.
5734 * config/bfin/rtems.h: Likewise.
5735 * config/i386/rtemself.h: Likewise.
5736 * config/lm32/rtems.h: Likewise.
5737 * config/m32c/rtems.h: Likewise.
5738 * config/m68k/rtemself.h: Likewise.
5739 * config/microblaze/rtems.h: Likewise.
5740 * config/mips/rtems.h: Likewise.
5741 * config/moxie/rtems.h: Likewise.
5742 * config/nios2/rtems.h: Likewise.
5743 * config/powerpcspe/rtems.h: Likewise.
5744 * config/rs6000/rtems.h: Likewise.
5745 * config/rtems.h: Likewise.
5746 * config/sh/rtems.h: Likewise.
5747 * config/sh/rtemself.h: Likewise.
5748 * config/sparc/rtemself.h: Likewise.
5750 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
5753 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
5754 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
5755 * tree-ssa-structalias.c (alias_get_name): Same.
5757 2017-07-25 Bin Cheng <bin.cheng@arm.com>
5760 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
5761 instructions if no du chain is found.
5763 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
5765 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
5767 2017-07-25 Richard Biener <rguenther@suse.de>
5770 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
5773 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
5775 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
5777 (ISA_2_7_MASKS_SERVER): Likewise.
5778 (ISA_3_0_MASKS_IEEE): Likewise.
5779 (OTHER_P8_VECTOR_MASKS): Likewise.
5780 (OTHER_VSX_VECTOR_MASKS): Likewise.
5781 (POWERPC_MASKS): Likewise.
5782 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
5783 duplicate list of options.
5784 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
5785 explicit -mupper-regs options.
5786 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
5787 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
5788 alias for -mupper-regs-df.
5789 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
5790 (rs6000_init_hard_regno_mode_ok): Likewise.
5791 (rs6000_option_override_internal): Likewise.
5792 (rs6000_opt_masks): Likewise.
5793 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
5794 options in terms of whether -mvsx or -mpower8-vector was used.
5795 (TARGET_UPPER_REGS_DI): Likewise.
5796 (TARGET_UPPER_REGS_SF): Likewise.
5797 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
5798 -mupper-regs-* options.
5800 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
5802 * passes.c (emergency_dump_function): Print some empty lines and a
5803 header before the RTL dump.
5805 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
5807 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
5809 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
5812 * config/aarch64/aarch64.c (aarch64_classify_symbol):
5813 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
5815 2017-07-24 Carl Love <cel@us.ibm.com>
5817 * config/rs6000/rs6000-c.c: Add support for built-in functions
5818 vector float vec_extract_fp32_from_shorth (vector unsigned short);
5819 vector float vec_extract_fp32_from_shortl (vector unsigned short);
5820 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
5821 vec_extract_fp_from_shortl): Add defines for the two builtins.
5822 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
5823 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
5825 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
5826 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
5827 * doc/extend.texi: Update the built-in documentation file for the
5828 new built-in function.
5830 2017-07-24 Jakub Jelinek <jakub@redhat.com>
5833 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
5835 * doc/generic.texi: Likewise.
5836 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
5837 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
5839 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
5841 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
5842 (aarch64_mls_elt_merge<mode>): Likewise.
5844 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
5846 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
5847 having __cxa_atexit.
5849 2017-07-23 Michael Collison <michael.collison@arm.com>
5851 * config/arm/arm.c (arm_option_override): Deprecate
5852 use of -mstructure-size-boundary.
5853 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
5854 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
5856 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5859 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
5860 Reduce cost estimate for direct moves.
5862 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
5865 * config/i386/i386.c (ix86_option_override_internal): Disable
5866 BMI, BMI2 and TBM instructions for -m16.
5868 2017-07-21 Carl Love <cel@us.ibm.com>
5870 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5871 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5872 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5873 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5874 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5875 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5876 VMULOSW): New enum "unspec" values.
5877 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5878 altivec_vmulosw): New patterns.
5879 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5880 VMULOSW): Add definitions.
5882 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
5884 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
5885 (qdf24xx): Likewise.
5886 * config/aarch64/aarch64-options-extensions.def (rdma); New.
5887 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
5888 (AARCH64_FL_V8_1): Renumber.
5889 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
5890 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
5891 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
5892 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
5893 rdma to feature modifiers list.
5895 2017-07-21 Yury Gribov <tetra2005@gmail.com>
5898 * ipa-visibility (function_and_variable_visibility): Convert
5899 recursive PLT call to direct call if appropriate.
5901 2017-07-21 Andrew Pinski <apinski@cavium.com>
5903 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
5904 operand 1 to see if the types precision matches.
5905 * fold-const.c (operand_equal_p): Likewise.
5907 2017-07-21 Richard Biener <rguenther@suse.de>
5909 PR tree-optimization/81303
5910 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
5911 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
5912 (vect_peeling_hash_get_lowest_cost): Adjust.
5913 (vect_enhance_data_refs_alignment): Likewise. Use
5914 vect_get_peeling_costs_all_drs to compute the penalty for no
5915 peeling to match up costs.
5917 2017-07-21 Richard Biener <rguenther@suse.de>
5919 PR tree-optimization/81500
5920 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
5921 we didn't identify a reduction path.
5923 2017-07-21 Tom de Vries <tom@codesourcery.com>
5924 Cesar Philippidis <cesar@codesourcery.com>
5926 PR gcov-profile/81442
5927 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
5930 2017-07-21 Tom de Vries <tom@codesourcery.com>
5933 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
5935 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
5936 nvptx_override_options_after_change.
5938 2017-07-21 Ulrich Drepper <drepper@redhat.com>
5940 * dwarf2out.c (output_file_names): Avoid double testing for
5943 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
5945 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
5947 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5949 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
5951 (try_crossjump_to_edge): Do not punt on partitioned functions.
5953 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5955 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
5956 Put all BBs reachable only via paths crossing cold region to cold
5958 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
5960 2016-07-21 Richard Biener <rguenther@suse.de>
5962 PR tree-optimization/81303
5963 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
5964 into account prologue and epilogue iterations when raising
5965 min_profitable_iters to sth at least covering one vector iteration.
5967 2017-07-21 Tamar Christina <tamar.christina@arm.com>
5969 * config/arm/arm.c (arm_test_cpu_arch_dat):
5972 2017-07-20 Nathan Sidwell <nathan@acm.org>
5974 Remove TYPE_METHODS.
5975 * tree.h (TYPE_METHODS): Delete.
5976 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
5977 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
5978 (dbxout_type_methods): Scan TYPE_FIELDS.
5979 (dbxout_type): Don't check TYPE_METHODS here.
5980 * function.c (use_register_for_decl): Always ignore register for
5981 class types when not optimizing.
5982 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
5983 * tree.c (free_lang_data_in_type): Stitch out member functions and
5984 templates from TYPE_FIELDS.
5985 (build_distinct_type_copy, verify_type_variant,
5986 verify_type): Member fns are on TYPE_FIELDS.
5987 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
5988 * tree-pretty-print.c (dump_generic_node): Likewise.
5990 2017-07-20 Jakub Jelinek <jakub@redhat.com>
5993 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
5994 V2TImode and V4TImode.
5995 (ix86_expand_vector_extract): Likewise.
5996 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
5997 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
5998 (ssescalarmode): Handle V4TImode and V2TImode.
5999 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
6000 (*vec_extractv2ti, *vec_extractv4ti): New insns.
6001 (VEXTRACTI128_MODE): New mode iterator.
6002 (splitter for *vec_extractv?ti first element): New.
6003 (VEC_INIT_MODE): New mode iterator.
6004 (vec_init<mode>): Consolidate 3 expanders into one using
6005 VEC_INIT_MODE mode iterator.
6007 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
6009 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
6010 non_spilled_static_chain_regno_p.
6012 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
6014 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
6016 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
6018 * bb-reorder.c (connect_traces): Allow copying of blocks within
6021 2017-07-20 Richard Biener <rguenther@suse.de>
6023 * gimple.h (gimple_phi_result): Add gphi * overload.
6024 (gimple_phi_result_ptr): Likewise.
6025 (gimple_phi_arg): Likewise. Adjust index assert to only
6026 allow actual argument accesses rather than all slots available
6028 (gimple_phi_arg_def): Add gphi * overload.
6029 * tree-phinodes.c (make_phi_node): Initialize only actual
6031 (resize_phi_node): Clear memory not covered by old node,
6032 do not initialize excess argument slots.
6033 (reserve_phi_args_for_new_edge): Initialize new argument slot
6036 2017-07-20 Bin Cheng <bin.cheng@arm.com>
6038 PR tree-optimization/81388
6040 2016-07-21 Bin Cheng <bin.cheng@arm.com>
6042 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
6043 by removing computation of may_be_zero.
6045 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
6046 Tom de Vries <tom@codesourcery.com>
6049 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
6050 when gimple level profile disagrees with what RTL expander did.
6052 2017-07-20 Richard Biener <rguenther@suse.de>
6054 PR tree-optimization/61171
6055 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
6056 (vect_analyze_stmt): Add slp instance parameter.
6057 (vectorizable_reduction): Likewise.
6058 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
6059 (vect_is_simple_reduction): Deal with chains not detected
6060 as SLP reduction chain, specifically not properly associated
6061 chains containing a mix of plus/minus.
6062 (get_reduction_op): Remove.
6063 (get_initial_defs_for_reduction): Simplify, pass in whether
6064 this is a reduction chain, pass in the SLP node for the PHIs.
6065 (vect_create_epilog_for_reduction): Get the SLP instance as
6067 (vectorizable_reduction): Get the SLP instance as arg.
6068 During analysis remember the SLP node with the PHIs in the
6069 instance. Simplify getting at the vectorized reduction PHIs.
6070 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
6071 through SLP instance.
6072 (vect_slp_analyze_operations): Likewise.
6073 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
6074 (vect_transform_stmt): Likewise.
6076 2017-07-20 Tom de Vries <tom@codesourcery.com>
6078 PR tree-optimization/81489
6079 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
6080 read of phi arg location to before loop that modifies phi.
6082 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
6084 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
6087 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6090 * except.c (execute): Fix ordering issue.
6092 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
6094 PR rtl-optimization/81423
6095 * combine.c (make_compound_operation_int): Don't try to optimize
6096 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
6098 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
6100 PR rtl-optimization/81423
6101 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
6102 with a constant that is -1 in the truncated to mode.
6104 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6106 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
6107 (determine_unlikely_bbs): ... here.
6108 * predict.h (propagate_unlikely_bbs_forward): Declare.
6109 * cfgexpand.c (pass_expand::execute): Use it.
6110 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
6112 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
6113 propagate_unlikely_bbs_forward.
6115 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6118 * except.c (maybe_add_nop_after_section_switch): New function.
6121 2017-07-19 Tom de Vries <tom@codesourcery.com>
6123 * gimple.h (gimple_phi_set_arg): Make assert more strict.
6125 2017-07-19 Tom de Vries <tom@codesourcery.com>
6127 * gimple.h (gimple_phi_arg): Make assert more strict.
6129 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
6131 * config.gcc (powerpc*-*-*): Add mmintrin.h.
6132 * config/rs6000/mmintrin.h: New file.
6133 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
6135 2017-07-19 Jakub Jelinek <jakub@redhat.com>
6137 PR tree-optimization/81346
6138 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
6140 2017-07-19 Tom de Vries <tom@codesourcery.com>
6142 * config/nvptx/nvptx.md (VECIM): Add V2DI.
6144 2017-07-19 Tom de Vries <tom@codesourcery.com>
6146 * config/nvptx/nvptx-modes.def: Add V2DImode.
6147 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
6148 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
6149 (nvptx_output_mov_insn): Handle lack of mov.b128.
6150 (nvptx_print_operand): Handle 'H' and 'L' codes.
6151 (nvptx_vector_mode_supported): Allow V2DImode.
6152 (nvptx_preferred_simd_mode): New function.
6153 (nvptx_data_alignment): New function.
6154 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
6155 nvptx_preferred_simd_mode.
6156 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
6158 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
6160 2017-07-19 Tom de Vries <tom@codesourcery.com>
6162 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
6163 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
6164 (nvptx_vector_mode_supported): New function. Allow V2SImode.
6165 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
6166 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
6167 (mov<VECIM>_insn): New define_insn.
6168 (define_expand "mov<VECIM>): New define_expand.
6170 2017-07-19 Tom de Vries <tom@codesourcery.com>
6172 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
6174 2017-07-19 Jakub Jelinek <jakub@redhat.com>
6176 PR tree-optimization/81346
6177 * fold-const.h (fold_div_compare, range_check_type): Declare.
6178 * fold-const.c (range_check_type): New function.
6179 (build_range_check): Use range_check_type.
6180 (fold_div_compare): No longer static, rewritten into
6181 a match.pd helper function.
6182 (fold_comparison): Don't call fold_div_compare here.
6183 * match.pd (X / C1 op C2): New optimization using fold_div_compare
6186 2017-07-19 Nathan Sidwell <nathan@acm.org>
6188 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
6189 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
6190 * tree.c (find_decls_types_r, verify_type): Use
6191 TYPE_{MIN,MAX}_VALUE_RAW.
6192 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
6193 (hash_tree): Likewise.
6194 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
6196 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
6199 2017-07-18 Tom de Vries <tom@codesourcery.com>
6202 * omp-expand.c (expand_omp_for_static_chunk): Handle
6203 equal-argument loop exit phi.
6205 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
6208 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
6209 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
6210 operand 2 predicate.
6211 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
6212 operand 2 predicate.
6213 (ror,rol -> rorx splitters): Use const_int_operand as
6214 operand 2 predicate.
6216 2017-06-18 Richard Biener <rguenther@suse.de>
6218 PR tree-optimization/81410
6219 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
6220 the gap in the ! slp_perm SLP case after each group.
6222 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
6225 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
6228 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
6231 * predict.c (set_even_probabilities): Cleanup; do not affect
6232 probabilities that are already known.
6233 (combine_predictions_for_bb): Call even when count is set.
6235 2017-07-18 Nathan Sidwell <nathan@acm.org>
6237 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
6240 2017-07-18 Bin Cheng <bin.cheng@arm.com>
6243 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
6244 optimization for loop niter analysis.
6246 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
6249 * config/avr/avr.c (avr_optimize_casesi): Don't use
6250 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
6252 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
6254 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
6255 body_cost_vec from _vect_peel_extended_info.
6256 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
6257 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
6260 2017-07-18 Bin Cheng <bin.cheng@arm.com>
6262 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
6264 2017-07-18 Richard Biener <rguenther@suse.de>
6266 PR tree-optimization/80620
6267 PR tree-optimization/81403
6268 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
6269 info when re-using a VN table entry.
6271 2017-07-18 Richard Biener <rguenther@suse.de>
6273 PR tree-optimization/81418
6274 * tree-vect-loop.c (vectorizable_reduction): Properly compute
6275 vectype_in. Verify that with lane-reducing reduction operations
6276 we have a single def-use cycle.
6278 2017-07-17 Carl Love <cel@us.ibm.com>
6280 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
6282 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6283 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6284 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
6285 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
6286 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
6287 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
6288 VMULOSW): New enum "unspec" values.
6289 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
6290 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
6291 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
6292 altivec_vmulosw): New patterns.
6293 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
6294 VMULOSW): Add definitions.
6295 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
6297 * config/alpha/alpha.c: Include predict.h.
6299 2017-07-17 Yury Gribov <tetra2005@gmail.com>
6301 * tree-vrp.c (compare_assert_loc): Fix comparison function
6302 to return predictable results.
6304 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6306 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
6309 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
6310 * doc/invoke.texi (mexpand-adddi): Update text.
6312 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6314 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
6315 that also clobbers the CC register. The old expand code is moved
6317 (*arc_clzsi2): ... here.
6318 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
6319 the CC register. The old expand code is moved to ...
6320 (arc_ctzsi2): ... here.
6322 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6324 * config/arc/arc.opt (mindexed-loads): Use initial value
6325 TARGET_INDEXED_LOADS_DEFAULT.
6326 (mauto-modify-reg): Use initial value
6327 TARGET_AUTO_MODIFY_REG_DEFAULT.
6328 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
6329 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
6330 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
6331 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
6333 2017-07-17 Martin Liska <mliska@suse.cz>
6336 * opts.c (finish_options): Do not allow -fgnu-tm
6337 w/ -fsanitize={kernel-,}address. Say sorry.
6339 2017-07-17 Bin Cheng <bin.cheng@arm.com>
6342 * tree-loop-distribution.c (classify_partition): Only assert on
6343 numer of iterations.
6344 (merge_dep_scc_partitions): Delete prameter. Update function call.
6345 (distribute_loop): Remove code handling loop with unknown niters.
6346 (pass_loop_distribution::execute): Skip loop with unknown niters.
6348 2017-07-17 Bin Cheng <bin.cheng@arm.com>
6351 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
6352 function sort_partitions_by_post_order.
6354 2017-07-17 Bin Cheng <bin.cheng@arm.com>
6356 PR tree-optimization/81374
6357 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
6358 the max index of basic blocks, rather than number of basic blocks.
6360 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6362 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
6364 (arc_legitimate_pic_operand_p): Likewise.
6365 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
6367 (arc_needs_pcl_p): Likewise.
6368 (arc_legitimate_pc_offset_p): Likewise.
6369 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
6370 function is also used in constrains.md.
6371 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
6372 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
6373 PLUS. Only return true/false in known cases, otherwise assert.
6374 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
6375 is already called in arc_legitimate_constant_p.
6376 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
6378 (LEGITIMATE_PIC_OPERAND_P): Use
6379 arc_raw_symbolic_reference_mentioned_p function.
6380 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
6385 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6386 Andrew Burgess <andrew.burgess@embecosm.com>
6388 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
6389 (arc_return_address_register): New function.
6390 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
6391 (arc_handle_fndecl_attribute): Add naked attribute.
6392 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
6393 (TARGET_WARN_FUNC_RETURN): Likewise.
6394 (arc_allocate_stack_slots_for_args): New function.
6395 (arc_warn_func_return): Likewise.
6396 (machine_function): Change type fn_type.
6397 (arc_compute_function_type): Consider new naked function type,
6398 change function return type.
6399 (arc_must_save_register): Adapt to handle new
6400 arc_compute_function_type's return type.
6401 (arc_expand_prologue): Likewise.
6402 (arc_expand_epilogue): Likewise.
6403 (arc_return_address_regs): Delete.
6404 (arc_return_address_register): New function.
6405 (arc_epilogue_uses): Use above function.
6406 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
6407 (arc_function_type): Change encoding, add naked type.
6408 (ARC_INTERRUPT_P): Change to handle the new encoding.
6409 (ARC_FAST_INTERRUPT_P): Likewise.
6410 (ARC_NORMAL_P): Define.
6411 (ARC_NAKED_P): Likewise.
6412 (arc_compute_function_type): Delete prototype.
6413 * config/arc/arc.md (in_ret_delay_slot): Use
6414 arc_return_address_register function.
6415 (simple_return): Likewise.
6416 (p_return_i): Likewise.
6418 2017-07-17 Jakub Jelinek <jakub@redhat.com>
6420 PR tree-optimization/81428
6421 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
6422 can't be built for those types.
6424 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
6426 Remove stuff dead since r239246.
6428 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
6429 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
6430 (avr_inform_devices): Remove dead stuff.
6432 2017-07-17 Tamar Christina <tamar.christina@arm.com>
6434 * config/arm/arm_neon.h: Fix softp typo.
6436 2017-07-17 Jakub Jelinek <jakub@redhat.com>
6438 PR tree-optimization/81365
6439 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
6440 aggregate moves onto bb predecessor edges, make sure there are no
6441 loads that could alias the lhs in between the start of bb and the
6444 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
6447 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
6448 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
6449 [LSHIFTRT, outer_code = TRUNCATE]: Same.
6451 2017-07-17 Jakub Jelinek <jakub@redhat.com>
6453 PR tree-optimization/81396
6454 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
6455 (init_symbolic_number): Initialize it to 1.
6456 (perform_symbolic_merge): Add n_ops from both operands into the new
6458 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
6459 without base_addr as useless if they need more than one operation.
6460 (bswap_replace): Handle !bswap case for NULL base_addr.
6462 2017-07-17 Tom de Vries <tom@codesourcery.com>
6465 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
6468 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
6470 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
6471 conditional builtin define __FIX_LEON3FT_B2BST.
6473 2017-07-17 Daniel Cederman <cederman@gaisler.com>
6475 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
6476 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
6479 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
6481 PR rtl-optimization/81424
6482 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
6483 to remove potential trapping from operands if -fnon-call-exceptions.
6485 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6487 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
6488 profile_proability for scalling.
6489 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
6491 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6493 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
6495 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6497 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
6498 fixpoint arithmetics.
6500 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6502 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
6503 fixpoint arithmetics.
6505 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6507 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
6508 fixpoint arithmetics.
6510 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6512 * profile-count.h (profile_probability::from_reg_br_prob_note,
6513 profile_probability::to_reg_br_prob_note): New functions.
6514 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
6515 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
6516 * predict.c (probability_reliable_p): Update.
6517 (edge_probability_reliable_p): Update.
6518 (br_prob_note_reliable_p): Update.
6519 (invert_br_probabilities): Update.
6520 (add_reg_br_prob_note): New function.
6521 (combine_predictions_for_insn): Update.
6522 * asan.c (asan_clear_shadow): Update.
6523 * cfgbuild.c (compute_outgoing_frequencies): Update.
6524 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6525 (update_br_prob_note): Update.
6526 (rtl_verify_edges): Update.
6527 (purge_dead_edges): Update.
6528 (fixup_reorder_chain): Update.
6529 * emit-rtl.c (try_split): Update.
6530 * ifcvt.c (cond_exec_process_insns): Update.
6531 (cond_exec_process_if_block): Update.
6532 (dead_or_predicable): Update.
6533 * internal-fn.c (expand_addsub_overflow): Update.
6534 (expand_neg_overflow): Update.
6535 (expand_mul_overflow): Update.
6536 * loop-doloop.c (doloop_modify): Update.
6537 * loop-unroll.c (compare_and_jump_seq): Update.
6538 * optabs.c (emit_cmp_and_jump_insn_1): Update.
6539 * predict.h: Update.
6540 * reorg.c (mostly_true_jump): Update.
6542 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
6543 * config/alpha/alpha.c (emit_unlikely_jump): Update.
6544 * config/arc/arc.c: (emit_unlikely_jump): Update.
6545 * config/arm/arm.c: (emit_unlikely_jump): Update.
6546 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
6547 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
6548 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
6549 (ix86_print_operand): Update.
6550 (ix86_split_fp_branch): Update.
6551 (predict_jump): Update.
6552 * config/ia64/ia64.c (ia64_print_operand): Update.
6553 * config/mmix/mmix.c (mmix_print_operand): Update.
6554 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
6555 (rs6000_expand_split_stack_prologue): Update.
6556 * config/rs6000/rs6000.c: Update.
6557 * config/s390/s390.c (s390_expand_vec_strlen): Update.
6558 (s390_expand_vec_movstr): Update.
6559 (s390_expand_cs_tdsi): Update.
6560 (s390_expand_split_stack_prologue): Update.
6561 * config/sh/sh.c (sh_print_operand): Update.
6562 (expand_cbranchsi4): Update.
6563 (expand_cbranchdi4): Update.
6564 * config/sparc/sparc.c (output_v9branch): Update.
6565 * config/spu/spu.c (get_branch_target): Update.
6566 (ea_load_store_inline): Update.
6567 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
6568 * config/tilepro/tilepro.c: Update.
6570 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
6572 * gimplify.c (mostly_copy_tree_r): Revert latest change.
6573 (gimplify_save_expr): Likewise.
6575 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6577 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
6579 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6581 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
6583 * timevar.def (TV_IPA_FNSUMMARY): Define.
6585 2017-07-16 Daniel Cederman <cederman@gaisler.com>
6587 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
6588 to back store errata sensitive sequence from being generated.
6589 (sqrtdf2_fix): Likewise.
6591 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6593 * tree-ssa-threadupdate.c (compute_path_counts,
6594 update_joiner_offpath_counts): Use profile_probability.
6596 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
6599 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6601 * config/arm/arm-c.c (arm_cpu_builtins): Define
6602 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
6604 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
6606 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6607 array entries to represent __ieee128 versions of the
6608 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
6609 scalar_extract_sig, and scalar_insert_exp built-in functions.
6610 (altivec_resolve_overloaded_builtin): Add special case handling
6611 for the __builtin_scalar_insert_exp function, as represented by
6612 the P9V_BUILTIN_VEC_VSIEDP constant.
6613 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
6614 exponent support for __ieee128 argument.
6615 (VSESQP): Add scalar extract signature support for __ieee128
6617 (VSTDCNQP): Add scalar test negative support for __ieee128
6619 (VSIEQP): Add scalar insert exponent support for __int128 argument
6620 with __ieee128 result.
6621 (VSIEQPF): Add scalar insert exponent support for __ieee128
6622 argument with __ieee128 result.
6623 (VSTDCQP): Add scalar test data class support for __ieee128
6625 (VSTDCNQP): Add overload support for scalar test negative with
6627 (VSTDCQP): Add overload support for scalar test data class
6629 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
6631 (UNSPEC_VSX_SIEXPQP): New constant.
6632 (xsxexpqp): New insn for VSX scalar extract exponent quad
6634 (xsxsigqp): New insn for VSX scalar extract significand quad
6636 (xsiexpqpf): New insn for VSX scalar insert exponent quad
6637 precision with floating point argument.
6638 (xststdcqp): New expand for VSX scalar test data class quad
6640 (xststdcnegqp): New expand for VSX scalar test negative quad
6642 (xststdcqp): New insn to match expansions for VSX scalar test data
6643 class quad precision and VSX scalar test negative quad precision.
6644 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6645 special case operand checking to enforce that second operand of
6646 VSX scalar test data class with quad precision argument is a 7-bit
6648 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
6649 prototypes and descriptions of __ieee128 versions of
6650 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
6651 scalar_test_data_class, and scalar_test_neg built-in functions.
6653 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6655 PR tree-optimization/81162
6656 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
6657 replace a negate with an add.
6659 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
6661 * doc/invoke.texi (arm/-mcpu): Document +crypto.
6663 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6665 * config/arm/arm-c.c (arm_cpu_builtins): Define
6666 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
6668 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6670 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
6671 (armv8-r): Set ARM Cortex-R52 as default CPU.
6672 * config/arm/arm-tables.opt: Regenerate.
6673 * config/arm/arm-tune.md: Regenerate.
6674 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
6676 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
6677 extension for -mcpu=cortex-r52.
6679 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6681 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
6682 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
6683 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
6684 (fp-armv8): Define it as FP_ARMv8 only.
6685 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
6686 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
6688 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
6689 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
6690 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
6691 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
6692 than TARGET_FPU_ARMV8.
6693 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
6694 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
6695 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
6697 * config/arm/neon.md (neon_vrint): Likewise.
6698 (neon_vcvt): Likewise.
6699 (neon_<fmaxmin_op><mode>): Likewise.
6700 (<fmaxmin><mode>3): Likewise.
6701 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
6702 * config/arm/predicates.md (arm_cond_move_operator): Check against
6703 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
6705 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
6707 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
6710 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6712 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
6713 loop in comment with memset.
6715 2017-07-14 Martin Liska <mliska@suse.cz>
6717 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
6718 * dwarf2out.c (is_java): Remove the function.
6719 (output_pubname): Remove usage of the function.
6720 (lower_bound_default): Remove usage of DW_LANG_Java.
6721 (gen_compile_unit_die): Likewise.
6722 * gcc.c: Remove compiler defaults for .java and .zip files.
6723 * gimple-expr.c (remove_suffix): Change as there's no longer
6724 extension than 4-letter one.
6725 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
6726 (gimplify_save_expr): Likewise.
6727 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
6728 as it's possible even for other languages than Java.
6729 * langhooks.h (struct lang_hooks): Remove Java from a comment.
6730 * lto-opts.c (lto_write_options): Remove reference to Java.
6731 * opts.c (strip_off_ending): Update file extension handling.
6732 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
6733 * tree-eh.c (lower_resx): Likewise.
6734 * tree.c (free_lang_data_in_type): Remove dead code.
6735 (find_decls_types_r): Likewise.
6736 (build_common_builtin_nodes): Remove Java from a comment.
6737 (verify_type): Remove dead code.
6738 * varasm.c (assemble_external): Remove Java from a comment.
6740 2017-07-14 Martin Liska <mliska@suse.cz>
6742 * opts.c (finish_options): Add quotes.
6743 (common_handle_option): Likewise.
6745 2017-07-14 Martin Liska <mliska@suse.cz>
6747 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
6748 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
6750 * dwarf2out.c (lower_bound_default): Do not handle
6752 (gen_compile_unit_die): Likewise.
6753 * gcc.c: Remove default extension binding for GNU Pascal.
6754 * stmt.c: Remove Pascal language from a comment.
6755 * xcoffout.c: Likewise.
6757 2017-07-13 David Malcolm <dmalcolm@redhat.com>
6760 * diagnostic-show-locus.c (fixit_cmp): New function.
6761 (layout::layout): Sort m_fixit_hints.
6762 (column_range::column_range): Assert that the values are valid.
6763 (struct char_span): New struct.
6764 (correction::overwrite): New method.
6765 (struct source_line): New struct.
6766 (line_corrections::add_hint): Add assertions. Reimplement memcpy
6767 calls in terms of classes source_line and char_span, and
6768 correction::overwrite.
6769 (selftest::test_overlapped_fixit_printing_2): New function.
6770 (selftest::diagnostic_show_locus_c_tests): Call it.
6772 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6774 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
6775 early if there is no lhs.
6777 2017-07-13 Martin Liska <mliska@suse.cz>
6779 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
6780 (gen_reference_type_die): Likewise.
6781 * stor-layout.c: Remove Pascal-related comment.
6783 2017-07-13 Martin Liska <mliska@suse.cz>
6785 * opts.c (finish_options): Add quotes to error messages.
6786 (parse_sanitizer_options): Likewise.
6788 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
6790 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
6792 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
6794 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
6796 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
6798 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
6800 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
6802 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6805 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
6806 provides the hardware capability bits, define the macro
6807 __BUILTIN_CPU_SUPPORTS__.
6808 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
6809 if GLIBC does not provide the hardware capability bits. Add a
6810 gcc_unreachable call if the built-in cpu function is neither
6811 __builtin_cpu_is nor __builtin_cpu_supports.
6812 (rs6000_get_function_versions_dispatcher): Change the warning
6813 that an old GLIBC is used which does not export the capability
6814 bits to be an error.
6815 * doc/extend.texi (target_clones attribute): Document the
6816 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
6817 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
6818 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
6819 the macros defined by GCC if the newer GLIBC is available.
6821 2017-07-12 Jeff Law <law@redhat.com>
6823 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
6824 remaining includes slightly.
6825 * config/riscv/riscv-builtins.c: Include profile-count.h.
6827 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
6830 * config/avr/avr.c (avr_set_current_function): In diagnostic
6831 messages: Quote keywords and (parts of) identifiers.
6832 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
6835 2017-07-12 Carl Love <cel@us.ibm.com>
6837 * config/rs6000/rs6000-c.c: Add support for built-in functions
6838 vector bool char vec_revb (vector bool char);
6839 vector bool short vec_revb (vector short char);
6840 vector bool int vec_revb (vector bool int);
6841 vector bool long long vec_revb (vector bool long long);
6842 * doc/extend.texi: Update the built-in documentation file for the
6843 new built-in functions.
6845 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6847 * config/s390/s390.md: Remove movcc splitter.
6849 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6851 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
6852 load/store on condition.
6854 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
6857 * config/avr/avr.c (avr_encode_section_info)
6858 [progmem && !TREE_READONLY]: Error if progmem object needs
6861 2017-07-11 Michael Collison <michael.collison@arm.com>
6863 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
6866 2017-07-11 Carl Love <cel@us.ibm.com>
6868 * config/rs6000/rs6000-c.c: Add support for builtins
6869 vector unsigned int vec_parity_lsbb (vector signed int);
6870 vector unsigned int vec_parity_lsbb (vector unsigned int);
6871 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
6872 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
6873 vector unsigned long long vec_parity_lsbb (vector signed long long);
6874 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
6875 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
6876 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
6877 * doc/extend.texi: Update the built-in documentation file for the
6878 new built-in functions.
6880 2017-07-11 David Malcolm <dmalcolm@redhat.com>
6882 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
6883 (layout::m_primary_loc): New field.
6884 (layout::layout): Initialize new field. Move location filtering
6885 logic from here to...
6886 (layout::maybe_add_location_range): ...this new method. Add
6887 support for filtering to just the lines already specified by other
6889 (layout::will_show_line_p): New method.
6890 (gcc_rich_location::add_location_if_nearby): New method.
6891 (selftest::test_add_location_if_nearby): New test function.
6892 (selftest::diagnostic_show_locus_c_tests): Call it.
6893 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
6896 2017-07-11 Tom de Vries <tom@codesourcery.com>
6898 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
6899 (bb_first_real_insn): New function.
6900 (nvptx_single): Add extra initialization of broadcasted condition
6903 2017-07-11 Nathan Sidwell <nathan@acm.org>
6905 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
6907 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
6909 * doc/extend.texi (AVR Function Attributes): Remove weblink to
6910 Binutils doc as TEXI will mess them up.
6911 * doc/invoke.texi (AVR Options): Same here.
6913 2017-07-11 Daniel Cederman <cederman@gaisler.com>
6915 * config/sparc/sparc.opt (mfix-ut700): New option.
6916 (mfix-gr712rc): Likewise.
6917 (sparc_fix_b2bst): New variable.
6918 * doc/invoke.texi (SPARC options): Document them.
6919 (ARM options): Fix warnings.
6920 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
6921 instructions to prevent sequences that can trigger the store-store
6922 errata for certain LEON3FT processors.
6923 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
6924 (sparc_option_override): Set sparc_fix_b2bst appropriately.
6925 * config/sparc/sparc.md (fix_b2bst): New attribute.
6926 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
6928 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
6931 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
6933 (*rsqrtsf2_sse): Ditto.
6935 (div<mode>3): Macroize insn from divdf3 and divsf3
6936 using MODEF mode iterator.
6938 2017-07-10 Martin Sebor <msebor@redhat.com>
6940 PR tree-optimization/80397
6941 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
6942 instead of testing for equality to INTEGER_TYPE.
6944 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
6946 * config.gcc: Remove uclibc from arc target spec.
6948 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
6950 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
6952 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6955 * lto-wrapper.c (remove_option): New function.
6956 (merge_and_complain): Merge PIC/PIE options more realistically.
6958 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
6960 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
6964 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
6965 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
6966 * config.in: Regenerate.
6967 * configure: Regenerate.
6968 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
6969 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
6970 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
6971 (TARGET_GASISR_PROLOGUES): ...target mask.
6972 * common/config/avr/avr-common.c
6973 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
6974 Set -mgas-isr-prologues.
6975 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
6976 INSERT_PASS_BEFORE for it.
6977 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
6978 * config/avr/avr.c (avr_option_override)
6979 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
6980 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
6981 (avr_attribute_table) <no_gccisr>: Add new function attribute.
6982 (avr_set_current_function) <is_no_gccisr>: Init machine field.
6983 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
6985 (make_avr_pass_pre_proep): New function.
6986 (emit_push_sfr) <treg>: Add argument to function and use it
6988 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
6989 and set machine->gasisr.yes.
6990 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
6991 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
6992 __gcc_isr.n_pushed to .L__stack_usage.
6993 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
6994 (avr_asm_final_postscan_insn): ...this new static function.
6995 * config/avr/avr.h (machine_function)
6996 <is_no_gccisr, use_L__stack_usage>: New fields.
6997 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
6998 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
6999 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
7000 (gasisr, *gasisr): New expander and insn.
7001 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
7002 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
7003 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
7005 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
7007 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
7010 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
7012 Move jump-tables out of .text again.
7015 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
7016 (ASM_OUTPUT_ADDR_VEC): New function.
7017 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
7018 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
7019 INSN_ADDRESSes as asm comment.
7020 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
7021 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
7022 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
7023 * config/avr/avr.md (*tablejump): Adjust comment.
7024 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
7025 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
7027 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
7028 (avr_output_addr_vec): New proto.
7029 (avr_log_t) <insn_addresses>: New field.
7031 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
7034 * config/i386/i386.c (ix86_function_arg_advance): Set
7035 outgoing_args_on_stack to true if there are outgoing arguments
7037 (ix86_function_arg): Likewise.
7038 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
7039 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
7040 * config/i386/i386.h (machine_function): Add
7041 outgoing_args_on_stack.
7043 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
7045 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7047 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
7049 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
7051 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
7052 (REAL_H): Remove $(MACHMODE_H).
7053 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
7055 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
7056 $(MACHMODE_H) and double-int.h.
7057 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
7059 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
7062 2017-07-07 Andrew Pinski <apinski@cavium.com>
7064 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
7065 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
7067 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7069 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
7070 Add warning if GCC was not configured to link against a GLIBC that
7071 exports the hardware capability bits.
7072 (make_resolver_func): Make resolver function private and not a
7073 COMDAT function. Create the name with clone_function_name instead
7074 of make_unique_name.
7077 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
7078 correct operand in doing the split.
7080 2017-07-07 Carl Love <cel@us.ibm.com>
7082 * config/rs6000/rs6000-c: Add support for built-in function
7083 vector unsigned short vec_pack_to_short_fp32 (vector float,
7085 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
7086 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
7087 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
7088 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
7089 (convert_4f32_8i16): Add define_expand.
7090 * doc/extend.texi: Update the built-in documentation file for the
7091 new built-in function.
7093 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7095 * config/sparc/m8.md: New file.
7096 * config/sparc/sparc.md: Include m8.md.
7098 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7100 * config/sparc/sparc.opt: New option -mvis4b.
7101 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
7102 (sparc_option_override): Handle VIS4B.
7103 (enum sparc_builtins): Define
7104 SPARC_BUILTIN_DICTUNPACK{8,16,32},
7105 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
7106 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
7107 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
7108 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
7109 (check_constant_argument): New function.
7110 (sparc_vis_init_builtins): Define builtins
7111 __builtin_vis_dictunpack{8,16,32},
7112 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
7113 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
7114 __builtin_vis_fpcmpde{8,16,32}shl and
7115 __builtin_vis_fpcmpur{8,16,32}shl.
7116 (sparc_expand_builtin): Check that the constant operands to
7117 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
7118 constant and in range.
7119 * config/sparc/sparc-c.c (sparc_target_macros): Handle
7121 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
7122 (SPARC_IMM5_P): Likewise.
7123 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
7124 (enabled): Handle vis4b.
7125 (UNSPEC_DICTUNPACK): New unspec.
7126 (UNSPEC_FPCMPSHL): Likewise.
7127 (UNSPEC_FPUCMPSHL): Likewise.
7128 (UNSPEC_FPCMPDESHL): Likewise.
7129 (UNSPEC_FPCMPURSHL): Likewise.
7130 (cpu_feature): New CPU feature `vis4b'.
7131 (dictunpack{8,16,32}): New insns.
7132 (FPCSMODE): New mode iterator.
7133 (fpcscond): New code iterator.
7134 (fpcsucond): Likewise.
7135 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
7136 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
7137 (fpcmpde{8,16,32}{si,di}shl): Likewise.
7138 (fpcmpur{8,16,32}{si,di}shl): Likewise.
7139 * config/sparc/constraints.md: Define constraints `q' for unsigned
7140 2-bit integer constants and `t' for unsigned 5-bit integer
7142 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
7144 (imm5_operand_dictunpack16): Likewise.
7145 (imm5_operand_dictunpack32): Likewise.
7146 (imm2_operand): Likewise.
7147 * doc/invoke.texi (SPARC Options): Document -mvis4b.
7148 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
7149 ditunpack* and fpcmp*shl builtins.
7151 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7153 * config.gcc: Handle m8 in --with-{cpu,tune} options.
7154 * config.in: Add HAVE_AS_SPARC6 define.
7155 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
7157 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
7159 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
7160 (CPP_CPU_SPEC): Handle m8.
7161 (ASM_CPU_SPEC): Likewise.
7162 * config/sparc/sparc-opts.h (enum processor_type): Add
7164 * config/sparc/sparc.c (m8_costs): New struct.
7165 (sparc_option_override): Handle TARGET_CPU_m8.
7166 (sparc32_initialize_trampoline): Likewise.
7167 (sparc64_initialize_trampoline): Likewise.
7168 (sparc_issue_rate): Likewise.
7169 (sparc_register_move_cost): Likewise.
7170 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
7171 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
7172 (ASM_CPU64_DEFAULT_SPEC): Likewise.
7173 (CPP_CPU_SPEC): Handle M8.
7174 (ASM_CPU_SPEC): Likewise.
7175 (AS_M8_FLAG): Define.
7176 * config/sparc/sparc.md: Add m8 to the cpu attribute.
7177 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
7178 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
7180 * configure: Regenerate.
7181 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
7184 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7186 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
7188 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
7189 ("*movdi_insn_sp32"): Do not set v3pipe.
7190 ("*movsi_insn"): Likewise.
7191 ("*movdi_insn_sp64"): Likewise.
7192 ("*movsf_insn"): Likewise.
7193 ("*movdf_insn_sp32"): Likewise.
7194 ("*movdf_insn_sp64"): Likewise.
7195 ("*zero_extendsidi2_insn_sp64"): Likewise.
7196 ("*sign_extendsidi2_insn"): Likewise.
7197 ("*mov<VM32:mode>_insn"): Likewise.
7198 ("*mov<VM64:mode>_insn_sp64"): Likewise.
7199 ("*mov<VM64:mode>_insn_sp32"): Likewise.
7200 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
7201 ("<vlop:code><VL:mode>3"): Likewise.
7202 ("*not_<vlop:code><VL:mode>3"): Likewise.
7203 ("*nand<VL:mode>_vis"): Likewise.
7204 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
7205 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
7206 ("one_cmpl<VL:mode>2"): Likewise.
7207 ("faligndata<VM64:mode>_vis"): Likewise.
7208 ("alignaddrsi_vis"): Likewise.
7209 ("alignaddrdi_vis"): Likweise.
7210 ("alignaddrlsi_vis"): Likewise.
7211 ("alignaddrldi_vis"): Likewise.
7212 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
7213 ("bmaskdi_vis"): Likewise.
7214 ("bmasksi_vis"): Likewise.
7215 ("bshuffle<VM64:mode>_vis"): Likewise.
7216 ("cmask8<P:mode>_vis"): Likewise.
7217 ("cmask16<P:mode>_vis"): Likewise.
7218 ("cmask32<P:mode>_vis"): Likewise.
7219 ("pdistn<P:mode>_vis"): Likewise.
7220 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
7222 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7224 * config/sparc/sparc.md ("subtype"): New insn attribute.
7225 ("*wrgsr_sp64"): Set insn subtype.
7226 ("*rdgsr_sp64"): Likewise.
7227 ("alignaddrsi_vis"): Likewise.
7228 ("alignaddrdi_vis"): Likewise.
7229 ("alignaddrlsi_vis"): Likewise.
7230 ("alignaddrldi_vis"): Likewise.
7231 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
7232 ("fexpand_vis"): Likewise.
7233 ("fpmerge_vis"): Likewise.
7234 ("faligndata<VM64:mode>_vis"): Likewise.
7235 ("bshuffle<VM64:mode>_vis"): Likewise.
7236 ("cmask8<P:mode>_vis"): Likewise.
7237 ("cmask16<P:mode>_vis"): Likewise.
7238 ("cmask32<P:mode>_vis"): Likewise.
7239 ("fchksm16_vis"): Likewise.
7240 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
7241 ("fmean16_vis"): Likewise.
7242 ("fp<plusminus_insn>64_vis"): Likewise.
7243 ("<plusminus_insn>v8qi3"): Likewise.
7244 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
7245 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
7246 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
7247 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
7248 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
7249 ("*movqi_insn"): Likewise.
7250 ("*movhi_insn"): Likewise.
7251 ("*movsi_insn"): Likewise.
7252 ("movsi_pic_gotdata_op"): Likewise.
7253 ("*movdi_insn_sp32"): Likewise.
7254 ("*movdi_insn_sp64"): Likewise.
7255 ("movdi_pic_gotdata_op"): Likewise.
7256 ("*movsf_insn"): Likewise.
7257 ("*movdf_insn_sp32"): Likewise.
7258 ("*movdf_insn_sp64"): Likewise.
7259 ("*zero_extendhisi2_insn"): Likewise.
7260 ("*zero_extendqihi2_insn"): Likewise.
7261 ("*zero_extendqisi2_insn"): Likewise.
7262 ("*zero_extendqidi2_insn"): Likewise.
7263 ("*zero_extendhidi2_insn"): Likewise.
7264 ("*zero_extendsidi2_insn_sp64"): Likewise.
7265 ("ldfsr"): Likewise.
7266 ("prefetch_64"): Likewise.
7267 ("prefetch_32"): Likewise.
7268 ("tie_ld32"): Likewise.
7269 ("tie_ld64"): Likewise.
7270 ("*tldo_ldub_sp32"): Likewise.
7271 ("*tldo_ldub1_sp32"): Likewise.
7272 ("*tldo_ldub2_sp32"): Likewise.
7273 ("*tldo_ldub_sp64"): Likewise.
7274 ("*tldo_ldub1_sp64"): Likewise.
7275 ("*tldo_ldub2_sp64"): Likewise.
7276 ("*tldo_ldub3_sp64"): Likewise.
7277 ("*tldo_lduh_sp32"): Likewise.
7278 ("*tldo_lduh1_sp32"): Likewise.
7279 ("*tldo_lduh_sp64"): Likewise.
7280 ("*tldo_lduh1_sp64"): Likewise.
7281 ("*tldo_lduh2_sp64"): Likewise.
7282 ("*tldo_lduw_sp32"): Likewise.
7283 ("*tldo_lduw_sp64"): Likewise.
7284 ("*tldo_lduw1_sp64"): Likewise.
7285 ("*tldo_ldx_sp64"): Likewise.
7286 ("*mov<VM32:mode>_insn"): Likewise.
7287 ("*mov<VM64:mode>_insn_sp64"): Likewise.
7288 ("*mov<VM64:mode>_insn_sp32"): Likewise.
7290 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7292 * config/sparc/sparc.md ("type"): New insn type viscmp.
7293 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
7295 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
7296 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
7297 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
7298 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
7300 ("n7_vis_logical_11cycle"): Likewise.
7301 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
7302 * config/sparc/niagara2.md ("niag3_vis": Likewise.
7303 * config/sparc/niagara.md ("niag_vis"): Likewise.
7304 * config/sparc/ultra3.md ("us3_fga"): Likewise.
7305 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
7307 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7309 * config/sparc/sparc.md: New instruction type `bmask'.
7310 (bmaskdi_vis): Use the `bmask' type.
7311 (bmasksi_vis): Likewise.
7312 * config/sparc/ultra3.md (us3_array): Likewise.
7313 * config/sparc/niagara7.md (n7_array): Likewise.
7314 * config/sparc/niagara4.md (n4_array): Likewise.
7315 * config/sparc/niagara2.md (niag2_vis): Likewise.
7316 (niag3_vis): Likewise.
7317 * config/sparc/niagara.md (niag_vis): Likewise.
7319 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7321 * ipa-comdats.c: Remove optimize check from gate.
7322 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
7323 for functions not optimized.
7324 (ipa_fn_summary_read): Skip optimize check.
7325 (ipa_fn_summary_write): Likewise.
7326 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
7328 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
7330 (can_inline_edge_p): Check flag_pcc_struct_return for match.
7331 (check_callers): Give up on caller which is not optimized.
7332 (inline_small_functions): Likewise.
7333 (ipa_inline): Do not give up when not optimizing.
7334 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
7335 away unoptimizes cdtors.
7336 (whole_program_function_and_variable_visibility): Do
7337 ipa_discover_readonly_nonaddressable_vars in LTO mode.
7338 * ipa.c (process_references): Do not check optimize.
7339 (symbol_table::remove_unreachable_nodes): Update optimize check.
7340 (set_writeonly_bit): Update optimize check.
7341 (pass_ipa_cdtor_merge::gate): Do not check optimize.
7342 (pass_ipa_single_use::gate): Remove.
7344 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7346 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
7347 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
7348 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
7349 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
7350 permute_load, permute_store, adjust_extract, adjust_splat,
7351 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
7352 replace_swap_with_copy, dump_swap_insn_table,
7353 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
7354 recombine_lvx_pattern, recombine_stvx_pattern,
7355 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
7356 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
7357 to file rs6000-p8swap.c.
7358 * config/rs6000/rs6000-p8swap.c: New file.
7359 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
7360 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
7361 and rs6000*-*-* targets.
7363 2017-07-06 David Malcolm <dmalcolm@redhat.com>
7365 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
7367 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7369 * lto-wrapper.c (merge_and_complain): Do not merge
7370 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
7371 fsigned_zeros, ftrapping_math, fwrapv.
7372 (append_compiler_options): Do not track these options.
7373 (append_linker_options): Likewie
7375 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7377 * cgraphunit.c (cgraph_node::finalize_function): When
7378 !flag_toplevel_reorde set no_reorder flag.
7379 (varpool_node::finalize_decl): Likewise.
7380 (symbol_table::compile): Drop no toplevel reorder path.
7382 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7384 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
7385 edges; zero probability is not better than uninitialized.
7387 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
7389 * asan.h (asan_sanitize_allocas_p): Declare.
7390 * asan.c (asan_sanitize_allocas_p): New function.
7391 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
7392 (handle_builtin_alloca): Likewise.
7393 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
7394 if !asan_sanitize_allocas_p.
7395 * params.def (asan-instrument-allocas): Add new option.
7396 * params.h (ASAN_PROTECT_ALLOCAS): Define.
7397 * opts.c (common_handle_option): Disable allocas sanitization for
7400 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
7402 * asan.c: Include gimple-fold.h.
7403 (get_last_alloca_addr): New function.
7404 (handle_builtin_stackrestore): Likewise.
7405 (handle_builtin_alloca): Likewise.
7406 (asan_emit_allocas_unpoison): Likewise.
7407 (get_mem_refs_of_builtin_call): Add new parameter, remove const
7408 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
7409 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
7410 (instrument_builtin_call): Pass gimple iterator to
7411 get_mem_refs_of_builtin_call.
7412 (last_alloca_addr): New global.
7413 * asan.h (asan_emit_allocas_unpoison): Declare.
7414 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
7415 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
7416 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
7417 if function calls alloca.
7418 * gimple-fold.c (replace_call_with_value): Remove static keyword.
7419 * gimple-fold.h (replace_call_with_value): Declare.
7420 * internal-fn.c: Include asan.h.
7421 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
7422 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
7424 2017-07-06 David Malcolm <dmalcolm@redhat.com>
7426 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
7427 (C_SELFTEST_FLAGS): New.
7428 (CPP_SELFTEST_FLAGS): New.
7429 (SELFTEST_DEPS): New, from deps of s-selftest.
7430 (C_SELFTEST_DEPS): New, from deps of s-selftest.
7431 (CPP_SELFTEST_DEPS): New.
7432 (selftest): Add dependency on s-selftest-c++.
7433 (s-selftest): Rename to...
7434 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
7435 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
7436 than SELFTEST_FLAGS.
7437 (selftest-gdb): Rename to...
7438 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
7440 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
7441 (selftest-valgrind): Rename to...
7442 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
7444 (selftest-valgrind): Reintroduce as an alias for
7445 selftest-c-valgrind.
7446 (s-selftest-c++): New.
7447 (selftest-c++-gdb): New.
7448 (selftest-c++-valgrind): New.
7450 2017-07-06 Olivier Hainque <hainque@adacore.com>
7452 * gcc.c (process_command): When deciding if undefined variables
7453 should be ignored when processing specs, accept "gcc -v" as well.
7455 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
7457 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
7458 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
7460 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
7462 * config/arm/arm-cpus.in (armv8-r): Add new entry.
7463 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
7464 * config/arm/arm-tables.opt: Regenerate.
7465 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
7467 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
7469 2017-07-06 Carl Love <cel@us.ibm.com>
7471 * ChangeLog: Clean up from mid air collision
7473 2017-07-06 Carl Love <cel@us.ibm.com>
7475 * config/rs6000/rs6000-c.c: Add support for built-in functions
7476 vector signed int vec_subc (vector signed int, vector signed int);
7477 vector signed __int128 vec_subc (vector signed __int128,
7478 vector signed __int128);
7479 vector unsigned __int128 vec_subc (vector unsigned __int128,
7480 vector unsigned __int128);
7481 vector signed int vec_sube (vector signed int, vector signed int,
7483 vector unsigned int vec_sube (vector unsigned int,
7484 vector unsigned int,
7485 vector unsigned int);
7486 vector signed __int128 vec_sube (vector signed __int128,
7487 vector signed __int128,
7488 vector signed__int128);
7489 vector unsigned __int128 vec_sube (vector unsigned __int128,
7490 vector unsigned __int128,
7491 vector unsigned __int128);
7492 vector signed int vec_subec (vector signed int, vector signed int,
7494 vector unsigned int vec_subec (vector unsigned int,
7495 vector unsigned int,
7496 vector unsigned int);
7497 vector signed __int128 vec_subec (vector signed __int128,
7498 vector signed __int128,
7499 vector signed__int128);
7500 vector unsigned __int128 vec_subec (vector unsigned __int128,
7501 vector unsigned __int128,
7502 vector unsigned __int128);
7503 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
7504 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
7505 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
7506 BU_ALTIVEC_OVERLOAD_X definitions.
7507 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
7508 * doc/extend.texi: Update the built-in documentation file for the new
7511 2017-07-06 David Malcolm <dmalcolm@redhat.com>
7514 * diagnostic-show-locus.c (layout::layout): Use start and finish
7515 spelling location for the start and finish of each range.
7516 * genmatch.c (linemap_client_expand_location_to_spelling_point):
7517 Add unused aspect param.
7518 * input.c (expand_location_1): Add "aspect" param, and use it
7519 to access the correct part of the location.
7520 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
7522 (expand_location_to_spelling_point): Likewise.
7523 (linemap_client_expand_location_to_spelling_point): Add "aspect"
7524 param, and pass it to expand_location_1.
7526 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
7528 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
7529 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
7530 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
7531 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
7532 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
7533 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
7534 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
7535 _mm_maskz_getmant_ss): New intrinsics.
7536 (__builtin_ia32_getexpss128_mask): Changed to ...
7537 __builtin_ia32_getexpss128_round ... this.
7538 (__builtin_ia32_getexpsd128_mask): Changed to ...
7539 __builtin_ia32_getexpsd128_round ... this.
7540 * config/i386/i386-builtin-types.def
7541 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
7542 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
7543 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
7544 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
7545 __builtin_ia32_getmantss_mask_round): New builtins.
7546 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
7547 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
7548 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
7549 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
7550 * config/i386/sse.md
7551 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
7552 avx512f_sgetexp<mode><mask_scalar_name>
7553 <round_saeonly_scalar_name> ... this.
7554 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
7555 %0, %1, %2<round_saeonly_op3>}): Changed to ...
7556 vgetexp<ssescalarmodesuffix>
7557 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7558 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
7559 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
7560 avx512f_vgetmant<mode><mask_scalar_name>
7561 <round_saeonly_scalar_name> ... this.
7562 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
7563 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
7564 vgetmant<ssescalarmodesuffix>
7565 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
7566 %0<mask_scalar_operand4>, %1, %2
7567 <round_saeonly_scalar_mask_op4>, %3} ... this.
7568 * config/i386/subst.md (mask_scalar_operand4,
7569 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
7570 round_saeonly_scalar_nimm_predicate): New subst attributes.
7572 2017-07-06 Julia Koval <julia.koval@intel.com>
7574 * config/i386/i386.c (ix86_erase_embedded_rounding):
7575 Remove code for old rounding pattern.
7577 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
7579 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
7581 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
7583 * doc/sourcebuild.texi (Test Directives, Variants of
7584 dg-require-support): Add documentation for dg-require-stack-check.
7586 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
7588 * config/i386/subst.md (mask_scalar, round_scalar,
7589 round_saeonly_scalar): New meta-templates.
7590 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
7591 round_scalar_mask_operand3, round_scalar_mask_op3,
7592 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
7593 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
7594 round_saeonly_scalar_constraint,
7595 round_saeonly_scalar_prefix): New subst attribute.
7596 * config/i386/sse.md
7597 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
7598 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
7599 <round_scalar_name> ... this.
7600 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
7601 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
7602 <round_scalar_name> ... this.
7603 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
7604 <sse>_vm<code><mode>3<mask_scalar_name>
7605 <round_saeonly_scalar_name> ... this.
7606 (v<plusminus_mnemonic><ssescalarmodesuffix>
7607 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
7608 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
7609 v<plusminus_mnemonic><ssescalarmodesuffix>
7610 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7611 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
7612 (v<multdiv_mnemonic><ssescalarmodesuffix>
7613 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
7614 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
7615 v<multdiv_mnemonic><ssescalarmodesuffix>
7616 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7617 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
7618 (v<maxmin_float><ssescalarmodesuffix>
7619 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
7620 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
7621 v<maxmin_float><ssescalarmodesuffix>
7622 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7623 %0<mask_scalar_operand3>, %1, %<iptr>2
7624 <round_saeonly_scalar_mask_op3>} ... this.
7626 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
7628 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
7629 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
7631 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
7632 Alan Hayward <alan.hayward@arm.com>
7633 David Sherwood <david.sherwood@arm.com>
7635 * combine.c (simplify_if_then_else): Remove "enum" before
7637 * compare-elim.c (can_eliminate_compare): Likewise.
7638 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
7640 (aarch64_lookup_simd_builtin_type): Likewise.
7641 (aarch64_simd_builtin_type): Likewise.
7642 (aarch64_init_simd_builtin_types): Likewise.
7643 (aarch64_simd_expand_args): Likewise.
7644 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
7646 (aarch64_reverse_mask): Likewise.
7647 (aarch64_simd_emit_reg_reg_move): Likewise.
7648 (aarch64_gen_adjusted_ldpstp): Likewise.
7649 (aarch64_ccmp_mode_to_code): Likewise.
7650 (aarch64_operands_ok_for_ldpstp): Likewise.
7651 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7652 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
7654 (aarch64_min_divisions_for_recip_mul): Likewise.
7655 (aarch64_reassociation_width): Likewise.
7656 (aarch64_get_condition_code_1): Likewise.
7657 (aarch64_simd_emit_reg_reg_move): Likewise.
7658 (aarch64_simd_attr_length_rglist): Likewise.
7659 (aarch64_reverse_mask): Likewise.
7660 (aarch64_operands_ok_for_ldpstp): Likewise.
7661 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7662 (aarch64_gen_adjusted_ldpstp): Likewise.
7663 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
7665 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
7666 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
7667 (arm_lookup_simd_builtin_type): Likewise.
7668 (arm_simd_builtin_type): Likewise.
7669 (arm_init_simd_builtin_types): Likewise.
7670 (arm_expand_builtin_args): Likewise.
7671 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
7672 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
7673 (ft32_setup_incoming_varargs): Likewise.
7674 (ft32_function_arg): Likewise.
7675 (ft32_function_arg_advance): Likewise.
7676 (ft32_pass_by_reference): Likewise.
7677 (ft32_arg_partial_bytes): Likewise.
7678 (ft32_valid_pointer_mode): Likewise.
7679 (ft32_addr_space_pointer_mode): Likewise.
7680 (ft32_addr_space_legitimate_address_p): Likewise.
7681 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
7683 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
7684 (ix86_emit_outlined_ms2sysv_restore): Likewise.
7685 (iamcu_alignment): Likewise.
7686 (canonicalize_vector_int_perm): Likewise.
7687 (ix86_noce_conversion_profitable_p): Likewise.
7688 (ix86_mpx_bound_mode): Likewise.
7689 (ix86_operands_ok_for_move_multiple): Likewise.
7690 * config/microblaze/microblaze-protos.h
7691 (microblaze_expand_conditional_branch_reg): Likewise.
7692 * config/microblaze/microblaze.c
7693 (microblaze_expand_conditional_branch_reg): Likewise.
7694 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
7696 (rs6000_reassociation_width): Likewise.
7697 (rs6000_invalid_binary_op): Likewise.
7698 (fusion_p9_p): Likewise.
7699 (emit_fusion_p9_load): Likewise.
7700 (emit_fusion_p9_store): Likewise.
7701 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
7703 (riscv_hard_regno_mode_ok_p): Likewise.
7704 (riscv_address_insns): Likewise.
7705 (riscv_split_symbol): Likewise.
7706 (riscv_legitimize_move): Likewise.
7707 (riscv_function_value): Likewise.
7708 (riscv_hard_regno_nregs): Likewise.
7709 (riscv_expand_builtin): Likewise.
7710 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
7711 (riscv_build_integer): Likewise.
7712 (riscv_split_integer): Likewise.
7713 (riscv_legitimate_constant_p): Likewise.
7714 (riscv_cannot_force_const_mem): Likewise.
7715 (riscv_regno_mode_ok_for_base_p): Likewise.
7716 (riscv_valid_base_register_p): Likewise.
7717 (riscv_valid_offset_p): Likewise.
7718 (riscv_valid_lo_sum_p): Likewise.
7719 (riscv_classify_address): Likewise.
7720 (riscv_legitimate_address_p): Likewise.
7721 (riscv_address_insns): Likewise.
7722 (riscv_load_store_insns): Likewise.
7723 (riscv_force_binary): Likewise.
7724 (riscv_split_symbol): Likewise.
7725 (riscv_force_address): Likewise.
7726 (riscv_legitimize_address): Likewise.
7727 (riscv_move_integer): Likewise.
7728 (riscv_legitimize_const_move): Likewise.
7729 (riscv_legitimize_move): Likewise.
7730 (riscv_address_cost): Likewise.
7731 (riscv_subword): Likewise.
7732 (riscv_output_move): Likewise.
7733 (riscv_canonicalize_int_order_test): Likewise.
7734 (riscv_emit_int_order_test): Likewise.
7735 (riscv_function_arg_boundary): Likewise.
7736 (riscv_pass_mode_in_fpr_p): Likewise.
7737 (riscv_pass_fpr_single): Likewise.
7738 (riscv_pass_fpr_pair): Likewise.
7739 (riscv_get_arg_info): Likewise.
7740 (riscv_function_arg): Likewise.
7741 (riscv_function_arg_advance): Likewise.
7742 (riscv_arg_partial_bytes): Likewise.
7743 (riscv_function_value): Likewise.
7744 (riscv_pass_by_reference): Likewise.
7745 (riscv_setup_incoming_varargs): Likewise.
7746 (riscv_print_operand): Likewise.
7747 (riscv_elf_select_rtx_section): Likewise.
7748 (riscv_save_restore_reg): Likewise.
7749 (riscv_for_each_saved_reg): Likewise.
7750 (riscv_register_move_cost): Likewise.
7751 (riscv_hard_regno_mode_ok_p): Likewise.
7752 (riscv_hard_regno_nregs): Likewise.
7753 (riscv_class_max_nregs): Likewise.
7754 (riscv_memory_move_cost): Likewise.
7755 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
7756 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
7757 (rl78_addr_space_address_mode): Likewise.
7758 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7760 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7761 (rs6000_reassociation_width): Likewise.
7762 (rs6000_invalid_binary_op): Likewise.
7763 (fusion_p9_p): Likewise.
7764 (emit_fusion_p9_load): Likewise.
7765 (emit_fusion_p9_store): Likewise.
7766 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
7767 (ok_for_simple_move_operands): Likewise.
7768 (ok_for_simple_move_strict_operands): Likewise.
7769 (ok_for_simple_arith_logic_operands): Likewise.
7770 (visium_legitimize_reload_address): Likewise.
7771 (visium_select_cc_mode): Likewise.
7772 (output_cbranch): Likewise.
7773 (visium_split_double_move): Likewise.
7774 (visium_expand_copysign): Likewise.
7775 (visium_expand_int_cstore): Likewise.
7776 (visium_expand_fp_cstore): Likewise.
7777 * config/visium/visium.c (visium_pass_by_reference): Likewise.
7778 (visium_function_arg): Likewise.
7779 (visium_function_arg_advance): Likewise.
7780 (visium_libcall_value): Likewise.
7781 (visium_setup_incoming_varargs): Likewise.
7782 (visium_legitimate_constant_p): Likewise.
7783 (visium_legitimate_address_p): Likewise.
7784 (visium_legitimize_address): Likewise.
7785 (visium_secondary_reload): Likewise.
7786 (visium_register_move_cost): Likewise.
7787 (visium_memory_move_cost): Likewise.
7788 (prepare_move_operands): Likewise.
7789 (ok_for_simple_move_operands): Likewise.
7790 (ok_for_simple_move_strict_operands): Likewise.
7791 (ok_for_simple_arith_logic_operands): Likewise.
7792 (visium_function_value_1): Likewise.
7793 (rtx_ok_for_offset_p): Likewise.
7794 (visium_legitimize_reload_address): Likewise.
7795 (visium_split_double_move): Likewise.
7796 (visium_expand_copysign): Likewise.
7797 (visium_expand_int_cstore): Likewise.
7798 (visium_expand_fp_cstore): Likewise.
7799 (visium_split_cstore): Likewise.
7800 (visium_select_cc_mode): Likewise.
7801 (visium_split_cbranch): Likewise.
7802 (output_cbranch): Likewise.
7803 (visium_print_operand_address): Likewise.
7804 * expmed.c (flip_storage_order): Likewise.
7805 * expmed.h (emit_cstore): Likewise.
7806 (flip_storage_order): Likewise.
7807 * genrecog.c (validate_pattern): Likewise.
7808 * hsa-gen.c (gen_hsa_addr): Likewise.
7809 * internal-fn.c (expand_arith_overflow): Likewise.
7810 * ira-color.c (allocno_copy_cost_saving): Likewise.
7811 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7812 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
7813 (process_invariant_for_inheritance): Likewise.
7814 * lra-eliminations.c (move_plus_up): Likewise.
7815 * omp-low.c (lower_oacc_reductions): Likewise.
7816 * simplify-rtx.c (simplify_subreg): Likewise.
7817 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
7818 (TARGET_CHKP_BOUND_MODE): Likewise..
7819 * targhooks.c (default_chkp_bound_mode): Likewise.
7820 (default_setup_incoming_vararg_bounds): Likewise.
7821 * targhooks.h (default_chkp_bound_mode): Likewise.
7822 (default_setup_incoming_vararg_bounds): Likewise.
7823 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
7824 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
7825 (have_whole_vector_shift): Likewise.
7826 * tree-vect-stmts.c (vectorizable_load): Likewise.
7827 * doc/tm.texi: Regenerate.
7829 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
7831 Graceful degrade if Binutils PR21472 is not available.
7834 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
7835 .rodata in flash test fails.
7836 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
7837 * confgure: Regenerate.
7838 * config.in: Regenerate.
7839 * config/avr/avr.c (avr_asm_named_section)
7840 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
7841 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
7842 (avr_asm_init_sections): Same.
7844 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7846 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
7847 (fma<VH:mode>4_intrinsic): Likewise.
7848 (*fmsub<VCVTF:mode>4): Likewise.
7849 (*fmsub<VH:mode>4_intrinsic): Likewise.
7851 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
7854 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
7855 Don't depend on "optimize > 0".
7856 (out_movhi_r_mr, out_movqi_mr_r): Same.
7857 (out_movhi_mr_r, out_movqi_r_mr): Same.
7858 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
7859 io_address_operand on "optimize > 0".
7861 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7863 * tree-loop-distribution.c: Add general explanantion on the pass.
7864 (generate_loops_for_partition): Mark distributed loop.
7865 (pg_add_dependence_edges): New parameter. Handle alias data
7866 dependence specially and record it in the parameter if asked.
7867 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
7868 (init_partition_graph_vertices, add_partition_graph_edge): New.
7869 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
7870 (free_partition_graph_vdata, build_partition_graph): New.
7871 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
7872 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
7873 (data_ref_segment_size, latch_dominated_by_data_ref): New.
7874 (compute_alias_check_pairs, version_loop_by_alias_check): New.
7875 (version_for_distribution_p, finalize_partitions): New.
7876 (distribute_loop): Handle alias data dependence specially. Factor
7877 out loop fusion code as functions and call these functions.
7879 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7881 * tree-loop-distribution.c (classify_partition): New parameter and
7882 better handle reduction statement.
7883 (rdg_build_partitions): Revise comment.
7884 (distribute_loop): Compute statements in all partitions and pass it
7885 to classify_partition.
7887 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7889 * tree-loop-distribution.c (enum partition_type): New.
7890 (struct partition): New field type.
7891 (partition_merge_into): Add parameter. Update partition type.
7892 (data_dep_in_cycle_p, update_type_for_merge): New functions.
7893 (build_rdg_partition_for_vertex): Compute partition type.
7894 (rdg_build_partitions): Dump partition type.
7895 (distribute_loop): Update calls to partition_merge_into.
7897 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7899 * tree-loop-distribution.c (struct ddr_hasher): New.
7900 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
7902 (classify_partition): Call get_data_dependence.
7903 (pg_add_dependence_edges): Ditto.
7904 (distribute_loop): Release data dependence hash table.
7906 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7908 * tree-loop-distribution.c (ref_base_address): Delete.
7909 (similar_memory_accesses): Rename ...
7910 (share_memory_accesses): ... to this. Check if partitions access
7911 the same memory reference.
7912 (distribute_loop): Call share_memory_accesses.
7914 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7916 * tree-loop-distribution.c (struct partition): New field recording
7918 (partition_alloc, partition_free): Init and release data refs.
7919 (partition_merge_into): Merge data refs.
7920 (build_rdg_partition_for_vertex): Collect data refs for partition.
7921 (pg_add_dependence_edges): Change parameters from vector to bitmap.
7923 (distribute_loop): Remve data refs from vertice data of partition
7926 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7928 * tree-loop-distribution.c (params.h): Include header file.
7929 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
7930 (datarefs_vec): New global var.
7931 (create_rdg_vertices): Use datarefs_vec directly.
7932 (free_rdg): Don't free data references.
7933 (build_rdg): Update use. Don't free data references.
7934 (distribute_loop): Compute global variable for data references.
7935 Bail out if there are too many data references.
7937 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7939 * tree-loop-distribution.c (loop_nest): New global var.
7940 (build_rdg): Use loop directly, rather than loop nest.
7941 (pg_add_dependence_edges): Remove loop nest parameter. Use global
7943 (distribute_loop): Compute global variable loop nest. Update use.
7945 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7947 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
7948 (partition_merge_into): New parameter. Dump reason for fusion.
7949 (distribute_loop): Update use of partition_merge_into.
7951 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7953 * tree-loop-distribution.c (bb_top_order_index): New.
7954 (bb_top_order_index_size, bb_top_order_cmp): New.
7955 (stmts_from_loop): Use topological order.
7956 (pass_loop_distribution::execute): Compute and release topological
7957 order for basic blocks.
7959 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7961 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
7964 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7966 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
7967 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
7968 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
7969 * internal-fn.def (LOOP_DIST_ALIAS): New.
7970 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
7971 (fold_loop_internal_call): ... this.
7972 (vect_loop_dist_alias_call): New function.
7973 (set_uid_loop_bbs): Call fold_loop_internal_call.
7974 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
7977 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7980 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
7981 Require dead FLAGS_REG at the beginning of a peephole.
7983 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7986 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
7987 arguments in the call to __builtin_ia32_sbb_u32.
7988 (_subborrow_u64): Swap _X and _Y arguments in the call to
7989 __builtin_ia32_sbb_u64.
7991 2017-07-04 Jakub Jelinek <jakub@redhat.com>
7994 * tree-vrp.c (compare_assert_loc): Turn into a function template
7995 with stable template parameter. Only test if a->e is NULL,
7996 !a->e == !b->e has been verified already. Use e == NULL or
7997 e != NULL instead of e or ! e tests. If stable is true, don't use
7998 iterative_hash_expr, on the other side allow a or b or both NULL
7999 and sort the NULLs last.
8000 (process_assert_insertions): Sort using compare_assert_loc<false>
8001 instead of compare_assert_loc, later sort using
8002 compare_assert_loc<true> before calling process_assert_insertions_for
8003 in a loop. Use break instead of continue once seen NULL pointer.
8005 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
8007 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
8008 Cortex-R7 and Cortex-R8 processors.
8010 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8012 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
8013 uninitialized while src is not.
8015 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
8017 * common/config/arm/arm-common.c: Adjust include path for
8019 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
8020 (arm-cpu.h): Create in build directory. Adjust dependency rules.
8021 (arm-cpu-data.h): Likewise.
8022 (arm-cpu-cdata.h): Likewise.
8023 * config/arm/arm-cpu.h: Delete.
8024 * config/arm/arm-cpu-cdata.h: Delete.
8025 * config/arm/arm-cpu-data.h: Delete.
8027 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
8029 * config/arm/arm-cpus.in (cortex-a55): New.
8030 (cortex-a75): Likewise.
8031 (cortex-a75.cortex-a55): Likewise.
8032 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
8034 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
8035 * config/arm/arm-cpu-cdata.h: Regenerate.
8036 * config/arm/arm-cpu-data.h: Regenerate.
8037 * config/arm/arm-cpu.h: Regenerate.
8038 * config/arm/arm-tables.opt: Regenerate.
8039 * config/arm/arm-tune.md: Regenerate.
8041 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8043 * haifa-sched.c (sched_create_recovery_edges): Update profile.
8045 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8047 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
8050 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
8052 PR tree-optimization/81292
8053 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
8054 full_string_p, also call adjust_related_strinfos if the adjustment
8055 is simple, otherwise invalidate related strinfos.
8057 2017-07-04 Martin Liska <mliska@suse.cz>
8060 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
8061 newly created variable as DECL_IGNORED_P.
8063 2017-07-04 Martin Liska <mliska@suse.cz>
8066 * ipa-inline.c (inline_small_functions):
8067 Use xstrdup_for_dump.
8069 2017-07-04 Tom de Vries <tom@codesourcery.com>
8071 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
8073 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
8076 * config/darwin.c (darwin_function_switched_text_sections):
8079 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
8081 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
8083 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
8085 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
8087 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8089 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
8090 min_profitable_iters, and th as inclusive lower bounds.
8091 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
8092 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
8093 for min_profitable_iters and min_profitable_estimate.
8094 (vect_transform_loop): Treat th as an inclusive lower bound.
8095 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
8097 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
8100 * config/darwin.c (darwin_function_switched_text_sections):
8101 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
8102 in two pieces, and suppress the use of buf.
8104 2017-07-03 Nathan Sidwell <nathan@acm.org>
8106 * hash-table.h (hash_table_mod1): Fix indentation.
8108 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8111 * predict.c (force_edge_cold): Be more careful about propagation
8113 * profile-count.h (profile_probability::guessed,
8114 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
8116 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
8118 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
8120 * doc/invoke.texi (rcpc architecture extension): Document it.
8122 2017-07-03 Richard Biener <rguenther@suse.de>
8124 PR tree-optimization/60510
8125 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
8126 the scalar reduction PHI and use it.
8127 (vectorizable_reduction): Properly guard the single_defuse_cycle
8128 path for non-SLP reduction chains where we cannot use it.
8129 Rework reduc_def/index and vector type deduction. Rework
8130 vector operand gathering during reduction op code-gen.
8131 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
8132 chains dissolve the chain and leave it to non-SLP reduction
8135 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8137 * tree-data-ref.h (dr_alignment): Declare.
8138 * tree-data-ref.c (dr_alignment): New function.
8139 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
8140 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
8142 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
8144 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8146 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
8147 and base_misalignment fields.
8148 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
8149 * tree-data-ref.c: Include builtins.h.
8150 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
8151 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
8152 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
8153 * tree-vect-data-refs.c: Include tree-cfg.h.
8154 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
8155 fields instead of calculating an alignment here.
8156 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
8157 innermost_loop_behavior fields.
8159 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8161 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
8163 (DR_STEP_ALIGNMENT): New macro.
8164 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
8165 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
8166 (create_data_ref): Print it.
8167 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
8168 to tell whether the step preserves vector (mis)alignment.
8169 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
8170 Move the check for an integer step and generalise to all INTEGER_CST.
8171 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
8172 Print the outer step alignment.
8174 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8176 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
8177 with offset_alignment.
8178 (DR_ALIGNED_TO): Delete.
8179 (DR_OFFSET_ALIGNMENT): New macro.
8180 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
8181 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
8182 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
8183 (create_data_ref): Likewise.
8184 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
8185 (vect_analyze_data_refs): Likewise.
8186 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
8187 creating dummy innermost behavior.
8189 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8191 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
8192 with a "innermost_loop_behavior *" and refeence tree.
8193 * tree-data-ref.c (dr_analyze_innermost): Likewise.
8194 (create_data_ref): Update call accordingly.
8195 * tree-predcom.c (find_looparound_phi): Likewise.
8197 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8199 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
8200 fields with dr_wrt_vec_loop.
8201 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
8202 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
8203 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
8204 (vect_dr_behavior): New function.
8205 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
8206 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
8207 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
8208 track whether the step preserves the misalignment.
8209 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
8210 Use vect_dr_behavior.
8211 (vect_setup_realignment): Update call accordingly.
8212 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
8213 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
8214 call to vect_create_addr_base_for_vector_ref.
8215 (vect_create_cond_for_align_checks): Likewise.
8216 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
8217 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
8218 (vect_recog_mask_conversion_pattern): Likewise.
8219 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
8220 (new_stmt_vec_info): Remove redundant zeroing.
8222 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
8224 * common/config/arm/arm-common.c (arm_be8_option): New function.
8225 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
8226 (ISA_ARMv6): Add isa_bit_be8.
8227 * config/arm/arm.h (arm_be8_option): Add prototype.
8228 (BE8_SPEC_FUNCTION): New define.
8229 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
8230 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
8231 (mlittle-endian): Similarly.
8232 (mbe8, mbe32): New options.
8233 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
8234 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
8236 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8238 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
8240 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8242 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
8243 (cleanup_tree_cfg_bb): Use it.
8244 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
8246 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
8248 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8251 * loop-doloop.c (add_test): Update profile.
8253 2017-07-03 Martin Liska <mliska@suse.cz>
8256 * sanopt.c (rewrite_usage_of_param): New function.
8257 (sanitize_rewrite_addressable_params): Likewise.
8258 (pass_sanopt::execute): Call rewrite_usage_of_param.
8260 2017-07-03 Richard Biener <rguenther@suse.de>
8262 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
8263 back to using VIEW_CONVERT_EXPR.
8265 2017-07-03 Martin Liska <mliska@suse.cz>
8268 * doc/extend.texi: Document when a resolver function is
8269 generated for target_clones.
8271 2017-07-03 Martin Liska <mliska@suse.cz>
8273 * asan.c (asan_emit_stack_protection): Unpoison just red zones
8274 and shadow memory of auto variables which are subject of
8275 use-after-scope sanitization.
8276 (asan_expand_mark_ifn): Add do set only when is_poison.
8278 2016-07-03 Richard Biener <rguenther@suse.de>
8280 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
8282 (vect_force_simple_reduction): Record reduction def -> phi mapping.
8283 (vectorizable_reduction): Perform reduction PHI creation when
8284 visiting a reduction PHI and adjust and simplify code generation
8285 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
8286 (vect_transform_loop): Visit reduction PHIs.
8287 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
8288 defs into the SLP tree.
8289 (vect_build_slp_tree): Reduction defs terminate the recursion.
8290 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
8292 (vect_get_vec_defs_for_stmt_copy): Export.
8293 (vect_get_vec_defs): Likewise.
8294 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
8296 (vect_get_vec_defs_for_stmt_copy): Declare.
8297 (vect_get_vec_defs): Likewise.
8299 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8301 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
8302 parameter with a "loop" parameter and use it instead of the
8303 loop containing DR_STMT. Don't check simple_iv when doing
8304 BB analysis. Describe the two analysis modes in the comment.
8306 2017-07-03 Tom de Vries <tom@codesourcery.com>
8308 PR tree-optimization/69468
8309 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
8310 (find_same_succ_bb): Handle ignore_edge_flags.
8312 2017-07-03 Tom de Vries <tom@codesourcery.com>
8314 PR tree-optimization/81192
8315 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
8317 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
8319 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
8321 2017-07-03 Tom de Vries <tom@codesourcery.com>
8323 PR tree-optimization/81192
8324 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
8325 BB_SAME_SUCC (bb) == NULL.
8327 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8329 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
8332 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8334 * dumpfile.c: Include profile-count.h
8335 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
8337 (insert_cond_bb): Update profile.
8338 * tree-cfg.h (insert_cond_bb): Update prototype.
8339 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
8340 * tree-dump.c: Do not include tree-cfg.
8342 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8344 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
8346 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8348 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
8351 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8353 * tree-complex.c (expand_complex_div_wide): update profile.
8355 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8356 Alan Hayward <alan.hayward@arm.com>
8357 David Sherwood <david.sherwood@arm.com>
8359 * Makefile.in (MACHMODE_H): Remove insn-modes.h
8360 (CORETYPES_H): New define.
8361 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
8362 (insn-modes-inline.h, s-modes-inline-h): New rules.
8363 (generated_files): Add insn-modes-inline.h.
8364 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
8365 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
8366 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
8367 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
8368 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
8369 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
8370 (build/gencodes.o, build/genconditions.o): Likewise.
8371 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
8372 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
8373 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
8374 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
8375 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
8376 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
8377 * coretypes.h: Include everything up to real.h for generators.
8378 Include insn-modes.h first. Include wide-int-print.h after
8379 wide-int.h. Include insn-modes-inline.h and then machmode.h.
8380 * machmode.h: Don't include insn-modes.h here.
8381 * function-tests.c: Remove includes of signop.h, machmode.h,
8382 double-int.h and wide-int.h.
8384 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
8386 * optc-save-gen.awk: Likewise.
8387 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
8388 * godump.c: Remove include of wide-int-print.h.
8389 * pretty-print.h: Likewise.
8390 * wide-int-print.cc: Likewise.
8391 * wide-int.cc: Likewise.
8392 * hash-map-tests.c: Remove include of signop.h.
8393 * hash-set-tests.c: Likewise.
8394 * rtl-tests.c: Likewise.
8395 * mkconfig.sh: Remove include of machmode.h.
8396 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
8398 (emit_insn_modes_inline_h): ...this new function. Emit the code
8399 into an insn-modes-inline.h header file, adding appropriate
8400 include guards and end comments.
8401 (emit_insn_modes_c_header): Remove include of machmode.h.
8402 (emit_min_insn_modes_c_header): Include coretypes.h rather than
8404 (main): Handle -i flag and call emit_insn_modes_inline_h when
8407 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8409 * tree-ssa-strlen.c (strinfo): Rename the length field to
8410 nonzero_chars. Add a full_string_p field.
8411 (compare_nonzero_chars, zero_length_string_p): New functions.
8412 (get_addr_stridx): Add an offset_out parameter.
8413 Use compare_nonzero_chars.
8414 (get_stridx): Update accordingly. Use compare_nonzero_chars.
8415 (new_strinfo): Update after above changes to strinfo.
8416 (set_endptr_and_length): Set full_string_p.
8417 (get_string_length): Update after above changes to strinfo.
8418 (unshare_strinfo): Update call to new_strinfo.
8419 (maybe_invalidate): Likewise.
8420 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
8421 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
8422 as a uhwi instead of an shwi. Update after above changes to
8423 strinfo and new_strinfo.
8424 (zero_length_string): Assert that chainsi contains full strings.
8425 Use zero_length_string_p. Update call to new_strinfo.
8426 (adjust_related_strinfos): Update after above changes to strinfo.
8427 Copy full_string_p from origsi.
8428 (adjust_last_stmt): Use zero_length_string_p.
8429 (handle_builtin_strlen): Update after above changes to strinfo and
8430 new_strinfo. Install the lhs as the string length if the previous
8431 entry didn't describe a full string.
8432 (handle_builtin_strchr): Update after above changes to strinfo
8434 (handle_builtin_strcpy): Likewise.
8435 (handle_builtin_strcat): Likewise.
8436 (handle_builtin_malloc): Likewise.
8437 (handle_pointer_plus): Likewise.
8438 (handle_builtin_memcpy): Likewise. Track nonzero characters
8439 that aren't necessarily followed by a nul terminator.
8440 (handle_char_store): Likewise.
8442 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8444 PR tree-optimization/80769
8445 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
8446 for malloc and calloc. Document the new invariant that all related
8447 strinfos have delayed lengths or none do.
8448 (verify_related_strinfos): Move earlier in file.
8449 (set_endptr_and_length): New function, split out from...
8450 (get_string_length): ...here. Also set the lengths of related
8452 (zero_length_string): Assert that chainsi has known (rather than
8454 (adjust_related_strinfos): Likewise.
8456 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
8458 PR tree-optimization/81136
8459 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
8460 assert that two references with the same misalignment have the same
8461 compile-time misalignment if those compile-time misalignments
8464 2017-07-01 Andi Kleen <ak@linux.intel.com>
8466 * print-tree.c (print_node): Print all attributes.
8468 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8470 * cfg.c (scale_bbs_frequencies): New function.
8471 * cfg.h (scale_bbs_frequencies): Declare it.
8472 * cfgloopanal.c (single_likely_exit): Cleanup.
8473 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
8475 (scale_loop_profile): Likewise.
8476 (loop_version): Likewise.
8477 (create_empty_loop_on_edge): Update.
8478 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
8479 scale_loop_frequencies, scale_loop_profile, loopify,
8480 loop_version): Update prototypes.
8481 * modulo-sched.c (sms_schedule): Update.
8482 * predict.c (unlikely_executed_edge_p): Also check probability.
8483 (probably_never_executed_edge_p): Fix typo.
8484 * tree-if-conv.c (version_loop_for_if_conversion): Update.
8485 * tree-parloops.c (gen_parallel_loop): Update.
8486 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
8487 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8488 * tree-ssa-loop-split.c (split_loop): Update.
8489 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
8490 * tree-vect-loop-manip.c (vect_do_peeling): Update.
8491 (vect_loop_versioning): Update.
8492 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8494 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8496 * trans-mem.c (split_bb_make_tm_edge): Update profile.
8498 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8500 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
8501 to keep profile consistent.
8503 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8505 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
8506 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
8507 * profile-count.h (max_safe_multiplier): Make unsigned.
8508 (profile_count::guessed_zero): New.
8510 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8512 * bb-reorder.c (fix_up_crossing_landing_pad,
8513 fix_crossing_conditional_branches): Use make_single_succ_edge
8514 to keep profile consistent.
8516 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8518 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
8521 2017-07-01 Jakub Jelinek <jakub@redhat.com>
8524 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
8525 the right scopes, make sure cond_jump isn't preserved between multiple
8526 iterations. Search for fallthru edge whenever there are 3+ edges and
8527 use find_fallthru_edge for it.
8529 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8531 Patch by Alexander Monakov <amonakov@ispras.ru>
8532 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
8533 probabilities consistently.
8535 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8537 * pa.c (pa_expand_compare_and_swap_loop): Update call of
8538 emit_cmp_and_jump_insns.
8540 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8543 * tree-inline.c (expand_call_inline): Combine profile statuses.
8545 2017-06-30 Andrew Pinski <apinski@cavium.com>
8547 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
8548 fold_stmt returned true.
8550 2017-06-30 Nathan Sidwell <nathan@acm.org>
8552 * ggc.h (empty_string): Delete.
8553 * cfgexpand.c (expand_asm_stmt): Use plain "".
8554 * optabs.c (expand_asm_memory_barrier): Likewise.
8555 * stringpool.c (empty_string): Delete.
8556 (digit_vector, digit_string): Delete.
8557 (ggc_alloc_string): Use plain "", don't optimize single digit
8558 strings. Use ggc_alloc_atomic.
8560 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
8562 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
8563 comparison set and one other set, use the cost of the non-comparison
8566 2017-06-30 Nathan Sidwell <nathan@acm.org>
8568 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
8571 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
8573 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
8574 loops. Remove now unneeded calls to gimple_switch_set_label() that
8575 just set removed labels to NULL_TREE.
8577 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
8579 * tree-ssanames.c (set_range_info_raw): Abstract from ...
8580 (set_range_info): ...here. Only call set_range_info_raw if domain
8582 (set_nonzero_bits): Call set_range_info_raw.
8583 * tree-ssanames.h (set_range_info_raw): New.
8585 2017-06-30 Jakub Jelinek <jakub@redhat.com>
8588 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
8589 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
8590 of nonimmediate_operand and <store_mask_constraint> instead of m
8591 for the input operand. For V8FI iterator, always split if input
8592 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
8593 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
8594 <store_mask_predicate> instead of register_operand and
8595 <store_mask_constraint> instead of v for the input operand. Make
8596 sure both operands aren't MEMs for if not <mask_applied>.
8598 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
8600 * lto-wrapper.c (copy_file) Close both file descriptors before
8603 2017-06-30 Martin Liska <mliska@suse.cz>
8606 * multiple_target.c (create_dispatcher_calls): Make ifunc
8607 also for function that don't have calls or are not referenced.
8609 2017-06-30 Richard Biener <rguenther@suse.de>
8611 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
8612 analyze the first scalar stmt. Move vector type computation
8613 for the BB case here from ...
8614 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
8615 live operation processing in the SLP case properly.
8617 2017-06-30 Richard Biener <rguenther@suse.de>
8619 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
8621 2017-06-30 Martin Liska <mliska@suse.cz>
8624 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
8625 before BUILT_IN_UNWIND_RESUME when ASAN is used.
8627 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
8629 * doc/invoke.texi (AArch64): Add missing options and remove redundant
8632 2017-06-30 Richard Biener <rguenther@suse.de>
8634 PR tree-optimization/81249
8635 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
8636 condition reduction result to original scalar type.
8638 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8640 * profile-count.h (enum profile_quality): Fix typos and whitespace
8643 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8645 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
8646 type for branch probabilities.
8648 2017-06-29 Julian Brown <julian@codesourcery.com>
8649 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8651 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
8652 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
8653 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
8654 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
8656 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8658 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
8659 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
8660 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
8661 CC usage from generic code to here.
8662 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
8663 CC usage into the target macros.
8665 2017-06-29 Maya Rashish <coypu@sdf.org>
8667 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
8670 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8672 * arm/arm-builtins.c: Include profile-count.h
8673 * except.c (sjlj_emit_function_enter): Use
8674 profile_probability::unlikely.
8676 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8678 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
8679 and tocrel_offset be pointer args rather than implicitly using
8681 (legitimate_constant_pool_address_p, rs6000_emit_move,
8682 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
8683 tocrel_offset and use in toc_relative_expr_p call.
8684 (print_operand, print_operand_address): Use static tocrel_base_oac
8685 and tocrel_offset_oac.
8686 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
8689 2017-06-29 Maya Rashish <coypu@sdf.org>
8691 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
8693 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
8695 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
8696 objects, take into account only the alignment of 'op0' and 'mode1' if
8699 2017-06-29 Steve Ellcey <sellcey@cavium.com>
8701 * ccmp.c (ccmp_tree_comparison_p): New function.
8702 (ccmp_candidate_p): Update to use above function.
8703 (get_compare_parts): New function.
8704 (expand_ccmp_next): Update to use new functions.
8705 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
8707 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
8708 take mode as argument.
8709 * ccmp.h (expand_ccmp_expr): Add mode as argument.
8710 * expr.c (expand_expr_real_1): Pass mode as argument.
8712 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
8714 * combine.c (combine_instructions): Print insns to dump_file, together
8717 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8719 * asan.c (asan_emit_stack_protection): Update.
8720 (create_cond_insert_point): Update.
8721 * auto-profile.c (afdo_propagate_circuit): Update.
8722 * basic-block.h (struct edge_def): Turn probability to
8723 profile_probability.
8724 (EDGE_FREQUENCY): Update.
8725 * bb-reorder.c (find_traces_1_round): Update.
8726 (better_edge_p): Update.
8727 (sanitize_hot_paths): Update.
8728 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
8729 (make_single_succ_edge): Update.
8730 (check_bb_profile): Update.
8731 (dump_edge_info): Update.
8732 (update_bb_profile_for_threading): Update.
8733 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
8734 probabilitycount to 0.
8735 * cfgbuild.c (compute_outgoing_frequencies): Update.
8736 * cfgcleanup.c (try_forward_edges): Update.
8737 (outgoing_edges_match): Update.
8738 (try_crossjump_to_edge): Update.
8739 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
8740 (expand_gimple_tailcall): Update.
8741 (construct_init_block): Use make_single_succ_edge.
8742 (construct_exit_block): Use make_single_succ_edge.
8743 * cfghooks.c (verify_flow_info): Update.
8744 (redirect_edge_succ_nodup): Update.
8745 (split_edge): Update.
8746 (account_profile_record): Update.
8747 * cfgloopanal.c (single_likely_exit): Update.
8748 * cfgloopmanip.c (scale_loop_profile): Update.
8749 (set_zero_probability): Remove.
8750 (duplicate_loop_to_header_edge): Update.
8751 * cfgloopmanip.h (loop_version): Update prototype.
8752 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8753 (force_nonfallthru_and_redirect): Update.
8754 (update_br_prob_note): Update.
8755 (rtl_verify_edges): Update.
8756 (purge_dead_edges): Update.
8757 (rtl_lv_add_condition_to_bb): Update.
8758 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
8759 * cgraphunit.c (init_lowered_empty_function): Update.
8760 (cgraph_node::expand_thunk): Update.
8761 * cilk-common.c: Include profile-count.h
8762 * dojump.c (inv): Remove.
8763 (jumpifnot): Update.
8764 (jumpifnot_1): Update.
8765 (do_jump_1): Update.
8767 (do_jump_by_parts_greater_rtx): Update.
8768 (do_compare_rtx_and_jump): Update.
8769 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
8770 do_jump_1. do_compare_rtx_and_jump): Update prototype.
8771 * dwarf2cfi.c: Include profile-count.h
8772 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
8773 (sjlj_emit_dispatch_table): Likewise.
8774 * explow.c: Include profile-count.h
8775 * expmed.c (emit_store_flag_force): Update.
8776 (do_cmp_and_jump): Update.
8777 * expr.c (compare_by_pieces_d::generate): Update.
8778 (compare_by_pieces_d::finish_mode): Update.
8779 (emit_block_move_via_loop): Update.
8780 (store_expr_with_bounds): Update.
8781 (store_constructor): Update.
8782 (expand_expr_real_2): Update.
8783 (expand_expr_real_1): Update.
8784 * expr.h (try_casesi, try_tablejump): Update prototypes.
8785 * gimple-pretty-print.c (dump_probability): Update.
8786 (dump_profile): New.
8787 (dump_gimple_label): Update.
8788 (dump_gimple_bb_header): Update.
8789 * graph.c (draw_cfg_node_succ_edges): Update.
8790 * hsa-gen.c (convert_switch_statements): Update.
8791 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
8792 (find_if_case_1): Update.
8793 (find_if_case_2): Update.
8794 * internal-fn.c (expand_arith_overflow_result_store): Update.
8795 (expand_addsub_overflow): Update.
8796 (expand_neg_overflow): Update.
8797 (expand_mul_overflow): Update.
8798 (expand_vector_ubsan_overflow): Update.
8799 * ipa-cp.c (good_cloning_opportunity_p): Update.
8800 * ipa-split.c (split_function): Use make_single_succ_edge.
8801 * ipa-utils.c (ipa_merge_profiles): Update.
8802 * loop-doloop.c (add_test): Update.
8803 (doloop_modify): Update.
8804 * loop-unroll.c (compare_and_jump_seq): Update.
8805 (unroll_loop_runtime_iterations): Update.
8806 * lra-constraints.c (lra_inheritance): Update.
8807 * lto-streamer-in.c (input_cfg): Update.
8808 * lto-streamer-out.c (output_cfg): Update.
8809 * mcf.c (adjust_cfg_counts): Update.
8810 * modulo-sched.c (sms_schedule): Update.
8811 * omp-expand.c (expand_omp_for_init_counts): Update.
8812 (extract_omp_for_update_vars): Update.
8813 (expand_omp_ordered_sink): Update.
8814 (expand_omp_for_ordered_loops): Update.
8815 (expand_omp_for_generic): Update.
8816 (expand_omp_for_static_nochunk): Update.
8817 (expand_omp_for_static_chunk): Update.
8818 (expand_cilk_for): Update.
8819 (expand_omp_simd): Update.
8820 (expand_omp_taskloop_for_outer): Update.
8821 (expand_omp_taskloop_for_inner): Update.
8822 * omp-simd-clone.c (simd_clone_adjust): Update.
8823 * optabs.c (expand_doubleword_shift): Update.
8824 (expand_abs): Update.
8825 (emit_cmp_and_jump_insn_1): Update.
8826 (expand_compare_and_swap_loop): Update.
8827 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
8828 * predict.c (predictable_edge_p): Update.
8829 (edge_probability_reliable_p): Update.
8830 (set_even_probabilities): Update.
8831 (combine_predictions_for_insn): Update.
8832 (combine_predictions_for_bb): Update.
8833 (propagate_freq): Update.
8834 (estimate_bb_frequencies): Update.
8835 (force_edge_cold): Update.
8836 * profile-count.c (profile_count::dump): Add missing space into dump.
8837 (profile_count::debug): Add newline.
8838 (profile_count::differs_from_p): Explicitly convert to unsigned.
8839 (profile_count::stream_in): Update.
8840 (profile_probability::dump): New member function.
8841 (profile_probability::debug): New member function.
8842 (profile_probability::differs_from_p): New member function.
8843 (profile_probability::differs_lot_from_p): New member function.
8844 (profile_probability::stream_in): New member function.
8845 (profile_probability::stream_out): New member function.
8846 * profile-count.h (profile_count_quality): Rename to ...
8847 (profile_quality): ... this one.
8848 (profile_probability): New.
8849 (profile_count): Update.
8850 * profile.c (compute_branch_probabilities): Update.
8851 * recog.c (peep2_attempt): Update.
8852 * sched-ebb.c (schedule_ebbs): Update.
8853 * sched-rgn.c (find_single_block_region): Update.
8854 (compute_dom_prob_ps): Update.
8855 (schedule_region): Update.
8856 * sel-sched-ir.c (compute_succs_info): Update.
8857 * stmt.c (struct case_node): Update.
8858 (do_jump_if_equal): Update.
8859 (get_outgoing_edge_probs): Update.
8860 (conditional_probability): Update.
8861 (emit_case_dispatch_table): Update.
8862 (expand_case): Update.
8863 (expand_sjlj_dispatch_table): Update.
8864 (emit_case_nodes): Update.
8865 * targhooks.c: Update.
8866 * tracer.c (better_p): Update.
8867 (find_best_successor): Update.
8868 * trans-mem.c (expand_transaction): Update.
8869 * tree-call-cdce.c: Update.
8870 * tree-cfg.c (gimple_split_edge): Upate.
8871 (move_sese_region_to_fn): Upate.
8872 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
8873 * tree-eh.c (lower_resx): Upate.
8874 (cleanup_empty_eh_move_lp): Upate.
8875 * tree-if-conv.c (version_loop_for_if_conversion): Update.
8876 * tree-inline.c (copy_edges_for_bb): Update.
8877 (copy_cfg_body): Update.
8878 * tree-parloops.c (gen_parallel_loop): Update.
8879 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
8880 (gimple_gen_time_profiler): Update.
8881 * tree-ssa-dce.c (remove_dead_stmt): Update.
8882 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8883 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8884 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8885 (unloop_loops): Update.
8886 (try_peel_loop): Update.
8887 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8888 * tree-ssa-loop-split.c (connect_loops): Update.
8889 (split_loop): Update.
8890 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
8891 (hoist_guard): Update.
8892 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8893 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8894 (value_replacement): Update.
8895 * tree-ssa-reassoc.c (branch_fixup): Update.
8896 * tree-ssa-tail-merge.c (replace_block_by): Update.
8897 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8898 (create_edge_and_update_destination_phis): Update.
8899 (compute_path_counts): Update.
8900 (recompute_probabilities): Update.
8901 (update_joiner_offpath_counts): Update.
8902 (freqs_to_counts_path): Update.
8903 (duplicate_thread_path): Update.
8904 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8905 (struct switch_conv_info): Update.
8906 (gen_inbound_check): Update.
8907 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8908 (vect_do_peeling): Update.
8909 (vect_loop_versioning): Update.
8910 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8911 (optimize_mask_stores): Update.
8912 * ubsan.c (ubsan_expand_null_ifn): Update.
8913 * value-prof.c (gimple_divmod_fixed_value): Update.
8914 (gimple_divmod_fixed_value_transform): Update.
8915 (gimple_mod_pow2): Update.
8916 (gimple_mod_pow2_value_transform): Update.
8917 (gimple_mod_subtract): Update.
8918 (gimple_mod_subtract_transform): Update.
8919 (gimple_ic): Update.
8920 (gimple_stringop_fixed_value): Update.
8921 (gimple_stringops_transform): Update.
8922 * value-prof.h: Update.
8924 2017-06-29 Carl Love <cel@us.ibm.com>
8926 * config/rs6000/rs6000-c.c: Add support for built-in functions
8927 vector signed int vec_signed (vector float);
8928 vector signed long long vec_signed (vector double);
8929 vector signed int vec_signed2 (vector double, vector double);
8930 vector signed int vec_signede (vector double);
8931 vector signed int vec_signedo (vector double);
8932 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
8933 instruction generator.
8934 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8935 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
8936 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
8938 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
8939 vunsignede_v2df): Add define_expands.
8940 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
8941 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
8942 VEC_UNSIGNEDO): Add definitions.
8943 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8944 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
8945 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
8946 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
8947 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
8948 * config/rs6000/altivec.h (vec_signed, vec_signed2,
8949 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
8950 vec_unsignede, vec_unsignedo): Add builtin defines.
8951 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
8953 * doc/extend.texi: Update the built-in documentation file for the
8954 new built-in functions.
8956 2017-06-29 Richard Biener <rguenther@suse.de>
8958 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
8959 reduction chains to LOOP_VINFO_REDUCTIONS.
8960 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
8961 SLP reductions after processing reduction chains.
8963 2017-06-29 Nathan Sidwell <nathan@acm.org>
8965 * builtins.c (fold_builtin_FUNCTION): Use
8966 lang_hooks.decl_printable_name.
8968 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
8971 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
8972 with only one label.
8973 * stmt.c (expand_case): Assert NCASES is greater than one.
8975 2017-06-29 Richard Biener <rguenther@suse.de>
8977 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
8979 (group_case_labels): Likewise.
8980 (find_taken_edge): Push sanity checking on val to workers...
8981 (find_taken_edge_cond_expr): ... here
8982 (find_taken_edge_switch_expr): ... and here, handle cases
8983 with just a default label.
8984 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
8985 (group_case_labels): Likewise.
8986 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
8987 group_case_labels does anything cleanup the CFG again.
8989 2017-06-29 Bin Cheng <bin.cheng@arm.com>
8991 PR tree-optimization/81196
8992 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
8993 exit condition comparing two IVs.
8995 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
8997 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
8998 profile to the dummy entry at the end of the list of architectures.
8999 * config/arm/arm-cpu-cdata.h: Regenerated.
9001 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9002 Michael Collison <michael.collison@arm.com>
9005 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
9007 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
9008 (*aarch64_reg_<mode>3_minus_mask): New pattern.
9009 (*aarch64_<optab>_reg_di3_mask2): New pattern.
9010 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
9011 of shift when the shift amount is masked with constant equal to
9012 the size of the mode.
9013 * config/aarch64/predicates.md (subreg_lowpart_operator): New
9016 2017-06-29 Martin Liska <mliska@suse.cz>
9018 * config/i386/i386.opt: Change range from [1,5] to [0,5].
9020 2017-06-29 Yury Gribov <tetra2005@gmail.com>
9023 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
9026 (edge_growth_cache_entry::edge_growth_cache_entry): New
9028 (reset_edge_growth_cache): Update to use constructor.
9030 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9032 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
9033 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
9034 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
9036 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
9038 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
9039 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
9041 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
9043 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
9044 (*-linux-uclibc*): Add t-uclibc tmake_file.
9045 * config/t-musl: New.
9046 * config/t-uclibc: New.
9048 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
9050 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
9052 (gen_comm_data): Emit architectural setting of arch_prof.
9053 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
9055 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
9056 (armv8-m.base, armv8-m.main): Likewise.
9057 * arm-protos.h (arm_build_target): Add profile field.
9058 (arch_option): Likewise.
9059 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
9061 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
9062 arm_active_target.profile.
9064 2017-06-28 Richard Biener <rguenther@suse.de>
9067 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
9068 TYPE_OVERFLOW_WRAPS.
9069 * match.pd (negate_expr_p): Likewise.
9070 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
9071 fold_build2, not fold_binary.
9073 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
9075 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
9076 Convert memory address to Pmode.
9077 (aarch64_print_operand): Assert MEM operands are always Pmode.
9079 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
9082 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
9083 Remove redundant if.
9084 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
9085 * config/arm/aarch-common-protos.h
9086 (aarch_forward_to_shift_is_not_shifted_re): Remove.
9087 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
9089 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
9092 * multiple_target.c (create_dispatcher_calls): Set the default
9093 clone to be static, not public.
9095 2017-06-28 Richard Biener <rguenther@suse.de>
9097 * tree-vect-loop.c (vectorizable_reduction): Move special
9098 cond reduction IV var creation ...
9099 (vect_create_epilog_for_reduction): ... here. Remove induction_index
9100 parameter. Use STMT_VINFO_VECTYPE.
9101 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
9104 2017-06-28 Martin Liska <mliska@suse.cz>
9107 * ipa-visibility.c (non_local_p): Handle visibility.
9109 2017-06-28 Martin Liska <mliska@suse.cz>
9112 * common.opt: Add IntegerRange to various options.
9113 * opt-functions.awk (integer_range_info): New function.
9114 * optc-gen.awk: Add integer_range_info to cl_options struct.
9115 * opts-common.c (decode_cmdline_option): Handle
9116 CL_ERR_INT_RANGE_ARG.
9117 (cmdline_handle_error): Likewise.
9118 * opts.c (print_filtered_help): Show valid interval in
9119 when --help is provided.
9120 * opts.h (struct cl_option): Add range_min and range_max fields.
9121 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
9123 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
9125 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
9126 (x * C EQ/NE y * C): New transformation.
9128 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
9130 * genmultilib (combination_space): Accept '+' in option names.
9132 2017-06-28 Martin Liska <mliska@suse.cz>
9135 * asan.c (instrument_derefs): Bail out inner references
9136 that are hard register variables.
9138 2017-06-28 Jakub Jelinek <jakub@redhat.com>
9141 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
9142 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
9144 2017-06-28 Richard Biener <rguenther@suse.de>
9146 * tree-vectorizer.h (vect_get_vec_defs): Remove.
9147 (vect_get_slp_defs): Adjust.
9148 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
9150 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
9152 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9153 get_initial_defs_for_reduction instead of vect_get_vec_defs.
9154 (vectorizable_reduction): Adjust.
9155 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
9157 (vect_get_slp_defs): Likewise.
9158 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
9159 (vectorizable_bswap): Adjust.
9160 (vectorizable_call): Likewise.
9161 (vectorizable_conversion): Likewise.
9162 (vectorizable_assignment): Likewise.
9163 (vectorizable_shift): Likewise.
9164 (vectorizable_operation): Likewise.
9165 (vectorizable_store): Likewise.
9166 (vectorizable_condition): Likewise.
9167 (vectorizable_comparison): Likewise.
9169 2017-06-28 Michael Collison <michael.collison@arm.com>
9172 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
9174 (arm_gen_movmemqi): Removed unused variable 'i'.
9175 Convert 'for' loop into 'while' loop.
9176 (arm_expand_prologue): Remove last unnecessary set of insn.
9177 (thumb_pop): Remove unused variable 'pushed_words'.
9178 (thumb_exit): Remove last unnecessary set of regs_to_pop.
9180 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9182 * config/s390/predicates.md: Use s390_rel_address_ok_p.
9183 * config/s390/s390-protos.h: Add prototype of
9184 s390_rel_address_ok_p.
9185 * config/s390/s390.c (s390_got_symbol): New function.
9186 (s390_rel_address_ok_p): New function.
9187 (legitimize_pic_address): Use s390_rel_address_ok_p.
9188 (s390_load_got): Use s390_got_symbol.
9189 (s390_option_override): Issue error if
9190 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
9191 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
9193 * config/s390/s390.opt: New option mpic-data-is-text-relative.
9195 2017-06-27 Andrew Pinski <apinski@cavium.com>
9197 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
9198 (X * copysign (1.0, X)): New pattern.
9199 (X * copysign (1.0, -X)): New pattern.
9200 (copysign (-1.0, CST)): New pattern.
9202 2017-06-27 Joseph Myers <joseph@codesourcery.com>
9204 * genmultilib (combination_space): Remove variable.
9205 Validate reuse rules against regular expression for any sequence
9206 of multilib options in any order.
9208 2017-06-27 Michael Collison <michael.collison@arm.com>
9210 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
9211 call aarch64_split_simd_combine.
9212 * (aarch64_combine_internal<mode>): Delete pattern.
9213 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
9214 Allow register and subreg operands.
9216 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9218 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
9219 specific need, just fallback on defaults.
9220 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
9222 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9223 Olivier Hainque <hainque@adacore.com>
9225 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
9227 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
9228 targets. Pick a default if no particular attempt applied.
9229 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
9232 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9234 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
9235 (x86_64-wrs-vxworks7): Likewise.
9237 2017-06-27 Marek Polacek <polacek@redhat.com>
9240 * ubsan.c (instrument_null): Check get_base_address's result for null.
9242 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
9244 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
9246 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
9248 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
9249 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
9250 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
9251 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
9253 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
9254 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
9255 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
9256 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
9257 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
9258 BUILT_IN_FEUPDATEENV): New builtins.
9259 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
9260 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
9261 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
9262 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
9264 (builtin_structptr_types): Adjust size.
9265 * tree.c (builtin_structptr_types): Add four entries.
9267 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9268 Olivier Hainque <hainque@adacore.com>
9270 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
9271 (TLS_SYM): New local macro, forcing reference to __tls__ on
9272 link command lines for VxWorks 7 RTPs, triggering initialization
9274 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
9275 OS features TLS support, true for RTPs on VxWorks 7.
9276 * config/vxworks.c (vxworks_override_options): Setup emutls
9279 2017-06-27 Jakub Jelinek <jakub@redhat.com>
9281 * predict.c (test_prediction_value_range): Use -1U instead of -1
9282 to avoid narrowing conversion warning.
9283 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
9284 to avoid narrowing conversion warning.
9285 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
9287 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
9289 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
9291 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
9292 64bit configurations.
9293 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
9294 (SIZE_TYPE): Likewise.
9295 * config/vxworks.c (vxworks_emutls_var_fields): Use
9296 long_unsigned_type_node instead of unsigned_type_node as the offset
9297 field type, which is "pointer" mode in emutls.c.
9299 2017-06-27 Jakub Jelinek <jakub@redhat.com>
9302 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
9305 * gimple-fold.c (replace_call_with_call_and_fold): Handle
9306 gimple_vuse copying separately from gimple_vdef copying.
9308 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9310 * value-prof.c (free_hist): Remove call to memset and the enclosing if
9313 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
9314 Olivier Hainque <hainque@adacore.com>
9316 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
9317 for all vxworks7 targets.
9318 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
9319 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
9320 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
9321 variations for VX6/VX7 and 32/64bits later on in ...
9322 (VXWORKS_LIB_SPEC): Leverage new macros.
9323 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
9324 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
9326 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
9328 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
9329 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
9331 2017-06-26 Carl Love <cel@us.ibm.com>
9333 * config/rs6000/rs6000-c.c: Add support for built-in functions
9334 vector bool char vec_reve (vector bool char);
9335 vector signed char vec_reve (vector signed char);
9336 vector unsigned char vec_reve (vector unsigned char);
9337 vector bool int vec_reve (vector bool int);
9338 vector signed int vec_reve (vector signed int);
9339 vector unsigned int vec_reve (vector unsigned int);
9340 vector bool long long vec_reve (vector bool long long);
9341 vector signed long long vec_reve (vector signed long long);
9342 vector unsigned long long vec_reve (vector unsigned long long);
9343 vector bool short vec_reve (vector bool short);
9344 vector signed short vec_reve (vector signed short);
9345 vector double vec_reve (vector double);
9346 vector float vec_reve (vector float);
9347 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
9348 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
9349 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
9350 (altivec_vreve): New pattern.
9351 * config/rs6000/altivec.h (vec_reve): New define.
9352 * doc/extend.texi (vec_rev): Update the built-in documentation file
9353 for the new built-in functions.
9355 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9357 PR tree-optimization/71815
9358 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
9360 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
9362 (slsr_process_phi): Likewise.
9363 (replace_uncond_cands_and_profitable_phis): Don't replace a
9364 multiply candidate with a stride of 1 (copy or cast).
9365 (phi_incr_cost): Call uses_consumed_by_stmt rather than
9367 (lowest_cost_path): Likewise.
9368 (total_savings): Likewise.
9370 2017-06-26 Richard Biener <rguenther@suse.de>
9373 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
9374 Use def_builtin_pure for all gather builtins.
9376 2017-06-26 Richard Biener <rguenther@suse.de>
9378 PR tree-optimization/81203
9379 * tree-tailcall.c (find_tail_calls): Do not move stmts into
9382 2017-06-26 Marek Polacek <polacek@redhat.com>
9385 * doc/invoke.texi: Document -Wmultistatement-macros.
9387 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
9389 * doc/sourcebuild.texi (ARM-specific attributes): Document new
9390 arm_neon_ok_no_float_abi effective target.
9392 2017-06-26 Richard Biener <rguenther@suse.de>
9394 PR tree-optimization/80928
9395 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
9396 (copy_bbs): Set BB_DUPLICATED flag early.
9397 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
9399 (execute_on_shrinking_pred): Likewise.
9400 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
9401 BB_DUPLICATED blocks.
9402 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
9403 iterate over all PHIs considering removal of *gsi.
9405 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
9407 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
9410 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9412 * config/rs6000/rs6000-string.c: (expand_block_clear,
9413 do_load_for_compare, select_block_compare_mode,
9414 compute_current_alignment, expand_block_compare,
9415 expand_strncmp_align_check, expand_strn_compare,
9416 expand_block_move, rs6000_output_load_multiple)
9417 Move functions related to string/block move/compare
9419 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
9420 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
9421 for this function which is now used in two files.
9422 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
9423 * config.gcc: Add rs6000-string.o to extra_objs for
9424 targets powerpc*-*-* and rs6000*-*-*.
9426 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
9429 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
9430 32-bit, since indexed is not valid for DImode.
9431 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
9432 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
9433 (define_peephole2 for Altivec d-form load): Add 32-bit support.
9434 (define_peephole2 for Altivec d-form store): Likewise.
9437 * multiple_target.c (create_dispatcher_calls): Only create the
9438 dispatcher call if the function is the default clone of a
9441 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
9444 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
9445 a call, force the call to not be a tail call.
9447 2017-06-23 Jeff Law <law@redhat.com>
9449 * doc/contrib.texi: Add entry for Steven Pemberton's work on
9452 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
9454 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
9455 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
9456 handling for early expansion of vector shifts (sl,sr,sra,rl).
9457 (builtin_function_type): Add vector shift right instructions
9458 to the unsigned argument list.
9460 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9462 rtl-optimizatoin/79286
9463 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
9464 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
9465 trap. PIC register plus a const unspec without offset can never trap.
9467 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
9469 * tree.h (builtin_structptr_type): New type.
9470 (builtin_structptr_types): Declare new array.
9471 * tree.c (builtin_structptr_types): New array.
9472 (free_lang_data, build_common_tree_nodes): Use it.
9474 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
9477 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
9480 2017-06-22 Matt Turner <mattst88@gmail.com>
9482 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
9483 Lake models to skylake case. Assume skylake for unknown
9484 models with clflushopt.
9486 2017-06-22 Jeff Law <law@redhat.com>
9488 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
9489 frame sizes that do not satisfy aarch64_uimm12_shift.
9491 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
9493 * profile-count.h (apply_probability,
9494 apply_scale, probability_in): Fix checks for zero.
9496 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9498 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
9499 * doc/cppdiropts.texi (-I @var{dir}): Document it.
9501 2016-06-22 Richard Biener <rguenther@suse.de>
9503 * tree-vect-loop.c (vect_model_reduction_cost): Handle
9504 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
9505 REDUC_MAX_EXPR support.
9506 (vectorizable_reduction): Likewise.
9507 (vect_create_epilog_for_reduction): Likewise.
9509 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
9511 * match.pd (A / (1 << B) -> A >> B): New.
9512 * generic-match-head.c: Include optabs-tree.h.
9513 * gimple-match-head.c: Likewise.
9514 * optabs-tree.h (target_supports_op_p): New.
9515 * optabs-tree.c (target_supports_op_p): New.
9517 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9519 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
9520 $gcc_cv_ld --help output.
9521 (gcc_cv_ld_demangle): Likewise.
9522 (gcc_cv_ld_eh_frame_hdr): Likewise.
9523 (gcc_cv_ld_pie): Likewise.
9524 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
9525 (gcc_cv_ld_buildid): Likewise.
9526 (gcc_cv_ld_sysroot): Likewise.
9527 (ld_bndplt_support): Likewise.
9528 (ld_pushpopstate_support): Likewise.
9529 * configure: Regenerate.
9530 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
9532 2017-06-21 Jakub Jelinek <jakub@redhat.com>
9535 * config/i386/sse.md (round<mode>2): Renumber match_dup and
9536 operands indexes to avoid gap between operands and match_dups.
9538 2017-06-21 Andrew Pinski <apinski@cavium.com>
9540 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
9541 Increment Arith_shift and Arith_shift_reg by 1.
9542 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
9544 * config/aarch64/aarch64.c (thunderx_tunings): Enable
9545 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
9546 (aarch64_strip_extend): Add new argument and test for it.
9547 (aarch64_cheap_mult_shift_p): New function.
9548 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
9549 add a cost if it is true.
9550 Update calls to aarch64_strip_extend.
9551 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
9553 2017-06-21 Andrew Pinski <apinski@cavium.com>
9555 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
9557 (thunderxt88): Likewise.
9558 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
9559 (thunderx_prefetch_tune): New variable.
9560 (thunderx2t99_prefetch_tune): Update for the correct values.
9561 (thunderxt88_tunings): New variable.
9562 (thunderx_tunings): Use thunderx_prefetch_tune instead of
9563 generic_prefetch_tune.
9564 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
9566 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9568 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
9569 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
9570 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
9571 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
9572 (aarch64_atomic_cas<mode>, GPI): Likewise.
9574 2017-06-21 Martin Liska <mliska@suse.cz>
9576 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
9577 statements on cold and hot labels.
9578 * predict.c (tree_estimate_probability_bb): Remove the
9579 prediction from this place.
9581 2017-06-21 Martin Liska <mliska@suse.cz>
9583 PR tree-optimization/79489
9584 * gimplify.c (maybe_add_early_return_predict_stmt): New
9586 (gimplify_return_expr): Call the function.
9587 * predict.c (tree_estimate_probability_bb): Remove handling
9589 * predict.def: Update comment about early return predictor.
9590 * gimple-predict.h (is_gimple_predict): New function.
9591 * predict.def: Change default value of early return to 66.
9592 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
9594 * passes.def: Put pass_strip_predict_hints to the beginning of
9597 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
9599 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
9600 FUNCTION_DECL declarations.
9601 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
9603 (dwaf2out_decl): Likewise.
9604 * godump.c (go_early_global_decl): Skip call to the real debug hook
9605 for FUNCTION_DECL declarations.
9606 * passes.c (rest_of_decl_compilation): Skip call to the
9607 early_global_decl debug hook for FUNCTION_DECL declarations, unless
9608 -fdump-go-spec is passed.
9610 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
9612 * config/i386/i386.c (struct builtin_isa): New field pure_p.
9613 Reorder for compactness.
9614 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
9615 (def_builtin_pure, def_builtin_pure2): New functions.
9616 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
9618 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
9620 * match.pd (nop_convert): New predicate.
9621 ((A +- CST1) +- CST2): Allow some NOP conversions.
9623 2017-06-21 Jakub Jelinek <jakub@redhat.com>
9626 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
9627 with ctors/dtors if GOVD_SHARED is set.
9629 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9631 * config/aarch64/aarch64.md (movti_aarch64):
9632 Emit mov rather than orr.
9633 (movtf_aarch64): Likewise.
9634 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
9635 Emit mov rather than orr.
9637 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9639 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
9640 Swap alternatives, make integer dup more expensive.
9642 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9644 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9645 Return true for non-tls symbols.
9647 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
9649 * config/aarch64/aarch64-cores.def (cortex-a55): New.
9650 (cortex-a75): Likewise.
9651 (cortex-a75.cortex-a55): Likewise.
9652 * config/aarch64/aarch64-tune.md: Regenerate.
9653 * doc/invoke.texi (-mtune): Document new values for -mtune.
9655 2017-06-21 Tom de Vries <tom@codesourcery.com>
9657 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
9659 (Effective-Target Keywords, Other attributes): Suggest using
9660 dg-add-options stack_size feature to get stack limit in stack_size
9661 effective target documentation.
9663 2017-06-21 Julian Brown <julian@codesourcery.com>
9664 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9666 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
9667 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
9668 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
9670 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
9671 attribute type list for neon_multiply.
9672 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
9673 attribute type list for neon_multiply.
9674 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
9675 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
9676 attribute type list for neon_multiply.
9677 * config/arm/types.md (crypto_pmull): Add.
9678 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
9679 attribute type list.
9681 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
9683 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
9686 2017-06-20 Jakub Jelinek <jakub@redhat.com>
9688 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
9689 to make sure not to dereference a NULL cost_classes_ptr pointer.
9691 2017-06-20 Carl Love <cel@us.ibm.com>
9693 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9694 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9695 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9696 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9697 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9698 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9699 VMULOSW): New enum "unspec" values.
9700 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
9701 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
9702 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9703 altivec_vmulosw): New patterns.
9704 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9705 VMULOSW): Add definitions.
9707 2017-06-20 Julia Koval <julia.koval@intel.com>
9709 * config/i386/i386.c: Fix rounding expand for new pattern.
9710 * config/i386/subst.md: Fix pattern (parallel -> unspec).
9712 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9714 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
9715 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
9717 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9719 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
9722 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9724 * config/aarch64/aarch64-cores.def: Rearrange to sort by
9725 architecture, then by implementer ID.
9726 * config/aarch64/aarch64-tune.md: Regenerate.
9728 2017-06-20 Richard Biener <rguenther@suse.de>
9731 * fold-const.c (split_tree): Fold to type before negating.
9733 2017-06-20 David Malcolm <dmalcolm@redhat.com>
9735 * diagnostic-show-locus.c
9736 (selftest::test_fixit_deletion_affecting_newline): New function.
9737 (selftest::diagnostic_show_locus_c_tests): Call it.
9739 2017-06-20 Andreas Schwab <schwab@suse.de>
9742 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
9745 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9747 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
9748 __ARM_FEATURE_COPROC according to support.
9750 2017-06-20 Jakub Jelinek <jakub@redhat.com>
9752 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
9753 Rewritten to avoid overflow for > 32-bit pointers.
9756 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
9757 by removing enum keyword.
9758 (ubsan_type_descriptor): Likewise. Formatting fix.
9761 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
9762 splitter): Require TARGET_SSE2 in the condition.
9764 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
9767 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
9768 for doing vector set of SFmode on ISA 3.0.
9769 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
9770 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
9772 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
9773 SFmode value into a V4SF variable that was extracted from another
9774 V4SF variable without converting the element to double precision
9775 and back to single precision vector format.
9776 (vsx_insert_extract_v4sf_p9_2): Likewise.
9778 2017-06-19 Jakub Jelinek <jakub@redhat.com>
9780 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
9781 in UWHI to avoid undefined overflow.
9784 * ubsan.h (enum ubsan_encode_value_phase): New.
9785 (ubsan_encode_value): Change second argument to
9786 enum ubsan_encode_value_phase with default value of
9787 UBSAN_ENCODE_VALUE_GENERIC.
9788 * ubsan.c (ubsan_encode_value): Change second argument to
9789 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
9790 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
9791 create_tmp_var_raw instead of create_tmp_var and use a
9793 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
9794 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
9795 ubsan_encode_value callers.
9798 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
9799 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
9800 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
9802 2017-06-19 Richard Biener <rguenther@suse.de>
9805 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
9806 estimates if we changed anything.
9808 2017-06-19 Richard Biener <rguenther@suse.de>
9810 PR tree-optimization/80887
9811 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
9812 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
9813 simplified lookups, then reset mprts_hook.
9814 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
9816 (try_to_simplify): Likewise.
9818 2017-06-19 Martin Liska <mliska@suse.cz>
9821 * gimplify.c (gimplify_switch_expr):
9822 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
9824 2017-06-19 Martin Liska <mliska@suse.cz>
9826 * doc/install.texi: Document that PGO runs in 4 stages.
9828 2017-06-19 Martin Liska <mliska@suse.cz>
9831 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
9832 to dispatcher function name.
9833 * multiple_target.c (replace_function_decl): New function.
9834 (create_dispatcher_calls): Redirect both edges and references.
9836 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
9838 * profile-count.c (profile_count::dump): Dump quality.
9839 (profile_count::differs_from_p): Update for unsigned val.
9840 * profile-count.h (profile_count_quality): New enum.
9841 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
9843 2017-06-19 Richard Biener <rguenther@suse.de>
9845 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
9846 struct function as arg.
9847 (estimate_numbers_of_iterations): Export overload with loop arg.
9848 (free_numbers_of_iterations_estimates_loop): Use an overload of
9849 free_numbers_of_iterations_estimates instead.
9850 * tree-cfg.c (remove_bb): Adjust.
9851 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
9852 * tree-parloops.c (gen_parallel_loop): Likewise.
9853 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
9855 (tree_unroll_loops_completely): Likewise.
9856 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
9857 Use an overload instead and export.
9858 (estimated_loop_iterations): Adjust.
9859 (max_loop_iterations): Likewise.
9860 (likely_max_loop_iterations): Likewise.
9861 (estimate_numbers_of_iterations): Take struct function as arg
9863 (loop_exits_before_overflow): Adjust.
9864 (free_numbers_of_iterations_estimates_loop): Use an overload.
9865 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
9866 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
9868 2017-06-19 Richard Biener <rguenther@suse.de>
9871 * ipa-prop.c (find_constructor_constant_at_offset): Handle
9872 RANGE_EXPR conservatively.
9874 2017-06-16 Carl Love <cel@us.ibm.com>
9876 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9877 definitions for vec_float, vec_float2, vec_floato,
9878 vec_floate built-ins.
9879 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
9880 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
9882 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
9883 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
9884 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
9885 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
9886 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
9887 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
9888 vec_floato): Add builtin defines.
9889 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
9890 Update the built-in documentation file for the new built-in
9893 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9895 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
9896 (mthumb): Mark as the negative of -marm.
9898 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9900 * doc/invoke.texi (ARM Options, -mcpu): Document supported
9902 (ARM Options, -mtune): Document that this accepts the same
9903 extension options as -mcpu.
9904 (ARM Options, -mfpu): Document addition of -mfpu=auto.
9906 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9908 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
9909 permitted extensions.
9911 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9913 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
9914 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
9915 (armv8-m.main): Add option +nodsp.
9916 * config/arm/arm-cpu-cdata.h: Regenerated.
9918 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9920 * config/arm/t-fuchsia: New file.
9921 * config.gcc (arm*-*-fuchsia*): Use it.
9923 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9925 * config/arm/t-symbian: Rewrite for new option infrastructure.
9927 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9929 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
9930 (MULTILIB_REQUIRED): Likewise.
9932 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9934 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
9935 (MULTILIB_RESUE): Likewise.
9936 (MULTILIB_MATCHES): Likewise.
9937 (MULTLIB_REQUIRED): Likewise.
9939 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9941 * config/arm/t-rtems: Rewrite for new option framework.
9943 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9945 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
9946 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
9947 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
9948 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
9949 * config/arm/t-multilib: ... here.
9950 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
9951 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
9952 armv7-a and armv8*-a when A-profile libraries have not been built.
9953 * config/arm/t-rmprofile: Rewrite.
9955 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9957 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
9958 with a backslash. Remove the backslash after substituting unescaped
9960 * doc/fragments.texi (MULTILIB_REUSE): Document it.
9962 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9964 * config.gcc: (arm*-*-*): When building a-profile libraries, force
9965 the driver to pass through the default setting of -mfloat-abi.
9966 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
9968 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
9969 (all_feat_combs): New rule.
9970 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
9972 * config/arm/t-aprofile: Rewrite.
9974 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9976 * config/arm/arm.h (FPUTYPE_AUTO): Define.
9977 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
9978 fpu is not specified by the user/command-line.
9979 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
9980 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
9981 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
9982 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
9983 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
9984 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
9986 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9988 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
9989 * config/arm/t-arm-elf: Rewritten.
9991 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9993 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
9994 have some floating-point instructions.
9995 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
9996 (TARGET_MAYBE_HARD_FLOAT): New macro.
9997 * config/arm/arm-builtins.c (arm_init_builtins): Use
9998 TARGET_MAYBE_HARD_FLOAT.
9999 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
10001 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10003 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
10004 (configargs.h): Include it.
10005 (arm_print_hint_for_fpu_option): New function.
10006 (arm_parse_fpu_option): New function.
10007 (candidate_extension): New class.
10008 (arm_canon_for_multilib): New function.
10009 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
10010 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
10011 (ARCH_CANONICAL_SPECS): New macro.
10012 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
10014 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10016 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
10017 are set after handling multilib fragments. Set target_cpu_default2
10020 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10022 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
10024 (arm*-*-*): Set target_cpu_default2 to a quoted string.
10025 * config/arm/parsecpu.awk (check_cpu): Validate any extension
10027 (check_arch): Likewise.
10028 * config/arm/arm.c (arm_configure_build_target): Handle
10029 TARGET_CPU_DEFAULT being a string constant. Scan any feature
10030 options in the default.
10032 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10034 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
10035 when an option is an alias of another.
10036 * config/arm/parsecpu.awk (optalias): New parser token.
10037 (gen_comm_data): Mark non-alias options as such. Emit entries
10038 for extension aliases.
10039 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
10040 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
10041 (armv6kz, armv6zk, armv6t2): Likewise.
10042 (armv7): Make vfpv3-d16 an alias.
10043 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
10045 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
10046 Sort in canonical order.
10047 (armv8-a): Sort in canonical order.
10048 (armv8.1-a, armv8.2-a): Likewise.
10049 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
10051 (cortex-a9): Sort in canonical order.
10052 * config/arm/arm.c (selftests.h): Include it.
10053 (arm_test_cpu_arch_data): New function.
10054 (arm_run_self_tests): New function.
10055 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
10056 (targetm): Move declaration to the end of the file.
10057 * arm-cpu-cdata.h: Regenerated.
10059 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10061 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
10062 call to target_mode_check describing the type of option passed.
10063 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
10064 (arm_target_thumb_only): Use arm_parse_arch_option_name or
10065 arm_parse_cpu_option_name to match parameters against list of
10067 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
10068 arm_arch_core_flags data structure.
10069 * config/arm/arm-cpu_cdata.h: Regenerated.
10071 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10073 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
10075 (arm_print_hint_for_cpu_option): Likewise.
10076 (arm_print_hint_for_arch_option): Likewise.
10077 (arm_parse_cpu_option_name): Likewise.
10078 (arm_parse_arch_option_name): Likewise.
10079 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
10080 of entries in the all_fpus list.
10081 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
10082 (arm_parse_cpu_option_name): Declare.
10083 (arm_parse_arch_option_name): Declare.
10084 (arm_parse_option_features): Declare.
10085 (arm_intialize_isa): Declare.
10086 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
10088 (gen_comm_data): ... here. Make definitions non-static.
10089 * config/arm/arm-cpu-data.h: Regenerated.
10090 * config/arm/arm-cpu-cdata.h: Regenerated.
10092 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10094 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
10095 (cpu_arch_extension): New structure.
10096 (cpu_arch_option, arch_option, cpu_option): New structures.
10097 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
10098 architecture types.
10099 (gen_data): Generate new format data tables.
10100 * config/arm/arm.c (cpu_tune): New structure.
10101 (cpu_option, processors): Delete.
10102 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
10103 (arm_print_hint_for_cpu_option): ... this and ...
10104 (arm_print_hint_for_arch_option): ... this.
10105 (arm_parse_arch_cpu_name): Delete. Replace with ...
10106 (arm_parse_cpu_option_name): ... this and ...
10107 (arm_parse_arch_option_name): ... this.
10108 (arm_unrecognized_feature): Change type of target parameter to
10110 (arm_parse_arch_cpu_features): Delete. Replace with ...
10111 (arm_parse_option_features): ... this.
10112 (arm_configure_build_target): Rework to use new configuration data
10114 (arm_print_tune_info): Rework for new configuration data tables.
10115 * config/arm/arm-cpu-data.h: Regenerated.
10116 * config/arm/arm-cpu.h: Regenerated.
10118 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10120 * Makefile.in (OBJS): Move sbitmap.o from here ...
10121 (OBJS-libcommon): ... to here.
10123 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10125 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
10126 (ISA_ALL_CRYPTO): New macro.
10127 (ISA_ALL_SIMD): New macro
10128 (ISA_ALL_FP): New macro.
10129 * config/arm/arm.c (fpu_bitlist): Update initializer.
10130 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
10132 (arm9e): Add fpu. Add option for nofp
10133 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
10134 (arm926ej-s, arm1026ej-s): Likewise.
10135 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
10136 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
10137 neon-fp16, neon-vfpv4, nofp and nosimd.
10138 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
10139 (cortex-a8): Add fpu. Add option for nofp.
10140 (cortex-a9): Add fpu. Add options for nosimd and nofp.
10141 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
10142 (cortex-r4f): Add fpu.
10143 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
10144 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
10146 (cortex-r8): Likewise.
10147 (cortex-m4): Add fpu. Add option for nofp.
10148 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
10149 (cortex-a17.cortex-a7): Likewise.
10150 (cortex-a32): Add fpu. Add options for crypto and nofp.
10151 (cortex-a35, cortex-a53): Likewise.
10152 (cortex-a57): Add fpu. Add option for crypto.
10153 (cortex-a72, cortex-a73): Likewise.
10154 (exynos-m1): Likewise.
10155 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
10156 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
10157 (cortex-m33): Add fpu. Add option for nofp.
10158 * config/arm/arm-cpu-cdata.h: Regenerated
10159 * config/arm/arm-cpu-data.h: Regenerated.
10161 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10163 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
10164 (armv5te, armv5tej): Likewise.
10165 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
10166 (armv7): Add options fp and vfpv3-d16.
10167 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
10168 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
10170 (armv7ve): Likewise.
10171 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
10172 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
10173 (armv8-a): Add nocrypto option.
10174 (armv8.1-a, armv8.2-a): Likewise.
10175 (armv8-m.main): add options fp, fp.dp and nofp.
10177 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10179 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
10181 (armv8-a+crc): Delete.
10182 (armv8.1-a): Add options simd, crypto and nofp.
10183 (armv8.2-a): Add options fp16, simd, crypto and nofp.
10184 (armv8.2-a+fp16): Delete.
10185 (armv8-m.main): Add option dsp.
10186 (armv8-m.main+dsp): Delete.
10187 (cortex-a8): Add fpu. Add nofp option.
10188 (cortex-a9): Add fpu. Add nofp and nosimd options.
10189 * config/arm/parsecpu.awk (gen_data): Generate option tables and
10190 link to main cpu and architecture data structures.
10191 (gen_comm_data): Only put isa attributes from the main architecture
10193 (option): New statement for architecture and CPU entries.
10194 * arm.c (struct cpu_option): New structure.
10195 (struct processors): Add entry for options.
10196 (arm_unrecognized_feature): New function.
10197 (arm_parse_arch_cpu_name): Ignore any characters after the first
10199 (arm_parse_arch_cpu_feature): New function.
10200 (arm_configure_build_target): Separate out any CPU and architecture
10201 features and parse separately. Don't error out if -mfpu=auto is
10202 used with only an architecture string.
10203 (arm_print_asm_arch_directives): New function.
10204 (arm_file_start): Call it.
10205 * config/arm/arm-cpu-cdata.h: Regenerated.
10206 * config/arm/arm-cpu-data.h: Likewise.
10207 * config/arm/arm-tables.opt: Likewise.
10209 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10211 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
10212 assembler when it is not -mfpu=auto.
10214 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10216 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
10217 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
10218 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
10219 (ASM_CPU_SPEC): Rewrite.
10220 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
10221 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
10222 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
10223 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
10224 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
10225 copied string is NUL-terminated. Also strip any characters prefixed
10227 (arm_rewrite_selected_arch): New function.
10228 (arm_rewrite_march): New function.
10230 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10232 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
10233 (x_arm_cpu_string, x_arm_tune_string): Likewise.
10234 (march, mcpu, mtune): Convert to string-based options.
10235 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
10236 (arm_parse_arch_cpu_name): New function.
10237 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
10238 identify selected architecture or CPU.
10239 (arm_option_save): New function.
10240 (TARGET_OPTION_SAVE): Redefine.
10241 (arm_option_restore): Restore string options.
10242 (arm_option_print): Print string options.
10244 2017-06-16 Martin Sebor <msebor@redhat.com>
10246 PR tree-optimization/80933
10247 PR tree-optimization/80934
10248 * builtins.c (fold_builtin_3): Do not handle bcmp here.
10249 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
10250 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
10251 (gimple_fold_builtin): Call them.
10253 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10255 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
10256 as unlikely; update profile.
10258 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10260 * predict.c (force_edge_cold): Handle declaring edges impossible
10263 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10265 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
10267 (try_unroll_loop_completely): Fix reporting.
10269 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10271 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
10273 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
10276 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
10277 if given a non-constant argument for an intrinsic which requires a
10280 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10282 * profile.c (compare_freqs): New function.
10283 (branch_prob): Sort edge list.
10284 (find_spanning_tree): Assume that the list is priority sorted.
10286 2017-06-16 Richard Biener <rguenther@suse.de>
10288 PR tree-optimization/81090
10289 * passes.def (pass_record_bounds): Remove.
10290 * tree-pass.h (make_pass_record_bounds): Likewise.
10291 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
10292 make_pass_record_bounds): Likewise.
10293 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
10294 not free niter estimates at the beginning but at the end.
10295 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
10297 2017-06-16 Richard Biener <rguenther@suse.de>
10299 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
10300 initializer to workaround ICE in host GCC 4.8.
10302 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
10304 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
10306 (clone_inlined_nodes): Update.
10308 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10310 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
10311 prefetch settings, and enable prefetching by default at -O3.
10313 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10315 * config/aarch64/aarch64.c (aarch64_override_options_internal):
10316 Set flag_prefetch_loop_arrays according to tuning data.
10318 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10320 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
10321 New tune structure.
10322 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
10323 [Unrelated to main purpose of the patch] Place the pointer field last
10324 to enable type checking errors when tune structure are wrongly merged.
10325 * config/aarch64/aarch64.c (generic_prefetch_tune,)
10326 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
10327 (thunderx2t99_prefetch_tune): New tune constants.
10328 (tune_params *_tunings): Update all tunings (no functional change).
10329 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
10330 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
10331 from tunings structures.
10333 2017-06-16 Jakub Jelinek <jakub@redhat.com>
10336 * ubsan.c (instrument_null): Add T argument, use it instead
10337 of computing it based on IS_LHS.
10338 (instrument_object_size): Likewise.
10339 (pass_ubsan::execute): Adjust instrument_null and
10340 instrument_object_size callers to pass gimple_get_lhs or
10341 gimple_assign_rhs1 result to it. Use instrument_null instead of
10342 calling get_base_address and instrument_mem_ref. Handle
10343 aggregate call arguments for object-size sanitization.
10345 2017-06-16 Yury Gribov <tetra2005@gmail.com>
10347 PR tree-optimization/81089
10348 * tree-vrp.c (is_masked_range_test): Validate operands of
10351 2017-06-15 Martin Sebor <msebor@redhat.com>
10354 * dumpfile.c (dump_register): Avoid calling memset to initialize
10355 a class with a default ctor.
10356 * gcc.c (struct compiler): Remove const qualification.
10357 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
10358 * hash-table.h: Ditto.
10359 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
10361 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
10362 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
10364 * params.h (struct param_info): Make struct members non-const.
10365 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
10366 with default initialization.
10367 * vec.h (vec_copy_construct, vec_default_construct): New helper
10369 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
10370 with vec_copy_construct.
10371 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
10372 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
10373 * doc/invoke.texi (-Wclass-memaccess): Document.
10375 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10377 * emit-rtl.h (is_leaf): Update comment about local
10378 register allocator.
10380 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10383 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
10384 for a variable to have a section before checking if the section has a
10386 Set section to.persistent if persistent attribute is set.
10387 Warn if .persistent attribute is used on an automatic variable.
10389 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
10391 PR rtl-optimization/80474
10392 * reorg.c (update_block): Do not ignore instructions in a delay slot.
10394 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
10396 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
10399 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
10401 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
10402 (casesi): Emit bounds checking as RTL.
10403 (casesi_internal_mips16_<mode>): Remove bounds checking.
10405 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
10407 * config/xtensa/xtensa.c (xtensa_option_override): Append
10408 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
10409 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
10410 xtensa_doloop_hooks): Define unconditionally.
10411 (xtensa_reorg_loops): Only call reorg_loops in the presence of
10413 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
10414 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
10415 for it in xtensa_option_override.
10416 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
10417 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
10419 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
10421 * doc/cppopts.texi: Document '-' special value to -MF.
10423 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
10425 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
10426 (cortex_a53_fconst): Likewise.
10427 (cortex_a53_fpmul): Likewise.
10428 (cortex_a53_f_load_64): Likewise.
10429 (cortex_a53_f_load_many): Likewise.
10430 (cortex_a53_advsimd_alu): Likewise.
10431 (cortex_a53_advsimd_alu_q): Likewise.
10432 (cortex_a53_advsimd_mul): Likewise.
10433 (cortex_a53_advsimd_mul_q): Likewise.
10434 (fpmac bypass): Add new bypass for fpmac-fpmac case.
10435 Add missing fmul, r2f_cvt and fconst cases.
10437 2017-06-14 Richard Biener <rguenther@suse.de>
10439 PR middle-end/81088
10440 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
10442 (fold_binary_loc): When associating do not treat pre-existing
10443 TREE_OVERFLOW on literal constants as a reason to allow
10444 TREE_OVERFLOW on associated literal constants.
10446 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
10448 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
10449 (MASK_FEATURES): New macro.
10450 * config/sparc/sparc.c (sparc_option_override): Remove the special
10451 handling of -mfpu and generalize it to all MASK_FEATURES switches.
10453 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
10455 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
10456 a division of 0 if non-call exceptions are enabled.
10458 2017-06-14 Andrew Pinski <apinski@cavium.com>
10459 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10462 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
10463 Improve vector initialization code gen for only variable case.
10465 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
10467 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
10469 2017-06-14 Richard Biener <rguenther@suse.de>
10471 PR tree-optimization/81083
10472 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
10475 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10477 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
10478 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
10479 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
10480 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
10481 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
10482 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
10484 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10486 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
10487 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
10489 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10491 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
10493 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10495 * config/rs6000/t-rtems: Don't handle SPE.
10497 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10499 * config/rs6000/t-linux: Don't handle SPE.
10501 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10503 * config/rs6000/eabispe.h: Delete file.
10505 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10507 * config/rs6000/t-spe: Delete file.
10509 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10511 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
10512 (rs6000_legitimate_offset_address_p): Return false for anything in
10513 V2SImode or V2SFmode.
10515 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10517 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
10518 except V2SF and V2SI. Rearrange the vector modes, and add comments.
10519 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
10521 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
10522 (rs6000_legitimate_offset_address_p): Ditto.
10523 (rs6000_emit_move): Ditto.
10524 (rs6000_init_builtins): Remove V4HI_type_node.
10526 2017-06-13 Martin Liska <mliska@suse.cz>
10529 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
10530 (gate_asan): Likewise.
10531 * asan.h (asan_no_sanitize_address_p): Remove the function.
10532 (sanitize_flags_p): New function.
10533 * builtins.def: Fix coding style.
10534 * common.opt: Use renamed enum value.
10535 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
10536 * doc/extend.texi: Document no_sanitize attribute.
10537 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
10538 to SANITIZE_UNDEFINED_NONDEFAULT.
10539 * gcc.c (sanitize_spec_function): Use the renamed enum value.
10540 * gimple-fold.c (optimize_atomic_compare_exchange_p):
10541 Use sanitize_flags_p.
10542 * gimplify.c (gimplify_function_tree): Likewise.
10543 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
10544 * opts.c (parse_no_sanitize_attribute): New function.
10545 (common_handle_option): Use renamed enum value.
10546 * opts.h (parse_no_sanitize_attribute): Declare.
10547 * tree.c (sanitize_flags_p): New function.
10548 * tree.h: Declared here.
10549 * tsan.c: Use sanitize_flags_p.
10550 * ubsan.c (ubsan_expand_null_ifn): Likewise.
10551 (instrument_mem_ref): Likewise.
10552 (instrument_bool_enum_load): Likewise.
10553 (do_ubsan_in_current_function): Remove the function.
10554 (pass_ubsan::execute): Use sanitize_flags_p.
10555 * ubsan.h: Remove do_ubsan_in_current_function
10556 * tree-cfg.c (print_no_sanitize_attr_value): New function.
10557 (dump_function_to_file): Use it here.
10559 2017-06-13 Martin Jambor <mjambor@suse.cz>
10561 PR tree-optimization/80803
10562 PR tree-optimization/81063
10563 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
10564 (propagate_subaccesses_across_link): Enqueue subtree whenever
10565 necessary instead of relying on the caller.
10567 2017-06-13 Martin Jambor <mjambor@suse.cz>
10569 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
10570 that have a first_link.
10571 (sort_and_splice_var_accesses): Do not check first_link before
10573 (subtree_mark_written_and_enqueue): Likewise.
10574 (propagate_all_subaccesses): Likewise and do not stop at first
10575 parent with a first_link.
10577 2017-06-13 Martin Jambor <mjambor@suse.cz>
10579 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
10582 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10584 * match.pd: New pattern.
10586 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10588 * tree-vrp.c (is_masked_range_test): New function.
10589 (register_edge_assert_for): Determine ranges for
10592 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10594 PR tree-optimization/67328
10595 * fold-const.c (maskable_range_p): New function.
10596 (build_range_check): Generate bittests if possible.
10598 2017-06-13 Martin Liska <mliska@suse.cz>
10600 * gimple-pretty-print.c (dump_probability): Add new argument.
10601 (dump_edge_probability): Dump both probability and count.
10602 (dump_gimple_label): Likewise.
10603 (dump_gimple_bb_header): Likewise.
10605 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
10608 * config/avr/avr-devices.c: Fix indentation.
10609 * config/avr/gen-avr-mmcu-specs.c: Dito.
10611 2017-06-13 Richard Biener <rguenther@suse.de>
10613 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
10614 instead get vector type from stmt_info.
10615 (vectorizable_reduction): Adjust. Remove dead code.
10617 2017-06-13 Richard Biener <rguenther@suse.de>
10619 PR middle-end/81065
10620 * fold-const.c (extract_muldiv_1): Remove bogus distribution
10621 case of C * (x * C2 + C3).
10622 (fold_addr_of_array_ref_difference): Properly fold index difference.
10624 2017-06-12 David S. Miller <davem@davemloft.net>
10627 * config/sparc/sparc.md (return expander): Emit frame blockage if
10628 function uses alloca.
10630 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
10632 * combine.c (make_field_assignment): Check len rather than the mode
10633 precision when calling force_to_mode.
10635 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
10637 Support multilibs and devices that see flash in RAM address range.
10640 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
10641 (avr_mcu_t) <flash_pm_offset>: New field.
10642 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
10643 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
10644 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
10645 (AVR_TINY_PM_OFFSET): Remove macro.
10646 * config/avr/avr.opt (-mshort-calls): New option.
10647 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
10648 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
10649 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
10650 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
10651 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
10652 instead of avr_arch->have_jmp_call.
10653 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
10654 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
10655 avr_arch->flash_pm_offset to define.
10656 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
10657 new field flash_pm_offset. Add entry for avrxmega3.
10658 (avr_texinfo): Add entry for avrxmega3.
10659 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
10660 attiny212, attiny214,
10661 attiny412, attiny414, attiny416, attiny417,
10662 attiny814, attiny816, attiny817,
10663 attiny1614, attiny1616, attiny1617,
10664 attiny3214, attiny3216, attiny3217.
10665 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
10666 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
10667 (avr_print_operand_address) [AVR_TINY]: Same.
10668 (avr_asm_init_sections) <readonly_data_section>: Only patch
10669 callback if avr_arch->flash_pm_offset = 0.
10670 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
10671 for rodata if avr_arch->flash_pm_offset != 0.
10672 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
10673 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
10674 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
10675 (m_options): Append opt_rcall.
10676 (m_dirnames): Append dir_rcall.
10677 * config/avr/t-multilib: Regenerate.
10679 * configure.ac [target=avr]: Check whether avrxmega3 default
10680 linker description file works as needed.
10681 * configure: Regenerate.
10682 * doc/avr-mmcu.texi: Regenerate.
10683 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
10684 <__AVR_ARCH__>: Document avrxmega3 and 103.
10685 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
10686 <__AVR_SHORT_CALLS__>: Document it.
10687 <__AVR_PM_BASE_ADDRESS__>: Document it.
10688 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
10689 (AVR Variable Attributes) <progmem>: Document this is
10690 not needed for avrxmega3.
10691 (AVR Named Address Spaces) <__flash>: Dito.
10693 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
10695 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
10697 2017-06-12 Doug Rupp <rupp@adacore.com>
10699 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
10700 Append vxworks-stdint.h to the tm_file list.
10701 * config/vxworks-stdint.h: New file.
10703 2017-06-12 Martin Liska <mliska@suse.cz>
10705 PR tree-optimization/81041
10706 * tree-profile.c (gimple_gen_ic_func_profiler):
10707 Create an extra BB in profile-generate
10708 (gimple_gen_time_profiler): Likewise.
10710 2017-06-12 Jakub Jelinek <jakub@redhat.com>
10712 PR tree-optimization/81003
10713 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
10714 (update_range_test): Use it instead of force_gimple_operand_gsi.
10716 2017-06-12 Richard Biener <rguenther@suse.de>
10718 PR tree-optimization/81053
10719 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
10720 with backedge value not defined in loop. Simplify def stmt
10723 2017-06-11 Tom de Vries <tom@codesourcery.com>
10726 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
10728 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
10729 nvptx_cannot_force_const_mem.
10731 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10733 * opts.c (finish_options): Move test for flag_split_stack after
10734 it has been initialized.
10736 2017-06-11 Jason Merrill <jason@redhat.com>
10738 * tree.h (id_equal): New.
10739 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
10740 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
10741 instead of strcmp of IDENTIFIER_POINTER.
10743 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10745 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
10746 (mark_all_inlined_calls_cdtor): Fix formating.
10747 (inline_transform): Rescale profile before inlining.
10749 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10751 * cgraph.h (cgraph_edge::clone): Update prototype.
10752 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
10753 (cgraph_node::create_clone): Update.
10754 (cgraph_node::create_version_clone): Update.
10755 * tree-inline.c (copy_bb): Update.
10756 (expand_call_inline): Update.
10758 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
10760 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
10761 factored out from ...
10762 (rs6000_emit_prologue): ... here.
10764 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
10766 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
10767 factored out from ...
10768 (rs6000_emit_prologue): ... here.
10770 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10772 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
10774 (handle_missing_profiles): Fix computation of tp_first_run.
10775 (counts_to_freqs): Do not touch freqs when count is 0.
10777 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10779 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
10782 2017-06-10 Tom de Vries <tom@codesourcery.com>
10784 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
10785 attributes): Document signal effective target.
10787 2017-06-10 Tom de Vries <tom@codesourcery.com>
10789 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
10790 Document effective target stack_size.
10792 2017-06-09 David Malcolm <dmalcolm@redhat.com>
10794 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
10795 to the edit_context if they can be auto-applied.
10797 2017-06-9 Ian Lance Taylor <iant@golang.org>
10799 * opts.c (finish_options): If -fsplit-stack, disable implicit
10800 -forder-blocks-and-partition.
10801 * doc/invoke.texi (Optimize Options): Document that when using
10802 -fsplit-stack -forder-blocks-and-partition is not implicitly
10805 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10807 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
10808 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
10809 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
10810 * builtins.def (abort, trap, unreachable): Declare cold.
10811 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
10812 * tree-core.h (ECF_COLD): New.
10813 * tree.c (set_call_expr_flags): Handle ECF_COLD.
10814 (build_common_builtin_nodes): Mark unreachable and abort as cold.
10816 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10818 * predict.c (unlikely_executed_stmt_p): Cleanup.
10820 2017-06-09 Richard Biener <rguenther@suse.de>
10822 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
10823 model if the ref is always written to.
10825 2017-06-09 Tamar Christina <tamar.christina@arm.com>
10827 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
10829 2017-06-09 Tamar Christina <tamar.christina@arm.com>
10831 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
10834 2017-06-09 Tom de Vries <tom@codesourcery.com>
10837 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
10838 "target cannot support label values" when encountering LABEL_REF.
10840 2017-06-09 Martin Liska <mliska@suse.cz>
10842 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
10843 (gimple_gen_ic_func_profiler): Emit direct comparison
10844 of __gcov_indirect_call_callee with NULL.
10845 (gimple_gen_time_profiler): Change probability from
10846 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
10848 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10850 * profile.c (edge_gcov_counts): Turn to pointer.
10851 (compute_branch_probabilities, compute_branch_probabilities): Update.
10852 (branch_prob): Do not clear edge_gcov_count.
10853 * profile.h (edge_gcov_counts): Turn to pointer.
10854 (edge_gcov_count): Update.
10856 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10858 * gimple.h (gimple_check_failed): Mark cold.
10860 2017-06-09 Richard Biener <rguenther@suse.de>
10862 PR tree-optimization/66623
10863 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
10864 refactor check_reduction into two parts, properly computing
10865 whether we have to check reduction validity for outer loop
10868 2017-06-09 Richard Biener <rguenther@suse.de>
10870 PR tree-optimization/79483
10871 * graphite-scop-detection.c (order): New global.
10872 (get_order): Compute bb to order mapping that satisfies code
10873 generation constraints.
10874 (cmp_pbbs): New helper.
10875 (build_scops): Start domwalk at entry block, sort generated
10878 2017-06-09 Richard Biener <rguenther@suse.de>
10880 PR middle-end/81007
10881 * ipa-polymorphic-call.c
10882 (ipa_polymorphic_call_context::restrict_to_inner_class):
10883 Skip FIELD_DECLs with error_mark_node type.
10884 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
10887 2017-06-09 Martin Liska <mliska@suse.cz>
10889 * predict.c (struct branch_predictor): New struct.
10890 (test_prediction_value_range): New test.
10891 (predict_c_tests): New function.
10892 * selftest-run-tests.c (selftest::run_tests): Run the function.
10893 * selftest.h: Declare new tests.
10895 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
10898 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
10899 gen_add3_insn did not fail.
10900 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
10901 r0, construct that number in a temporary reg and add that reg to r0.
10902 If asked to put the result in r0 as well, fail.
10904 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
10906 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10907 for early expansion of vec_eqv.
10909 2017-06-08 Jakub Jelinek <jakub@redhat.com>
10911 PR middle-end/81005
10912 * ubsan.c (instrument_null): Avoid pointless code temporary.
10913 (pass_ubsan::execute): Instrument aggregate arguments of calls.
10915 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
10919 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
10922 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
10925 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10927 * predict.c (unlikely_executed_edge_p): Move ahead.
10928 (probably_never_executed_edge_p): Use it.
10930 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10932 PR middle-end/79988
10933 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
10934 gimple_call_builtin_p call.
10936 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10938 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
10939 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
10940 rtl_check_failed_type2, rtl_check_failed_code1,
10941 rtl_check_failed_code2, rtl_check_failed_code_mode,
10942 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
10943 rtvec_check_failed_bounds, rtl_check_failed_flag,
10944 _fatal_insn_not_found, _fatal_insn): Likewise.
10945 * tree.h (tree_contains_struct_check_failed,
10946 tree_check_failed, tree_not_check_failed,
10947 tree_class_check_failed, tree_range_check_failed,
10948 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
10949 tree_vec_elt_check_failed, phi_node_elt_check_failed,
10950 tree_operand_check_failed, omp_clause_check_failed,
10951 omp_clause_operand_check_failed, omp_clause_range_check_failed):
10954 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10956 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
10957 flag_branch_probabilities.
10958 * ipa-inline.c (edge_badness): Likewise.
10959 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
10960 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
10961 * predict.c (maybe_hot_frequency_p): Likewise.
10962 (probably_never_executed): Likewise.
10963 * sched-ebb.c (schedule_ebbs): Likewise.
10964 * sched-rgn.c (find_single_block_region): Likewise.
10965 * tracer.c (tail_duplicate): Likewise.
10967 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10969 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
10970 longer requires x_flag_profile_use.
10972 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10974 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
10975 instead of flag_reorder_blocks_and_partition.
10976 * dbxout.c (dbxout_function_end): Likewise.
10977 * dwarf2out.c (gen_subprogram_die): Likewise.
10978 * haifa-sched.c (sched_create_recovery_edges): Likewise.
10979 * hw-doloop.c (reorg_loops): Likewise.
10980 * varasm.c (assemble_start_function,
10981 assemble_end_function): Likewise.
10982 (decide_function_section): Do not check for
10983 flag_reorder_blocks_and_partition.
10985 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10987 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10989 (chkp_get_hard_register_fake_addr_expr): Ditto.
10990 (chkp_build_addr_expr): Add check for hard reg case.
10991 (chkp_parse_array_and_component_ref): Ditto.
10992 (chkp_find_bounds_1): Ditto.
10993 (chkp_process_stmt): Don't generate bounds store for
10996 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10998 * predict.c (maybe_hot_bb_p): Do not check profile status.
10999 (maybe_hot_edge_p): Likewise.
11000 (probably_never_executed): Check for zero counts even if profile
11002 (unlikely_executed_edge_p): New function.
11003 (unlikely_executed_stmt_p): New function.
11004 (unlikely_executed_bb_p): New function.
11005 (set_even_probabilities): Use unlikely predicates.
11006 (combine_predictions_for_bb): Likewise.
11007 (predict_paths_for_bb): Likewise.
11008 (predict_paths_leading_to_edge): Likewise.
11009 (determine_unlikely_bbs): New function.
11010 (estimate_bb_frequencies): Use it.
11011 (compute_function_frequency): Use zero counts even if profile is
11013 * profile-count.h: Fix typo.
11015 2017-08-08 Julia Koval <julia.koval@intel.com>
11017 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
11018 _mm512_mask_cvtsepi16_storeu_epi8,
11019 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
11020 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
11021 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
11022 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
11023 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
11024 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
11025 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
11026 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
11027 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
11028 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
11029 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
11030 __builtin_ia32_pmovuswb256mem_mask,
11031 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
11032 __builtin_ia32_pmovwb512mem_mask): New builtins.
11034 2017-08-08 Julia Koval <julia.koval@intel.com>
11036 PR target/73350,80862
11037 * config/i386/subst.md (round): Fix round pattern.
11038 * config/i386/i386.c (ix86_erase_embedded_rounding):
11039 Fix erasing rounding for the fixed pattern.
11041 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11043 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
11045 2017-06-08 Martin Liska <mliska@suse.cz>
11047 PR gcov-profile/80911
11048 * gcov.c (block_info::block_info): New constructor.
11050 2017-06-07 Carl Love <cel@us.ibm.com>
11052 * config/rs6000/rs6000-c: The return type of the following
11053 built-in functions was implemented as int not long long. Fix sign
11054 of return value for the unsigned version of vec_mulo and vec_mule.
11055 vector unsigned long long vec_bperm (vector unsigned long long,
11056 vector unsigned char)
11057 vector signed long long vec_mule (vector signed int,
11059 vector unsigned long long vec_mule (vector unsigned int,
11060 vector unsigned int)
11061 vector signed long long vec_mulo (vector signed int,
11063 vector unsigned long long vec_mulo (vector unsigned int,
11064 vector unsigned int)
11065 * doc/extend.texi: Fix the documentation for the built-in
11068 2017-06-07 Carl Love <cel@us.ibm.com>
11071 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
11074 2017-06-07 Carl Love <cel@us.ibm.com>
11076 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
11077 support, Generate doublehv for signed int/float for BE case only.
11079 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
11081 * doc/invoke.texi (mcx16): Rewrite.
11083 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11085 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
11086 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
11087 *mov<mode>_softfloat, and an anonymous splitter): Use
11088 nonimmediate_operand instead of rs6000_nonimmediate_operand.
11090 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11092 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
11094 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
11095 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
11096 (rs6000_debug_reg_global): Adjust.
11097 (rs6000_init_hard_regno_mode_ok): Adjust.
11098 (rs6000_dbx_register_number): Adjust.
11099 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
11100 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
11101 Remove SPE_ACC and SPEFSCR.
11102 (REG_ALLOC_ORDER): Ditto.
11103 (FRAME_POINTER_REGNUM): Change to 111.
11104 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
11105 (REG_CLASS_NAMES): Ditto.
11106 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
11107 (REGISTER_NAMES): Ditto.
11108 (ADDITIONAL_REG_NAMES): Ditto.
11109 (rs6000_reg_names): Ditto.
11110 * config/rs6000/rs6000.md: Renumber some register number
11113 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11115 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
11117 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
11118 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
11120 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
11121 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
11122 Delete the SPE high registers.
11123 (REG_ALLOC_ORDER): Ditto.
11124 (enum reg_class): Remove SPE_HIGH_REGS.
11125 (REG_CLASS_NAMES): Ditto.
11126 (REG_CLASS_CONTENTS): Delete the SPE high registers.
11127 (REGISTER_NAMES): Ditto.
11128 (rs6000_reg_names): Ditto.
11129 * doc/tm.texi.in: Remove SPE as example.
11130 * doc/tm.texi: Regenerate.
11132 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11134 * config/rs6000/8540.md (ppc8540_brinc): Delete.
11135 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
11136 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
11137 * config/rs6000/rs6000.md (type): Remove "brinc".
11139 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11141 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
11142 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
11143 * config/rs6000/linuxspe.h: Delete file.
11144 * config/rs6000/rs6000.md: Don't include spe.md.
11145 * config/rs6000/spe.h: Delete file.
11146 * config/rs6000/spe.md: Delete file.
11147 * config/rs6000/t-rs6000: Remove spe.md.
11149 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11151 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
11152 (reg_or_none500mem_operand): Delete.
11153 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
11154 instead of reg_or_none500mem_operand.
11156 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11158 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
11159 handling of SPE flags.
11160 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
11162 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11164 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
11166 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
11167 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
11168 paired_divv2sf3): Similar.
11169 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
11170 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
11171 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
11173 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
11174 Rename the paired_* instruction patterns.
11175 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
11177 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
11178 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
11179 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
11180 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
11181 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
11182 PAIRED_VECTOR_MODE.
11183 (struct machine_function): Delete field spe_insn_chain_scanned_p.
11184 (spe_func_has_64bit_regs_p): Delete.
11185 (spe_expand_predicate_builtin): Delete.
11186 (spe_expand_evsel_builtin): Delete.
11187 (TARGET_DWARF_REGISTER_SPAN): Do not define.
11188 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
11189 (invalid_e500_subreg): Delete.
11190 (rs6000_legitimize_address): Always force_reg op2 as well, for
11191 paired single memory accesses.
11192 (rs6000_member_type_forces_blk): Delete.
11193 (rs6000_spe_function_arg): Delete.
11194 (rs6000_expand_unop_builtin): Delete SPE handling.
11195 (rs6000_expand_binop_builtin): Ditto.
11196 (spe_expand_stv_builtin): Delete.
11197 (bdesc_2arg_spe): Delete.
11198 (spe_expand_builtin): Delete.
11199 (spe_expand_predicate_builtin): Delete.
11200 (spe_expand_evsel_builtin): Delete.
11201 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
11202 (spe_init_builtins): Delete.
11203 (spe_func_has_64bit_regs_p): Delete.
11204 (savres_routine_name): Delete "info" parameter. Adjust callers.
11205 (rs6000_emit_stack_reset): Ditto.
11206 (rs6000_dwarf_register_span): Delete.
11207 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
11208 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
11209 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
11211 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
11213 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
11214 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
11215 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
11216 mulv2sf3, divv2sf3): Delete expanders.
11218 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11220 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
11222 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11224 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
11225 * config/rs6000/rs6000.c: Ditto.
11227 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11229 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
11230 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
11231 comparison_operator.
11233 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11235 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
11236 * config/rs6000/rs6000.opt: Ditto.
11237 * config/rs6000/t-rtems: Ditto.
11239 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11241 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
11242 TARGET_E500_SINGLE by 0, simplify.
11243 * config/rs6000/rs6000.c: Ditto.
11244 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
11245 (spe_build_register_parallel): Delete.
11246 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
11247 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
11248 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
11249 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
11250 (E500_CONVERT): Delete.
11251 * config/rs6000/spe.md: Remove many patterns and all define_constants.
11253 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11255 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
11256 * config/rs6000/dfp.md: Ditto.
11257 (negdd2, *negdd2_fpr): Merge.
11258 (absdd2, *absdd2_fpr): Merge.
11259 (negtd2, *negtd2_fpr): Merge.
11260 (abstd2, *abstd2_fpr): Merge.
11261 * config/rs6000/e500.h: Delete file.
11262 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
11263 TARGET_FPRS by 1 and simplify.
11264 * config/rs6000/rs6000-c.c: Ditto.
11265 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
11266 TARGET_DF_SPE by 0.
11267 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
11269 * config/rs6000/rs6000.md: Ditto.
11270 (floatdidf2, *floatdidf2_fpr): Merge.
11271 (move_from_CR_gt_bit): Delete.
11272 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
11273 (E500_CR_IOR_COMPARE): Delete.
11274 (All patterns that require !TARGET_FPRS): Delete.
11275 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
11277 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11279 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
11281 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11283 * graphds.c (add_edge): Intitialize edge's attached data.
11284 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
11285 pointer parameter. Call pointed function on each edge during
11286 graph traversing. Skip traversing the edge when the function
11288 (graphds_dfs, graphds_scc): Ditto.
11289 (for_each_edge): New parameter. Pass the new parameter to callback
11291 * graphds.h (skip_edge_callback): New function pointer type.
11292 (graphds_dfs, graphds_scc): New function pointer parameter.
11293 (graphds_edge_callback, for_each_edge): New parameter.
11295 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11297 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
11298 out code checking if runtime alias check is possible to below ...
11299 Call the new function.
11300 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
11301 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
11303 2017-06-07 Marek Polacek <polacek@redhat.com>
11306 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
11307 TYPE_OVERFLOW_WRAPS check.
11309 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11311 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
11312 if versioning is required.
11313 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
11314 peeling with the check for versioning.
11316 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11318 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
11319 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
11320 Set true to new parameter if new ssa variable is defined.
11321 (vect_gen_vector_loop_niters): Refactor. Set range information
11322 for the new vector loop bound variable.
11323 (vect_do_peeling): Ditto.
11325 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11327 * tree-affine.c (ssa.h): Include header file.
11328 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
11329 has wrapping overflow behavior.
11331 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11333 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
11335 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11337 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
11338 (tree_to_aff_combination): ... here.
11340 2017-06-07 Bin Cheng <bin.cheng@arm.com>
11342 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
11343 reg_pressure model function.
11344 (ivopts_global_cost_for_size): Delete.
11345 (determine_set_costs, iv_ca_recount_cost): Call new model function
11346 ivopts_estimate_reg_pressure.
11348 2017-06-07 Tamar Christina <tamar.christina@arm.com>
11350 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
11351 expensive than udiv. Remove floating point cases from mod.
11353 2017-06-07 Tamar Christina <tamar.christina@arm.com>
11355 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
11356 Increase idiv cost.
11358 2017-06-07 Tamar Christina <tamar.christina@arm.com>
11360 * config/aarch64/aarch64.md
11361 (copysignsf3): Fix mask generation.
11363 2017-06-07 Jakub Jelinek <jakub@redhat.com>
11365 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
11367 (class dump_manager): Add register_dumps method.
11368 * dumpfile.c: Include langhooks.h.
11369 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
11370 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
11371 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
11372 (dump_manager::dump_register): Start with 512 entries instead of 32.
11373 (dump_manager::register_dumps): New method.
11374 * toplev.c (general_init): Instead of invoking register_dumps
11375 langhook, invoke register_dumps method on the dump manager.
11376 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
11379 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
11381 * doc/md.texi: Clarify the restrictions on a define_insn condition.
11382 Say that # requires an associated define_split to exist, and that
11383 the define_split must be suitable for use after register allocation.
11385 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11387 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
11388 (compute_outgoing_frequencies): Also initialize zero counts.
11389 (find_many_sub_basic_blocks): Do not produce uninitialized profile
11390 around loops; preserve more of profile when nothing changes.
11392 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
11394 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
11396 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
11397 * config/arm/arm-cpu-cdata.h: Regenerate.
11398 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
11399 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
11400 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
11401 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
11402 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
11404 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
11406 * config/arm/t-rmprofile: Likewise.
11407 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
11409 2017-06-06 David S. Miller <davem@davemloft.net>
11412 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
11413 blockage if function uses alloca.
11415 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11417 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
11418 New "uid" fields to hold pretty-print IDs of group and ref.
11419 Memory references are now identified as <group_id>:<ref_id>
11420 instead of using [random] addresses.
11421 (dump_mem_details): Simplify, no functional change.
11422 (dump_mem_ref): Simplify and make output more concise.
11423 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
11424 (find_or_create_group): Initialize group uid.
11425 (record_ref): Initialize ref uid. Improve debug output.
11426 (prune_group_by_reuse, should_issue_prefetch_p,)
11427 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
11428 (mark_nontemporal_store, determine_loop_nest_reuse):
11429 Improve debug output.
11431 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11433 * dbgcnt.def (prefetch): New debug counter.
11434 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
11435 (schedule_prefetches): Stop issueing prefetches if debug counter
11438 2017-06-06 Tom de Vries <tom@codesourcery.com>
11440 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
11441 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
11443 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11445 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
11446 Use aarch64_reg_or_zero predicate for operand 4.
11447 (aarch64_compare_and_swap<mode> define_insn_and_split):
11448 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
11449 (aarch64_store_exclusive<mode>): Likewise for operand 2.
11451 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
11453 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
11454 (arm_compute_save_core_reg_mask): This.
11455 (thumb1_compute_save_reg_mask): Rename into ...
11456 (thumb1_compute_save_core_reg_mask): This.
11457 (arm_compute_save_reg0_reg12_mask): Adapt comment.
11458 (arm_compute_frame_layout): Likewise.
11460 2017-06-06 Richard Biener <rguenther@suse.de>
11462 PR tree-optimization/80974
11463 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
11464 keep or clear leaders SSA info.
11466 2017-06-06 Tom de Vries <tom@codesourcery.com>
11468 * config/nvptx/nvptx.c (split_mode_p): New function.
11469 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
11471 2017-06-06 Tom de Vries <tom@codesourcery.com>
11473 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
11475 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11478 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
11481 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11483 * shrink-wrap.c (handle_simple_exit): Update profile.
11484 (try_shrink_wrapping): Upate profile.
11486 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11488 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
11489 (tree_guess_outgoing_edge_probabilities): New.
11490 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
11491 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
11493 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11495 * ipa-split.c (split_function): Initialize return bb profile.
11497 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11499 * profile.c (compute_branch_probabilities): Also initialize
11500 EXIT_BLOCK profile.
11502 2017-06-06 Richard Biener <rguenther@suse.de>
11504 PR tree-optimization/80928
11505 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
11506 (vect_analyze_loop_operations): Properly guard analysis for
11508 (vect_transform_loop): Likewise.
11509 (vect_analyze_loop_2): Also reset SLP type on PHIs.
11510 (vect_model_induction_cost): Do not cost for pure SLP.
11511 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
11512 of induction in inner loop vectorization.
11513 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
11514 (vect_get_and_check_slp_defs): Handle vect_induction_def.
11515 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
11517 (vect_analyze_slp_cost_1): Cost induction.
11518 (vect_detect_hybrid_slp_stmts): Handle PHIs.
11519 (vect_get_slp_vect_defs): Likewise.
11520 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
11521 (vect_transform_stmt): Handle SLP reductions.
11522 * tree-vectorizer.h (vectorizable_induction): Adjust.
11524 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
11526 * config/rs6000/rs6000.c (make_resolver_func): Update
11527 init_lowered_empty_function call.
11529 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
11531 * doc/invoke.texi: Document the -fprofile-abs-path option.
11532 * common.opt (fprofile-abs-path): New option.
11533 * gcov-io.h (gcov_write_filename): Declare.
11534 * gcov-io.c (gcov_write_filename): New function.
11535 * coverage.c (coverage_begin_function): Use gcov_write_filename.
11536 * profile.c (output_location): Likewise.
11538 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11540 * shring-wrap.c: Revert accidental commit.
11542 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
11544 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
11546 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11548 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
11550 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
11551 profile in callgraph edge.
11552 * profile-count.h (apply_probability): If THIS is 0, then result is 0
11553 (apply_scale): Likewise.
11554 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
11555 Also scale profile when inlining function with zero profile.
11556 (initialize_cfun): Update exit block profile even when it is zero.
11557 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
11558 when profile is read.
11560 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
11562 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
11563 (CLONE_*): New constants to define the processors we can generate
11564 code for with the target_clone attribute.
11565 (rs6000_clone_map): New array to identify which clone processors
11566 the current program is running on.
11567 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
11568 target_clone attribute.
11569 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
11570 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
11571 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
11572 (cpu_expand_builtin): Add support for target_clone attribute.
11573 (rs6000_valid_attribute_p): Allow "default" attribute.
11574 (get_decl_name): New debug function to simplify printing the
11575 current function name in debugging statements.
11576 (rs6000_clone_priority): New functions to support the target_clone
11577 attribute, and be able to generate code to switch between ISA 2.05
11578 through ISA 3.0 (power6 through power9).
11579 (rs6000_compare_version_priority): Likewise.
11580 (rs6000_get_function_versions_dispatcher): Likewise.
11581 (make_resolver_func): Likewise.
11582 (add_condition_to_bb): Likewise.
11583 (dispatch_function_versions): Likewise.
11584 (rs6000_generate_version_dispatcher_body): Likewise.
11585 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
11586 (fusion_gpr_load_p): Fix a spacing issue.
11587 * doc/extend.texi (Common Function Attributes): Document that the
11588 PowerPC supports the target_clone attribute.
11590 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
11592 * config/arm/arm.h: explain F symbol found in description of ARM
11593 register allocation in its legend.
11595 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11597 * config/mips/frame-header-opt.c: Include profile-count.h.
11598 * config/riscv/riscv.c: Include profile-count.h
11600 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11602 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
11604 (sm_set_flag_if_changed): Add bbs field.
11605 (execute_sm_if_changed_flag_set): Pass BBS.
11606 (execute_sm): Update.
11608 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11610 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
11613 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11615 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
11616 (peephole2): New peephole2 to emit the above.
11617 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
11619 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11621 * config/aarch64/aarch64.c (define_peephole2 above
11622 *sub_<shift>_<mode>): New peephole.
11624 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11626 * config/i386/i386.c (make_resolver_func): Update.
11627 * Makefile.in: Add profile-count.h and profile-count.o
11628 * auto-profile.c (afdo_indirect_call): Update to new API.
11629 (afdo_set_bb_count): Update.
11630 (afdo_propagate_edge): Update.
11631 (afdo_propagate_circuit): Update.
11632 (afdo_calculate_branch_prob): Update.
11633 (afdo_annotate_cfg): Update.
11634 * basic-block.h: Include profile-count.h
11635 (struct edge_def): Turn count to profile_count.
11636 (struct basic_block_def): Likewie.
11637 (REG_BR_PROB_BASE): Move to profile-count.h
11638 (RDIV): Move to profile-count.h
11639 * bb-reorder.c (max_entry_count): Turn to profile_count.
11640 (find_traces): Update.
11641 (rotate_loop):Update.
11642 (connect_traces):Update.
11643 (sanitize_hot_paths):Update.
11644 * bt-load.c (migrate_btr_defs): Update.
11645 * cfg.c (RDIV): Remove.
11646 (init_flow): Use alloc_block.
11647 (alloc_block): Uninitialize count.
11648 (unchecked_make_edge): Uninitialize count.
11649 (check_bb_profile): Update.
11650 (dump_edge_info): Update.
11651 (dump_bb_info): Update.
11652 (update_bb_profile_for_threading): Update.
11653 (scale_bbs_frequencies_int): Update.
11654 (scale_bbs_frequencies_gcov_type): Update.
11655 (scale_bbs_frequencies_profile_count): New.
11656 * cfg.h (update_bb_profile_for_threading): Update.
11657 (scale_bbs_frequencies_profile_count): Declare.
11658 * cfgbuild.c (compute_outgoing_frequencies): Update.
11659 (find_many_sub_basic_blocks): Update.
11660 * cfgcleanup.c (try_forward_edges): Update.
11661 (try_crossjump_to_edge): Update.
11662 * cfgexpand.c (expand_gimple_tailcall): Update.
11663 (construct_exit_block): Update.
11664 * cfghooks.c (verify_flow_info): Update.
11665 (dump_bb_for_graph): Update.
11666 (split_edge): Update.
11667 (make_forwarder_block): Update.
11668 (duplicate_block): Update.
11669 (account_profile_record): Update.
11670 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11671 (get_estimated_loop_iterations): Update.
11672 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11673 (single_likely_exit): Update.
11674 * cfgloopmanip.c (scale_loop_profile): Update.
11676 (set_zero_probability): Update.
11677 (lv_adjust_loop_entry_edge): Update.
11678 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
11679 (purge_dead_edges): Update.
11680 (rtl_account_profile_record): Update.
11681 * cgraph.c (cgraph_node::create): Uninitialize count.
11682 (symbol_table::create_edge): Uninitialize count.
11683 (cgraph_update_edges_for_call_stmt_node): Update.
11684 (cgraph_edge::dump_edge_flags): Update.
11685 (cgraph_node::dump): Update.
11686 (cgraph_edge::maybe_hot_p): Update.
11687 * cgraph.h: Include profile-count.h
11688 (create_clone), create_edge, create_indirect_edge): Update.
11689 (cgraph_node): Turn count to profile_count.
11690 (cgraph_edge0: Likewise.
11691 (make_speculative, clone): Update.
11692 (create_edge): Update.
11693 (init_lowered_empty_function): Update.
11694 * cgraphclones.c (cgraph_edge::clone): Update.
11695 (duplicate_thunk_for_node): Update.
11696 (cgraph_node::create_clone): Update.
11697 * cgraphunit.c (cgraph_node::analyze): Update.
11698 (cgraph_node::expand_thunk): Update.
11699 * final.c (dump_basic_block_info): Update.
11700 * gimple-streamer-in.c (input_bb): Update.
11701 * gimple-streamer-out.c (output_bb): Update.
11702 * graphite.c (print_global_statistics): Update.
11703 (print_graphite_scop_statistics): Update.
11704 * hsa-brig.c: Include basic-block.h.
11705 * hsa-dump.c: Include basic-block.h.
11706 * hsa-gen.c (T sum_slice): Update.
11707 (convert_switch_statements):Update.
11708 * hsa-regalloc.c: Include basic-block.h.
11709 * ipa-chkp.c (chkp_produce_thunks): Update.
11710 * ipa-cp.c (struct caller_statistics): Update.
11711 (init_caller_stats): Update.
11712 (gather_caller_stats): Update.
11713 (ipcp_cloning_candidate_p): Update.
11714 (good_cloning_opportunity_p): Update.
11715 (get_info_about_necessary_edges): Update.
11716 (dump_profile_updates): Update.
11717 (update_profiling_info): Update.
11718 (update_specialized_profile): Update.
11719 (perhaps_add_new_callers): Update.
11720 (decide_about_value): Update.
11721 (ipa_cp_c_finalize): Update.
11722 * ipa-devirt.c (struct odr_type_warn_count): Update.
11723 (struct decl_warn_count): Update.
11724 (struct final_warning_record): Update.
11725 (possible_polymorphic_call_targets): Update.
11726 (ipa_devirt): Update.
11727 * ipa-fnsummary.c (redirect_to_unreachable): Update.
11728 * ipa-icf.c (sem_function::merge): Update.
11729 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
11730 * ipa-inline.c (compute_uninlined_call_time): Update.
11731 (compute_inlined_call_time): Update.
11732 (want_inline_small_function_p): Update.
11733 (want_inline_self_recursive_call_p): Update.
11734 (edge_badness): Update.
11735 (lookup_recursive_calls): Update.
11736 (recursive_inlining): Update.
11737 (inline_small_functions): Update.
11738 (dump_overall_stats): Update.
11739 (dump_inline_stats): Update.
11740 * ipa-profile.c (ipa_profile_generate_summary): Update.
11741 (ipa_propagate_frequency): Update.
11742 (ipa_profile): Update.
11743 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11744 * ipa-utils.c (ipa_merge_profiles): Update.
11745 * loop-doloop.c (doloop_modify): Update.
11746 * loop-unroll.c (report_unroll): Update.
11747 (unroll_loop_runtime_iterations): Update.
11748 * lto-cgraph.c (lto_output_edge): Update.
11749 (lto_output_node): Update.
11750 (input_node): Update.
11751 (input_edge): Update.
11752 (merge_profile_summaries): Update.
11753 * lto-streamer-in.c (input_cfg): Update.
11754 * lto-streamer-out.c (output_cfg): Update.
11755 * mcf.c (create_fixup_graph): Update.
11756 (adjust_cfg_counts): Update.
11757 (sum_edge_counts): Update.
11758 * modulo-sched.c (sms_schedule): Update.
11759 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11760 * predict.c (maybe_hot_count_p): Update.
11761 (probably_never_executed): Update.
11762 (dump_prediction): Update.
11763 (combine_predictions_for_bb): Update.
11764 (propagate_freq): Update.
11765 (handle_missing_profiles): Update.
11766 (counts_to_freqs): Update.
11767 (rebuild_frequencies): Update.
11768 (force_edge_cold): Update.
11769 * predict.h: Include profile-count.h
11770 (maybe_hot_count_p, counts_to_freqs): UPdate.
11771 * print-rtl-function.c: Do not include cfg.h
11772 * print-rtl.c: Include basic-block.h
11773 * profile-count.c: New file.
11774 * profile-count.h: New file.
11775 * profile.c (is_edge_inconsistent): Update.
11776 (correct_negative_edge_counts): Update.
11777 (is_inconsistent): Update.
11778 (set_bb_counts): Update.
11779 (read_profile_edge_counts): Update.
11780 (compute_frequency_overlap): Update.
11781 (compute_branch_probabilities): Update; Initialize and deinitialize
11783 (branch_prob): Update.
11784 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
11785 (edge_gcov_count): New.
11786 (bb_gcov_count): New.
11787 * shrink-wrap.c (try_shrink_wrapping): Update.
11788 * tracer.c (better_p): Update.
11789 * trans-mem.c (expand_transaction): Update.
11790 (ipa_tm_insert_irr_call): Update.
11791 (ipa_tm_insert_gettmclone_call): Update.
11792 * tree-call-cdce.c: Update.
11793 * tree-cfg.c (gimple_duplicate_sese_region): Update.
11794 (gimple_duplicate_sese_tail): Update.
11795 (gimple_account_profile_record): Update.
11796 (execute_fixup_cfg): Update.
11797 * tree-inline.c (copy_bb): Update.
11798 (copy_edges_for_bb): Update.
11799 (initialize_cfun): Update.
11800 (freqs_to_counts): Update.
11801 (copy_cfg_body): Update.
11802 (expand_call_inline): Update.
11803 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11804 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
11805 (try_unroll_loop_completely): Update.
11806 (try_peel_loop): Update.
11807 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11808 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
11809 * tree-ssa-loop-split.c (connect_loops): Update.
11810 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11811 * tree-ssa-reassoc.c (branch_fixup): Update.
11812 * tree-ssa-tail-merge.c (replace_block_by): Update.
11813 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
11814 (compute_path_counts): Update.
11815 (update_profile): Update.
11816 (recompute_probabilities): Update.
11817 (update_joiner_offpath_counts): Update.
11818 (estimated_freqs_path): Update.
11819 (freqs_to_counts_path): Update.
11820 (clear_counts_path): Update.
11821 (ssa_fix_duplicate_block_edges): Update.
11822 (duplicate_thread_path): Update.
11823 * tree-switch-conversion.c (case_bit_test_cmp): Update.
11824 (struct switch_conv_info): Update.
11825 * tree-tailcall.c (decrease_profile): Update.
11826 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11827 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11828 * value-prof.c (check_counter): Update.
11829 (gimple_divmod_fixed_value): Update.
11830 (gimple_mod_pow2): Update.
11831 (gimple_mod_subtract): Update.
11832 (gimple_ic_transform): Update.
11833 (gimple_stringop_fixed_value): Update.
11834 * value-prof.h (gimple_ic): Update.
11836 2017-06-02 Carl Love <cel@us.ibm.com>
11838 * config/rs6000/rs6000-c: Add support for built-in functions
11839 vector double vec_doublee (vector signed int);
11840 vector double vec_doublee (vector unsigned int);
11841 vector double vec_doublee (vector float);
11842 vector double vec_doubleh (vector signed int);
11843 vector double vec_doubleh (vector unsigned int);
11844 vector double vec_doubleh (vector float);
11845 vector double vec_doublel (vector signed int);
11846 vector double vec_doublel (vector unsigned int);
11847 vector double vec_doublel (vector float);
11848 vector double vec_doubleo (vector signed int);
11849 vector double vec_doubleo (vector unsigned int);
11850 vector double vec_doubleo (vector float);.
11851 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
11852 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
11854 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
11855 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
11856 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
11858 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
11859 vec_doublel, vec_doubleh.
11860 * doc/extend.texi: Update the built-in documentation file for the
11861 new built-in functions.
11863 2017-06-02 David Malcolm <dmalcolm@redhat.com>
11866 * ipa-inline-analysis.c (free_growth_caches): Set
11867 edge_removal_hook_holder to NULL after removing it.
11869 2017-06-02 Sudakshina Das <sudi.das@arm.com>
11871 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
11872 comparision with zero.
11874 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
11875 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11876 for early expansion of vec_min and vec_max builtins.
11877 (builtin_function_type): Add min/max unsigned variants to those
11878 identified as having unsigned arguments.
11880 2017-06-02 Olivier Hainque <hainque@adacore.com>
11882 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
11884 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11886 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
11887 Use VALL_F16 iterator rather than VALL.
11889 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11891 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
11892 Emit CBNZ inside loop when doing a strong exchange and comparing
11893 against zero. Generate the CC flags after the loop.
11895 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
11897 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
11898 (dl_section_ref): New.
11899 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
11900 On AIX, append an expression to subtract the size of the
11901 section length to dl_section_ref.
11903 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
11905 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11906 for early expansion of vector absolute builtins.
11908 2017-06-02 Richard Biener <rguenther@suse.de>
11910 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
11911 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
11913 2017-06-02 Richard Biener <rguenther@suse.de>
11915 PR tree-optimization/80948
11916 * tree-tailcall.c (find_tail_calls): Track stmts to move in
11917 stmt order as well.
11919 2017-06-02 Richard Biener <rguenther@suse.de>
11921 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
11923 * tree-vect-stmts.c (process_use): Do not mark backedge defs
11924 for inductions as relevant.
11926 2017-06-02 Richard Biener <rguenther@suse.de>
11928 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
11929 (vectorizable_induction): ... this. Remove dead code.
11931 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
11933 * builtins. (expand_builtin_alloca): Remove second parameter and
11934 infer its value from the first parameter instead.
11935 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
11937 2017-06-02 Jakub Jelinek <jakub@redhat.com>
11939 PR rtl-optimization/80903
11940 * loop-doloop.c (add_test): Unshare sequence.
11942 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
11944 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
11946 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
11948 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
11950 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
11951 xlogue_layout::get_instance, logue_layout::xlogue_layout,
11952 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
11953 (xlogue_layout::get_stub_rtx): Make static.
11954 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
11955 (xlogue_layout::compute_stub_managed_regs): Rename to...
11956 (xlogue_layout::count_stub_managed_regs): ...this.
11957 (xlogue_layout::is_stub_managed_reg): New function.
11958 (xlogue_layout::m_stub_names): Rename to...
11959 (xlogue_layout::s_stub_names): ...this, make static.
11960 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
11961 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
11962 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
11963 xlogue_layout::s_stub_names): Instantiate statics.
11964 (stub_managed_regs): Remove.
11965 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
11966 (disable_call_ms2sysv_xlogues): Rename to...
11967 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
11968 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
11970 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
11971 change after reload_completed.
11972 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
11974 (ix86_expand_prologue): Likewise.
11975 (ix86_expand_epilogue): Likewise.
11976 (ix86_expand_split_stack_prologue): Likewise.
11977 (ix86_compute_frame_layout): Remove frame parameter ...
11978 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
11979 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
11981 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
11982 (ix86_frame): Move from here ...
11983 * config/i386/i386.h (ix86_frame): ... to here.
11984 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
11985 complete ix86_frame data structure instead. Remove some_ld_name.
11987 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
11989 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
11990 symbols that hold a DECL_VALUE_EXPR.
11992 2017-06-01 Martin Jambor <mjambor@suse.cz>
11994 PR tree-optimization/80898
11995 * tree-sra.c (process_subtree_disqualification): Removed.
11996 (disqualify_candidate): Do not acll
11997 process_subtree_disqualification.
11998 (subtree_mark_written_and_enqueue): New function.
11999 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
12000 RHS has been disqualified and re-queue LHS if necessary. Apart
12001 from that, ignore disqualified RHS.
12003 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12005 * config/s390/s390.c (s390_emit_epilogue): Disable early return
12006 address fetch for z10 or later.
12008 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12010 * config/arc/arc.md (tst_movb): Add guard when splitting.
12012 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12014 * config/arc/arc.c (arc_can_eliminate): Test against
12015 arc_frame_pointer_needed.
12017 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12019 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
12020 to prevent store reordering.
12021 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
12022 (type): Add block type.
12023 (stack_tie): Define special instruction to be used in
12026 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12028 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
12029 constraint. It is not valid for the pattern.
12030 (noncommutative_binary_comparison): Likewise.
12032 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12034 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
12037 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12039 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
12040 be used by the reg-alloc.
12042 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12044 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
12045 reg-alloc when having mul64 or mul32x16 instructions.
12046 (mulsidi3): Likewise.
12047 (umulsidi3): Likewise.
12048 (mulsi32x16): New pattern.
12049 (mulsi64): Likewise.
12050 (mulsidi64): Likewise.
12051 (umulsidi64): Likewise.
12052 (MUL32x16_REG): Define.
12053 (mul64_600): Use MUL32x16_REG.
12054 (mac64_600): Likewise.
12055 (umul64_600): Likewise.
12056 (umac64_600): Likewise.
12058 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12060 * config/arc/arc.md (mulsi3_700): Make it commutative.
12062 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
12064 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
12066 (*sign_extendsidi2_insn): Likewise for movstosw.
12068 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
12070 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
12071 the type of the input discriminant value. Convert the
12072 discriminant value of signedness vary.
12074 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
12076 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
12077 Add to -Wall section.
12079 2017-06-01 Richard Biener <rguenther@suse.de>
12081 PR middle-end/66313
12082 * fold-const.c (fold_plusminus_mult_expr): If the factored
12083 factor may be zero use a wrapping type for the inner operation.
12084 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
12085 and handle moved defs.
12086 (process_assignment): Properly guard the unary op case. Return a
12087 tri-state indicating that moving the stmt before the call may allow
12088 to continue. Pass through to_move.
12089 (find_tail_calls): Handle moving unrelated defs before
12092 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
12095 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
12096 splitter result in the canonical way.
12098 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
12100 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
12101 also for 32bit target. Update insn attributes.
12102 (zero-extendsidi2 splitter): Allow all registers for operand 1.
12104 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
12106 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
12107 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
12108 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
12109 (_mm_maskz_min_ss): New intrinsics.
12111 2017-05-31 Martin Liska <mliska@suse.cz>
12113 * tree-vect-loop.c (vect_create_epilog_for_reduction):
12114 Change comment style to one we normally use.
12115 (vectorizable_reduction): Likewise.
12116 (vectorizable_induction): Likewise.
12117 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12118 (vectorizable_call): Likewise.
12119 (vectorizable_simd_clone_call): Likewise.
12120 (vectorizable_conversion): Likewise.
12121 (vectorizable_assignment): Likewise.
12122 (vectorizable_shift): Likewise.
12123 (vectorizable_operation): Likewise.
12124 (vectorizable_store): Likewise.
12125 (vectorizable_load): Likewise.
12126 * tree-vectorizer.h: Likewise.
12128 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
12130 * passes.c (emergency_dump_function): New.
12131 * tree-pass.h (emergency_dump_function): Declare.
12132 * plugin.c (plugins_internal_error_function): Remove.
12133 * plugin.h (plugins_internal_error_function): Remove declaration.
12134 * toplev.c (internal_error_function): New static function. Use it...
12135 (general_init): ...here.
12137 2017-05-31 Graham Markall <graham.markall@embecosm.com>
12139 * config/arc/arc.c (arc_print_operand): Handle constant operands.
12140 (arc_rtx_costs): Add costs for new patterns.
12141 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
12142 * config/arc/predicates.md: Add _1_2_3_operand predicate.
12144 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
12146 * tree-ssa-strlen.c (get_next_strinfo): New function.
12147 (get_stridx_plus_constant): Use it.
12148 (zero_length_string): Likewise.
12149 (adjust_related_strinfos): Likewise.
12150 (adjust_last_stmt): Likewise.
12152 2017-05-31 Richard Biener <rguenther@suse.de>
12155 * config/i386/i386.c (ix86_expand_builtin): Remove assert
12156 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
12158 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
12160 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
12161 loop_vinfo argument and use of dependence distance vectors.
12162 Check instead whether the two references differ only in their
12163 initial value and assume that they have the same alignment if the
12164 difference is a multiple of the vector alignment.
12165 (vect_analyze_data_refs_alignment): Update call accordingly.
12167 2017-05-31 Martin Liska <mliska@suse.cz>
12170 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
12172 2017-05-31 Bin Cheng <bin.cheng@arm.com>
12174 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
12175 (create_intersect_range_checks): Move from ...
12176 * tree-data-ref.c (create_intersect_range_checks_index)
12177 (create_intersect_range_checks): ... to here.
12178 (create_runtime_alias_checks): New function factored from ...
12179 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
12180 here. Call above function.
12181 * tree-data-ref.h (create_runtime_alias_checks): New function.
12183 2017-05-31 Bin Cheng <bin.cheng@arm.com>
12185 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
12186 segment length for dr_b and compute it in wide_int.
12188 2017-05-31 Richard Biener <rguenther@suse.de>
12190 PR tree-optimization/80906
12191 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
12192 and pass through iv_map.
12193 (copy_bb_and_scalar_dependences): Adjust.
12194 (translate_pending_phi_nodes): Likewise.
12195 (copy_loop_close_phi_args): Handle code-generating IVs instead
12198 2017-05-30 David Malcolm <dmalcolm@redhat.com>
12200 * diagnostic-color.c (color_dict): Add "type-diff".
12201 (parse_gcc_colors): Update comment.
12202 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
12203 -fdiagnostics-show-template-tree and -fno-elide-type.
12204 (GCC_COLORS): Add type-diff to example.
12206 (-fdiagnostics-show-template-tree): New.
12207 (-fno-elide-type): New.
12208 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
12209 the pp_format_decoder callback. Call any m_format_postprocessor's
12211 (pretty_printer::pretty_printer): Initialize
12212 m_format_postprocessor.
12213 (pretty_printer::~pretty_printer): Delete any
12214 m_format_postprocessor.
12215 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
12216 (class format_postprocessor): New class.
12217 (struct pretty_printer::format_decoder): Document the new parameters.
12218 (struct pretty_printer::m_format_postprocessor): New field.
12219 * tree-diagnostic.c (default_tree_printer): Update for new
12220 bool and const char ** params.
12221 * tree-diagnostic.h (default_tree_printer): Likewise.
12223 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
12225 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
12226 (lwa_operand): Delete rs6000_gen_cell_microcode test.
12227 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
12228 rs6000_gen_cell_microcode code.
12229 (rs6000_final_prescan_insn): Delete.
12230 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
12231 "warn-cell-microcode" entries.
12232 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
12233 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
12234 throughout. Change cc_reg_not_micro_cr0_operand to
12235 cc_reg_not_cr0_operand throughout.
12236 (*extendhi<mode>2_noload): Delete.
12237 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
12238 (mwarn-cell-microcode): Delete.
12239 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
12240 -mgen-cell-microcode and -mwarn-cell-microcode.
12242 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
12245 * config/i386/constraints.md (Yd): New constraint.
12247 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
12248 and (?Yd, r) alternatives. Update insn attributes.
12249 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
12250 and (?*Yd, r) alternatives. Update insn attributes.
12251 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
12253 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
12255 * gimplify.c (gimplify_modify_expr): Don't create a
12256 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
12259 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
12261 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
12263 2017-05-30 Richard Biener <rguenther@suse.de>
12265 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
12266 and reduc_def fields.
12267 (STMT_VINFO_REDUC_TYPE): New define.
12268 (STMT_VINFO_REDUC_DEF): Likewise.
12269 (vect_force_simple_reduction): Adjust prototype.
12270 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
12271 (vect_is_simple_reduction): Remove check_reduction argument.
12272 (vect_force_simple_reduction): Adjust and set
12273 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
12274 (vectorizable_reduction): Do not re-do reduction analysis
12275 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
12276 * tree-parloops.c (gather_scalar_reductions): Adjust.
12278 2017-05-30 Richard Biener <rguenther@suse.de>
12280 PR middle-end/80901
12281 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
12284 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
12286 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
12287 Introduce unknown_misalignment parameter and remove vf.
12288 (vect_peeling_hash_get_lowest_cost):
12289 Pass unknown_misalignment parameter.
12290 (vect_enhance_data_refs_alignment):
12291 Fix unsupportable data ref treatment.
12293 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12295 * tree-vect-data-refs.c (vect_get_data_access_cost):
12296 Workaround for SLP handling.
12297 (vect_enhance_data_refs_alignment):
12298 Compute costs for doing no peeling at all, compare to the best
12299 peeling costs so far and avoid peeling if cheaper.
12301 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12303 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
12304 Return peeling info and set costs to zero for unlimited cost
12306 (vect_enhance_data_refs_alignment): Also inspect all datarefs
12307 with unknown misalignment. Compute and costs for unknown
12308 misalignment, compare them to the costs for known misalignment
12309 and choose the cheapest for peeling.
12311 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12313 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
12314 (vect_get_peeling_costs_all_drs): Create function.
12315 (vect_peeling_hash_get_lowest_cost):
12316 Use vect_get_peeling_costs_all_drs.
12317 (vect_peeling_supportable): Create function.
12318 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
12320 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
12322 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
12323 DR_HAS_NEGATIVE_STEP.
12324 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
12325 (vect_enhance_data_refs_alignment): Use.
12326 (vect_duplicate_ssa_name_ptr_info): Use.
12327 * tree-vectorizer.h (dr_misalignment): Use.
12328 (known_alignment_for_access_p): Use.
12330 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12333 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
12335 (has_section_name): New function.
12337 2017-05-30 Martin Liska <mliska@suse.cz>
12340 * auto-profile.c (get_function_decl_from_block): Fix
12343 2017-05-30 Richard Biener <rguenther@suse.de>
12345 PR middle-end/80876
12346 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
12348 2017-05-30 Martin Liska <mliska@suse.cz>
12350 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
12351 * dumpfile.h (struct dump_file_info): Remove ctors.
12353 2017-05-30 Martin Liska <mliska@suse.cz>
12355 * predict.def: Fix GNU coding style.
12357 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
12359 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
12360 Mark 'to' argument with ATTRIBUTE_UNUSED.
12362 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
12364 * config/xtensa/xtensa.c (xtensa_emit_call): Use
12365 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
12366 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
12369 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
12371 * doc/install.texi (Options specification): Restore entry of
12372 --enable-sjlj-exceptions.
12374 2017-05-27 Michael Eager <eager@eagercon.com>
12377 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
12379 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
12381 * config/microblaze/microblaze.h
12382 (FIXED_REGISTERS): Update in macro.
12383 (CALL_USED_REGISTERS): Update in macro.
12385 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
12387 * doc/install.texi: Add links to macOS binary distributions.
12389 2017-05-27 Jakub Jelinek <jakub@redhat.com>
12393 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12395 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
12397 2017-05-26 Martin Liska <mliska@suse.cz>
12399 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
12401 2017-05-26 Martin Liska <mliska@suse.cz>
12403 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
12404 always leading ';; '.
12405 (dump_bb_info): Likewise.
12406 (brief_dump_cfg): Likewise.
12407 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
12408 * dumpfile.c: Remove usage of TDF_VERBOSE.
12409 * dumpfile.h (enum dump_kind): Likewise.
12410 (dump_gimple_bb_header): Do not use TDF_COMMENT.
12411 * print-tree.c (debug_verbose): Remove.
12412 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
12413 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
12414 * tree-diagnostic.c (default_tree_printer): Replace
12415 TDF_DIAGNOSTIC with TDF_SLIM.
12417 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12419 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
12420 in parameter loop, rather than loop_vinfo.
12421 (create_intersect_range_checks): Ditto.
12422 (vect_create_cond_for_alias_checks): Update call to above functions.
12424 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12426 PR tree-optimization/80815
12427 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
12428 for merging runtime alias checks. Handle negative DR_STEPs.
12430 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12432 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
12434 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
12435 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
12436 out code pruning runtime alias checks.
12437 * tree-data-ref.c (prune_runtime_alias_test_list): New function
12438 factored out from above.
12439 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
12441 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
12443 (prune_runtime_alias_test_list): New decalaration.
12445 2017-05-26 Bin Cheng <bin.cheng@arm.com>
12447 * tree-vect-data-refs.c (compare_tree): Rename and move ...
12448 * tree-data-ref.c (data_ref_compare_tree): ... to here.
12449 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
12450 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
12451 (operator==, comp_dr_with_seg_len_pair): Ditto.
12452 (vect_prune_runtime_alias_test_list): Ditto.
12454 2017-05-26 Martin Liska <mliska@suse.cz>
12457 * params.def: Bound partial-inlining-entry-probability param.
12459 2017-05-26 Marek Polacek <polacek@redhat.com>
12462 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
12465 2017-05-26 Richard Biener <rguenther@suse.de>
12467 PR tree-optimization/80842
12468 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
12471 2017-05-26 Richard Biener <rguenther@suse.de>
12473 PR tree-optimization/80844
12474 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
12476 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
12478 * doc/md.texi (Machine Constraints): Update x86 family
12479 machine constraints section to match 'config/i386/constraints.md'.
12481 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
12483 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
12485 2017-05-25 Nathan Sidwell <nathan@acm.org>
12487 * doc/invoke.texi (--enable-languages): Update documentation.
12489 2017-05-25 Martin Liska <mliska@suse.cz>
12491 * dumpfile.c: Add TDF_FOLDING.
12492 * dumpfile.h (enum dump_kind): Likewise.
12493 * genmatch.c (dt_simplify::gen_1): Use it.
12495 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12497 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
12499 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12501 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
12502 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
12504 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12506 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
12507 * match.pd (X == C): Rewrite it here.
12508 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
12509 with_certain_nonzero_bits2): New predicates.
12510 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
12512 2017-05-24 Nathan Sidwell <nathan@acm.org>
12514 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
12517 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
12520 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
12522 * config/powerpcspe: New port. Files are copied from the rs6000
12523 port, with "rs6000" in filenames replaced by "powerpcspe".
12525 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
12527 PR rtl-optimization/80754
12528 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
12530 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
12532 * config/sparc/sparc.md (length): Return the correct value for -mflat
12533 sibcalls to match output_sibcall.
12535 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
12539 * config/rs6000/rs6000.c (struct machine_function): Add new field
12541 (rs6000_get_separate_components): Init that field, use it.
12542 (rs6000_components_for_bb): Use the field.
12544 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
12546 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
12548 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
12550 PR middle-end/80823
12551 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
12553 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12556 * config/s390/s390.c (s390_check_qrst_address): Check incoming
12557 address against address_operand predicate.
12558 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
12560 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
12562 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
12563 parameters passed indirectly.
12565 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
12567 * config/i386/i386.md (*movdi_internal): Remove SSE4
12568 alternative 18 (?r, *v). Update insn attributes.
12569 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
12570 Update insn attributes.
12571 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
12572 Update insn attributes.
12573 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
12574 alternative 1 (r, v). Remove isa attribute.
12575 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
12576 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
12577 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
12579 2017-05-23 Tom de Vries <tom@codesourcery.com>
12581 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
12584 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12586 * cgraphunit.c (symbol_table::process_new_functions): Update.
12587 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
12588 (inline_generate_summary): Rename to ...
12589 (ipa_fn_summary_generate): ... this one.
12590 (inline_read_summary): Rename to ...
12591 (ipa_fn_summary_read): ... this one.
12592 (inline_write_summary): Rename to ...
12593 (ipa_fn_summary_write): ... this one.
12594 (inline_free_summary): Rename to ...
12595 (ipa_free_fn_summary): ... this one.
12596 (pass_data_local_fn_summary, pass_local_fn_summary,
12597 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
12598 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
12599 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
12600 make_pass_ipa_fn_summary): New.
12601 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
12602 inline_write_summary, inline_free_summary): Remove.
12603 (ipa_free_fn_summary) : New.
12604 * ipa-inline.c (ipa_inline): Update.
12605 (pass_ipa_inline): Do not generate summaries.
12606 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
12608 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
12609 and add pass_ipa_fn_summary.
12610 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
12612 (make_pass_inline_parameters): Remove.
12614 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
12616 * omp-low.c (struct omp_context): Remove "default_kind" member.
12619 * omp-offload.c (execute_oacc_device_lower): Remove the
12620 parallelism dimensions function attributes for unparallelized
12621 OpenACC kernels constructs.
12623 2017-05-23 Martin Liska <mliska@suse.cz>
12625 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
12627 (cgraph_edge::make_speculative): Likewise.
12628 (cgraph_edge::resolve_speculation): Likewise.
12629 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
12630 (cgraph_node::dump): Likewise.
12631 * cgraph.h: Likewise.
12632 * cgraphunit.c (analyze_functions): Likewise.
12633 (symbol_table::compile): Likewise.
12634 * ipa-cp.c (print_all_lattices): Likewise.
12635 (determine_versionability): Likewise.
12636 (initialize_node_lattices): Likewise.
12637 (ipcp_verify_propagated_values): Likewise.
12638 (estimate_local_effects): Likewise.
12639 (update_profiling_info): Likewise.
12640 (create_specialized_node): Likewise.
12641 (perhaps_add_new_callers): Likewise.
12642 (decide_about_value): Likewise.
12643 (decide_whether_version_node): Likewise.
12644 (identify_dead_nodes): Likewise.
12645 (ipcp_store_bits_results): Likewise.
12646 * ipa-devirt.c (dump_targets): Likewise.
12647 (ipa_devirt): Likewise.
12648 * ipa-icf.c (sem_item::dump): Likewise.
12649 (sem_function::equals): Likewise.
12650 (sem_variable::equals): Likewise.
12651 (sem_item_optimizer::read_section): Likewise.
12652 (sem_item_optimizer::execute): Likewise.
12653 (congruence_class::dump): Likewise.
12654 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
12655 (dump_inline_summary): Likewise.
12656 (estimate_node_size_and_time): Likewise.
12657 (inline_analyze_function): Likewise.
12658 * ipa-inline-transform.c (inline_call): Likewise.
12659 * ipa-inline.c (report_inline_failed_reason): Likewise.
12660 (want_early_inline_function_p): Likewise.
12661 (edge_badness): Likewise.
12662 (update_edge_key): Likewise.
12663 (inline_small_functions): Likewise.
12664 * ipa-profile.c (ipa_profile): Likewise.
12665 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
12666 (ipa_make_edge_direct_to_target): Likewise.
12667 (remove_described_reference): Likewise.
12668 (ipa_impossible_devirt_target): Likewise.
12669 (propagate_controlled_uses): Likewise.
12670 (ipa_print_node_params): Likewise.
12671 (ipcp_transform_function): Likewise.
12672 * ipa-pure-const.c (pure_const_read_summary): Likewise.
12673 (propagate_pure_const): Likewise.
12674 * ipa-reference.c (generate_summary): Likewise.
12675 (read_write_all_from_decl): Likewise.
12676 (propagate): Likewise.
12677 (ipa_reference_read_optimization_summary): Likewise.
12678 * ipa-utils.c (ipa_merge_profiles): Likewise.
12679 * ipa.c (walk_polymorphic_call_targets): Likewise.
12680 (symbol_table::remove_unreachable_nodes): Likewise.
12681 (ipa_single_use): Likewise.
12682 * passes.c (execute_todo): Likewise.
12683 * predict.c (drop_profile): Likewise.
12684 * symtab.c (symtab_node::get_dump_name): New function.
12685 (symtab_node::dump_name): Likewise.
12686 (symtab_node::dump_asm_name): Likewise.
12687 (symtab_node::dump_references): Likewise.
12688 (symtab_node::dump_referring): Likewise.
12689 (symtab_node::dump_base): Likewise.
12690 (symtab_node::debug_symtab): Likewise.
12691 * tree-sra.c (convert_callers_for_node): Likewise.
12692 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
12693 * value-prof.c (init_node_map): Likewise.
12695 2017-05-23 Martin Liska <mliska@suse.cz>
12697 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
12698 and symtab_node::debug_symtab to symbol_table::debug.
12699 * cgraphunit.c (analyze_functions): Use the renamed function.
12700 (symbol_table::compile): Likewise.
12701 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
12702 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
12703 * passes.c (execute_todo): Likewise.
12704 * symtab.c (symbol_table::dump): New function.
12705 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
12707 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12709 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
12710 that nonconst implies exec.
12712 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12714 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
12715 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
12716 (inline_edge_summary_vec): Turn into ...
12717 (ipa_call_summaries): ... this one.
12718 (redirect_to_unreachable, edge_set_predicate,
12719 evaluate_properties_for_edge, inline_summary_alloc,
12720 reset_ipa_call_summary, reset_inline_summary,
12721 inline_summary_t::duplicate): Update.
12722 (inline_edge_duplication_hook): Turn to ...
12723 (ipa_call_summary_t::duplicate): ... this one.
12724 (inline_edge_removal_hook): Turn to ...
12725 (ipa_call_summary_t::remove): ... this one.
12726 (dump_inline_edge_summary): Turn to ...
12727 (dump_ipa_call_summary): ... this one.
12728 (estimate_function_body_sizes): Update.
12729 (inline_update_callee_summaries): Update.
12730 (remap_edge_change_prob): Update.
12731 (remap_edge_summaries): Update.
12732 (inline_merge_summary): Update.
12733 (do_estimate_edge_time): Update.
12734 (inline_generate_summary): Update.
12735 (inline_read_section): Update.
12736 (inline_read_summary): Update.
12737 (inline_free_summary): Update.
12738 * ipa-inline.c (can_inline_edge_p): Update.
12739 (compute_inlined_call_time): Update.
12740 (want_inline_small_function_p): Update.
12741 (edge_badness): Update.
12742 (early_inliner): Update.
12743 * ipa-inline.h (inline_edge_summary): Turn to ...
12744 (ipa_call_summary): ... this one.
12745 (ipa_call_summary_t): New class.
12746 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
12747 (ipa_call_summaries): New.
12748 (inline_edge_summary): Remove.
12749 (estimate_edge_growth): Update.
12750 * ipa-profile.c (ipa_propagate_frequency_1): Update.
12751 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12752 * ipa-split.c (execute_split_functions): Update.
12753 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
12755 2017-05-23 Tom de Vries <tom@codesourcery.com>
12757 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
12758 attributes): Document rdrand effective target.
12760 2017-05-23 Tom de Vries <tom@codesourcery.com>
12762 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
12763 attributes): Sort alphabetically.
12765 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
12767 * config/avr/genmultilib.awk: Use gsub instead of gensub.
12769 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
12772 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
12773 V2DF/V2DI splat into two separate patterns, one that handles
12774 registers, and the other that only handles memory. Drop support
12775 for splatting from a GPR on ISA 2.07 and then splitting the
12776 splat into direct move and splat.
12777 (vsx_splat_<mode>_reg): Likewise.
12778 (vsx_splat_<mode>_mem): Likewise.
12780 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
12782 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
12784 2017-05-22 Jakub Jelinek <jakub@redhat.com>
12786 PR middle-end/80809
12787 * omp-low.c (finish_taskreg_remap): New function.
12788 (finish_taskreg_scan): If unit size of ctx->record_type
12789 is non-constant, unshare the size expression and replace
12790 decls in it with possible outer var refs.
12792 PR middle-end/80809
12793 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
12794 GOVD_SHARED rather than GOVD_PRIVATE with it.
12795 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
12796 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
12798 PR middle-end/80853
12799 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
12800 as last argument to build_outer_var_ref for pointer bases of array
12801 section reductions.
12803 2017-05-19 Martin Sebor <msebor@redhat.com>
12805 * print-tree.c (print_node): Print DECL_READ_P flag.
12807 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12809 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
12810 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
12811 * cgraph.c: Likewise.
12812 * cgraphunit.c: Likewise.
12813 * gengtype.c: Likewise.
12814 * ipa-cp.c: Likewise.
12815 * ipa-devirt.c: Likewise.
12816 * ipa-icf.c: Likewise.
12817 * ipa-predicate.c: Likewise.
12818 * ipa-profile.c: Likewise.
12819 * ipa-prop.c: Likewise.
12820 * ipa-split.c: Likewise.
12822 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
12823 edge_predicate_pool, dump_inline_hints,
12824 inline_summary::account_size_time, redirect_to_unreachable,
12825 edge_set_predicate, set_hint_predicate,
12826 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
12827 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
12828 inline_summary_t::remove, remap_hint_predicate_after_duplication,
12829 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
12830 ipa_call_summary_t::remove, initialize_growth_caches,
12831 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
12832 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
12833 mark_modified, unmodified_parm_1, unmodified_parm,
12834 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
12835 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
12836 compute_bb_predicates, will_be_nonconstant_expr_predicate,
12837 will_be_nonconstant_predicate, record_modified_bb_info,
12838 get_minimal_bb, record_modified, param_change_prob,
12839 phi_result_unknown_predicate, predicate_for_phi_result,
12840 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
12841 estimate_function_body_sizes, compute_inline_parameters,
12842 compute_inline_parameters_for_curren, pass_data_inline_parameters,
12843 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
12844 inline_update_callee_summaries, remap_edge_change_prob,
12845 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
12846 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
12847 inline_analyze_function, inline_summary_t::insert,
12848 inline_generate_summary, read_ipa_call_summary, inline_read_section,
12849 inline_read_summary, write_ipa_call_summary, inline_write_summary,
12850 inline_free_summary): Move to ipa-fnsummary.h
12851 (predicate_t): Remove.
12852 * ipa-fnsummary.c: New file.
12853 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
12854 (enum inline_hints_vals, inline_hints, agg_position_info,
12855 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
12856 inline_summaries, ipa_call_summary, ipa_call_summary_t,
12857 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
12858 dump_inline_summary, dump_inline_hints, inline_generate_summary,
12859 inline_read_summary, inline_write_summary, inline_free_summary,
12860 inline_analyze_function, initialize_inline_failed,
12861 inline_merge_summary, inline_update_overall_summary,
12862 compute_inline_parameters): Move to ipa-fnsummary.h
12863 * ipa-fnsummary.h: New file.
12864 * ipa-inline-transform.h: Include ipa-inline.h.
12865 * ipa-inline.c: LIkewise.
12867 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12869 * ipa-inline.c (edge_badness): Use inlined_time instead of
12870 inline_summaries->get.
12872 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12874 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
12876 2017-05-22 Nathan Sidwell <nathan@acm.org>
12878 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
12879 (fdump-lang): Document 'raw' option.
12880 * dumpfile.h (TDI_tu): Delete.
12881 * dumpfile.c (dump_files): Remove translation-unit.
12882 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
12884 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
12886 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
12887 command option from $(AWK) call.
12888 * config/avr/genmultilib.awk: Simplify and rewrite so that it
12889 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
12890 [FORMAT]: Remove handling of variable.
12891 * config/avr/t-multilib: Regenerate.
12893 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12895 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
12897 (dump_inline_summary): Do not print self_time.
12898 (estimate_function_body_sizes): Do not set self_time.
12899 (compute_inline_parameters): Likewise.
12900 (inline_read_section, inline_write_summary): Do not stream self_time.
12901 * ipa-inline.h (inline_summary): Drop self_time.
12903 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12905 * ipa-inline-analysis.c (account_size_time): Rename to ...
12906 (inline_summary::account_size_time): ... this one.
12907 (reset_ipa_call_summary): Turn to ...
12908 (ipa_call_summary::reset): ... this one.
12909 (reset_inline_summary): Turn to ...
12910 (inline_summary::reset): ... this one.
12911 (inline_summary_t::remove): Update.
12912 (inline_summary_t::duplicate): Update.
12913 (ipa_call_summary_t::remove): Update.
12914 (dump_inline_summary): Update.
12915 (estimate_function_body_sizes): Update.
12916 (compute_inline_parameters): Update.
12917 (estimate_node_size_and_time): Update.
12918 (inline_merge_summary): Update.
12919 (inline_update_overall_summary): Update.
12920 (inline_read_section): Update.
12921 (inline_write_summary): Update.
12922 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
12923 add account_size_time and reset member functions.
12924 (ipa_call_summary): Add reset function.
12925 * ipa-predicate.h (predicate::operator &): Constify.
12927 2017-05-22 Richard Biener <rguenther@suse.de>
12929 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
12931 2017-05-19 Jason Merrill <jason@redhat.com>
12933 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
12935 2017-05-19 Marek Polacek <polacek@redhat.com>
12938 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
12939 TYPE_OVERFLOW_WRAPS checks.
12941 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
12943 * tree-core.h (enum omp_clause_default_kind): Add
12944 "OMP_CLAUSE_DEFAULT_PRESENT".
12945 * tree-pretty-print.c (dump_omp_clause): Handle it.
12946 * gimplify.c (enum gimplify_omp_var_data): Add
12947 "GOVD_MAP_FORCE_PRESENT".
12948 (gimplify_adjust_omp_clauses_1): Map it to
12949 "GOMP_MAP_FORCE_PRESENT".
12950 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
12952 * gimplify.c (oacc_default_clause): Clarify.
12954 2017-05-19 Nathan Sidwell <nathan@acm.org>
12956 LANG_HOOK_REGISTER_DUMPS
12957 * toplev.c (general_init): Call register dump lang hook.
12958 * doc/invoke.texi: Document -fdump-lang option family.
12959 * dumpfile.c (dump_files): Remove class dump here.
12960 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
12961 * dumpfile.h (tree_dump_index): Remove TDI_class.
12962 * langhooks-def.h (lhd_register_dumps): Declare.
12963 (LANG_HOOKS_REGISTER_DUMPS): Define.
12964 (LANG_HOOKS_INITIALIZER): Add it.
12965 * langhooks.c (lhd_register_dumps): Define.
12966 * langhooks.h (struct lang_hooks): Add register_dumps.
12968 2017-05-19 Nathan Sidwell <nathan@acm.org>
12970 * context.h (context::set_passes): New.
12971 * context.c (context::context): Do not create pass manager.
12972 * toplev.c (general_init): Create pass manager here.
12974 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
12976 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
12977 use this splitter if two add or or instructions would also work for
12978 the constant we want to generate.
12980 2017-05-19 Richard Biener <rguenther@suse.de>
12983 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
12984 predicate evaluation.
12986 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12988 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
12990 * ipa-inline.c (want_inline_small_function_p): Do not cast to
12993 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12995 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
12996 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
12997 (inline_edge_summary_vec): Turn into ...
12998 (ipa_call_summaries): ... this one.
12999 (redirect_to_unreachable, edge_set_predicate,
13000 evaluate_properties_for_edge, inline_summary_alloc,
13001 reset_ipa_call_summary, reset_inline_summary,
13002 inline_summary_t::duplicate): Update.
13003 (inline_edge_duplication_hook): Turn to ...
13004 (ipa_call_summary_t::duplicate): ... this one.
13005 (inline_edge_removal_hook): Turn to ...
13006 (ipa_call_summary_t::remove): ... this one.
13007 (dump_inline_edge_summary): Turn to ...
13008 (dump_ipa_call_summary): ... this one.
13009 (estimate_function_body_sizes): Update.
13010 (inline_update_callee_summaries): Update.
13011 (remap_edge_change_prob): Update.
13012 (remap_edge_summaries): Update.
13013 (inline_merge_summary): Update.
13014 (do_estimate_edge_time): Update.
13015 (inline_generate_summary): Update.
13016 (inline_read_section): Update.
13017 (inline_read_summary): Update.
13018 (inline_free_summary): Update.
13019 * ipa-inline.c (can_inline_edge_p): Update.
13020 (compute_inlined_call_time): Update.
13021 (want_inline_small_function_p): Update.
13022 (edge_badness): Update.
13023 (early_inliner): Update.
13024 * ipa-inline.h (inline_edge_summary): Turn to ...
13025 (ipa_call_summary): ... this one.
13026 (ipa_call_summary_t): New class.
13027 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
13028 (ipa_call_summaries): New.
13029 (inline_edge_summary): Remove.
13030 (estimate_edge_growth): Update.
13031 * ipa-profile.c (ipa_propagate_frequency_1): Update.
13032 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13033 * ipa-split.c (execute_split_functions): Update.
13034 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
13036 2017-05-19 Richard Biener <rguenther@suse.de>
13038 PR middle-end/80764
13039 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
13041 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
13043 * config/rs6000/rs6000.c (struct machine_function): Add field
13044 fpr_is_wrapped_separately.
13045 (rs6000_get_separate_components): Use 64 components. Handle the
13046 new FPR components.
13047 (rs6000_components_for_bb): Handle the FPR components.
13048 (rs6000_emit_prologue_components): Handle the FPR components.
13049 (rs6000_emit_epilogue_components): Handle the FPR components.
13050 (rs6000_set_handled_components): Handle the FPR components.
13051 (rs6000_emit_prologue): Don't output prologue code for those FPRs
13052 that are already separately shrink-wrapped.
13053 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
13054 that are already separately shrink-wrapped.
13056 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
13059 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
13062 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
13063 (define_peephole2 for Altivec d-form load): Add peepholes to catch
13064 cases where the register allocator uses a move and an offsettable
13065 memory operation to/from a FPR register on ISA 2.06/2.07.
13066 (define_peephole2 for Altivec d-form store): Likewise.
13068 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
13071 * config/i386/mmx.md (*mov<mode>_internal): Enable
13072 alternatives 11, 12, 13 and 14 also for 32bit targets.
13073 Remove alternatives 15, 16, 17 and 18.
13074 * config/i386/sse.md (vec_concatv2di): Change
13075 alternative (!x, *y) to (x, ?!*Yn).
13077 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
13079 * dumpfile.h (enum dump_kind): Remove stray comma.
13081 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
13083 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
13084 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
13085 predicate::num_conditions
13086 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
13087 (CHANGED): turn into predicate::changed.
13088 (agg_position_info): Move to ipa-predicate.h
13089 (add_condition, predicate::add_clause, predicate::operator &=,
13090 predicate::or_with, predicate::evaluate, predicate::probability,
13091 dump_condition, dump_clause, predicate::dump,
13092 predicate::remap_after_duplication, predicate::remap_after_inlining,
13093 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
13094 (evaluate_conditions_for_known_args): Update.
13095 (set_cond_stmt_execution_predicate): Update.
13096 * ipa-inline.h: Include ipa-predicate.h
13097 (condition, inline_param_summary, conditions, agg_position_info,
13098 predicate): Move to ipa-predicate.h
13099 * ipa-predicate.c: New file.
13100 * ipa-predicate.h: New file.
13102 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
13104 * final.c (leaf_function_p): Check we are not in a sequence.
13106 2017-05-18 Martin Liska <mliska@suse.cz>
13108 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
13109 * dumpfile.c (dump_register): Use new enum dump_kind.
13110 (get_dump_file_name): Likewise.
13111 (dump_enable_all): Likewise.
13112 (dump_switch_p_1): Likewise.
13113 (enable_rtl_dump_file): Remove usage of TDF_RTL.
13114 * dumpfile.h (enum dump_kind): New enum type.
13115 (struct dump_file_info): Create constructor and
13116 format fields and comments.
13117 * passes.c (pass_manager::register_one_dump_file):
13119 * statistics.c (statistics_early_init): Likewise.
13120 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
13121 TDF_TREE with TDF_SLIM.
13122 (gather_memory_references_ref): Likewise.
13124 2017-05-18 Martin Liska <mliska@suse.cz>
13126 * vec.h (struct vnull): Use it.
13128 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
13130 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
13131 (true_predicate, false_predicate, true_predicate_p,
13132 false_predicate_p): Remove.
13133 (single_cond_predicate, not_inlined_predicate): Turn to member function
13135 (add_condition): Update.
13136 (add_clause): Turn to...
13137 (predicate::add_clause): ... this one; update; allow passing NULL
13139 (and_predicates): Turn to ...
13140 (predicate::operator &=): ... this one.
13141 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
13142 (or_predicates): Turn to ...
13143 (predicate::or_with): ... this one.
13144 (evaluate_predicate): Turn to ...
13145 (predicate::evaluate): ... this one.
13146 (predicate_probability): Turn to ...
13147 (predicate::probability): ... this one.
13148 (dump_condition): Update.
13149 (dump_predicate): Turn to ...
13150 (predicate::dump): ... this one.
13151 (account_size_time): Update.
13152 (edge_set_predicate): Update.
13153 (set_hint_predicate): UPdate.
13154 (evaluate_conditions_for_known_args): Update.
13155 (evaluate_properties_for_edge): Update.
13156 (remap_predicate_after_duplication): Turn to...
13157 (predicate::remap_after_duplication): ... this one.
13158 (remap_hint_predicate_after_duplication): Update.
13159 (inline_summary_t::duplicate): UPdate.
13160 (dump_inline_edge_summary): Update.
13161 (dump_inline_summary): Update.
13162 (set_cond_stmt_execution_predicate): Update.
13163 (set_switch_stmt_execution_predicate): Update.
13164 (compute_bb_predicates): Update.
13165 (will_be_nonconstant_expr_predicate): Update.
13166 (will_be_nonconstant_predicate): Update.
13167 (phi_result_unknown_predicate): Update.
13168 (predicate_for_phi_result): Update.
13169 (array_index_predicate): Update.
13170 (estimate_function_body_sizes): Update.
13171 (estimate_node_size_and_time): Update.
13172 (estimate_ipcp_clone_size_and_time): Update.
13173 (remap_predicate): Rename to ...
13174 (predicate::remap_after_inlining): ... this one.
13175 (remap_hint_predicate): Update.
13176 (inline_merge_summary): Update.
13177 (inline_update_overall_summary): Update.
13178 (estimate_size_after_inlining): Update.
13179 (read_predicate): Rename to ...
13180 (predicate::stream_in): ... this one.
13181 (read_inline_edge_summary): Update.
13182 (write_predicate): Rename to ...
13183 (predicate::stream_out): ... this one.
13184 (write_inline_edge_summary): Update.
13185 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
13186 (clause_t): Turn to uint32_t
13187 (predicate): Turn to class; implement constructor and operators
13189 (size_time_entry): Update.
13190 (inline_summary): Update.
13191 (inline_edge_summary): Update.
13193 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
13195 * fold-const.c (fold_binary_loc): Move transformation...
13196 * match.pd (C - X CMP X): ... here.
13198 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
13200 * config/sparc/sparc.c (sparc_option_override): Set function
13201 alignment for -mcpu=niagara7 to 64 to match the I$ line.
13202 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
13204 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
13206 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
13208 2017-05-18 Marek Polacek <polacek@redhat.com>
13211 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
13212 (pass_ubsan::execute): Call gimple_assign_single_p instead of
13213 gimple_assign_load_p.
13215 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
13217 PR middle-end/80692
13218 * real.c (do_compare): Give decimal_do_compare preference over
13219 comparing just the signs.
13221 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
13223 * doc/md.texi (Canonicalization of Instructions): Describe the
13224 canonical form of instructions that inherently set a condition
13227 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
13229 PR middle-end/80775
13230 * tree-cfg.c: Move deletion of unreachable case statements to after
13231 the merging of consecutive case labels.
13233 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
13235 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
13236 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
13237 restoring of callee-saved registers.
13239 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
13241 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
13242 * config/visium/visium.c (single_set_and_flags): Likewise.
13243 * config/visium/visium.md (Substitutions): Likewise.
13245 2017-05-17 Martin Liska <mliska@suse.cz>
13247 * cfg.c: Introduce dump_flags_t type and
13248 use it instead of int type.
13250 * cfghooks.c: Likewise.
13251 * cfghooks.h (struct cfg_hooks): Likewise.
13252 * cfgrtl.c: Likewise.
13253 * cfgrtl.h: Likewise.
13254 * cgraph.c (cgraph_node::get_body): Likewise.
13255 * coretypes.h: Likewise.
13256 * domwalk.c: Likewise.
13257 * domwalk.h: Likewise.
13258 * dumpfile.c (struct dump_option_value_info): Likewise.
13259 (dump_enable_all): Likewise.
13260 (dump_switch_p_1): Likewise.
13261 (opt_info_switch_p): Likewise.
13262 * dumpfile.h (enum tree_dump_index): Likewise.
13263 (struct dump_file_info): Likewise.
13264 * genemit.c: Likewise.
13265 * generic-match-head.c: Likewise.
13266 * gengtype.c (open_base_files): Likewise.
13267 * gimple-pretty-print.c: Likewise.
13268 * gimple-pretty-print.h: Likewise.
13269 * graph.c (print_graph_cfg): Likewise.
13270 * graphite-scop-detection.c (dot_all_sese): Likewise.
13271 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
13272 * loop-unroll.c (report_unroll): Likewise.
13273 * passes.c (pass_manager::register_one_dump_file): Likewise.
13274 * print-tree.c: Likewise.
13275 * statistics.c: Likewise.
13276 * tree-cfg.c: Likewise.
13277 * tree-cfg.h: Likewise.
13278 * tree-dfa.c: Likewise.
13279 * tree-dfa.h: Likewise.
13280 * tree-dump.c (dump_function): Likewise.
13281 * tree-dump.h (struct dump_info): Likewise.
13282 * tree-pretty-print.c: Likewise.
13283 * tree-pretty-print.h: Likewise.
13284 * tree-ssa-live.c: Likewise.
13285 * tree-ssa-live.h: Likewise.
13286 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
13287 * tree-vect-loop.c: Likewise.
13288 * tree-vect-slp.c: Likewise.
13290 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
13291 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13293 PR tree-optimization/80457
13294 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
13295 of all arguments to a statement as scalar_to_vec operations.
13296 (vectorizable_call): Adjust call to vect_model_simple_cost for
13298 (vectorizable_conversion): Likewise.
13299 (vectorizable_assignment): Likewise.
13300 (vectorizable_shift): Likewise.
13301 (vectorizable_operation): Likewise.
13302 (vectorizable_comparison): Likewise.
13303 (vect_is_simple_cond): Record the def types for operands.
13304 (vectorizable_condition): Likewise, call vect_model_simple_cost.
13305 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
13306 for statement argument count.
13308 2017-05-16 Carl Love <cel@us.ibm.com>
13310 * config/rs6000/rs6000-c: Add support for built-in functions
13311 vector unsigned long long vec_bperm (vector unsigned long long,
13312 vector unsigned char)
13313 vector signed long long vec_mule (vector signed int,
13315 vector unsigned long long vec_mule (vector unsigned int,
13316 vector unsigned int)
13317 vector signed long long vec_mulo (vector signed int,
13319 vector unsigned long long vec_mulo (vector unsigned int,
13320 vector unsigned int)
13321 vector signed char vec_sldw (vector signed char,
13322 vector signed char,
13324 vector unsigned char vec_sldw (vector unsigned char,
13325 vector unsigned char,
13327 vector signed short vec_sldw (vector signed short,
13328 vector signed short,
13330 vector unsigned short vec_sldw (vector unsigned short,
13331 vector unsigned short,
13333 vector signed int vec_sldw (vector signed int,
13336 vector unsigned int vec_sldw (vector unsigned int,
13337 vector unsigned int,
13339 vector signed long long vec_sldw (vector signed long long,
13340 vector signed long long,
13342 vector unsigned long long vec_sldw (vector unsigned long long,
13343 vector unsigned long long,
13345 * config/rs6000/rs6000-c: Add support for built-in functions
13346 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
13347 * config/rs6000/altivec.h: Add defintion for vec_sldw.
13348 * doc/extend.texi: Update the built-in documentation for the
13349 new built-in functions.
13351 2017-05-16 Marek Polacek <polacek@redhat.com>
13355 * tree.c (save_expr): Don't fold the expression.
13357 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
13359 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
13360 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
13361 and (?*y,m). Update insn attributes.
13363 2017-05-16 Martin Liska <mliska@suse.cz>
13365 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
13366 flags argument of print_gimple_stmt, print_gimple_expr,
13367 print_generic_stmt and print_generic_expr.
13368 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
13369 * coretypes.h: Likewise.
13370 * except.c (dump_eh_tree): Likewise.
13371 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
13372 * gimple-pretty-print.h: Likewise.
13373 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
13374 (backprop::push_to_worklist): Likewise.
13375 (backprop::pop_from_worklist): Likewise.
13376 (backprop::process_use): Likewise.
13377 (backprop::intersect_uses): Likewise.
13378 (note_replacement): Likewise.
13379 * gimple-ssa-store-merging.c
13380 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
13381 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
13382 (pass_store_merging::execute): Likewise.
13383 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
13384 (ssa_base_cand_dump_callback): Likewise.
13385 (dump_incr_vec): Likewise.
13386 (replace_refs): Likewise.
13387 (replace_mult_candidate): Likewise.
13388 (create_add_on_incoming_edge): Likewise.
13389 (create_phi_basis): Likewise.
13390 (insert_initializers): Likewise.
13391 (all_phi_incrs_profitable): Likewise.
13392 (introduce_cast_before_cand): Likewise.
13393 (replace_one_candidate): Likewise.
13394 * gimplify.c (gimplify_expr): Likewise.
13395 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
13396 (set_rename): Likewise.
13397 (rename_uses): Likewise.
13398 (copy_loop_phi_nodes): Likewise.
13399 (add_close_phis_to_merge_points): Likewise.
13400 (copy_loop_close_phi_args): Likewise.
13401 (copy_cond_phi_args): Likewise.
13402 (graphite_copy_stmts_from_block): Likewise.
13403 (translate_pending_phi_nodes): Likewise.
13404 * graphite-poly.c (print_pdr): Likewise.
13405 (dump_gbb_cases): Likewise.
13406 (dump_gbb_conditions): Likewise.
13407 (print_scop_params): Likewise.
13408 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
13409 (build_cross_bb_scalars_use): Likewise.
13410 (gather_bbs::before_dom_children): Likewise.
13411 * hsa-dump.c (dump_hsa_immed): Likewise.
13412 * ipa-cp.c (print_ipcp_constant_value): Likewise.
13413 (get_replacement_map): Likewise.
13414 * ipa-inline-analysis.c (dump_condition): Likewise.
13415 (estimate_function_body_sizes): Likewise.
13416 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
13417 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
13418 * ipa-prop.c (ipa_dump_param): Likewise.
13419 (ipa_print_node_jump_functions_for_edge): Likewise.
13420 (ipa_modify_call_arguments): Likewise.
13421 (ipa_modify_expr): Likewise.
13422 (ipa_dump_param_adjustments): Likewise.
13423 (ipa_dump_agg_replacement_values): Likewise.
13424 (ipcp_modif_dom_walker::before_dom_children): Likewise.
13425 * ipa-pure-const.c (check_stmt): Likewise.
13426 (pass_nothrow::execute): Likewise.
13427 * ipa-split.c (execute_split_functions): Likewise.
13428 * omp-offload.c (dump_oacc_loop_part): Likewise.
13429 (dump_oacc_loop): Likewise.
13430 * trans-mem.c (tm_log_emit): Likewise.
13431 (tm_memopt_accumulate_memops): Likewise.
13432 (dump_tm_memopt_set): Likewise.
13433 (dump_tm_memopt_transform): Likewise.
13434 * tree-cfg.c (gimple_verify_flow_info): Likewise.
13435 (print_loop): Likewise.
13436 * tree-chkp-opt.c (chkp_print_addr): Likewise.
13437 (chkp_gather_checks_info): Likewise.
13438 (chkp_get_check_result): Likewise.
13439 (chkp_remove_check_if_pass): Likewise.
13440 (chkp_use_outer_bounds_if_possible): Likewise.
13441 (chkp_reduce_bounds_lifetime): Likewise.
13442 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
13443 (chkp_mark_completed_bounds): Likewise.
13444 (chkp_register_incomplete_bounds): Likewise.
13445 (chkp_mark_invalid_bounds): Likewise.
13446 (chkp_maybe_copy_and_register_bounds): Likewise.
13447 (chkp_build_returned_bound): Likewise.
13448 (chkp_get_bound_for_parm): Likewise.
13449 (chkp_build_bndldx): Likewise.
13450 (chkp_get_bounds_by_definition): Likewise.
13451 (chkp_generate_extern_var_bounds): Likewise.
13452 (chkp_get_bounds_for_decl_addr): Likewise.
13453 * tree-chrec.c (chrec_apply): Likewise.
13454 * tree-data-ref.c (dump_data_reference): Likewise.
13455 (dump_subscript): Likewise.
13456 (dump_data_dependence_relation): Likewise.
13457 (analyze_overlapping_iterations): Likewise.
13458 * tree-inline.c (expand_call_inline): Likewise.
13459 (tree_function_versioning): Likewise.
13460 * tree-into-ssa.c (dump_defs_stack): Likewise.
13461 (dump_currdefs): Likewise.
13462 (dump_names_replaced_by): Likewise.
13463 (dump_update_ssa): Likewise.
13464 (update_ssa): Likewise.
13465 * tree-object-size.c (pass_object_sizes::execute): Likewise.
13466 * tree-parloops.c (build_new_reduction): Likewise.
13467 (try_create_reduction_list): Likewise.
13468 (ref_conflicts_with_region): Likewise.
13469 (oacc_entry_exit_ok_1): Likewise.
13470 (oacc_entry_exit_single_gang): Likewise.
13471 * tree-pretty-print.h: Likewise.
13472 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
13473 (get_scalar_evolution): Likewise.
13474 (add_to_evolution): Likewise.
13475 (get_loop_exit_condition): Likewise.
13476 (analyze_evolution_in_loop): Likewise.
13477 (analyze_initial_condition): Likewise.
13478 (analyze_scalar_evolution): Likewise.
13479 (instantiate_scev): Likewise.
13480 (number_of_latch_executions): Likewise.
13481 (gather_chrec_stats): Likewise.
13482 (final_value_replacement_loop): Likewise.
13483 (scev_const_prop): Likewise.
13484 * tree-sra.c (dump_access): Likewise.
13485 (disqualify_candidate): Likewise.
13486 (create_access): Likewise.
13487 (reject): Likewise.
13488 (maybe_add_sra_candidate): Likewise.
13489 (create_access_replacement): Likewise.
13490 (analyze_access_subtree): Likewise.
13491 (analyze_all_variable_accesses): Likewise.
13492 (sra_modify_assign): Likewise.
13493 (initialize_constant_pool_replacements): Likewise.
13494 (find_param_candidates): Likewise.
13495 (decide_one_param_reduction): Likewise.
13496 (replace_removed_params_ssa_names): Likewise.
13497 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
13498 * tree-ssa-copy.c (dump_copy_of): Likewise.
13499 (copy_prop_visit_cond_stmt): Likewise.
13500 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
13501 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
13502 (record_equivalences_from_stmt): Likewise.
13503 * tree-ssa-dse.c (compute_trims): Likewise.
13504 (delete_dead_call): Likewise.
13505 (delete_dead_assignment): Likewise.
13506 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
13507 (forward_propagate_into_cond): Likewise.
13508 (pass_forwprop::execute): Likewise.
13509 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13510 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
13512 (move_computations_worker): Likewise.
13513 (execute_sm): Likewise.
13514 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
13515 (remove_exits_and_undefined_stmts): Likewise.
13516 (remove_redundant_iv_tests): Likewise.
13517 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
13518 (adjust_iv_update_pos): Likewise.
13519 * tree-ssa-math-opts.c (bswap_replace): Likewise.
13520 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
13521 (value_replacement): Likewise.
13522 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
13523 * tree-ssa-pre.c (print_pre_expr): Likewise.
13524 (get_representative_for): Likewise.
13525 (create_expression_by_pieces): Likewise.
13526 (insert_into_preds_of_block): Likewise.
13527 (eliminate_insert): Likewise.
13528 (eliminate_dom_walker::before_dom_children): Likewise.
13529 (eliminate): Likewise.
13530 (remove_dead_inserted_code): Likewise.
13531 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
13532 * tree-ssa-reassoc.c (get_rank): Likewise.
13533 (eliminate_duplicate_pair): Likewise.
13534 (eliminate_plus_minus_pair): Likewise.
13535 (eliminate_not_pairs): Likewise.
13536 (undistribute_ops_list): Likewise.
13537 (eliminate_redundant_comparison): Likewise.
13538 (update_range_test): Likewise.
13539 (optimize_range_tests_var_bound): Likewise.
13540 (optimize_vec_cond_expr): Likewise.
13541 (rewrite_expr_tree): Likewise.
13542 (rewrite_expr_tree_parallel): Likewise.
13543 (linearize_expr): Likewise.
13544 (break_up_subtract): Likewise.
13545 (linearize_expr_tree): Likewise.
13546 (attempt_builtin_powi): Likewise.
13547 (attempt_builtin_copysign): Likewise.
13548 (transform_stmt_to_copy): Likewise.
13549 (transform_stmt_to_multiply): Likewise.
13550 (dump_ops_vector): Likewise.
13551 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
13552 (print_scc): Likewise.
13553 (set_ssa_val_to): Likewise.
13554 (visit_reference_op_store): Likewise.
13555 (visit_use): Likewise.
13556 (sccvn_dom_walker::before_dom_children): Likewise.
13557 (run_scc_vn): Likewise.
13558 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
13560 (expr_hash_elt::print): Likewise.
13561 (const_and_copies::pop_to_marker): Likewise.
13562 (const_and_copies::record_const_or_copy_raw): Likewise.
13563 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
13564 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
13565 (dump_predicates): Likewise.
13566 (find_uninit_use): Likewise.
13567 (warn_uninitialized_phi): Likewise.
13568 (pass_late_warn_uninitialized::execute): Likewise.
13569 * tree-ssa.c (verify_vssa): Likewise.
13570 (verify_ssa): Likewise.
13571 (maybe_optimize_var): Likewise.
13572 * tree-vrp.c (dump_value_range): Likewise.
13573 (dump_all_value_ranges): Likewise.
13574 (dump_asserts_for): Likewise.
13575 (register_edge_assert_for_2): Likewise.
13576 (vrp_visit_cond_stmt): Likewise.
13577 (vrp_visit_switch_stmt): Likewise.
13578 (vrp_visit_stmt): Likewise.
13579 (vrp_visit_phi_node): Likewise.
13580 (simplify_cond_using_ranges_1): Likewise.
13581 (fold_predicate_in): Likewise.
13582 (evrp_dom_walker::before_dom_children): Likewise.
13583 (evrp_dom_walker::push_value_range): Likewise.
13584 (evrp_dom_walker::pop_value_range): Likewise.
13585 (execute_early_vrp): Likewise.
13587 2017-05-16 Richard Biener <rguenther@suse.de>
13589 * dwarf2out.c (loc_list_from_tree_1): Do not create
13590 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
13592 2017-05-16 Richard Biener <rguenther@suse.de>
13594 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
13596 (note_variable_value_in_expr): If we resolved the decl ref
13597 do not push to the stack.
13599 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
13601 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
13602 operations in fast-math mode.
13603 (vaddq_f16): Likewise.
13604 (vmul_f16): Likewise.
13605 (vmulq_f16): Likewise.
13606 (vsub_f16): Likewise.
13607 (vsubq_f16): Likewise.
13608 * config/arm/neon.md (add<mode>3): New.
13610 (fma:<VH:mode>3): New. Also remove outdated comment.
13613 2017-05-16 Martin Liska <mliska@suse.cz>
13617 * ipa-devirt.c (warn_types_mismatch): Fix typo.
13618 (odr_types_equivalent_p): Likewise.
13620 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
13622 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
13624 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
13627 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
13628 non-interunit SSE move alternatives with '?'.
13629 (zero-extendsidi peephole2): New peephole to skip intermediate
13630 general register in SSE zero-extend sequence.
13632 2017-05-15 Jeff Law <law@redhat.com>
13634 * reorg.c (relax_delay_slots): Create a new variable to hold
13635 the temporary target rather than clobbering TARGET_LABEL.
13637 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
13638 missing argument to extract_bit_field call.
13639 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
13641 2017-05-15 Martin Liska <mliska@suse.cz>
13644 * gcc.c (process_command): Do not allow empty argument of -o option.
13646 2017-05-15 Renlin Li <renlin.li@arm.com>
13648 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
13649 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
13650 * config/aarch64/constraints.md (Usf): Add long call check.
13651 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
13652 (call_value): Likewise.
13653 (sibcall): Likewise.
13654 (sibcall_value): Likewise.
13656 (call_value_insn): New.
13657 (sibcall_insn): Update rtx pattern.
13658 (sibcall_value_insn): Likewise.
13659 (call_internal): Remove.
13660 (call_value_internal): Likewise.
13661 (sibcall_internal): Likewise.
13662 (sibcall_value_internal): Likewise.
13663 (call_reg): Likewise.
13664 (call_symbol): Likewise.
13665 (call_value_reg): Likewise.
13666 (call_value_symbol): Likewise.
13668 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
13671 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
13673 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
13675 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
13676 compatible with CCGOCmode and with CCZmode.
13678 2017-05-14 Martin Sebor <msebor@redhat.com>
13680 PR middle-end/77671
13681 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
13682 (gimple_fold_builtin_snprintf): Same.
13683 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
13684 (gimple_fold_builtin_snprintf): Same.
13685 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
13686 of character types.
13687 (is_call_safe): New function.
13688 (try_substitute_return_value): Call it.
13689 (try_simplify_call): New function.
13690 (pass_sprintf_length::handle_gimple_call): Call it.
13692 2017-05-14 Martin Sebor <msebor@redhat.com>
13694 PR middle-end/80669
13695 * builtins.c (expand_builtin_stpncpy): Simplify.
13697 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
13699 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
13700 * config/i386/i386.h
13701 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
13702 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
13703 (struct machine_function): Add new members call_ms2sysv,
13704 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
13705 (struct machine_frame_state): New fields sp_realigned and
13706 sp_realigned_offset.
13707 * config/i386/i386.c
13708 (enum xlogue_stub): New enum.
13709 (enum xlogue_stub_sets): New enum.
13710 (class xlogue_layout): New class.
13711 (struct ix86_frame): New fields stack_realign_allocate_offset,
13712 stack_realign_offset and outlined_save_offset. Modify comments to
13713 detail stack layout when using out-of-line stubs.
13714 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
13715 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
13716 -mcall-ms2sysv-xlogues.
13717 (stub_managed_regs): New static variable.
13718 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
13719 registers managed by out-of-line stub.
13720 (disable_call_ms2sysv_xlogues): New function.
13721 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
13722 m->call_ms2sysv when appropriate and compute frame layout for
13724 (sp_valid_at, fp_valid_at): New inline functions.
13725 (choose_basereg): New function.
13726 (choose_baseaddr): Add align parameter, use choose_basereg and modify
13728 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
13729 Use align parameter of choose_baseaddr to generated aligned SSE movs
13731 (pro_epilogue_adjust_stack): Modify to track
13732 machine_frame_state::sp_realigned.
13733 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
13734 (ix86_nsaved_sseregs): Likewise.
13735 (ix86_emit_save_regs): Likewise.
13736 (ix86_emit_save_regs_using_mov): Likewise.
13737 (ix86_emit_save_sse_regs_using_mov): Likewise.
13738 (get_scratch_register_on_entry): Likewise.
13739 (gen_frame_set): New function.
13740 (gen_frame_load): Likewise.
13741 (gen_frame_store): Likewise.
13742 (emit_outlined_ms2sysv_save): Likewise.
13743 (emit_outlined_ms2sysv_restore): Likewise.
13744 (ix86_expand_prologue): Modify stack re-alignment code and call
13745 emit_outlined_ms2sysv_save when appropriate.
13746 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
13747 parameter rtx_insn *insn, which allows the function to be used to only
13748 generate the notes.
13749 (ix86_expand_epilogue): Modify validity checks of frame and stack
13750 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
13751 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
13752 * config/i386/predicates.md
13753 (save_multiple): New predicate.
13754 (restore_multiple): Likewise.
13755 * config/i386/sse.md
13756 (save_multiple<mode>): New pattern.
13757 (save_multiple_realign<mode>): Likewise.
13758 (restore_multiple<mode>): Likewise.
13759 (restore_multiple_and_return<mode>): Likewise.
13760 (restore_multiple_leave_return<mode>): Likewise.
13761 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
13763 2017-05-14 Julia Koval <julia.koval@intel.com>
13765 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
13766 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
13767 (__builtin_ia32_xsetbv): New builtins.
13768 * config/i386/i386.c (ix86_expand_special_args_builtin):
13770 (ix86_expand_builtin): Special expand for new intrinsics.
13771 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
13772 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
13773 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
13775 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13777 * cfganal.c (inverted_post_order_compute): Change argument type
13779 * cfganal.h (inverted_post_order_compute): Adjust prototype.
13780 * df-core.c (rest_of_handle_df_initialize): Adjust.
13781 (rest_of_handle_df_finish): Likewise.
13782 (df_analyze_1): Likewise.
13783 (df_analyze): Likewise.
13784 (loop_inverted_post_order_compute): Change argument to be a vec *.
13785 (df_analyze_loop): Adjust.
13786 (df_get_n_blocks): Likewise.
13787 (df_get_postorder): Likewise.
13788 * df.h (struct df_d): Change field to be a vec.
13789 * lcm.c (compute_laterin): Adjust.
13790 (compute_available): Likewise.
13791 * lra-lives.c (lra_create_live_ranges_1): Likewise.
13792 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
13793 * tree-ssa-pre.c (compute_antic): Likewise.
13795 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13797 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
13798 (depth_first_search::depth_first_search): Change structure init
13799 function to this constructor.
13800 (depth_first_search::add_bb): Rename function to this member.
13801 (depth_first_search::execute): Likewise.
13802 (flow_dfs_compute_reverse_finish): Adjust.
13804 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13806 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
13807 (longest_simple_path): Likewise.
13808 * shrink-wrap.c (spread_components): Likewise.
13809 (disqualify_problematic_components): Likewise.
13810 (emit_common_heads_for_components): Likewise.
13811 (emit_common_tails_for_components): Likewise.
13812 (insert_prologue_epilogue_for_components): Likewise.
13814 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13816 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
13819 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13821 * df-core.c (df_set_blocks): Start using auto_bitmap.
13822 (df_compact_blocks): Likewise.
13823 * df-problems.c (df_rd_confluence_n): Likewise.
13824 * df-scan.c (df_insn_rescan_all): Likewise.
13825 (df_process_deferred_rescans): Likewise.
13826 (df_update_entry_block_defs): Likewise.
13827 (df_update_exit_block_uses): Likewise.
13828 (df_entry_block_bitmap_verify): Likewise.
13829 (df_exit_block_bitmap_verify): Likewise.
13830 (df_scan_verify): Likewise.
13831 * lra-constraints.c (lra_constraints): Likewise.
13832 (undo_optional_reloads): Likewise.
13833 (lra_undo_inheritance): Likewise.
13834 * lra-remat.c (calculate_gen_cands): Likewise.
13835 (do_remat): Likewise.
13836 * lra-spills.c (assign_spill_hard_regs): Likewise.
13837 (spill_pseudos): Likewise.
13838 * tree-ssa-pre.c (bitmap_set_and): Likewise.
13839 (bitmap_set_subtract_values): Likewise.
13841 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13843 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
13844 management with auto_bitmap.
13845 (fix_inter_tick): Likewise.
13846 (fix_recovery_deps): Likewise.
13847 * ira.c (add_store_equivs): Likewise.
13848 (find_moveable_pseudos): Likewise.
13849 (split_live_ranges_for_shrink_wrap): Likewise.
13850 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
13851 (rtx_reuse_manager::seen_def_p): Likewise.
13852 (rtx_reuse_manager::set_seen_def): Likewise.
13853 * print-rtl.h (class rtx_reuse_manager): Likewise.
13855 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13857 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
13859 (migrate_btr_def): Likewise.
13860 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
13861 * df-core.c (loop_post_order_compute): Likewise.
13862 (loop_inverted_post_order_compute): Likewise.
13863 * hsa-common.h: Likewise.
13864 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
13865 * init-regs.c (initialize_uninitialized_regs): Likewise.
13866 * ipa-inline.c (resolve_noninline_speculation): Likewise.
13867 (inline_small_functions): Likewise.
13868 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
13869 * ira.c (combine_and_move_insns): Likewise.
13870 (build_insn_chain): Likewise.
13871 * loop-invariant.c (find_invariants): Likewise.
13872 * lower-subreg.c (propagate_pseudo_copies): Likewise.
13873 * predict.c (tree_predict_by_opcode): Likewise.
13874 (predict_paths_leading_to): Likewise.
13875 (predict_paths_leading_to_edge): Likewise.
13876 (estimate_loops_at_level): Likewise.
13877 (estimate_loops): Likewise.
13878 * shrink-wrap.c (try_shrink_wrapping): Likewise.
13879 (spread_components): Likewise.
13880 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
13881 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
13882 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
13883 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
13884 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
13885 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
13886 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
13887 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
13888 (mark_threaded_blocks): Likewise.
13889 (thread_through_all_blocks): Likewise.
13890 * tree-ssa.c (verify_ssa): Likewise.
13891 (execute_update_addresses_taken): Likewise.
13892 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
13894 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13896 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
13898 (post_order_compute): Likewise.
13899 (inverted_post_order_compute): Likewise.
13900 (pre_and_rev_post_order_compute_fn): Likewise.
13902 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13904 * genrecog.c (int_set::int_set): Explicitly construct our
13905 auto_vec base class.
13906 * vec.h (auto_vec::auto_vec): New constructor.
13908 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13910 * bitmap.h (class auto_bitmap): New constructor taking
13911 bitmap_obstack * argument.
13913 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13915 * bitmap.h (class auto_bitmap): Change type of m_bits to
13916 bitmap_head, and adjust ctor / dtor and member operators.
13918 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
13920 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
13921 when returned register mode doesn't match original mode.
13923 2017-05-12 Jeff Law <law@redhat.com>
13924 Jakub Jelinek <jakub@redhat.com>
13926 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
13927 we look for cc setter after the compare-elim changes.
13928 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
13929 within the vector to match what compare-elim now expects.
13930 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
13931 (xorsi3_flags, one_cmplsi2_flags): Likewise.
13933 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
13934 after the compare-elim changes.
13935 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
13936 the vector to match what compare-elim now expects.
13937 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
13938 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
13939 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
13940 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
13941 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
13943 * config/visium/visium.c (single_set_and_flags): Fix where
13944 we look for cc setter after the compare-elim changes.
13945 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
13946 with the vector to match what compare-elim now expects.
13947 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
13948 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
13949 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
13950 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
13951 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
13952 (neg<mode>2_insn_set_overflow): Likewise.
13954 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
13956 PR middle-end/79794
13957 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
13958 maybe_expand_insn call, set ops[0].target. If still set after call,
13959 set alt_rtl. Add extra arg to recursive calls.
13960 (extract_bit_field): Add alt_rtl argument. Pass to
13962 * expmed.h (extract_bit_field): Fix prototype.
13963 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
13964 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
13965 to extract_bit_field_calls.
13966 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
13967 Pass alt_rtl to extract_bit_field calls.
13968 * calls.c (store_unaligned_arguments_into_psuedos)
13969 load_register_parameters): Pass extra NULL to extract_bit_field calls.
13970 * optabs.c (maybe_legitimize_operand): Clear op->target when call
13972 * optabs.h (struct expand_operand): Add target bitfield.
13974 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13976 * compare-elim.c (try_eliminate_compare): Canonicalize
13977 operation with embedded compare to
13978 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
13979 (set (reg) (operation)].
13981 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
13983 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13986 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
13987 cost of adding a carry flag for ADC instruction.
13988 [case MINUS]: Ignore the cost of subtracting a carry flag
13989 for SBB instruction.
13991 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
13993 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
13995 * config/rs6000/bmiintrin.h: New file.
13996 * config/rs6000/bmi2intrin.h: New file.
13997 * config/rs6000/x86intrin.h: New file.
13999 2017-05-12 Jeff Law <law@redhat.com>
14001 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
14004 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
14006 PR middle-end/80707
14007 * tree-cfg.c: Remove cfg edges of unreachable case statements.
14009 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14011 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
14012 early expansion of vector divide builtins.
14013 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
14014 builtins identified as having unsigned arguments.
14016 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14018 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
14019 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
14020 expansion of vector logical operations (and, andc, or, xor,
14023 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14025 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
14026 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
14028 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14030 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
14031 early GIMPLE expansion of vector multiplies.
14033 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14035 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
14036 TARGET_HAVE_MOVT conditional.
14037 (movt splitter): Likewise.
14039 2017-05-12 Richard Biener <rguenther@suse.de>
14041 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
14042 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14043 Fold all stmts not inplace.
14045 2017-05-12 Richard Biener <rguenther@suse.de>
14047 PR tree-optimization/80713
14048 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
14049 inserted_exprs bit for not removed stmts.
14051 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
14053 PR middle-end/69921
14054 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
14055 parallelized" attribute for parallelized OpenACC kernels.
14056 * omp-offload.c (execute_oacc_device_lower): Use it.
14058 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
14059 Set "oacc kernels" attribute.
14060 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
14061 parameter. Adjust all users.
14062 (oacc_fn_attrib_kernels_p): Remove function.
14063 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
14064 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
14065 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
14066 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
14067 assert "oacc kernels" attribute is set.
14069 2017-05-11 Carl Love <cel@us.ibm.com>
14071 * config/rs6000/rs6000-c: Add support for built-in functions
14072 vector unsigned char vec_popcnt (vector signed char)
14073 vector unsigned char vec_popcnt (vector unsigned char)
14074 vector unsigned short vec_popcnt (vector signed short)
14075 vector unsigned short vec_popcnt (vector unsigned short)
14076 vector unsigned int vec_popcnt (vector signed int)
14077 vector unsigned int vec_popcnt (vector unsigned int)
14078 vector unsigned long long vec_popcnt (vector signed long long)
14079 vector unsigned long long vec_popcnt (vector unsigned long long)
14080 vector signed long long vec_slo (vector signed long long,
14081 vector signed char)
14082 vector signed long long vec_slo (vector signed long long,
14083 vector unsigned char)
14084 vector unsigned long long vec_slo (vector unsigned long long,
14085 vector signed char)
14086 vector unsigned long long vec_slo (vector unsigned long long,
14087 vector unsigned char)
14088 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
14089 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
14090 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
14091 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
14092 * doc/extend.texi: Update the built-in documentation file for the
14093 new built-in functions.
14095 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
14097 * attribs.h (sorted_attr_string): Move machine independent
14098 functions for target clone support from the i386 port to common
14099 code. Rename ix86_function_versions to common_function_versions.
14100 Rename make_name to make_unique_name.
14101 (common_function_versions): Likewise.
14102 (make_unique_name): Likewise.
14103 (make_dispatcher_decl): Likewise.
14104 (is_function_default_version): Likewise.
14105 * attribs.c (attr_strcmp): Likewise.
14106 (sorted_attr_string): Likewise.
14107 (common_function_versions): Likewise.
14108 (make_unique_name): Likewise.
14109 (make_dispatcher_decl): Likewise.
14110 (is_function_default_version): Likewise.
14111 * config/i386/i386.c (attr_strcmp): Likewise.
14112 (sorted_attr_string): Likewise.
14113 (ix86_function_versions): Likewise.
14114 (make_name): Likewise.
14115 (make_dispatcher_decl): Likewise.
14116 (is_function_default_version): Likewise.
14117 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
14119 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14122 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
14123 Account for direct move costs for vec_construct of integer
14126 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
14129 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
14130 (UNSPEC_STX_ATOMIC): Ditto.
14131 (loaddi_via_sse): New insn.
14132 (storedi_via_sse): Ditto.
14133 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
14134 Update corresponding peephole2 patterns.
14135 (atomic_storedi_fpu): Ditto.
14137 2017-05-11 Julia Koval <julia.koval@intel.com>
14139 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
14140 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
14142 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
14143 (__builtin_ia32_rsqrt14ss_mask): New builtins.
14144 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
14146 2017-05-11 Nathan Sidwell <nathan@acm.org>
14148 * graphite-poly.c: Include dumpfile.h.
14150 * dumpfle.h (dump_function): Declare here ...
14151 * tree-dump.h (dump_function): ... not here.
14152 * dumpfile.c: #include tree-cfg.h.
14153 (dump_function): Move here from ...
14154 * tree-dump.c (dump_function): ... here.
14155 * gimplify.c: #include splay-tree.h, not tree-dump.h.
14156 * graphite-poly.c: Don't include tree-dump.h.
14157 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
14158 * print-tree.c: Likewise.
14159 * stor-layout.c: Likewise.
14160 * tree-nested.c: Likewise.
14162 * dumpfile.c (dump_start): Use TDF_FLAGS.
14163 (dump_enable_all): Fix TDF_KIND check thinko.
14165 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
14167 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14168 array entries to represent two legal parameterizations of the
14169 overloaded __builtin_cmpb function, as represented by the
14170 P6_OV_BUILTIN_CMPB constant.
14171 (altivec_resolve_overloaded_builtin): Add special case handling
14172 for the __builtin_cmpb function, as represented by the
14173 P6_OV_BUILTIN_CMPB constant.
14174 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
14175 (BU_P6_64BIT_2): New macro.
14176 (BU_P6_OVERLOAD_2): New macro
14177 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
14178 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
14179 (CMPB): Add overload support to represent both 32-bit and 64-bit
14180 compare-bytes function.
14181 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14182 support for TARGET_CMPB.
14183 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
14184 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
14185 documentation of the __builtin_cmpb overloaded built-in function.
14187 2017-05-11 Richard Biener <rguenther@suse.de>
14189 PR tree-optimization/80705
14190 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
14191 bases are not vectorizable.
14193 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14195 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
14196 when counting register pressure.
14198 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14200 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
14201 (iv_ca_more_deps): Renamed to ...
14202 (iv_ca_compare_deps): ... this.
14203 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
14205 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14207 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
14209 (determine_group_iv_costs): ... here.
14210 (find_inv_vars_cb): Record inv var if it's not recorded before.
14212 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14214 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
14215 (get_shiftadd_cost): Ditto.
14217 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14219 * tree-ssa-address.c: Include header file.
14220 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
14222 (add_to_parts): Refactor.
14223 (addr_to_parts): New parameter. Update use of move_hint_to_base.
14224 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
14227 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14229 PR tree-optimization/53090
14230 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
14232 (extract_cond_operands): Detect condition with IV on both sides
14233 and return COMP_IV_EXPR_2.
14234 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
14235 (rewrite_use_compare): Simplify by removing call to function
14236 extract_cond_operands.
14238 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14240 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
14241 (extract_cond_operands): Detect condition comparing against non-
14242 invariant bound and return appropriate enum value.
14243 (find_interesting_uses_cond): Update use of extract_cond_operands.
14244 Handle its return value accordingly.
14245 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
14247 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14249 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
14250 nonlinear iv_use computation in loop invariant sensitive way.
14252 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14254 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
14255 (find_iv_candidates): Call relate_compare_use_with_all_cands.
14257 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14259 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
14260 (dump_cand): Support iv_cand.inv_exprs.
14261 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
14263 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
14266 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14268 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
14270 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
14271 as local function. Include necessary header files.
14272 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
14274 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14276 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
14278 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14280 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
14281 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
14282 RSHIFT_EXPR and BIT_NOT_EXPR.
14284 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14286 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
14287 (adjust_setup_cost): New parameter supporting round up adjustment.
14288 (struct address_cost_data): Delete.
14289 (force_expr_to_var_cost): Don't bound cost with spill_cost.
14290 (split_address_cost, ptr_difference_cost): Delete.
14291 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
14292 (struct ainc_cost_data): New struct.
14293 (get_address_cost_ainc): New function.
14294 (get_address_cost, get_computation_cost): Reimplement.
14295 (determine_group_iv_cost_address): Record inv_expr for all uses of
14297 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
14298 (iv_ca_has_deps): Reimplemented to ...
14299 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
14301 (iv_ca_extend): Call iv_ca_more_deps.
14303 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14305 * tree-ssa-address.c (struct mem_address): Move to header file.
14306 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
14307 * tree-ssa-address.h (struct mem_address): Move from C file.
14308 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
14310 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14312 * tree-affine.h (aff_combination_type): New interface.
14313 (aff_combination_zero_p): Remove static.
14314 (aff_combination_const_p): New interface.
14315 (aff_combination_singleton_var_p): New interfaces.
14317 2017-05-11 Richard Biener <rguenther@suse.de>
14319 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14320 Skip unreachable blocks and destinations.
14321 (eliminate): Move stmt removal and fixup ...
14322 (fini_eliminate): ... here. Skip inserted exprs.
14323 (pass_pre::execute): Move fini_pre after fini_eliminate.
14324 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
14325 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
14326 PRE to get rid of dead code that has invalid SSA form and
14327 split critical edges again.
14329 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14331 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14333 2017-05-11 Richard Biener <rguenther@suse.de>
14335 * passes.c (execute_function_todo): Verify loops if they are
14336 said to be up-to-date.
14337 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
14338 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
14340 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
14343 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
14344 handle calling assemble_external ourself.
14347 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
14348 modes with zero size. Enhance comment.
14350 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14352 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
14353 built-ins for vec_xl and vec_xst with short and char pointer
14356 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
14358 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
14359 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
14360 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
14361 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
14362 (_mm_maskz_min_round_ss): New intrinsics.
14363 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
14364 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
14365 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
14366 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
14367 (__builtin_ia32_minss_mask_round): New builtins.
14368 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14369 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
14370 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
14372 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
14373 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
14375 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
14378 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
14380 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
14381 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
14382 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
14383 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
14384 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
14385 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
14386 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
14387 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14388 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
14389 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
14390 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
14391 (__builtin_ia32_mulss_mask_round): New builtins.
14392 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14393 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
14394 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
14396 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
14397 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
14399 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
14402 2017-05-10 Julia Koval <julia.koval@intel.com>
14404 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
14405 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
14406 (_mm256_setr_m128i): New intrinsics.
14408 2017-05-10 Julia Koval <julia.koval@intel.com>
14410 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
14411 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
14412 (_mm_maskz_rcp14_ss): New intrinsics.
14413 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
14414 (__builtin_ia32_rcp14ss_mask): New builtins.
14415 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
14417 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
14419 PR tree-optimization/51513
14420 * tree-cfg.c (gimple_seq_unreachable_p): New function.
14421 (assert_unreachable_fallthru_edge_p): Use it.
14422 (group_case_labels_stmt): Likewise.
14423 * tree-cfg.h: Prototype it.
14424 * stmt.c: Include cfghooks.h and tree-cfg.h.
14425 (emit_case_dispatch_table) <gap_label>: New local variable.
14426 Use it to fill dispatch table gaps.
14427 Test for default_label before updating probabilities.
14428 (expand_case) <default_label>: Remove unneeded initialization.
14429 Test for unreachable default case statement and remove its edge.
14430 Set default_label accordingly.
14431 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
14433 2017-05-10 Carl Love <cel@us.ibm.com>
14435 * config/rs6000/rs6000-c: Add support for built-in functions
14436 vector signed char vec_neg (vector signed char)
14437 vector signed short int vec_neg (vector short int)
14438 vector signed int vec_neg (vector signed int)
14439 vector signed long long vec_neg (vector signed long long)
14440 vector float vec_neg (vector float)
14441 vector double vec_neg (vector double)
14442 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
14444 * config/rs6000/altivec.h: Add define for vec_neg
14445 * doc/extend.texi: Update the built-in documentation for the
14446 new built-in functions.
14448 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14450 PR tree-optimization/77644
14451 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
14453 2017-05-10 Nathan Sidwell <nathan@acm.org>
14455 * dumpfile.h (TDI_lang_all): New.
14456 (TDF_KIND): New. Renumber others
14457 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
14459 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
14461 (get_dump_file_name): Adjust suffix generation.
14462 (dump_enable_all): Use TDF_KIND.
14463 * doc/invoke.texi (-fdump-lang-all): Document.
14465 * dumpfile.h: Tabify.
14467 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
14470 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
14471 Move member access before delete.
14473 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
14475 * tree-inline.c (expand_call_inline): Split block at stmt
14478 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
14481 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
14482 are now unused after splitting mov{sf,sd}_hardfloat.
14483 (f32_lr2): Likewise.
14484 (f32_lm): Likewise.
14485 (f32_lm2): Likewise.
14486 (f32_li): Likewise.
14487 (f32_li2): Likewise.
14488 (f32_lv): Likewise.
14489 (f32_sr): Likewise.
14490 (f32_sr2): Likewise.
14491 (f32_sm): Likewise.
14492 (f32_sm2): Likewise.
14493 (f32_si): Likewise.
14494 (f32_si2): Likewise.
14495 (f32_sv): Likewise.
14496 (f32_dm): Likewise.
14497 (f32_vsx): Likewise.
14498 (f32_av): Likewise.
14499 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
14500 For movsf, order stores so the VSX stores occur before the GPR
14501 store which encourages the register allocator to use a traditional
14502 FPR instead of a GPR. For movsd, order the stores so that the GPR
14503 store comes before the VSX stores to allow the power6 to work.
14504 This is due to the power6 not having a 32-bit integer store
14505 instruction from a FPR.
14506 (movsf_hardfloat): Likewise.
14507 (movsd_hardfloat): Likewise.
14509 2017-05-09 Martin Sebor <msebor@redhat.com>
14511 PR translation/80280
14512 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
14515 PR translation/80280
14516 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
14517 data member added in r247778.
14518 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
14520 2017-05-09 Nathan Sidwell <nathan@acm.org>
14522 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
14524 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
14527 2017-05-09 Marek Polacek <polacek@redhat.com>
14529 * doc/invoke.texi: Fix typo.
14531 2017-05-09 Richard Biener <rguenther@suse.de>
14533 * tree-vrp.c (vrp_val_is_max): Adjust comment.
14534 (vrp_val_is_min): Likewise.
14535 (set_value_range_to_value): Likewise.
14536 (set_value_range_to_nonnegative): Likewise.
14537 (gimple_assign_nonzero_p): Likewise.
14538 (gimple_stmt_nonzero_p): Likewise.
14539 (vrp_int_const_binop): Likewise. Remove unreachable case.
14540 (adjust_range_with_scev): Adjust comments.
14541 (compare_range_with_value): Likewise.
14542 (extract_range_from_phi_node): Likewise.
14543 (test_for_singularity): Likewise.
14545 2017-05-09 Richard Biener <rguenther@suse.de>
14547 * tree-vrp.c (get_single_symbol): Add assert that we don't
14548 get overflowed constants as invariant part.
14549 (compare_values_warnv): Add comment before the TREE_NO_WARNING
14550 checks. Use wi::cmp instead of recursing for integer constants.
14551 (compare_values): Just ignore whether we assumed undefined
14552 overflow instead of failing the compare.
14553 (extract_range_for_var_from_comparison_expr): Add comment before the
14554 TREE_NO_WARNING sets.
14555 (test_for_singularity): Likewise.
14556 (extract_range_from_comparison): Do not disable optimization
14557 when we assumed undefined overflow.
14558 (extract_range_basic): Remove init of unused var.
14560 2017-05-09 Richard Biener <rguenther@suse.de>
14562 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
14563 (extract_range_from_multiplicative_op_1): Adjust.
14564 (extract_range_from_binary_expr_1): Use int_const_binop.
14566 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
14569 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
14570 rs6000_store_data_bypass_p in seven define_bypass directives and
14571 in several comments.
14572 * config/rs6000/rs6000-protos.h: Add prototype for
14573 rs6000_store_data_bypass_p function.
14574 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
14575 function implements slightly different (rs6000-specific) semantics
14576 than store_data_bypass_p, returning false rather than aborting
14577 with assertion error when arguments do not satisfy the
14578 requirements of store data bypass.
14579 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
14580 rs6000_store_data_bypass_p.
14582 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
14584 * config/xtensa/xtensa-protos.h
14585 (xtensa_initial_elimination_offset): New declaration.
14586 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
14587 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
14588 macro definition, add case for FRAME_POINTER_REGNUM when
14589 FRAME_GROWS_DOWNWARD.
14590 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
14591 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
14592 xtensa_initial_elimination_offset.
14594 2017-05-08 Nathan Sidwell <nathan@acm.org>
14596 * doc/invoke.texi: Alphabetize -fdump options.
14598 2017-05-08 Martin Sebor <msebor@redhat.com>
14600 PR translation/80280
14601 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
14603 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
14605 * target.def (compute_frame_layout): New optional target hook.
14606 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
14607 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
14608 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
14610 * reload1.c (verify_initial_elim_offsets): Likewise.
14611 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
14612 (use_simple_return_p): Call arm_compute_frame_layout if needed.
14613 (arm_get_frame_offsets): Split up into this ...
14614 (arm_compute_frame_layout): ... and this function.
14616 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
14618 * config/aarch64/constraints.md (Usa): New constraint.
14619 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
14621 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
14623 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
14624 with_multilib_list after it has been checked.
14626 2017-05-08 Richard Biener <rguenther@suse.de>
14628 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
14629 (bitmap_set_subtract_values): Likewise.
14631 2017-05-08 Richard Biener <rguenther@suse.de>
14633 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
14634 (gimple_assign_nonzero): ... this and remove strict_overflow_p
14636 (gimple_stmt_nonzero_warnv_p): Rename to ...
14637 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
14639 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
14640 (extract_range_basic): Adjust, do not disable propagation on
14641 strict overflow sensitive simplification.
14642 (vrp_visit_cond_stmt): Likewise.
14644 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
14646 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
14647 body size unconditionally.
14649 2017-05-07 Jeff Law <law@redhat.com>
14652 2017-05-06 Jeff Law <law@redhat.com>
14653 PR tree-optimization/78496
14654 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
14657 PR tree-optimization/78496
14658 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
14659 (simplify_stmt_using_ranges): Call it.
14660 (vrp_dom_walker::before_dom_children): Extract equivalences
14661 from an ASSERT_EXPR with an equality comparison against a
14664 2017-05-06 Jeff Law <law@redhat.com>
14666 PR tree-optimization/78496
14667 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
14670 PR tree-optimization/78496
14671 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
14672 (simplify_stmt_using_ranges): Call it.
14673 (vrp_dom_walker::before_dom_children): Extract equivalences
14674 from an ASSERT_EXPR with an equality comparison against a
14677 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
14679 * lra-constraints.c (lra_copy_reg_equiv): New function.
14680 (split_reg): Use it to copy equivalence information from the
14681 original register to the spill register.
14683 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
14685 PR rtl-optimization/75964
14686 * simplify-rtx.c (simplify_const_relational_operation): Remove
14687 invalid handling of comparisons of integer ABS.
14689 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
14691 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
14692 initialize to zero.
14693 (init_regs): Remove declaration.
14694 (function_arg_advance_32): Initialize error_p as boolean variable.
14696 2017-05-05 Nathan Sidwell <nathan@acm.org>
14698 * store-motion.c (remove_reachable_equiv_notes): Reformat long
14699 lines. Use for (;;).
14701 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14703 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
14704 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
14705 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
14706 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
14707 VF=2 that require versioning.
14709 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14711 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
14714 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14716 * diagnostic.h (diagnostic_override_option_index): Convert from
14717 macro to inline function.
14719 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14721 * diagnostic.c (last_module_changed_p): New function.
14722 (set_last_module): New function.
14723 (diagnostic_report_current_module): Convert macro usage to
14724 the above functions.
14725 * diagnostic.h (diagnostic_context::last_module): Strengthen
14726 from const line_map * to const line_map_ordinary *.
14727 (diagnostic_last_module_changed): Delete macro.
14728 (diagnostic_set_last_module): Delete macro.
14730 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14732 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
14733 with diagnostic_report_diagnostic.
14734 (diagnostic_n_impl_richloc): Likewise.
14735 * diagnostic.h (report_diagnostic): Delete macro.
14736 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
14737 with diagnostic_report_diagnostic.
14738 * substring-locations.c (format_warning_va): Likewise.
14740 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14742 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
14743 save/restor of format_spec. Move option-printing code to...
14744 (print_option_information): ...this new function, and
14745 reimplement by simply printing to the pretty_printer,
14746 rather than appending to the format string.
14748 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14750 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
14751 handling logic into...
14752 (update_effective_level_from_pragmas): ...this new function.
14754 2017-05-04 Andrew Waterman <andrew@sifive.com>
14756 * config/riscv/riscv.opt (mstrict-align): New option.
14757 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
14758 (SLOW_UNALIGNED_ACCESS): Define.
14759 (riscv_slow_unaligned_access): Declare.
14760 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
14762 (riscv_slow_unaligned_access): New variable.
14763 (rocket_tune_info): Set slow_unaligned_access to true.
14764 (optimize_size_tune_info): Set slow_unaligned_access to false.
14765 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
14766 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
14767 (riscv_option_override): Set riscv_slow_unaligned_access.
14768 * doc/invoke.texi: Add -mstrict-align to RISC-V.
14770 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
14772 * config/riscv/riscv.md: Unify indentation.
14774 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14779 * config/rs6000/rs6000.md (u code attribute): Add FIX and
14781 (extendsi<mode>2): Add support for doing sign extension via
14782 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
14783 don't have ISA 3.0 instructions.
14784 (extendsi<mode>2 splitter): Likewise.
14785 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
14786 generate the normal insns since SImode can now go in vector
14787 registers. Disallow the special UNSPECs needed for previous
14788 machines to hide SImode being used. Add new insns
14789 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
14790 (fix_trunc<mode>si2_stfiwx): Likewise.
14791 (fix_trunc<mode>si2_internal): Likewise.
14792 (fixuns_trunc<mode>si2): Likewise.
14793 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14794 (fctiw<u>z_<mode>_smallint): Likewise.
14795 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
14796 of floating point to 32-bit integer from doing a direct move to
14797 the GPR registers to do a store.
14798 (fctiwz_<mode>): Break long line.
14800 2017-05-05 Bin Cheng <bin.cheng@arm.com>
14802 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
14803 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
14804 (addr_list, addr_offset_valid_p): New.
14805 (split_address_groups): Check offset validity with above function.
14806 (gt-tree-ssa-loop-ivopts.h): Include header file.
14808 2017-05-05 Nathan Sidwell <nathan@acm.org>
14810 * config.gcc (arm*-*-*): Add missing 'fi'.
14812 2017-05-05 Steve Ellcey <sellcey@cavium.com>
14814 * doc/invoke.texi (-fopt-info): Explicitly say order of options
14815 included in -fopt-info does not matter.
14816 * doc/optinfo.texi (-fopt-info): Fix description of default
14817 behavour. Explicitly say order of options included in -fopt-info
14820 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
14822 * config.gcc: Allow combinations of aprofile and rmprofile values for
14823 --with-multilib-list.
14824 * config/arm/t-multilib: New file.
14825 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
14826 variables. Remove setting of ISA and floating-point ABI in
14827 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
14828 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
14829 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
14830 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
14832 * config/arm/t-rmprofile: Likewise except for the matches changes.
14833 * doc/install.texi (--with-multilib-list): Document the combination of
14834 aprofile and rmprofile values and warn about pitfalls in doing that.
14836 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
14838 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
14839 (movdi_aarch64): Likewise.
14841 2017-05-05 Jakub Jelinek <jakub@redhat.com>
14843 PR tree-optimization/80632
14844 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
14846 (build_arrays): Initialize it for virtual phis.
14847 (fix_phi_nodes): Use it for virtual phis.
14849 PR tree-optimization/80558
14850 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
14851 [x, y] op z into [x op, y op z] for op & or | if conditions
14854 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
14855 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14858 * config/arm/arm.md (use_literal_pool): Remove.
14859 (64-bit immediate split): No longer takes cost into consideration
14860 if arm_disable_literal_pool is enabled.
14861 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
14862 used when arm_disable_literal_pool is enabled.
14863 (arm_max_const_double_inline_cost): Remove use of
14864 arm_disable_literal_pool.
14865 (push_minipool_fix): Add assert.
14866 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
14867 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
14868 (no_literal_pool_sf_immediate): New.
14870 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14872 PR tree-optimization/80613
14873 * tree-ssa-dce.c (propagate_necessity): Remove cases for
14874 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
14876 2017-05-05 Richard Biener <rguenther@suse.de>
14878 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
14880 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
14882 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
14883 of this flag from insn conditions due to removal from r247495.
14885 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
14887 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
14889 (arm_early_store_addr_dep_ptr): Likewise.
14890 * config/arm/aarch-common-protos.h
14891 (arm_early_load_addr_dep_ptr): Add prototype.
14892 (arm_early_store_addr_dep_ptr): Likewise.
14893 * config/arm/cortex-a53.md: Add new bypasses.
14895 2017-05-05 Jakub Jelinek <jakub@redhat.com>
14897 * tree.c (next_type_uid): Change type to unsigned.
14898 (type_hash_canon): Decrement back next_type_uid if
14899 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
14900 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
14903 2017-05-04 Martin Sebor <msebor@redhat.com>
14905 * builtins.c: Fix a trivial typo in a comment.
14907 PR middle-end/79234
14908 * builtins.c (check_sizes): Adjust to handle reading past the end.
14909 Avoid printing excessive upper bound of ranges. Use %E to print
14910 tree nodes instead of converting them to %wu.
14911 (expand_builtin_memchr): New function.
14912 (compute_dest_size): Rename...
14913 (compute_objsize): ...to this.
14914 (expand_builtin_memcpy): Adjust.
14915 (expand_builtin_mempcpy): Adjust.
14916 (expand_builtin_strcat): Adjust.
14917 (expand_builtin_strcpy): Adjust.
14918 (check_strncat_sizes): Adjust.
14919 (expand_builtin_strncat): Adjust.
14920 (expand_builtin_strncpy): Adjust and simplify.
14921 (expand_builtin_memset): Adjust.
14922 (expand_builtin_bzero): Adjust.
14923 (expand_builtin_memcmp): Adjust.
14924 (expand_builtin): Handle memcmp.
14925 (maybe_emit_chk_warning): Check strncat just once.
14927 2017-05-04 Martin Sebor <msebor@redhat.com>
14929 PR preprocessor/79214
14930 PR middle-end/79222
14931 PR middle-end/79223
14932 * builtins.c (check_sizes): Add inlining context and issue
14933 warnings even when -Wno-system-headers is set.
14934 (check_strncat_sizes): Same.
14935 (expand_builtin_strncat): Same.
14936 (expand_builtin_memmove): New function.
14937 (expand_builtin_stpncpy): Same.
14938 (expand_builtin): Handle memmove and stpncpy.
14940 2017-05-04 Bin Cheng <bin.cheng@arm.com>
14942 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
14943 which is not used any more.
14945 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14947 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
14949 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14951 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
14952 (cortexa53_tunings): Likewise.
14953 (cortexa57_tunings): Likewise.
14954 (cortexa72_tunings): Likewise.
14955 (cortexa73_tunings): Likewise.
14957 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14959 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
14960 Set loop alignment to 8.
14962 2017-05-04 Martin Sebor <msebor@redhat.com>
14964 PR translation/80280
14965 * builtins.c (expand_builtin_object_size): Add missing quoting to
14966 %D and like directives.
14967 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
14968 (hsa_type_for_tree_type): Same.
14969 (verify_function_arguments): Same.
14970 * symtab.c (symbol_table::change_decl_assembler_name): Same.
14971 * varasm.c (get_section): Same.
14974 2017-05-04 Martin Sebor <msebor@redhat.com>
14976 PR translation/80280
14977 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
14979 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14981 * config/aarch64/aarch64.c (generic_addrcost_table):
14982 Change HI/TI mode setting.
14984 2017-05-04 Martin Jambor <mjambor@suse.cz>
14986 PR tree-optimization/80622
14987 * tree-sra.c (comes_initialized_p): New function.
14988 (build_accesses_from_assign): Only set write lazily when
14989 comes_initialized_p is false.
14990 (analyze_access_subtree): Use comes_initialized_p.
14991 (propagate_subaccesses_across_link): Assert !comes_initialized_p
14992 instead of testing for PARM_DECL.
14994 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14996 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
14997 constraint on operand 0 to allow more general addressing modes.
14998 Adjust output template.
14999 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
15001 * config/aarch64/aarch64-protos.h
15002 (aarch64_address_valid_for_prefetch_p): Declare prototype.
15003 * config/aarch64/constraints.md (Dp): New address constraint.
15004 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
15007 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
15009 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
15010 update use of estimate_ipcp_clone_size_and_time.
15011 (estimate_local_effects): Update use of
15012 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
15013 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
15014 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
15015 Return nonspecialized time.
15017 2017-05-04 Richard Biener <rguenther@suse.de>
15019 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
15020 for the last VUSE which def dominates the PHI. Directly call
15022 (get_continuation_for_phi_1): Remove.
15024 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
15026 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
15027 to explain the use of truncating division. Cap the number of
15028 iterations to the maximum given by nb_iterations_upper_bound,
15031 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15033 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
15034 * configure: Regenerate.
15035 * config.in: Regenerate.
15036 * config/i386/driver-mingw32.c: new file.
15037 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
15038 * config.host: Link driver-mingw32.o on MinGW host.
15039 * doc/install.texi: Document new --enable-mingw-wildcard configure
15042 2017-05-04 Marek Polacek <polacek@redhat.com>
15044 PR tree-optimization/80612
15045 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
15047 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15048 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
15050 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
15051 (movt splitter): Likewise.
15052 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
15053 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
15054 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
15055 block for Thumb-1 with MOVT.
15056 (thumb2_legitimate_address_p): Move code block ...
15057 (can_avoid_literal_pool_for_label_p): ... into this new function.
15058 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
15060 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
15061 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
15062 "M-profile targets with the MOVT instruction".
15064 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15066 * config/arm/arm-builtins.c (arm_init_builtins): Rename
15067 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
15068 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
15070 2017-05-04 Martin Liska <mliska@suse.cz>
15072 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
15073 variable cond_code.
15075 2017-05-04 Richard Biener <rguenther@suse.de>
15077 * tree.c (array_at_struct_end_p): Handle arrays at struct
15078 end with flexarrays more conservatively. Refactor and treat
15079 arrays of arrays or aggregates more strict. Fix
15080 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
15081 * tree.c (array_at_struct_end_p): Adjust prototype.
15082 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
15083 * gimple-fold.c (get_range_strlen): Likewise.
15084 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
15086 2017-05-04 Richard Biener <rguenther@suse.de>
15088 PR tree-optimization/31130
15089 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
15091 (supports_overflow_infinity): Likewise.
15092 (is_negative_overflow_infinity): Likewise.
15093 (is_positive_overflow_infinity): Likewise.
15094 (is_overflow_infinity): Likewise.
15095 (stmt_overflow_infinity): Likewise.
15096 (overflow_infinity_range_p): Likewise.
15097 (usable_range_p): Remove as always returning true.
15098 (make_overflow_infinity): Remove.
15099 (negative_overflow_infinity): Likewise.
15100 (positive_overflow_infinity): Likewise.
15101 (avoid_overflow_infinity): Likewise.
15102 (set_value_range): Adjust accordingly.
15103 (set_value_range_to_nonnegative): Likewise, remove now unused
15104 overflow_infinity arg.
15105 (vrp_operand_equal_p): Adjust.
15106 (update_value_range): Likewise.
15107 (range_int_cst_singleton_p): Likewise.
15108 (operand_less_p): Likewise.
15109 (compare_values_warnv): Likewise.
15110 (extract_range_for_var_from_comparison_expr): Likewise.
15111 (vrp_int_const_binop): Likewise.
15112 (zero_nonzero_bits_from_vr): Likewise.
15113 (extract_range_from_multiplicative_op_1): Likewise.
15114 (extract_range_from_binary_expr_1): Likewise.
15115 (extract_range_from_unary_expr): Likewise.
15116 (extract_range_from_comparison): Likewise.
15117 (extract_range_basic): Likewise.
15118 (adjust_range_with_scev): Likewise.
15119 (compare_ranges): Likewise.
15120 (compare_range_with_value): Likewise.
15121 (dump_value_range): Likewise.
15122 (test_for_singularity): Likewise, remove strict_overflow_p parameter
15124 (simplify_cond_using_ranges): Adjust.
15126 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
15128 * brig-builtins.def: Added a builtin for class_f64.
15129 * builtin-types.def: Added a builtin type needed by class_f64.
15131 2017-05-03 Jason Merrill <jason@redhat.com>
15133 * timevar.def: Add TV_CONSTEXPR.
15135 2017-05-03 David Malcolm <dmalcolm@redhat.com>
15137 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
15139 2017-05-03 Martin Jambor <mjambor@suse.cz>
15141 * ipa-prop.c (ipa_update_after_lto_read): Removed.
15142 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
15143 * ipa-cp.c (ipcp_propagate_stage): Do not call
15144 ipa_update_after_lto_read.
15145 * ipa-inline.c (ipa_inline): Likewise.
15147 2017-05-03 Martin Jambor <mjambor@suse.cz>
15149 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
15150 tag. Added a default constructor and a destructor.
15151 (ipa_edge_args_sum_t): New class;
15152 (ipa_edge_args_sum): Declare.
15153 (ipa_edge_args_vector): Remove declaration.
15154 (IPA_EDGE_REF): Use ipa_edge_args_sum.
15155 (ipa_free_edge_args_substructures): Remove declaration.
15156 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
15157 (ipa_edge_args_info_available_for_edge_p): Likewise.
15158 * ipa-prop.c (ipa_edge_args_vector): Removed.
15159 (edge_removal_hook_holder): Likewise.
15160 (edge_duplication_hook_holder): Likewise.
15161 (ipa_edge_args_sum): New variable.
15162 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
15163 ipa_edge_args_vector.
15164 (ipa_free_edge_args_substructures): Likewise.
15165 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
15166 ipa_edge_args_vector.
15167 (ipa_edge_removal_hook): Turned into method
15168 ipa_edge_args_sum_t::remove.
15169 (ipa_edge_duplication_hook): Turned into method
15170 ipa_edge_args_sum_t::duplicate.
15171 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
15172 registering edge hooks.
15173 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
15174 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
15175 ipa_edge_args_sum instead of ipa_edge_args_vector.
15176 * ipa-profile.c (ipa_profile): Likewise.
15178 2017-05-03 Martin Jambor <mjambor@suse.cz>
15180 * symbol-summary.h (function_summary): New method exists.
15181 (function_summary::symtab_removal): Deallocate through release.
15182 (call_summary): New class.
15183 (gt_ggc_mx): New overload.
15184 (gt_pch_nx): Likewise.
15185 (gt_pch_nx): Likewise.
15187 2017-05-03 Jeff Law <law@redhat.com>
15189 PR tree-optimization/78496
15190 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
15191 from simplify_cond_using_ranges. Split off code to walk
15192 backwards through casts into ...
15193 (simplify_cond_using_ranges_2): New function.
15194 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
15195 (execute_vrp): After identifying jump threads, call
15196 simplify_cond_using_ranges_2.
15198 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
15201 * ipa-inline.h (inline_summary): Add ctor.
15202 (create_ggc): Do not use ggc_cleared_alloc.
15204 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
15205 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15207 * gcc.c (handle_braces): Support escaping in switch matching
15209 * doc/invoke.texi (Spec Files): Document it.
15210 Remove superfluous @code markup in items.
15212 2017-05-03 David Malcolm <dmalcolm@redhat.com>
15214 * diagnostic-show-locus.c (struct column_range): New struct.
15215 (get_affected_columns): New function.
15216 (get_printed_columns): New function.
15217 (struct correction): New struct.
15218 (correction::ensure_capacity): New function.
15219 (correction::ensure_terminated): New function.
15220 (struct line_corrections): New struct.
15221 (line_corrections::~line_corrections): New dtor.
15222 (line_corrections::add_hint): New function.
15223 (layout::print_trailing_fixits): Reimplement in terms of the new
15225 (selftest::test_overlapped_fixit_printing): New function.
15226 (selftest::diagnostic_show_locus_c_tests): Call it.
15228 2017-05-03 Nathan Sidwell <nathan@acm.org>
15230 Canonicalize canonical type hashing
15231 * tree.h (type_hash_canon_hash): Declare.
15232 * tree.c (type_hash_list, attribute_hash_list): Move into
15233 type_hash_canon_hash.
15234 (build_type_attribute_qual_variant): Break out hash code calc into
15235 type_hash_canon_hash.
15236 (type_hash_canon_hash): New. Generic type hash computation.
15237 (build_range_type_1, build_array_type_1, build_function_type,
15238 build_method_type_directly, build_offset_type, build_complex_type,
15239 make_vector_type): Call it.
15241 2017-05-03 Richard Biener <rguenther@suse.de>
15243 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
15244 When all DRs have unknown misaligned do not always peel
15245 when there is a store but apply the same costing model as if
15246 there were only loads.
15248 2017-05-03 Richard Biener <rguenther@suse.de>
15251 PR tree-optimization/80492
15252 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
15253 compare_base_decls returning dont-know properly.
15255 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
15257 * config/arm/iterators.md (CCSI): New mode iterator.
15258 (arch): New mode attribute.
15259 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
15260 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
15261 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
15262 code iterator for success result mode.
15263 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
15264 the corresponding new insn generators.
15266 2017-05-03 Bin Cheng <bin.cheng@arm.com>
15269 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15270 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15272 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
15274 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
15275 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
15276 (DDR_A): Wrap DDR argument in brackets.
15277 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
15278 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
15279 (DDR_REVERSED_P): Likewise.
15281 2017-05-03 Jakub Jelinek <jakub@redhat.com>
15283 PR tree-optimization/79472
15284 * tree-switch-conversion.c (struct switch_conv_info): Add
15285 contiguous_range and default_case_nonstandard fields.
15286 (collect_switch_conv_info): Compute contiguous_range and
15287 default_case_nonstandard fields, don't clear final_bb if
15288 contiguous_range and only the default case doesn't have the required
15290 (check_all_empty_except_final): Set default_case_nonstandard instead
15291 of failing if contiguous_range and the default case doesn't have empty
15293 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
15294 and only the default case doesn't have the required constants. Skip
15296 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
15297 if default_case_nonstandard.
15298 (build_constructors): Build constant 1 just once. Assert that default
15299 values aren't inserted in between cases if contiguous_range. Skip
15301 (build_arrays): Skip virtual phis.
15302 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
15303 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
15304 Handle virtual phis.
15305 (gen_inbound_check): Handle default_case_nonstandard case.
15306 (process_switch): Adjust check_final_bb caller. Call
15307 gather_default_values with the first non-default case instead of
15308 default case if default_case_nonstandard.
15310 2017-05-02 Nathan Sidwell <nathan@acm.org>
15312 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
15313 check. Fix formatting.
15315 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
15317 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
15318 errors when comparing specialized and unspecialized times.
15320 2017-05-02 David Malcolm <dmalcolm@redhat.com>
15322 * diagnostic-show-locus.c
15323 (layout::should_print_annotation_line_p): Make private.
15324 (layout::print_annotation_line): Make private.
15325 (layout::annotation_line_showed_range_p): Make private.
15326 (layout::show_ruler): Make private.
15327 (layout::print_source_line): Make private. Pass in line and
15328 line_width, rather than calling location_get_source_line. Drop
15330 (layout::print_leading_fixits): New method.
15331 (layout::print_any_fixits): Rename to...
15332 (layout::print_trailing_fixits): ...this, and make private.
15333 Don't print newline fixits.
15334 (diagnostic_show_locus): Move logic for printing one row into...
15335 (layout::print_line): ...this new function. Move the
15336 location_get_source_line call and error-handling from
15337 print_source_line to here. Call print_leading_fixits, and rename
15338 print_any_fixits to print_trailing_fixits.
15339 (selftest::test_fixit_insert_containing_newline): Update now that
15340 newlines are partially supported.
15341 (selftest::test_fixit_insert_containing_newline_2): New test.
15342 (selftest::test_fixit_replace_containing_newline): Update comments.
15343 (selftest::diagnostic_show_locus_c_tests): Call the new test.
15344 * edit-context.c (class added_line): New class.
15345 (class edited_line): Describe newline handling in comment.
15346 (edited_line::actually_edited_p): New method.
15347 (edited_line::print_content): Delete redundant decl.
15348 (edited_line::m_predecessors): New field.
15349 (edited_file::print_content): Call edited_line::print_content.
15350 (edited_file::print_diff): Update to support newlines.
15351 (edited_file::print_diff_hunk): Likewise.
15352 (edited_file::print_run_of_changed_lines): New function.
15353 (edited_file::print_diff_line): Convert to...
15354 (print_diff_line): ...this.
15355 (edited_file::get_effective_line_count): New function.
15356 (edited_line::edited_line): Initialize new field m_predecessors.
15357 (edited_line::~edited_line): Clean up m_predecessors.
15358 (edited_line::apply_fixit): Handle newlines.
15359 (edited_line::get_effective_line_count): New function.
15360 (edited_line::print_content): New function.
15361 (edited_line::print_diff_lines): New function.
15362 (selftest::test_applying_fixits_insert_containing_newline): New
15364 (selftest::test_applying_fixits_replace_containing_newline): New
15366 (selftest::insert_line): New function.
15367 (selftest::test_applying_fixits_multiple_lines): Add example of
15369 (selftest::edit_context_c_tests): Call the new tests.
15371 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15373 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
15374 parameter cand. Update dump information.
15375 (get_computation_cost): Update uses.
15377 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15379 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
15380 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
15381 (get_computation_at, rewrite_use_address): Update use of
15382 get_computation_aff.
15384 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15386 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
15387 (get_computation): Delete.
15388 (get_computation_cost): Implement like get_computation_cost_at.
15389 Use get_computation_at.
15390 (get_computation_cost_at): Delete.
15391 (rewrite_use_nonlinear_expr): Use get_computation_at.
15392 (rewrite_use_compare, remove_unused_ivs): Ditto.
15394 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15396 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
15398 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15400 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
15401 (ivopts_global_cost_for_size): Rename parameter and update uses.
15402 (iv_ca_recount_cost): Update uses.
15403 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
15404 candidates seperately in n_invs and n_cands.
15405 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
15407 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15409 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
15410 (find_inv_vars_cb): New.
15411 (find_depends): Renamed to ...
15412 (find_inv_vars): ... this.
15413 (add_candidate_1, force_var_cost): Call find_inv_vars.
15414 (split_address_cost, determine_group_iv_cost_cond): Ditto.
15416 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15418 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
15419 inv_vars. Add inv_exprs.
15420 (struct iv_cand): Rename depends_on to inv_vars.
15421 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
15422 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
15423 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
15424 (dump_cand): Dump inv_vars.
15425 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
15426 (record_invariant, find_depends, add_candidate_1): Ditto.
15427 (set_group_iv_cost, force_var_cost): Ditto.
15428 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
15429 (get_computation_cost_at, get_computation_cost): Ditto.
15430 (determine_group_iv_cost_generic): Ditto.
15431 (determine_group_iv_cost_address): Ditto.
15432 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
15433 (determine_group_iv_costs): Ditto.
15434 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
15435 (iv_ca_set_remove_invariants): Renamed to ...
15436 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
15437 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
15438 (iv_ca_set_add_invariants): Renamed to ...
15439 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
15440 (iv_ca_set_cp): Use iv_ca_set_add_invs.
15441 (iv_ca_has_deps): Support inv_vars and inv_exprs.
15442 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
15443 (create_new_ivs): Remove useless dump.
15445 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15447 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
15449 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
15450 (iv_ca_set_no_cp, create_new_iv): Ditto.
15452 2017-05-02 Bin Cheng <bin.cheng@arm.com>
15454 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15456 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
15458 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
15459 function tree_check2.
15461 2017-05-02 Martin Liska <mliska@suse.cz>
15463 * doc/gcov.texi: Add missing preposition.
15464 * gcov.c (function_info::function_info): Properly fill up
15465 all member variables.
15467 2017-05-02 Tamar Christina <tamar.christina@arm.com>
15469 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
15471 2017-05-02 Tamar Christina <tamar.christina@arm.com>
15473 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
15475 2017-05-02 Martin Liska <mliska@suse.cz>
15478 * lto-streamer-in.c (lto_read_tree_1): Remove
15479 LTO_STREAMER_DEBUG.
15480 * lto-streamer.c (struct tree_hash_entry): Likewise.
15481 (struct tree_entry_hasher): Likewise.
15482 (tree_entry_hasher::hash): Likewise.
15483 (tree_entry_hasher::equal): Likewise.
15484 (lto_streamer_init): Likewise.
15485 (lto_orig_address_map): Likewise.
15486 (lto_orig_address_get): Likewise.
15487 (lto_orig_address_remove): Likewise.
15488 * lto-streamer.h: Likewise.
15489 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
15490 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
15492 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
15494 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
15495 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
15496 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
15497 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
15498 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
15499 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
15500 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
15501 (mm_maskz_sub_ss): New intrinsics.
15502 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15503 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
15504 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
15505 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
15506 (__builtin_ia32_subss_mask_round): New builtins.
15507 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15508 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15509 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
15511 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
15512 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
15514 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
15517 2017-05-02 Martin Jambor <mjambor@suse.cz>
15519 PR tree-optimization/78687
15520 * tree-sra.c (access): New field parent.
15521 (process_subtree_disqualification): New function.
15522 (disqualify_candidate): Call it.
15523 (build_accesses_from_assign): Reset write flag if creating an
15525 (build_access_subtree): Fill in parent field and also prpagate
15526 down grp_write flag.
15527 (create_artificial_child_access): New parameter set_grp_write, set
15528 grp_write to its value.
15529 (propagate_subaccesses_across_link): Also propagate grp_write flag
15531 (propagate_all_subaccesses): Push the closest parent back to work
15532 queue if add_access_to_work_queue returned true.
15534 2017-05-02 Richard Biener <rguenther@suse.de>
15536 * common.opt (fstrict-overflow): Alias negative to fwrapv.
15537 * doc/invoke.texi (fstrict-overflow): Remove all traces of
15538 -fstrict-overflow documentation.
15539 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
15540 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
15541 flag_strict_overflow.
15542 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
15543 * lto-opts.c (lto_write_options): Do not stream it.
15544 * lto-wrapper.c (merge_and_complain): Do not handle it.
15545 * opts.c (default_options_table): Do not set -fstrict-overflow.
15546 (finish_options): Likewise do not clear it when sanitizing.
15547 * simplify-rtx.c (simplify_const_relational_operation): Do not
15548 test flag_strict_overflow.
15550 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
15552 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
15553 using enabled attribute.
15554 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
15555 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
15556 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
15557 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
15558 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
15559 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
15560 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
15561 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
15562 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
15563 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
15565 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
15567 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
15569 2017-05-02 Richard Biener <rguenther@suse.de>
15571 PR tree-optimization/80591
15573 2017-04-10 Richard Biener <rguenther@suse.de>
15575 * tree-ssa-structalias.c (find_func_aliases): Properly handle
15578 2017-05-02 Richard Biener <rguenther@suse.de>
15580 PR tree-optimization/80549
15581 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
15582 (cleanup_tree_cfg_noloop): Create forwarders to known loop
15583 headers if they do not have a preheader.
15585 2017-05-02 Martin Liska <mliska@suse.cz>
15588 * common.opt: Fix typo.
15589 * doc/invoke.texi: Likewise.
15591 2017-05-01 Jan Beulich <jbeulich@suse.com>
15593 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
15594 swapping, add (x,x,m,x,n) alternative.
15596 2017-05-01 Nathan Sidwell <nathan@acm.org>
15598 * calls.c (combine_pending_stack_adjustment_and_call): Remove
15599 unnecessary unadjusted_alignment check.
15601 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
15604 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
15606 * gimplify.c (gimplify_cilk_detach): New function.
15607 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
15608 * tree-core.h: Document EXPR_CILK_SPAWN.
15609 * tree.h (EXPR_CILK_SPAWN): Define.
15611 2017-05-01 David Malcolm <dmalcolm@redhat.com>
15613 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
15614 to use new fixit_hint representation, using the "replace" logic.
15615 (get_line_span_for_fixit_hint): Likewise.
15616 (layout::print_any_fixits): Likewise.
15617 (selftest::test_one_liner_many_fixits): Rename to...
15618 (selftest::test_one_liner_many_fixits_1): ...this, and update
15619 comment and expected output to reflect that the multiple fix-it
15620 hints are now consolidated into one insertion.
15621 (selftest::test_one_liner_many_fixits_2): New test.
15622 (selftest::test_diagnostic_show_locus_one_liner): Update for
15624 (selftest::test_fixit_consolidation): Update for fix-it API
15626 * diagnostic.c (print_parseable_fixits): Likewise.
15627 * edit-context.c (edited_line::m_line_events): Convert from
15628 auto_vec <line_event *> to auto_vec <line_event>.
15629 (class line_event): Convert from abstract base class to a concrete
15630 class, taking over the role of replace_event.
15631 (class insert_event): Delete.
15632 (class replace_event): Rename to class line_event. Convert to
15634 (edit_context::add_fixits): Reimplement.
15635 (edit_context::apply_insert): Delete.
15636 (edit_context::apply_replace): Rename to...
15637 (edit_context::apply_fixit): ...this. Convert to half-open range.
15638 (edited_file::apply_insert): Delete.
15639 (edited_file::apply_replace): Rename to...
15640 (edited_file::apply_fixit): ...this.
15641 (edited_line::~edited_line): Drop deletion of events.
15642 (edited_line::apply_insert): Delete.
15643 (edited_line::apply_replace): Rename to...
15644 (edited_line::apply_fixit): ...this. Convert to half-open range.
15645 Update for change to type of m_line_events.
15646 * edit-context.h (edit_context::apply_insert): Delete.
15647 (edit_context::apply_replace): Rename to...
15648 (edit_context::apply_fixit): ...this.
15650 2017-05-01 Martin Sebor <msebor@redhat.com>
15652 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
15655 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
15658 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
15659 __get_cpuid_max returns 0.
15660 (__get_cpuid_count): Ditto.
15662 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
15664 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
15665 replacement expression is another instance of one of its arguments.
15667 2017-05-01 Jakub Jelinek <jakub@redhat.com>
15670 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
15671 check for stack push/pop autoinc.
15672 * config/i386/i386.c (ix86_agi_dependent): Return false
15673 if the only reason why modified_in_p returned true is that
15674 addr is SP based and set_insn is a push or pop.
15676 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
15678 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
15681 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
15684 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
15685 (account_size_time): Use two predicates - exec_pred and
15687 (evaluate_conditions_for_known_args): Compute both clause and
15689 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
15690 (inline_summary_t::duplicate): Update.
15691 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
15693 (compute_inline_parameters): Likewise.
15694 (estimate_edge_size_and_time): Update caluclation of time.
15695 (estimate_node_size_and_time): Compute both time and nonspecialized
15697 (estimate_ipcp_clone_size_and_time): Update.
15698 (inline_merge_summary): Update.
15699 (do_estimate_edge_time): Update.
15700 (do_estimate_edge_size): Update.
15701 (do_estimate_edge_hints): Update.
15702 (inline_read_section, inline_write_summary): Stream both new predicates.
15703 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
15705 (compute_inlined_call_time): Cleanup.
15706 (big_speedup_p): Update.
15707 (edge_badness): Update.
15708 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
15709 (size_time_entry): Replace predicate by exec_predicate and
15710 nonconst_predicate.
15711 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
15712 (estimate_edge_time): Return also nonspec_time.
15713 (reset_edge_growth_cache): Update.
15715 2017-04-29 Jakub Jelinek <jakub@redhat.com>
15717 PR rtl-optimization/80491
15718 * ifcvt.c (noce_process_if_block): When looking for x setter
15719 with missing else_bb, don't check only the insn right before
15720 cond_earliest, but look for the last insn that x is modified in
15721 within the same bb.
15723 PR rtl-optimization/80491
15724 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
15726 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
15728 PR tree-optimization/80487
15729 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
15731 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15733 PR tree-optimization/79697
15734 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
15735 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
15736 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
15738 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
15739 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
15741 2017-04-28 Martin Sebor <msebor@redhat.com>
15743 PR tree-optimization/80523
15744 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
15745 (init_target_to_host_charmap, target_to_host, target_strtol10): New
15747 (maybe_warn, format_directive, parse_directive): Use new functions.
15748 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
15750 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
15752 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
15754 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
15756 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
15757 target_header_dir): Set correctly.
15758 * configure: Regenerated.
15759 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15760 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15761 instead of SYSTEM_HEADER_DIR.
15763 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
15765 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
15766 (estimate_local_effects): Likewise.
15767 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
15768 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
15769 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
15770 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
15771 do_estimate_edge_time, estimate_edge_time): Likewise.
15772 * ipa-inline-analysis.c (estimate_node_size_and_time,
15773 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
15774 (estimate_time_after_inlining): Remove.
15776 2017-04-28 Martin Liska <mliska@suse.cz>
15778 * doc/gcov.texi: Enhance documentation of gcov.
15780 2017-04-28 Martin Liska <mliska@suse.cz>
15782 * doc/gcov.texi: Sort options in alphabetic order.
15783 * doc/gcov-dump.texi: Likewise.
15784 * doc/gcov-tool.texi: Likewise.
15785 * gcov.c (print_usage): Likewise.
15786 * gcov-dump.c (print_usage): Likewise.
15787 * gcov-tool.c (print_merge_usage_message): Likewise.
15788 (print_rewrite_usage_message): Likewise.
15789 (print_overlap_usage_message): Likewise.
15791 2017-04-28 Martin Liska <mliska@suse.cz>
15793 PR gcov-profile/53915
15794 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
15796 2017-04-28 Martin Liska <mliska@suse.cz>
15798 PR gcov-profile/79891
15799 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
15800 is marked by compiler as living on a line.
15801 (get_cycles_count): Remove usage of the union.
15802 (output_intermediate_file): Likewise.
15803 (find_source): Fix GNU coding style.
15804 (accumulate_line_counts): Remove old non-all block mode.
15805 (output_lines): Remove usage of the union.
15806 * profile.c (output_location): Include all BBs, even if
15807 belonging to a same line (and file) as a previous BB.
15809 2017-04-28 Martin Liska <mliska@suse.cz>
15811 * gcov.c (process_args): Handle new argument 'w'.
15812 (read_graph_file): Assign ID to BBs.
15813 (output_branch_count): Display BB # if verbose flag is set.
15814 (output_lines): Likewise for arcs.
15815 (print_usage): Add '--verbose' option help.
15816 * doc/gcov.texi: Document --verbose (-w) option.
15818 2017-04-28 Martin Liska <mliska@suse.cz>
15820 * gcov.c (struct block_location_info): New struct.
15821 (process_file): Fill up the new structure.
15822 (read_graph_file): Replace usage of encoding by the newly added
15824 (add_line_counts): Likewise.
15825 (accumulate_line_counts): Remove usage of the union.
15826 (function_info::function_info): New function.
15827 (function_info::~function_info): Likewise.
15828 (process_file): Call delete instead of release_function.
15829 (release_function): Release the function.
15830 (release_structures): Call delete instead of release_function.
15831 (solve_flow_graph): Replace usage of num_blocks.
15832 (find_exception_blocks): Likewise.
15833 (output_lines): Fix GNU coding style.
15835 2017-04-28 Martin Liska <mliska@suse.cz>
15838 * coverage.c (coverage_remove_note_file): New function.
15839 * coverage.h: Declare the function.
15840 * toplev.c (finalize): Clean if an error has been seen.
15842 2017-04-28 Martin Liska <mliska@suse.cz>
15844 PR gcov-profile/80031
15845 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
15846 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
15847 * gcov.c (read_graph_file): Read just number of blocks.
15848 * profile.c (branch_prob): Do not stream 0 flags per a basic
15851 2017-04-28 Martin Liska <mliska@suse.cz>
15853 * gcov-dump.c (tag_*): Add new argument to declarations.
15854 (dump_gcov_file): Likewise.
15855 (tag_blocks): Add and use new argument depth.
15856 (tag_arcs): Likewise.
15857 (tag_lines): Likewise.
15858 (tag_counters): Likewise.
15859 (tag_summary): Likewise.
15860 (dump_working_sets): Use depth to do a proper indentation.
15862 2017-04-28 Jakub Jelinek <jakub@redhat.com>
15865 * cgraph.h (symtab_node::debug_symtab): No longer inline.
15866 * symtab.c (symtab_node::debug_symtab): Move definition here.
15868 2017-04-28 Richard Biener <rguenther@suse.de>
15870 * lto-streamer.h (LTO_major_version): Bump to 7.
15872 2017-04-28 Richard Biener <rguenther@suse.de>
15874 * tree-vrp.c (assert_info): New struct.
15875 (add_assert_info): New helper.
15876 (register_edge_assert_for_2): Refactor to add asserts to a vector
15878 (register_edge_assert_for_1): Likewise.
15879 (register_edge_assert_for): Likewise.
15880 (finish_register_edge_assert_for): New helper actually registering
15881 asserts where live on edge.
15882 (find_conditional_asserts): Adjust.
15883 (find_switch_asserts): Likewise.
15884 (evrp_dom_walker::try_find_new_range): Generalize.
15885 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
15887 2017-04-27 Marek Polacek <polacek@redhat.com>
15890 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
15891 arg10 and arg11 to itype.
15893 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
15895 * doc/extend.texi (Object Size Checking): Improve grammar.
15897 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
15900 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
15901 that the logic for permitting reciprocal estimates matches that
15904 2017-04-27 Jakub Jelinek <jakub@redhat.com>
15907 * tree.c (type_cache_hasher::equal): Only compare
15908 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
15909 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
15910 non-aggregate element types.
15911 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
15912 about the flag on ARRAY_TYPEs in the comment, formatting fix.
15914 2017-04-27 Richard Biener <rguenther@suse.de>
15916 PR middle-end/80533
15917 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
15918 stripping ARRAY_REFs from MEM_EXPR make sure we're not
15919 keeping a reference to a trailing array.
15921 2017-04-27 Richard Biener <rguenther@suse.de>
15923 PR middle-end/80539
15924 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
15925 being in loop-closed SSA form conservatively.
15926 (chrec_fold_multiply_poly_poly): Likewise.
15928 2017-04-27 Tamar Christina <tamar.christina@arm.com>
15930 PR middle-end/79665
15931 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
15932 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
15934 2017-04-27 Jakub Jelinek <jakub@redhat.com>
15937 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
15938 (aarch64_function_arg_alignment): Return unsigned int again, but still
15939 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
15940 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
15941 Don't emit -Wpsabi note.
15942 (aarch64_function_arg_boundary): Likewise.
15943 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
15946 2017-04-26 Nathan Sidwell <nathan@acm.org>
15948 * tree.h (crc32_unsigned_n): Declare.
15949 (crc32_unsigned, crc32_unsigned): Make inline.
15950 * tree.c (crc32_unsigned_bits): Replace with ...
15951 (crc32_unsigned_n): ... this.
15952 (crc32_unsigned, crc32_byte): Remove.
15953 (crc32_string): Remove unnecessary braces.
15955 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15957 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
15958 * ipa-inline-analysis.c (MAX_TIME): Remove.
15959 (account_size_time): Use sreal for time.
15960 (dump_inline_summary): Update.
15961 (estimate_function_body_sizes): Update.
15962 (estimate_edge_size_and_time): Update.
15963 (estimate_calls_size_and_time): Update.
15964 (estimate_node_size_and_time): Update.
15965 (inline_merge_summary): Update.
15966 (inline_update_overall_summary): Update.
15967 (estimate_time_after_inlining): Update.
15968 (inline_read_section): Update.
15969 (inline_write_summary): Update.
15970 * ipa-inline.c (compute_uninlined_call_time): Update.
15971 (compute_inlined_call_time): Update.
15972 (recursive_inlining): Update.
15973 (inline_small_functions): Update.
15974 (dump_overall_stats): Update.
15975 * ipa-inline.h: Include sreal.h.
15976 (size_time_entry): Turn time to sreal.
15977 (inline_summary): Turn self_time nad time to sreal.
15979 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15981 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
15983 (sreal::stream_out, sreal::stream_in): New.
15984 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
15986 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15988 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
15991 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
15994 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
15996 (dimode_scalar_chain::compute_convert_gain): Ditto.
15997 (dimode_scalar_chain::make_vector_copies): Ditto.
15998 (dimode_scalar_chain::convert_reg): Ditto.
15999 (dimode_scalar_chain::convert_insn): Ditto.
16000 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
16001 (VI248_AVX512BW_1): New mode iterator.
16002 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
16003 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
16006 2017-04-25 Martin Sebor <msebor@redhat.com>
16008 PR tree-optimization/80497
16009 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
16010 constants are representable in HOST_WIDE_INT.
16011 (parse_directive): Ditto.
16013 2017-04-25 Martin Sebor <msebor@redhat.com>
16016 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
16017 (new_zero_array): Adjust signature.
16018 (dom_info::dom_init): Used unsigned rather that size_t.
16019 (dom_info::dom_info): Same.
16021 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16022 Jakub Jelinek <jakub@redhat.com>
16025 * config/arm/arm.c: Include gimple.h.
16026 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
16027 returns negative, increment ncrn only if it returned positive.
16028 (arm_needs_doubleword_align): Return int instead of bool,
16029 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
16030 members, but if there is any such non-FIELD_DECL
16031 > PARM_BOUNDARY aligned decl, return -1 instead of false.
16032 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
16033 returns negative, increment nregs only if it returned positive.
16034 (arm_setup_incoming_varargs): Likewise.
16035 (arm_function_arg_boundary): Emit -Wpsabi note if
16036 arm_needs_doubleword_align returns negative, return
16037 DOUBLEWORD_ALIGNMENT only if it returned positive.
16039 2017-04-25 Marek Polacek <polacek@redhat.com>
16042 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
16043 first argument to type.
16045 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
16048 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
16049 type checks to test for compatibility instead of equality.
16051 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16052 Jakub Jelinek <jakub@redhat.com>
16055 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
16057 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
16058 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
16059 the alignment computation, but return their maximum in warn_alignment.
16060 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
16061 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
16063 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
16064 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
16067 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16069 * config/arc/simdext.md (dmpyh): Fix typo.
16071 2017-04-25 Richard Biener <rguenther@suse.de>
16073 PR tree-optimization/80492
16074 * alias.c (compare_base_decls): Handle registers with asm
16075 specification conservatively.
16076 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16077 compare_base_decls returning dont-know properly.
16079 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16081 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
16082 (legitimate_offset_address_p): New function.
16083 (arc_legitimate_address_p): Use above function.
16085 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16087 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
16089 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16091 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
16092 ACCH registers whenever they are available.
16094 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16096 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
16097 double regs fix when not used.
16099 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16101 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
16103 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
16104 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
16106 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16108 * config/arc/arc.c (arc_output_addsi): Check for h-register class
16109 when emitting short ADD instructions.
16111 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16113 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
16115 (cmpsi_cc_c_insn): Likewise.
16116 (cbranchsi4_scratch): Compute proper instruction length using
16117 compact_hreg_operand.
16118 * config/arc/predicates.md (compact_hreg_operand): New predicate.
16120 2017-04-25 Richard Biener <rguenther@suse.de>
16122 PR middle-end/80509
16123 * passes.c (pass_manager::pass_manager): Initialize
16124 m_name_to_pass_map.
16126 2017-04-25 Richard Biener <rguenther@suse.de>
16128 PR tree-optimization/79201
16129 * tree-ssa-sink.c (statement_sink_location): Handle calls.
16131 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16134 * config/s390/vector.md: Split MEM->GPR vector moves for
16135 non-s_operand addresses.
16137 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16140 * config/s390/predicates.md (reload_const_wide_int_operand): New
16142 * config/s390/s390.md ("movti"): Remove d/P alternative.
16143 ("movti_bigconst"): New pattern definition.
16145 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16148 * s390-protos.h (s390_expand_cs_hqi): Removed.
16149 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
16150 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
16151 modes as well as CCZ1mode and CCZmode.
16152 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
16153 signature of s390_emit_compare_and_swap.
16154 (s390_expand_cs_hqi): Likewise, make static.
16155 (s390_expand_cs_tdsi): Generate an explicit compare before trying
16156 compare-and-swap, in some cases.
16157 (s390_expand_cs): Wrapper function.
16158 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
16160 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
16161 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
16162 patterns for small and large integers. Forbid symref memory operands.
16163 Move expander to s390.c. Require cc register.
16164 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
16165 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
16166 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
16167 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
16168 symref memory operands. Remove CC mode and call s390_match_ccmode
16170 ("atomic_exchange<mode>"): Allow and implement all integer modes.
16172 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16174 * config/s390/s390.md (define_peephole2): New peephole to help
16175 combining the load-and-test pattern with volatile memory.
16177 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16179 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
16180 with CCZmode for TARGET_Z196.
16182 2017-04-25 Jakub Jelinek <jakub@redhat.com>
16184 PR rtl-optimization/80501
16185 * combine.c (make_compound_operation_int): Set subreg_code to SET
16186 even for AND with mask of the sign bit of mode.
16188 PR rtl-optimization/80500
16189 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
16190 sum's initial value.
16192 2017-04-25 Julian Brown <julian@codesourcery.com>
16193 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16195 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
16197 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
16199 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
16201 2017-04-25 Julian Brown <julian@codesourcery.com>
16202 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16204 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
16205 (thunderx2t99_sha): New Reservation.
16207 2017-04-25 Julian Brown <julian@codesourcery.com>
16208 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16210 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
16211 type for 1-element load.
16213 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
16215 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
16217 2017-04-24 Martin Jambor <mjambor@suse.cz>
16219 PR tree-optimization/80293
16220 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
16221 char arrays not totally scalarizable if it is false.
16222 (analyze_all_variable_accesses): Pass correct value in the new
16223 parameter. Add a statistics counter.
16225 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
16227 PR middle-end/79931
16228 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
16230 2017-04-24 Richard Biener <rguenther@suse.de>
16232 PR tree-optimization/80494
16233 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
16234 out for complex types.
16236 2017-04-24 Richard Biener <rguenther@suse.de>
16238 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
16239 * tree-ssa-sccvn.c (print_scc): Print SCC size.
16240 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
16241 (DFS): Adjust and never fail.
16242 (sccvn_dom_walker::fail): Remove.
16243 (sccvn_dom_walker::before_dom_children): Adjust.
16244 (run_scc_vn): Likewise and never fail.
16245 * tree-ssa-pre.c (pass_pre::execute): Adjust.
16246 (pass_fre::execute): Likewise.
16248 2017-04-24 Richard Biener <rguenther@suse.de>
16250 PR tree-optimization/79725
16251 * tree-ssa-sink.c (statement_sink_location): Return whether
16252 failure reason was zero uses. Move that check later.
16253 (sink_code_in_bb): Deal with zero uses by removing the stmt
16256 2017-04-24 Richard Biener <rguenther@suse.de>
16259 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
16260 pointer-based references.
16262 2017-04-24 Richard Biener <rguenther@suse.de>
16265 * pass_manager.h (pass_manager::operator new): Remove.
16266 (pass_manager::operator delete): Likewise.
16267 * passes.c (pass_manager::operator new): Remove.
16268 (pass_manager::operator delete): Likewise.
16269 (pass_manager::pass_manager): Zero individual pass members.
16271 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
16274 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
16275 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
16276 Check "XEXP (src, 1)" operand here.
16277 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
16278 Check "XEXP (src, 1)" operand here.
16279 (dimode_scalar_chain::make_vector_copies): Detect count register
16280 of a shift instruction. Zero extend count register from QImode
16281 to DImode to satisfy vector shift pattern count operand predicate.
16282 Substitute vector shift count operand with a DImode copy.
16283 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
16286 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
16288 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
16289 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
16290 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
16291 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
16292 (UNSPEC_NOREX_MEM): Remove definition.
16294 2017-04-21 Richard Biener <rguenther@suse.de>
16296 PR tree-optimization/79547
16297 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
16298 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
16299 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
16300 without any constraints.
16302 2017-04-21 Richard Biener <rguenther@suse.de>
16304 PR tree-optimization/78847
16305 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
16307 2017-04-21 Richard Biener <rguenther@suse.de>
16309 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
16310 (build_distinct_type_copy): Likewise.
16311 (build_variant_type_copy): Likewise.
16312 * tree.c (build_qualified_type): Pass down mem-stat info.
16313 (build_distinct_type_copy): Likewise.
16314 (build_variant_type_copy): Likewise.
16316 2017-04-21 Richard Biener <rguenther@suse.de>
16318 PR tree-optimization/80237
16319 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
16321 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
16322 for a simplified result.
16324 2016-04-21 Richard Biener <rguenther@suse.de>
16326 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
16327 sth as strict as a simple_iv but a chrec without symbols and an
16328 operand defined in the loop we are peeling (and not some subloop).
16329 (propagate_constants_for_unrolling): Propagate all constants.
16331 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
16334 * config/i386/i386.c (print_reg): Remove assert for disalowed
16335 regno values, call output_operand_lossage instead.
16337 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
16340 * config/i386/constraints.md (Yc): New register constraint.
16341 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
16342 Use Yc constraint for alternative 2 of operand 0. Remove
16343 preferred_for_speed attribute.
16345 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
16347 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
16348 lastprivate clauses in SIMT case.
16350 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
16352 * doc/invoke.texi (-Wextra-semi): Document new warning option.
16354 2017-04-20 Richard Biener <rguenther@suse.de>
16356 PR tree-optimization/57796
16357 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
16358 as N scalar stores.
16359 (vect_model_load_cost): Cost gathers as N scalar loads.
16361 2017-04-20 Richard Biener <rguenther@suse.de>
16363 * ggc-page.c (ggc_allocated_p): Rename to ...
16364 (safe_lookup_page_table_entry): ... this and return the lookup
16366 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
16368 2017-04-20 Richard Biener <rguenther@suse.de>
16370 PR tree-optimization/80453
16371 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
16372 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
16373 from the conditions.
16374 (vn_phi_eq): Pass them down.
16375 (vn_phi_lookup): Record them.
16376 (vn_phi_insert): Likewise.
16378 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
16380 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
16381 uninitialized variable warning to avoid buffer overrun.
16383 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
16386 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
16387 is suppressed for '{ 0 }' in C.
16389 2017-04-20 Jakub Jelinek <jakub@redhat.com>
16391 * BASE-VER: Set to 8.0.0.
16393 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
16395 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
16396 priority .init_array and .fini_array section with SECTION_NOTYPE
16399 2017-04-20 Jakub Jelinek <jakub@redhat.com>
16401 PR middle-end/80423
16402 * tree.h (build_array_type): Add typeless_storage default argument.
16403 * tree.c (type_cache_hasher::equal): Also compare
16404 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
16405 (build_array_type): Add typeless_storage argument, set
16406 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
16408 (build_nonshared_array_type): Adjust build_array_type_1 caller.
16409 (build_array_type): Likewise. Add typeless_storage argument.
16411 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
16412 Jakub Jelinek <jakub@redhat.com>
16414 PR tree-optimization/80426
16415 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
16416 operation on symbolic operands, also compute the overflow for the
16417 invariant part when the operation degenerates into a negation.
16419 2017-04-19 Jakub Jelinek <jakub@redhat.com>
16422 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
16423 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
16426 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
16428 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
16431 * config/avr/avr.c (tree.h): Include it.
16432 (cgraph.h): Include it.
16433 (avr_encode_section_info): Don't warn for uninitialized progmem
16434 variable if it's just an alias.
16436 2017-04-19 Richard Biener <rguenther@suse.de>
16439 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
16440 when needed by AutoPGO.
16442 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
16445 * doc/lto.texi: Remove an extra 'that'.
16447 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
16449 PR rtl-optimization/80429
16450 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
16451 are only used in debug insns.
16453 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
16454 Vladimir Makarov <vmakarov@redhat.com>
16456 * config/sparc/predicates.md (input_operand): Add comment. Return
16457 true for any memory operand when LRA is in progress.
16458 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
16460 2017-04-18 Jeff Law <law@redhat.com>
16463 * mips.md ({return,simple_return}_internal): Do not overwrite
16466 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16468 PR tree-optimization/80443
16469 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
16470 instead of adding 1, subtract -1 and similarly instead of subtracting
16473 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
16475 PR rtl-optimization/80357
16476 * haifa-sched.c (tmp_bitmap): New variable.
16477 (model_recompute): Handle duplicate use records.
16478 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
16479 (free_global_sched_pressure_data): Free it.
16481 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
16484 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
16485 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
16486 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
16487 instead of SYSTEM_HEADER_DIR.
16489 2017-04-18 Jeff Law <law@redhat.com>
16491 PR middle-end/80422
16492 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
16493 predecessors after walking up the insn chain.
16495 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16498 * dwarf2out.c (modified_type_die): Try harder not to emit internal
16499 sizetype type into debug info.
16501 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
16504 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
16505 unneeded test for TARGET_UPPER_REGS_SF.
16506 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
16508 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16511 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
16512 instead of gsi_after_labels.
16514 2017-04-18 Jeff Law <law@redhat.com>
16516 * regcprop.c (maybe_mode_change): Avoid creating copies of the
16520 2017-04-13 Jeff Law <law@redhat.com>
16521 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
16522 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
16524 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
16527 * config/avr/avr.c (intl.h): Include it.
16528 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
16530 2017-04-18 Martin Liska <mliska@suse.cz>
16532 PR gcov-profile/78783
16533 * gcov-tool.c (gcov_output_files): Validate that destination
16534 file is either removed by the tool or by a user.
16536 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
16537 Guy Benyei <guybe@mellanox.com>
16539 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
16540 block, and do not negate it, the stored id is already negative.
16542 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
16544 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
16546 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
16549 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
16550 masks of options that should be turned off if the VSX vector
16551 options are turned off.
16552 (OTHER_P8_VECTOR_MASKS): Likewise.
16553 (OTHER_VSX_VECTOR_MASKS): Likewise.
16554 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
16555 rs6000_disable_incompatible_switches to validate no type switches
16557 (rs6000_incompatible_switch): New function to disallow turning on
16558 other vector options if -mno-vsx, -mno-power8-vector, or
16559 -mno-power9-vector are specified.
16561 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16563 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
16565 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16567 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
16568 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
16569 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
16570 (ARG_POINTER_CFA_OFFSET): Likewise.
16572 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16574 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
16575 conditions to take advantage of various optimizations.
16577 2017-04-13 Jeff Law <law@redhat.com>
16579 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
16580 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
16581 (zero_extendsidi2_dext): Likewise.
16583 2017-04-13 Jakub Jelinek <jakub@redhat.com>
16586 * fold-const.c (fold_ternary_loc): Revert
16587 use op0 instead of fold_convert_loc (loc, type, arg0) part of
16590 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
16592 PR rtl-optimization/80343
16593 * lra-remat.c (update_scratch_ops): Assign original hard reg to
16594 new scratch pseudo.
16596 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
16599 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
16600 to ubsan_encode_value.
16602 2017-04-13 Jeff Law <law@redhat.com>
16604 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
16605 appearing in DEBUG_INSNs.
16607 2017-04-13 Martin Liska <mliska@suse.cz>
16609 PR gcov-profile/80413
16610 * gcov-io.c (gcov_write_string): Copy to buffer just when
16611 allocated size is greater than zero.
16613 2017-04-13 Jakub Jelinek <jakub@redhat.com>
16616 * dwarf2out.c (decls_for_scope): Ignore declarations of
16617 current_function_decl in BLOCK_NONLOCALIZED_VARS.
16619 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
16622 * ipa-visibility.c (non_local_p): Fix typos.
16623 (localize_node): When localizing symbol in same comdat group,
16624 dissolve the group only when we know external symbols are going
16626 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
16628 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16630 PR tree-optimization/79390
16631 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
16632 order does not result in usable sequence, retry with reversed operand
16638 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
16639 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
16640 op0 instead of fold_convert_loc (loc, type, arg0).
16642 2017-04-12 Jeff Law <law@redhat.com>
16644 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
16645 has a delay slot in the generated code.
16647 * config/cris/cris.md (cris_preferred_reload_class): Return
16648 GENNONACR_REGS rather than GENERAL_REGS.
16650 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16653 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
16654 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
16655 signedness of the result type.
16657 2017-04-12 Richard Biener <rguenther@suse.de>
16658 Jeff Law <law@redhat.com>
16660 PR tree-optimization/80359
16661 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
16662 trim stores to TARGET_MEM_REFs.
16664 2017-04-12 Richard Biener <rguenther@suse.de>
16666 PR tree-optimization/79390
16667 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
16668 threading case even more.
16670 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
16673 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
16674 for quad_address_p for TImode, instead of just not indexed_address.
16676 2017-04-12 Richard Biener <rguenther@suse.de>
16677 Bernd Edlinger <bernd.edlinger@hotmail.de>
16679 PR middle-end/79671
16680 * alias.c (component_uses_parent_alias_set_from): Handle
16681 TYPE_TYPELESS_STORAGE.
16682 (get_alias_set): Likewise.
16683 * tree-core.h (tree_type_common): Add typeless_storage flag.
16684 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
16685 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
16686 for types containing members with TYPE_TYPELESS_STORAGE.
16687 (place_field): Likewise.
16688 (layout_type): Likewise for ARRAY_TYPE.
16689 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
16690 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
16691 TYPE_TYPELESS_STORAGE.
16692 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
16694 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16697 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
16698 first argument to type.
16700 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16704 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
16705 CONST0_RTX (mode) rather than const0_rtx where appropriate.
16706 (rs6000_expand_binop_builtin): Likewise.
16707 (rs6000_expand_ternop_builtin): Likewise; also add missing
16708 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
16709 vshasigma built-ins.
16710 * doc/extend.texi: Document that vec_xxpermdi's third argument
16711 must be a constant.
16713 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
16715 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
16716 Use shift_const cost parameter when calculating gain of STV shifts.
16718 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
16720 PR rtl-optimization/70478
16721 * lra-constraints.c (process_alt_operands): Check memory for
16722 disfavoring memory insn operand.
16724 2017-04-11 Jakub Jelinek <jakub@redhat.com>
16726 PR middle-end/80100
16727 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
16728 left shift in unsigned HOST_WIDE_INT type.
16730 PR rtl-optimization/80385
16731 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
16732 (not (neg X)) into (plus X -1) for complex or non-integral modes.
16735 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
16736 if they have any depend clauses.
16738 2017-04-11 Martin Liska <mliska@suse.cz>
16741 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
16742 * ipa-split.c (split_function): Create a local comdat symbol
16743 if caller is in a comdat group.
16745 2017-04-11 Martin Liska <mliska@suse.cz>
16748 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
16751 2017-04-11 Martin Sebor <msebor@redhat.com>
16753 PR middle-end/80364
16754 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
16755 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
16757 (directive::set_width, directive::set_precision, format_character):
16759 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
16762 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
16765 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
16766 conflict, set target->arch_name instead of target->cpu_name.
16768 2017-04-11 Richard Biener <rguenther@suse.de>
16770 PR tree-optimization/80374
16771 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
16772 build_zero_cst, remove fold_convertible_p check again.
16774 2017-04-11 Martin Liska <mliska@suse.cz>
16777 * ubsan.c (instrument_object_size): Do not instrument register
16780 2017-04-11 Jakub Jelinek <jakub@redhat.com>
16783 * config/i386/i386-builtin-types.def
16784 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
16785 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
16786 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
16787 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
16788 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
16789 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
16790 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
16791 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
16792 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
16793 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
16794 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
16795 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
16796 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
16797 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
16798 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
16799 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
16800 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
16801 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
16802 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
16803 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
16804 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
16805 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
16806 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
16807 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
16808 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
16809 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
16810 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
16811 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
16812 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
16813 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
16814 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
16815 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
16816 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
16817 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
16818 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
16819 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
16820 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
16821 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
16822 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
16823 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
16824 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
16825 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
16826 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
16827 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
16828 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
16829 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
16831 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
16832 flag to second_arg_count, handle 4 argument function type _COUNT
16833 aliases, handle second_arg_count on second argument rather than last.
16835 2017-04-10 Jeff Law <law@redhat.com>
16837 PR tree-optimization/80374
16838 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
16839 record anything if we can not convert integer_zero_node to the
16842 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
16845 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16846 Enhance special handling given to the TARGET_P9_MINMAX option in
16847 relation to certain other options.
16849 2017-04-10 Bin Cheng <bin.cheng@arm.com>
16851 PR tree-optimization/80153
16852 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
16853 remove POINTER_PLUS_EXPR's base part directly, rather than through
16856 2017-04-10 Richard Biener <rguenther@suse.de>
16857 Bin Cheng <bin.cheng@arm.com>
16859 PR tree-optimization/80153
16860 * tree-affine.c (aff_combination_to_tree): Get base pointer from
16861 the first element of pointer type aff_tree. Build result expr in
16863 (add_elt_to_tree): Convert to type unconditionally. Remove other
16864 fold_convert calls.
16865 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
16866 (rewrite_use_nonlinear_expr): Check invariant using iv information.
16868 2017-04-10 Richard Biener <rguenther@suse.de>
16870 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16873 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
16875 PR rtl-optimization/70478
16876 * lra-constraints.c (curr_small_class_check): New.
16877 (update_and_check_small_class_inputs): New.
16878 (process_alt_operands): Update curr_small_class_check. Disfavor
16879 alternative insn memory operands. Check available regs for small
16882 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
16885 * config/mips/mips.opt (-mvirt): Update description.
16886 * doc/invoke.texi (-mvirt): Likewise.
16888 2017-04-10 Richard Biener <rguenther@suse.de>
16890 PR middle-end/80362
16891 * fold-const.c (fold_binary_loc): Look at unstripped ops when
16892 looking for NEGATE_EXPR in -A / -B to A / B folding.
16894 2017-04-10 Martin Liska <mliska@suse.cz>
16896 PR gcov-profile/80224
16897 * gcov.c (print_usage): Fix usage string.
16898 (get_gcov_intermediate_filename): Remove.
16899 (output_gcov_file): Use both for normal and intermediate format.
16900 (generate_results): Do not initialize special file for
16901 intermediate format.
16903 2017-04-10 Richard Biener <rguenther@suse.de>
16905 PR tree-optimization/80304
16906 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
16909 2017-04-10 Nathan Sidwell <nathan@acm.org>
16912 * config/rs6000/rs6000.c (rs6000_vector_type): New.
16913 (rs6000_init_builtins): Use it.
16915 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16917 * config/arm/arm.md (<mrc>): Add mode to SET source.
16918 (<mrrc>): Likewise.
16920 2017-04-10 Richard Biener <rguenther@suse.de>
16922 PR middle-end/80344
16923 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
16925 2017-04-10 Jakub Jelinek <jakub@redhat.com>
16928 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
16929 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
16930 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
16931 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
16932 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
16933 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
16934 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
16935 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
16936 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
16937 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
16938 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
16939 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
16940 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
16941 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
16942 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
16943 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
16944 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
16945 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
16946 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
16947 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
16948 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
16949 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
16950 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
16952 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
16954 PR rtl-optimization/70478
16955 * lra-constraints.c: Reverse the last patch.
16957 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
16959 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
16960 Add comment for WCHAR_T.
16962 2017-04-08 Martin Liska <mliska@suse.cz>
16965 2017-04-07 Martin Liska <mliska@suse.cz>
16968 * ipa-split.c (split_function): Add function part to a same comdat
16971 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16974 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
16976 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
16978 * rs6000/rs6000.c (vec_load_pendulum): Rename...
16979 (vec_pairing): ...to this.
16980 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
16981 (rs6000_sched_init): Adjust for name change.
16982 (struct rs6000_sched_context): Likewise.
16983 (rs6000_init_sched_context): Likewise.
16984 (rs6000_set_sched_context): Likewise.
16986 2017-04-07 Jakub Jelinek <jakub@redhat.com>
16992 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
16994 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
16995 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
16996 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
16998 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
17000 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
17002 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
17004 PR rtl-optimization/70703
17005 * ira-color.c (update_conflict_hard_regno_costs): Use
17006 int64_t instead of HOST_WIDE_INT.
17008 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
17010 PR rtl-optimization/70478
17011 * lra-constraints.c (process_alt_operands): Disfavor alternative
17012 insn memory operands.
17014 2017-04-07 Jeff Law <law@redhat.com>
17016 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
17017 CALL and NOTE_INSN_CALL_ARG_LOCATION.
17019 2017-04-07 Martin Liska <mliska@suse.cz>
17022 * config/aarch64/aarch64.c (aarch64_process_target_attr):
17023 Show error message instead of an ICE.
17025 2017-04-07 Martin Liska <mliska@suse.cz>
17028 * ipa-split.c (split_function): Add function part to a same comdat
17031 2017-04-07 Richard Biener <rguenther@suse.de>
17033 PR middle-end/80341
17034 * tree.c (get_unwidened): Also handle ! for_type case for
17036 * convert.c (do_narrow): Split out from ...
17037 (convert_to_integer_1): ... here. Do not pass final truncation
17038 type to get_unwidened for TRUNC_DIV_EXPR.
17040 2017-04-07 Richard Biener <rguenther@suse.de>
17042 * tree-affine.c (wide_int_ext_for_comb): Take type rather
17044 (aff_combination_const): Adjust.
17045 (aff_combination_scale): Likewise.
17046 (aff_combination_add_elt): Likewise.
17047 (aff_combination_add_cst): Likewise.
17048 (aff_combination_convert): Likewise.
17049 (add_elt_to_tree): Likewise. Remove unused argument.
17050 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
17052 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
17054 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
17056 * config/arm/arm.c (arm_default_short_enums): Use
17057 ARM_DEFAULT_SHORT_ENUMS.
17058 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
17060 2017-04-06 Jakub Jelinek <jakub@redhat.com>
17063 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
17064 members with redundant out-of-class redeclaration.
17066 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
17069 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
17070 * config/i386/i386.md (*zero_extendsidi2):
17071 Add (?*x,*x) and (?*v,*v) alternatives.
17073 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
17076 * config/i386/i386.c (ix86_expand_builtin)
17077 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
17078 mode from insn data. Convert operands to insn operand mode.
17079 Copy operands that don't satisfy insn predicate to a register.
17081 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
17083 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
17086 2017-04-06 Richard Biener <rguenther@suse.de>
17088 PR tree-optimization/80334
17089 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
17090 preserve alignment of accesses.
17092 2017-04-06 Richard Biener <rguenther@suse.de>
17094 PR tree-optimization/80262
17095 * tree-sra.c (build_ref_for_offset): Preserve address-space
17097 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
17098 Drop useless address-space information on MEM_REF offsets.
17100 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
17102 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
17104 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
17106 PR rtl-optimization/70703
17107 * ira-color.c (update_conflict_hard_regno_costs): Use
17108 HOST_WIDE_INT instead of long.
17110 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
17113 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
17114 not defined for x86_64 target. Add -mmmx target option when __SSE2__
17116 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
17117 for x86_64 target. Handle -m3dnowa option.
17119 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
17121 PR rtl-optimization/70703
17122 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
17123 (update_conflict_hard_regno_costs): Use long instead of unsigned
17124 arithmetic for cost calculation.
17126 2017-04-05 Jakub Jelinek <jakub@redhat.com>
17127 Bernd Edlinger <bernd.edlinger@hotmail.de>
17130 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
17133 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
17136 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
17137 ptr_mode with Pmode throughout.
17138 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
17139 into probe_stack_range and use DImode.
17141 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
17144 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
17145 call_eh_return is true.
17147 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17149 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
17150 Initialize last_match_fntype_index.
17152 2017-04-05 Jakub Jelinek <jakub@redhat.com>
17155 * tree-nvr.c: Include internal-fn.h.
17156 (pass_return_slot::execute): Ignore internal calls without
17159 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17160 Richard Biener <rguenther@suse.de>
17163 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
17164 captures used multiple times, except for the last use.
17165 * generic-match-head.c: Include gimplify.h.
17167 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17169 PR tree-optimization/79390
17170 * target.h (struct noce_if_info): Declare.
17171 * targhooks.h (default_noce_conversion_profitable_p): Declare.
17172 * target.def (noce_conversion_profitable_p): New target hook.
17173 * ifcvt.h (struct noce_if_info): New type, moved from ...
17174 * ifcvt.c (struct noce_if_info): ... here.
17175 (noce_conversion_profitable_p): Renamed to ...
17176 (default_noce_conversion_profitable_p): ... this. No longer
17178 (noce_try_store_flag_constants, noce_try_addcc,
17179 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
17180 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
17181 instead of noce_conversion_profitable_p.
17182 * config/i386/i386.c: Include ifcvt.h.
17183 (ix86_option_override_internal): Don't override
17184 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
17185 (ix86_noce_conversion_profitable_p): New function.
17186 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
17187 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
17188 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
17189 * doc/tm.texi: Regenerated.
17191 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17193 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
17196 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17199 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
17200 instructions for small multiply cores.
17202 2017-04-04 Jeff Law <law@redhat.com>
17204 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
17206 (mips_expand_vec_perm_const): Initialize elements in orig_perm
17207 that are not set by the loop over the elements.
17209 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17212 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
17213 int mode, convert_modes it to mode as unsigned, otherwise use
17214 lowpart_subreg to mode rather than SImode.
17215 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
17216 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
17217 Use DImode instead of SImode for the shift count operand.
17218 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
17221 2017-04-04 Richard Biener <rguenther@suse.de>
17223 PR middle-end/80281
17224 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
17225 arithmetic done for the negate or the plus. Simplify.
17226 (A - (-B) -> A + B): Likewise.
17227 * fold-const.c (split_tree): Make sure to not negate pointers.
17229 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
17231 PR rtl-optimization/60818
17232 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
17233 a compare of comparisons with the thing compared if this results
17234 in a different machine mode.
17236 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
17238 * alias.c (base_alias_check): Fix typo in comment.
17239 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
17240 * cgraphunit.c (symbol_table::compile): Likewise.
17241 * collect2.c (maybe_run_lto_and_relink): Likewise.
17242 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
17243 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
17244 * config/avr/avr.c (avr_map_op_t): Likewise.
17245 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
17246 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
17247 * config/epiphany/epiphany.md (movcc): Likewise.
17248 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
17249 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
17251 * config/mips/mips.c (mips_save_restore_reg): Likewise.
17252 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
17253 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
17254 * config/sh/sh.c (sh_rtx_costs): Likewise.
17255 * fold-const.c (fold_truth_andor): Likewise.
17256 * genautomata.c (collapse_flag): Likewise.
17257 * gengtype.h (struct type::u::s): Likewise.
17258 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
17259 * input.c (FORMAT_AMOUNT): Likewise.
17260 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
17261 (known_aggs_to_agg_replacement_list): Likewise.
17262 * ipa-inline-analysis.c: Likewise.
17263 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
17264 * ipa-polymorphic-call.c
17265 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
17266 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
17267 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
17269 * modulo-sched.c (apply_reg_moves): Likewise.
17270 * omp-expand.c (build_omp_regions_1): Likewise.
17271 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
17272 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
17273 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
17274 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
17275 * value-prof.c: Likewise.
17276 * var-tracking.c (val_reset): Likewise.
17278 2017-04-03 Richard Biener <rguenther@suse.de>
17280 PR tree-optimization/80275
17281 * fold-const.c (split_address_to_core_and_offset): Handle
17284 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
17286 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
17287 descriptors is at least equal to that of functions.
17289 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
17291 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
17293 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
17296 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
17297 (mov<IMOD4:mode>): New expander.
17298 (*mov<IMOD4:mode>_internal): New insn and split pattern.
17300 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
17302 PR rtl-optimization/79405
17303 * fwprop.c (propagations_left): New variable.
17304 (forward_propagate_into): Decrement it.
17305 (fwprop_init): Initialize it.
17306 (fw_prop): If the variable has reached zero, stop propagating.
17307 (fwprop_addr): Ditto.
17309 2017-03-31 Jakub Jelinek <jakub@redhat.com>
17312 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
17313 a FUNCTION_DECL, pass it as decl instead of origin to
17316 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
17318 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
17321 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
17324 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
17325 TARGET_VSX_SMALL_INTEGER.
17327 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17329 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17330 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
17332 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
17334 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
17335 extraction from odd-numbered MSA register.
17337 2017-03-31 Jakub Jelinek <jakub@redhat.com>
17339 PR middle-end/80173
17340 * expmed.c (store_bit_field_1): Don't attempt to create
17341 a word subreg out of hard registers wider than word if they
17342 have HARD_REGNO_NREGS of 1 for their mode.
17344 PR middle-end/80163
17345 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
17346 conversions to integer types wider than word and pointer.
17349 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
17350 (rtx_equal_for_cselib_p): Pass 0 to it.
17351 * cselib.c (cselib_hasher::equal): Likewise.
17352 (rtx_equal_for_cselib_1): Add depth argument. If depth
17353 is 128, don't look up VALUE locs and punt. Increment
17354 depth in recursive calls when walking VALUE locs.
17356 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
17358 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
17359 (make_gcov_file_name): Use the canonical path name for generating
17361 (read_line): Fix handling of files with ascii null bytes.
17363 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
17365 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
17366 to initialise a vector register instead
17367 of using a const_int.
17369 2017-03-30 Jakub Jelinek <jakub@redhat.com>
17371 PR translation/80189
17372 * gimplify.c (omp_default_clause): Use %qs instead of %s in
17373 diagnostic messages.
17375 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
17378 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
17379 (dfp_diex_<mode>): Update mode of operand 1.
17380 * doc/extend.texi (dxex, dxexq): Document change to return type.
17381 (diex, diexq): Document change to argument type.
17383 2017-03-30 Martin Jambor <mjambor@suse.cz>
17386 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
17387 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
17388 it reflects the signature changes performed at the callee side.
17389 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
17390 to cgraph_build_function_type_skip_args.
17391 (build_function_decl_skip_args): Adjust call to the above function.
17393 2017-03-30 Jakub Jelinek <jakub@redhat.com>
17396 * config/i386/sse.md
17397 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
17398 register as dest whenever it is a MEM not rtx_equal_p to the
17399 corresponding dup operand, and when forcing into reg move the
17400 reg into the memory afterwards.
17401 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
17402 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
17403 for the force_reg mode.
17404 (avx512vl_vextractf128<mode>): Use register as dest either
17405 always when a MEM, or when it is a MEM not rtx_equal_p to the
17406 corresponding dup operand, or even not when it is a CONST_VECTOR
17407 depending on the mode and lo vs. hi.
17408 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
17410 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
17411 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
17412 Likewise. Require that operands[2] is even.
17413 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
17414 Remove extraneous parens. Require that operands[2] is a multiple
17416 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
17417 operands[0] is a MEM if <mask_applied>, the predicates/constraints
17418 disallow memory then.
17420 2017-03-30 Richard Biener <rguenther@suse.de>
17422 PR tree-optimization/77498
17423 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
17424 to non-constants over backedges.
17426 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
17428 PR rtl-optimization/80233
17429 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
17430 as last_combined_insn. Do not test for BARRIER_P separately.
17432 2017-03-29 Andreas Schwab <schwab@suse.de>
17435 * calls.c (prepare_call_address): Convert funexp to Pmode before
17436 copying to temp reg.
17438 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17440 PR tree-optimization/80158
17441 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
17442 Handle possible future case of more than one alternate
17444 (replace_rhs_if_not_dup): Likewise.
17445 (replace_one_candidate): Likewise.
17447 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
17449 PR rtl-optimization/80193
17450 * ira.c (ira): Do not check allocation for LRA.
17452 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
17454 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
17455 (nvptx_output_simt_exit): Declare.
17456 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
17457 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
17458 (init_softstack_frame): Move initialization of crtl->is_leaf to...
17459 (nvptx_declare_function_name): ...here. Emit declaration of local
17460 memory space buffer for omp_simt_enter insn.
17461 (nvptx_output_unisimt_switch): New.
17462 (nvptx_output_softstack_switch): New.
17463 (nvptx_output_simt_enter): New.
17464 (nvptx_output_simt_exit): New.
17465 * config/nvptx/nvptx.h (struct machine_function): New fields
17466 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
17467 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
17468 (UNSPECV_SIMT_EXIT): Ditto.
17469 (omp_simt_enter_insn): New insn.
17470 (omp_simt_enter): New expansion.
17471 (omp_simt_exit): New insn.
17472 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
17474 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
17475 (expand_GOMP_SIMT_ENTER_ALLOC): New.
17476 (expand_GOMP_SIMT_EXIT): New.
17477 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
17478 (GOMP_SIMT_ENTER_ALLOC): Ditto.
17479 (GOMP_SIMT_EXIT): Ditto.
17480 * target-insns.def (omp_simt_enter): New insn.
17481 (omp_simt_exit): Ditto.
17482 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
17484 (lower_rec_simd_input_clauses): Implement SIMT privatization.
17485 (lower_rec_input_clauses): Likewise.
17486 (lower_lastprivate_clauses): Handle SIMT privatization.
17488 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
17489 (ompdevlow_adjust_simt_enter): New.
17490 (find_simtpriv_var_op): New.
17491 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
17492 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
17494 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
17495 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
17496 (copy_decl_for_dup_finish): Ditto.
17498 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
17500 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
17503 * config/i386/i386.c (ix86_option_override_internal): Always
17504 allow -mpreferred-stack-boundary=3 for 64-bit targets.
17506 2017-03-28 Bin Cheng <bin.cheng@arm.com>
17508 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
17510 2017-03-28 Bin Cheng <bin.cheng@arm.com>
17512 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
17513 mark new edge's irreducible flag accordign to it.
17514 (vect_do_peeling): Check loop preheader edge's irreducible flag
17515 and pass it to function slpeel_add_loop_guard.
17517 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
17519 PR tree-optimization/80218
17520 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
17521 Update block frequencies and counts.
17523 2017-03-28 Richard Biener <rguenther@suse.de>
17525 PR tree-optimization/78644
17526 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
17527 of a simplification result we may not use it at all.
17529 2017-03-28 Richard Biener <rguenther@suse.de>
17532 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
17533 without arguments, generate default definition of a SSA name.
17535 2017-03-28 Richard Biener <rguenther@suse.de>
17537 PR middle-end/80222
17538 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
17539 TYPE_REF_CAN_ALIAS_ALL references.
17540 * fold-const.c (fold_indirect_ref_1): Likewise.
17542 2017-03-28 Martin Liska <mliska@suse.cz>
17545 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
17546 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
17548 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
17549 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
17551 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
17552 (EXTRA_SPECS): Define.
17553 (SUBTARGET_EXTRA_SPECS): Likewise.
17554 (SUBTARGET_CPP_SPEC): Likewise.
17555 * config/arc/elf.h (EXTRA_SPECS): Renamed to
17556 SUBTARGET_EXTRA_SPECS.
17557 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
17559 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
17561 * config/arc/simdext.md (vst64_insn): Update pattern.
17562 (vld32wh_insn): Likewise.
17563 (vld32wl_insn): Likewise.
17564 (vld64_insn): Likewise.
17565 (vld32_insn): Likewise.
17567 2017-03-28 Marek Polacek <polacek@redhat.com>
17570 * fold-const.c (fold_comparison): Use protected_set_expr_location
17571 instead of SET_EXPR_LOCATION.
17573 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
17575 * tree.c (add_expr): Avoid name lookup warning.
17577 2017-03-27 Jeff Law <law@redhat.com>
17579 PR tree-optimization/80216
17580 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
17581 function name. Limit recursion depth.
17582 (record_temporary_equivalences): Corresponding changes.
17584 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
17586 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
17589 2017-03-27 Jakub Jelinek <jakub@redhat.com>
17592 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
17594 * cfgcleanup.c (reg_note_cfa_p): New array.
17595 (insns_have_identical_cfa_notes): New function.
17596 (old_insns_match_p): Don't cross-jump in between /f
17597 and non-/f instructions. If both i1 and i2 are frame related,
17598 verify all CFA notes, their order and content.
17600 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
17603 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
17604 HImode and SImode with zero extend to DImode to one insn.
17605 (bswap<mode>2_extenddi): Likewise.
17606 (bswapsi2_extenddi): Likewise.
17607 (bswaphi2_extendsi): Likewise.
17608 (bswaphi2): Combine bswap HImode and SImode into one insn.
17609 Separate memory insns from swapping register.
17610 (bswapsi2): Likewise.
17611 (bswap<mode>2): Likewise.
17612 (bswaphi2_internal): Delete, no longer used.
17613 (bswapsi2_internal): Likewise.
17614 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
17615 store, and gpr<-gpr swap insns.
17616 (bswap<mode>2_store): Likewise.
17617 (bswaphi2_reg): Register only splitter, combine with the splitter.
17618 (bswaphi2 splitter): Likewise.
17619 (bswapsi2_reg): Likewise.
17620 (bswapsi2 splitter): Likewise.
17621 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
17622 the insns into load, store, and register/register insns.
17623 (bswapdi2_ldbrx): Likewise.
17624 (bswapdi2_load): Likewise.
17625 (bswapdi2_store): Likewise.
17626 (bswapdi2_reg): Likewise.
17628 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
17630 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
17631 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
17633 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
17636 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
17638 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
17639 special handling for target option conflicts between dform
17640 options (-mpower9-dform, -mpower9-dform-vector,
17641 -mpower9-dform-scalar) and -mno-direct-move.
17643 2017-03-27 Richard Biener <rguenther@suse.de>
17645 PR tree-optimization/80181
17646 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
17648 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17650 * config/arc/predicates.md (move_double_src_operand): Replace the
17651 call to move_double_src_operand with a call to address_operand.
17653 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17655 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
17656 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
17657 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
17659 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17661 * config/arc/predicates.md (long_immediate_loadstore_operand):
17662 Consider scaled addresses cases.
17664 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17666 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
17667 restored when in interrupt.
17668 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
17669 doesn't have delay slot.
17671 2017-03-27 Richard Biener <rguenther@suse.de>
17674 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
17675 inlined thunk clones.
17677 2017-03-27 Jakub Jelinek <jakub@redhat.com>
17680 * asan.c (instrument_derefs): Copy over last operand from
17681 original COMPONENT_REF to the new COMPONENT_REF with
17682 DECL_BIT_FIELD_REPRESENTATIVE.
17683 * ubsan.c (instrument_object_size): Likewise.
17685 2017-03-27 Richard Biener <rguenther@suse.de>
17687 PR tree-optimization/80170
17688 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
17689 sure DR/SCEV didnt fold in constants we do not see when looking
17690 at the reference base alignment.
17692 2017-03-27 Richard Biener <rguenther@suse.de>
17694 PR middle-end/80171
17695 * gimple-fold.c (fold_ctor_reference): Properly guard against
17696 NULL return value from canonicalize_constructor_val.
17698 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
17701 * config/i386/i386.c (ix86_expand_builtin)
17702 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
17703 flags reg setting and flags reg using instructions.
17704 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
17705 clobbering instructions to zero extend op2.
17707 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
17709 * doc/install.texi (Configuration) <--with-aix-soname>:
17710 Update link to AIX ld.
17712 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
17714 PR rtl-optimization/80160
17715 PR rtl-optimization/80159
17716 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
17717 reg_alternate_class into account.
17719 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
17722 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
17723 to consider in curr_insn_transform.
17725 2017-03-24 Jakub Jelinek <jakub@redhat.com>
17727 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
17728 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
17729 and emit_mode_inner.
17731 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17733 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
17734 argument to the overloaded builtin variants. Use the new flag to
17735 deprecate certain builtin variants.
17736 * config/s390/s390-builtin-types.def: Add new builtin types.
17737 * config/s390/s390-builtins.h: Support new flags field for
17738 overloaded builtins.
17739 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
17740 (s390_macro_to_expand): Enable vector float data type.
17741 (s390_cpu_cpp_builtins_internal): Indicate support of the new
17742 builtins by incrementing the __VEC__ version number.
17743 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
17745 (s390_resolve_overloaded_builtin): Emit error messages depending
17746 on the builtin flags.
17747 * config/s390/s390.c (s390_expand_builtin): Support additional
17748 flags argument. Change error message to match the messages
17749 emitted in s390-c.c.
17750 * config/s390/s390.md: New UNSPEC_* constants.
17751 (op_type): Add new instruction types.
17752 * config/s390/vecintrin.h: Add new builtins and test data class
17754 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
17755 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
17756 (VEC_INEXACT, VEC_NOINEXACT): New constants.
17757 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
17758 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
17759 ("vec_mergel<mode>"): V_HW -> VEC_HW.
17761 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
17762 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
17763 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
17764 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
17766 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
17767 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
17768 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
17769 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
17771 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
17772 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
17773 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
17774 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
17775 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
17776 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
17777 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
17779 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
17780 ("vec_scatter_element<V_HW_4:mode>_DI")
17781 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
17782 ("vec_fpint<mode>", "vflls")
17783 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
17784 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
17785 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
17786 ("*vec_cmphe<mode>_cc"): ... these.
17788 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
17789 mode constant instead of magic value.
17791 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17793 * config/s390/s390.c (s390_expand_vec_compare): Support other
17794 vector floating point modes than just V2DF.
17795 (s390_expand_vcond): Likewise.
17796 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
17797 (s390_cannot_change_mode_class): Prevent mode changes between TF
17798 and V1TF in vector registers.
17799 * config/s390/s390.md (DF, SF): New mode attributes.
17800 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
17801 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
17802 SFmode support for VRs.
17803 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
17805 (VFT, VF_HW): New mode iterators.
17806 (vw, sdx): New mode attributes.
17807 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
17808 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
17809 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
17810 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
17811 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
17812 also the new vector floating point modes. Renaming to ...
17814 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
17815 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
17816 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
17817 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
17818 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
17819 ("vec_unordered<mode>"): ... these.
17821 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
17822 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
17823 ("*vec_extendv2df"): New insn definitions.
17825 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17827 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
17828 ("mulditi3_2", "*muldi3_sign"): New patterns.
17829 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
17830 rename the pattern definition.
17832 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17834 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
17836 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
17838 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17840 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
17841 instruction if possible.
17842 * config/s390/vector.md (vec_halfnumelts): New mode
17844 ("*vec_vllezlf<mode>"): New pattern.
17846 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17848 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
17849 ("popcountv4si2", "popcountv2di2"): Rename to ...
17850 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
17851 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
17853 ("popcount<mode>2_vxe"): New pattern.
17855 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17857 * common/config/s390/s390-common.c (processor_flags_table): Add
17859 * config.gcc: Add arch12.
17860 * config/s390/driver-native.c (s390_host_detect_local_cpu):
17861 Default to arch12 for unknown CPU model numbers.
17862 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
17863 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
17864 PROCESSOR_max sanity check.
17865 * config/s390/s390-opts.h (enum processor_type): Add
17867 * config/s390/s390.c (processor_table): Add arch12.
17868 (s390_expand_builtin): Add check for B_VXE flag.
17869 (s390_issue_rate): Add PROCESSOR_ARCH12.
17870 (s390_get_sched_attrmask): Likewise.
17871 (s390_get_unit_mask): Likewise.
17872 (s390_sched_score): Enable z13 scheduling for arch12.
17873 (s390_sched_reorder): Likewise.
17874 (s390_sched_variable_issue): Likewise.
17875 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
17877 (s390_tune_attr): Use z13 scheduling also for arch12.
17878 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
17879 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
17880 (TARGET_VXE_P): New macros.
17881 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
17882 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
17883 * config/s390/s390.opt: Add arch12 as processor_type.
17885 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17887 * config/s390/s390.md
17888 ("fixuns_truncdddi2", "fixuns_trunctddi2")
17889 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
17890 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
17892 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
17893 Rename expanders to ...
17895 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
17896 ("fixuns_truncdddi2_emu"): ... these.
17898 ("fixuns_trunc<mode>si2_emu"): New expander.
17900 ("*fixuns_truncdfdi2_z13"): Rename to ...
17901 ("*fixuns_truncdfdi2_vx"): ... this.
17903 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17905 * config/s390/2964.md: Remove the single element vector compare
17906 instructions which are no longer used.
17907 * config/s390/s390.c (s390_select_ccmode): Remove handling of
17909 (s390_canonicalize_comparison): Remove handling of DFmode
17911 (s390_expand_vec_compare_scalar): Remove function.
17912 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
17913 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
17915 ("*cmp<mode>_ccs"): Add wfcdb instruction.
17917 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17919 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
17921 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
17922 will anyway by matched by mov<mode>_64dfp.
17924 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17926 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
17927 vlef/vstef. Add missing operand to vleif.
17929 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17931 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
17932 pair for all vector types with 64 bit elements.
17933 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
17934 * config/s390/vector.md (V_HW_64): ... here.
17935 (V_128_NOSINGLE): New mode iterator.
17936 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
17937 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
17938 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
17939 ("*vec_load_pairv2di"): Change to ...
17940 ("*vec_load_pair<mode>"): ... this one.
17942 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17944 * config/s390/constraints.md: Add comments.
17945 (jKK): Reject element sizes > 8 bytes.
17946 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
17948 * config/s390/s390.md: Add the s_operand checks formerly in
17949 s390_split_ok_p to various splitters where they are still
17951 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
17952 for 128 bit vectors. Plus two splitters.
17954 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17956 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
17959 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17962 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
17963 error if the boundary argument is not constant.
17965 2017-03-24 Jakub Jelinek <jakub@redhat.com>
17967 PR rtl-optimization/80112
17968 * loop-doloop.c (doloop_condition_get): Don't check condition
17969 if cmp isn't SET with IF_THEN_ELSE src.
17971 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17973 PR tree-optimization/80158
17974 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
17975 replacing a candidate statement, also replace it for the
17976 candidate's alternate interpretation.
17977 (replace_rhs_if_not_dup): Likewise.
17978 (replace_one_candidate): Likewise.
17980 2017-03-24 Richard Biener <rguenther@suse.de>
17982 PR tree-optimization/80167
17983 * graphite-isl-ast-to-gimple.c
17984 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
17986 (translate_isl_ast_to_gimple::get_rename): Likewise.
17988 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
17990 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
17991 handling of certain combinations of target options, including the
17992 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
17993 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
17995 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17998 * config/arm/arm.md (*load_multiple): Add reload_completed to
17999 matching condition.
18001 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18002 Richard Biener <rguenth@suse.de>
18004 PR tree-optimization/79908
18005 PR tree-optimization/80136
18006 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
18007 been cast away, gimplify_and_add suffices.
18009 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
18011 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
18013 2017-03-23 Richard Biener <rguenther@suse.de>
18015 PR tree-optimization/80032
18016 * gimplify.c (gimple_push_cleanup): Forced unconditional
18017 cleanups still have to go to the conditional_cleanups
18020 2017-03-22 Jakub Jelinek <jakub@redhat.com>
18022 PR tree-optimization/80072
18023 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
18025 (next_operand_entry_id): Change type to unsigned int.
18026 (sort_by_operand_rank): Make sure to return the right return value
18027 even if unsigned fields are bigger than INT_MAX.
18028 (struct oecount): Change cnt and id type to unsigned int.
18029 (oecount_hasher::equal): Formatting fix.
18030 (oecount_cmp): Make sure to return the right return value
18031 even if unsigned fields are bigger than INT_MAX.
18032 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
18035 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
18036 TREE_READONLY on result if writing it more than once.
18039 * doc/invoke.texi (-fsanitize=thread): Document that with
18040 -fnon-call-exceptions atomics are not able to throw
18044 * tsan.c: Include tree-eh.h.
18045 (instrument_builtin_call): Call maybe_clean_eh_stmt or
18046 maybe_clean_or_replace_eh_stmt where needed.
18047 (instrument_memory_accesses): Add cfg_changed argument.
18048 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
18049 if it returned true.
18050 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
18052 PR rtl-optimization/63191
18053 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
18054 wrapper function, moved the whole old content into ...
18055 (ix86_delegitimize_address_1): ... this. New inline function.
18056 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
18057 true as last argument instead of ix86_delegitimize_address.
18059 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
18061 * config/aarch64/aarch64.c (generic_branch_cost): Copy
18062 cortexa57_branch_cost.
18064 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
18066 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
18068 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18071 * doc/md.texi (Constraints): Document wA constraint.
18072 * config/rs6000/constraints.md (wA): New.
18073 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
18074 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
18075 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
18076 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
18078 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
18081 * gimplify.c (is_oacc_declared): New function.
18082 (oacc_default_clause): Use it to set default flags for acc declared
18083 variables inside parallel regions.
18084 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
18085 declared variables.
18086 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
18087 declare attribute to any decl as necessary.
18089 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
18092 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
18093 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
18094 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
18095 (arm_arch_lpae): This.
18096 * config/arm/arm.c (arm_arch7ve): Rename into ...
18097 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
18098 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
18101 2017-03-22 Martin Liska <mliska@suse.cz>
18104 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
18105 error message instead of an ICE.
18107 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18109 * doc/extend.texi (6.11 Additional Floating Types): Revise.
18111 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
18113 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
18115 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18118 2017-03-21 Martin Sebor <msebor@redhat.com>
18120 * doc/extend.texi: Use "cannot" instead of "can't."
18121 * doc/hostconfig.texi: Same.
18122 * doc/install.texi: Same.
18123 * doc/invoke.texi: Same.
18124 * doc/loop.texi: Same.
18125 * doc/md.texi: Same.
18126 * doc/objc.texi: Same.
18127 * doc/rtl.texi: Same.
18128 * doc/tm.texi: Same.
18129 * doc/tm.texi.in: Same.
18130 * doc/trouble.texi: Same.
18132 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
18135 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
18136 (collect_checksum_attributes): Set it.
18137 (die_checksum_ordered): Use it.
18139 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18141 PR tree-optimization/79908
18142 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
18143 change: For a VA_ARG whose LHS has been cast away, use
18144 force_gimple_operand to construct the side effects.
18146 2017-03-21 David Malcolm <dmalcolm@redhat.com>
18148 PR translation/80001
18149 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
18150 more amenable to translation.
18151 (oacc_loop_auto_partitions): Likewise.
18153 2017-03-21 Marek Polacek <polacek@redhat.com>
18154 Martin Sebor <msebor@redhat.com>
18156 PR tree-optimization/80109
18157 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
18158 on INTEGRAL_TYPE_P.
18160 2017-03-21 Jakub Jelinek <jakub@redhat.com>
18161 Segher Boessenkool <segher@kernel.crashing.org>
18164 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
18165 check reg_used_between_p between insn and one of succ or succ2
18166 depending on if succ is artificial insn not inserted into insn
18169 2017-03-21 Martin Liska <mliska@suse.cz>
18171 PR gcov-profile/80081
18172 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
18173 * doc/gcc.texi: Include gcov-dump stuff.
18174 * doc/gcov-dump.texi: New file.
18176 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
18178 PR rtl-optimization/79150
18179 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
18180 conditional jump, if the jump is the last insn of the loop.
18182 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18183 Richard Biener <rguenth@suse.de>
18185 PR tree-optimization/79908
18186 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
18187 been cast away, use force_gimple_operand to construct the side
18190 2017-03-21 Martin Liska <mliska@suse.cz>
18192 PR libfortran/79956
18193 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
18196 2017-03-21 Brad Spengler <spender@grsecurity.net>
18199 * plugin.c (htab_hash_plugin): New function.
18200 (add_new_plugin): Use it and adjust.
18201 (parse_plugin_arg_opt): Adjust.
18202 (init_one_plugin): Likewise.
18204 2017-03-21 Richard Biener <rguenther@suse.de>
18206 PR tree-optimization/80032
18207 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
18208 if set force the cleanup to happen unconditionally.
18209 (gimplify_target_expr): Push inserted clobbers with force_uncond
18210 to avoid them being removed by control-dependent DCE.
18212 2017-03-21 Richard Biener <rguenther@suse.de>
18214 PR tree-optimization/80122
18215 * tree-inline.c (copy_bb): Do not expans va-arg packs or
18216 va_arg_pack_len when the inlined call stmt requires pack
18218 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
18220 2017-03-21 Jakub Jelinek <jakub@redhat.com>
18223 * tsan.c (instrument_builtin_call): If the memory model argument
18224 is not a constant, assume it is valid.
18227 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
18230 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
18232 PR rtl-optimization/79910
18233 * combine.c (can_combine_p): Do not allow combining an I0 or I1
18234 if its dest is used by an insn before I2 (other than the combined
18235 insns themselves, which are properly handled already).
18237 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
18240 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
18242 * combine.c (record_used_regs): New static function.
18243 (try_combine): Handle situations where there is an additional
18244 instruction between I2 and I3 which needs to have a LOG_LINK
18248 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
18250 * combine.c (try_combine): Delete redundant i1 test. Call
18251 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
18253 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18256 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
18257 alternatives 13/14.
18259 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18261 PR tree-optimization/80054
18262 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
18263 the optimization if a PHI or any of its arguments is not dominated
18264 by the candidate's basis. Use gphi* rather than gimple* as
18266 (replace_profitable_candidates): Clean up a gimple* variable that
18267 should be a gphi* variable.
18269 2017-03-20 Martin Sebor <msebor@redhat.com>
18272 * doc/extend.texi (attribute constructor): Document present limitation.
18274 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
18277 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
18278 __POWER9_VECTOR__ #ifdef control, change template definition to
18279 use Power9-specific built-in function.
18280 (vec_any_eq): Likewise.
18281 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
18282 to control outcomes from this test.
18283 (vector_ae_<mode>p): For VEC_F modes, likewise.
18285 2017-03-20 Ian Lance Taylor <iant@google.com>
18287 * config/i386/i386.c (ix86_function_regparm): Save an extra
18288 register for -fsplit-stack with DECL_STATIC_CHAIN.
18290 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
18293 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
18294 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
18296 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
18298 * config/riscv/riscv.c (riscv_print_operand): Use "fence
18300 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
18303 2017-03-20 Marek Polacek <polacek@redhat.com>
18306 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
18308 2017-03-20 Richard Biener <rguenther@suse.de>
18310 PR tree-optimization/80113
18311 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
18312 allocate extra SSA name for PHI def.
18313 (add_close_phis_to_outer_loops): Likewise.
18314 (add_close_phis_to_merge_points): Likewise.
18315 (copy_loop_close_phi_args): Likewise.
18316 (copy_cond_phi_nodes): Likewise.
18318 2017-03-20 Martin Liska <mliska@suse.cz>
18320 PR middle-end/79753
18321 * tree-chkp.c (chkp_build_returned_bound): Do not build
18322 returned bounds for a LHS that's not a BOUNDED_P type.
18324 2017-03-20 Martin Liska <mliska@suse.cz>
18328 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
18329 COMPLEX_CST and VECTOR_CST.
18331 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18334 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
18335 target operand. A new splitter adds the clobber statement in case
18336 the target operand is dead anyway.
18338 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
18340 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
18341 to age-old versions of binutils and glibc.
18343 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
18345 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
18347 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
18349 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
18351 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
18353 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
18354 requirement for binutils 2.13.
18356 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
18358 * combine.c (try_combine): Delete redundant i1 test. Call
18359 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
18361 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
18363 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
18364 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
18366 <riscv64-*-elf>: Re-arrange section
18367 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
18368 <riscv32-*-linux>: Likewise.
18369 <riscv64-*-elf>: Likewise
18370 <riscv64-*-linux>: Likewise.
18372 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
18375 * aarch64.opt(verbose-cost-dump): Fix typo.
18377 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
18380 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
18381 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
18383 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
18385 * reload.c (find_reloads): When reloading a nonoffsettable address,
18386 use RELOAD_OTHER for it and its address reloads.
18388 PR rtl-optimization/79910
18389 * combine.c (record_used_regs): New static function.
18390 (try_combine): Handle situations where there is an additional
18391 instruction between I2 and I3 which needs to have a LOG_LINK
18394 2017-03-17 Jeff Law <law@redhat.com>
18396 PR tree-optimization/71437
18397 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
18398 conditional in the hash table first.
18399 (vrp_dom_walker::before_dom_children): Extract condition from
18400 ASSERT_EXPR. Record condition, its inverion and any implied
18401 conditions as well.
18403 2017-03-17 Marek Polacek <polacek@redhat.com>
18404 Markus Trippelsdorf <markus@trippelsdorf.de>
18406 PR tree-optimization/80079
18407 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
18410 2017-03-17 Richard Biener <rguenther@suse.de>
18412 PR middle-end/80075
18413 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
18414 Properly verify the LHS before the RHS possibly claims to be
18416 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
18419 2017-03-17 Martin Jambor <mjambor@suse.cz>
18421 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
18422 (List of -O2 options): Likewise.
18423 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
18426 2017-03-17 Tom de Vries <tom@codesourcery.com>
18428 * gcov-dump.c (print_usage): Print bug_report_url.
18430 2017-03-17 Richard Biener <rguenther@suse.de>
18432 PR middle-end/80050
18433 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
18434 (parser::peek): Likewise.
18436 2017-03-17 Richard Biener <rguenther@suse.de>
18438 PR tree-optimization/80048
18439 * sese.c (free_sese_info): Properly release rename_map and
18440 copied_bb_map elements.
18442 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
18444 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
18445 Add linked-list forward and backlinks. Insert on
18446 construction, remove on destruction.
18447 (class pass_store_merging): Add m_stores_head field.
18448 (pass_store_merging::terminate_and_process_all_chains):
18449 Iterate over m_stores_head list.
18450 (pass_store_merging::terminate_all_aliasing_chains):
18452 (pass_store_merging::execute): Check for debug stmts first.
18453 Push new chains onto the m_stores_head stack.
18455 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
18458 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
18459 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
18460 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
18462 2017-03-16 Jeff Law <law@redhat.com>
18464 PR tree-optimization/71437
18465 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
18466 member function. Implementation moved into after_dom_children
18467 member function and into the threader's thread_outgoing_edges
18469 (dom_opt_dom_walker::after_dom_children): Simplify by moving
18470 some code into new thread_outgoing_edges.
18471 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
18472 definition. Simplify marker handling (do it here). Assume we always
18473 have the available expression and the const/copies tables.
18474 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
18476 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
18477 * tree-vrp.c (equiv_stack): No longer file scoped.
18478 (vrp_dom_walker): New class.
18479 (vrp_dom_walker::before_dom_children): New member function.
18480 (vrp_dom_walker::after_dom_children): Likewise.
18481 (identify_jump_threads): Setup domwalker. Use it rather than
18482 walking edges in a random order by hand. Simplify setup/finalization.
18483 (finalize_jump_threads): Remove.
18484 (vrp_finalize): Do not call identify_jump_threads here.
18485 (execute_vrp): Do it here instead and call thread_through_all_blocks
18488 PR tree-optimization/71437
18489 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
18491 (simplify_stmt_for_jump_threading): Add basic_block argument. All
18493 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
18494 (dom_opt_dom_walker::thread_across_edge): Remove
18495 handle_dominating_asserts argument. All callers changed.
18496 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
18497 changes. Remove calls to lhs_of_dominating_assert. Other
18498 uses of handle_dominating_asserts turn into unconditional code
18499 (simplify_control_stmt_condition_1): Likewise.
18500 (simplify_control_stmt_condition): Likewise.
18501 (thread_through_normal_block, thread_across_edge): Likewise.
18502 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
18503 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
18504 object if it is not an SSA_NAME.
18505 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
18506 before calling into the VRP specific simplifiers.
18507 (identify_jump_threads): Remove handle_dominating_asserts
18510 2017-03-16 Jakub Jelinek <jakub@redhat.com>
18513 * tree-diagnostic.c (default_tree_printer): No longer static.
18514 * tree-diagnostic.h (default_tree_printer): New prototype.
18516 2017-03-16 Tamar Christina <tamar.christina@arm.com>
18518 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
18519 Change ins into fmov.
18521 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18523 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
18524 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
18525 Use h_con constraint for operand 1.
18526 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
18527 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
18529 2017-03-15 Jeff Law <law@redhat.com>
18531 PR tree-optimization/71437
18532 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
18533 (record_temporary_equivalences): Use it.
18535 PR tree-optimization/71437
18536 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
18537 tree-ssa-scopedtables.
18538 (lookup_avail_expr, build_and_record_new_cond): Likewise.
18539 (record_conditions, record_cond, vuse_eq): Likewise.
18540 (record_edge_info): Adjust to API tweak of record_conditions.
18541 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
18542 (record_temporary_equivalences, optimize_stmt): Likewise.
18543 (eliminate_redundant_computations): Likewise.
18544 (record_equivalences_from_stmt): Likewise.
18545 * tree-ssa-scopedtables.c: Include options.h and params.h.
18546 (vuse_eq): New function, moved from tree-ssa-dom.c
18547 (build_and_record_new_cond): Likewise.
18548 (record_conditions): Likewise. Accept vector of conditions rather
18549 than edge_equivalence structure for first argument.
18550 for the first argument.
18551 (avail_exprs_stack::lookup_avail_expr): New member function, moved
18552 from tree-ssa-dom.c.
18553 (avail_exprs_stack::record_cond): Likewise.
18554 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
18555 from tree-ssa-dom.c.
18556 (avail_exprs_stack): Add new member functions lookup_avail_expr
18558 (record_conditions): Declare.
18560 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
18563 * lra-constraints.c (process_alt_operands): Increase reject for
18564 reloading an input/output operand.
18566 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
18569 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
18570 insns to convert from signed/unsigned char/short to IEEE 128-bit
18572 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
18574 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
18577 * config/i386/i386.c (ix86_vector_duplicate_value): Create
18578 subreg of inner mode for values already in registers.
18580 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
18582 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
18583 iteration reg is used after the loop.
18585 2017-03-14 Martin Sebor <msebor@redhat.com>
18587 PR tree-optimization/79800
18588 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
18589 precision in negative-positive range.
18590 (format_floating): Call non-const overload with adjusted precision.
18592 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
18595 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
18598 2017-03-14 Martin Sebor <msebor@redhat.com>
18600 PR middle-end/80020
18601 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
18602 * builtins.def (aligned_alloc): Use it.
18605 * Makefile.in (GTFILES): Add calls.c.
18606 * calls.c: Include "gt-calls.h".
18608 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
18610 PR rtl-optimization/79728
18611 * regs.h (struct target_regs): New field
18612 x_contains_allocatable_regs_of_mode.
18613 (contains_allocatable_regs_of_mode): New macro.
18614 * reginfo.c (init_reg_sets_1): Initialize it, and change
18615 contains_reg_of_mode so it includes global regs as well.
18616 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
18617 rather than contains_regs_of_mode.
18619 2017-03-14 Martin Liska <mliska@suse.cz>
18621 * doc/invoke.texi: Document options that can't be combined with
18622 -fcheck-pointer-bounds.
18624 2017-03-14 Martin Liska <mliska@suse.cz>
18626 PR middle-end/79831
18627 * doc/invoke.texi (-Wchkp): Document the option.
18629 2017-03-14 Martin Liska <mliska@suse.cz>
18631 * Makefile.in: Install gcov-dump.
18633 2017-03-14 Martin Liska <mliska@suse.cz>
18635 * multiple_target.c (expand_target_clones): Bail out for
18636 an invalid attribute.
18638 2017-03-14 Richard Biener <rguenther@suse.de>
18640 * alias.c (struct alias_set_entry): Pack properly.
18641 * cfgloop.h (struct loop): Likewise.
18642 * cse.c (struct set): Likewise.
18643 * ipa-utils.c (struct searchc_env): Likewise.
18644 * loop-invariant.c (struct invariant): Likewise.
18645 * lra-remat.c (struct cand): Likewise.
18646 * recog.c (struct change_t): Likewise.
18647 * rtl.h (struct address_info): Likewise.
18648 * symbol-summary.h (function_summary): Likewise.
18649 * tree-loop-distribution.c (struct partition): Likewise.
18650 * tree-object-size.c (struct object_size_info): Likewise.
18651 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
18652 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
18653 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
18654 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
18655 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
18656 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
18657 (struct _stmt_vec_info): Likewise.
18659 2017-03-14 Martin Liska <mliska@suse.cz>
18662 * multiple_target.c (create_dispatcher_calls): Check that
18663 a target can create a function dispatcher.
18665 2017-03-14 Martin Liska <mliska@suse.cz>
18668 * multiple_target.c (expand_target_clones): Drop local.local
18669 flag for default implementation.
18671 2017-03-14 Richard Biener <rguenther@suse.de>
18673 PR tree-optimization/80030
18674 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
18676 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
18678 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
18679 gcc_fallthrough() instead of __attribute__((fallthrough));
18681 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
18683 * doc/gcc.texi: Remove "up" link to (DIR).
18684 * doc/gccint.texi: Ditto.
18686 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
18688 * doc/install.texi (Specific) <avr>: Remove reference to
18691 2017-03-13 Jeff Law <law@redhat.com>
18693 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
18694 attribute rather than comments.
18696 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
18697 match_scratch operand is highest.
18699 2017-03-13 Martin Liska <mliska@suse.cz>
18701 PR middle-end/78339
18702 * ipa-pure-const.c (warn_function_noreturn): If the declarations
18703 is a CHKP clone, use original declaration.
18705 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18707 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
18708 (arc_conditional_register_usage): Use a different allocation order
18709 when optimizing for size.
18710 * common/config/arc/arc-common.c (arc_option_optimization_table):
18711 Section anchors default on when optimizing for size.
18713 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18715 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
18717 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18719 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
18720 * config/arc/arc.md (cpu_facility): Add cd variant.
18721 (*movqi_insn): Add code density variant.
18722 (*movhi_insn): Likewise.
18723 (*movqi_insn): Likewise.
18724 (*addsi3_mixed): Likewise.
18725 (subsi3_insn): Likewise.
18727 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18729 * config/arc/arc.md (movsi_cond_exec): Update constraint.
18731 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18733 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
18734 expressions with MINUS and UNARY ops.
18736 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18739 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
18741 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
18742 between vec_select and vector argument.
18743 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
18744 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
18745 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
18746 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
18747 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
18748 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
18750 2017-03-13 Richard Biener <rguenther@suse.de>
18753 * params.def (vect-max-peeling-for-alignment): Fix typo.
18755 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18757 * doc/install.texi (Specific) <mips-*-*>: Remove description of
18758 issue that only occurred with binutils below 2.18.
18760 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18762 * doc/install.texi (Specific) <cris-axis-elf>: No longer
18763 refer to binutils 2.11/2.12 minimum.
18765 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18767 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
18768 ftp.kernel.org and simplify binutils requirement.
18770 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
18772 * doc/invoke.texi (Warning Options): Fix spelling of link-time
18774 (Optimize Options): Ditto. Also remove redundancy.
18776 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18778 PR translation/79848
18779 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
18781 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
18782 to G_ to avoid double translation.
18784 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18786 PR translation/79923
18787 * auto-profile.c (get_combined_location): Convert leading
18788 character of diagnostics to lower case and remove trailing period.
18789 (read_profile): Likewise for various diagnostics.
18790 * config/arm/arm.c (arm_option_override): Remove trailing period
18791 from various diagnostics.
18792 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
18793 (msp430_expand_delay_cycles): Likewise.
18795 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18798 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
18799 full command-line argument, rather than just "str".
18800 (aarch64_validate_march): Likewise.
18801 (aarch64_validate_mtune): Likewise.
18803 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
18805 PR rtl-optimization/78911
18806 * lra-assigns.c (must_not_spill_p): New function.
18807 (spill_for): Use it.
18809 2017-03-10 Jakub Jelinek <jakub@redhat.com>
18811 PR tree-optimization/79981
18812 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
18813 ATOMIC_COMPARE_EXCHANGE ifn result.
18814 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
18815 IFN_ATOMIC_COMPARE_EXCHANGE.
18817 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18820 * opts.c (parse_sanitizer_options): Add missing question mark to
18821 "did you mean" message.
18823 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18825 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
18827 (VMULEUH_UNS): Likewise.
18828 (VMULOUB_UNS): Likewise.
18829 (VMULOUH_UNS): Likewise.
18830 * config/rs6000/rs6000.c (builtin_function_type): Remove
18831 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
18833 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18836 * read-rtl-function.c (function_reader::read_rtx_operand): Update
18837 x with result of extra_parsing_for_operand_code_0.
18838 (function_reader::extra_parsing_for_operand_code_0): Convert
18839 return type from void to rtx, returning x. When reading
18840 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
18841 larger size containing struct block_symbol.
18843 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
18845 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
18846 -mfloat128-hardware without -m64.
18848 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
18851 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
18852 entries to the case statement that marks unsigned arguments to
18853 overloaded functions.
18855 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18857 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
18858 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
18860 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
18863 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
18864 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
18866 2017-03-10 Martin Liska <mliska@suse.cz>
18870 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
18871 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
18874 2017-03-10 Olivier Hainque <hainque@adacore.com>
18876 * tree-switch-conversion (array_value_type): Start by resetting
18877 candidate type to it's main variant.
18879 2017-03-10 Jakub Jelinek <jakub@redhat.com>
18881 PR rtl-optimization/79909
18882 * combine.c (try_combine): Use simplify_replace_rtx on individual
18883 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
18884 of the whole CALL_INSN_FUNCTION_USAGE.
18886 PR tree-optimization/79972
18887 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
18888 get_range_info on SSA_NAMEs. Formatting fixes.
18890 2017-03-10 Richard Biener <rguenther@suse.de>
18891 Jakub Jelinek <jakub@redhat.com>
18893 PR tree-optimization/77975
18894 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
18895 edge to be constant.
18896 (get_val_for): For constant x return it. Formatting fix.
18897 (loop_niter_by_eval): Avoid pointless looping if the next iteration
18898 would use the same bases as the current one.
18900 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18902 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
18903 instead of vec_select for V1TImode.
18904 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
18906 (VSX_LE_128): Add V1TI to this mode iterator.
18907 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
18908 (*vsx_le_perm_store_<mode>): Likewise.
18909 (pre-reload splitter for VSX stores): Likewise.
18910 (post-reload splitter for VSX stores): Likewise.
18911 (*vsx_xxpermdi2_le_<mode>): Likewise.
18912 (*vsx_lxvd2x2_le_<mode>): Likewise.
18913 (*vsx_stxvd2x2_le_<mode>): Likewise.
18915 2017-03-09 Michael Eager <eager@eagercon.com>
18917 Correct failures with --enable-checking=yes,rtl.
18919 * config/microblaze/microblaze.c (microblaze_expand_shift):
18920 Replace GET_CODE test with CONST_INT_P and INTVAL test with
18921 test for const0_rtx.
18922 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
18923 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
18925 2017-03-09 Richard Biener <rguenther@suse.de>
18927 PR tree-optimization/79977
18928 * graphite-scop-detection.c (scop_detection::merge_sese):
18929 Handle the case of extra exits to blocks dominating the entry.
18931 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
18933 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
18936 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
18938 PR rtl-optimization/79949
18939 * lra-constraints.c (process_alt_operands): Check memory when
18940 trying to predict a cycle. Print about the overall increase.
18942 2017-03-09 Richard Biener <rguenther@suse.de>
18944 PR middle-end/79971
18945 * gimple-expr.c (useless_type_conversion_p): Preserve
18946 TYPE_SATURATING for fixed-point types.
18948 2017-03-09 Richard Biener <rguenther@suse.de>
18951 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
18952 alignment of BLKmode params.
18954 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18957 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
18958 (VALL_NO_V2Q): Likewise.
18960 * config/aarch64/aarch64-simd.md
18961 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
18963 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
18964 VALL_NO_V2Q mode iterator.
18965 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
18967 2017-03-09 Martin Liska <mliska@suse.cz>
18969 PR tree-optimization/79631
18970 * tree-chkp-opt.c (chkp_is_constant_addr): Call
18971 tree_int_cst_sign_bit just for INTEGER constants.
18973 2017-03-09 Martin Liska <mliska@suse.cz>
18977 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
18980 2017-03-09 Marek Polacek <polacek@redhat.com>
18983 * tree.c (inchash::add_expr): Handle TREE_VEC.
18985 2017-03-09 Martin Liska <mliska@suse.cz>
18988 (chkp_narrow_size_and_offset): New function.
18989 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
18990 (void chkp_parse_bit_field_ref): New function.
18991 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
18992 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
18994 2017-03-09 Martin Liska <mliska@suse.cz>
18997 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
18998 (chkp_find_bounds_1): Remove gcc_unreachable.
19000 2017-03-09 Jakub Jelinek <jakub@redhat.com>
19003 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
19004 BUILT_IN_SYNC*, determine the access type from the size suffix and
19005 always build a MEM_REF with that type. Handle forgotten
19006 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
19009 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
19010 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
19011 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
19012 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
19013 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
19014 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
19015 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
19016 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
19017 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
19018 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
19019 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
19020 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
19021 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
19022 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
19023 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
19024 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
19025 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
19026 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
19027 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
19028 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
19029 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
19030 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
19031 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
19032 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
19033 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
19034 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
19035 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
19036 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
19037 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
19038 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
19039 definitions outside of __OPTIMIZE__ guarded section.
19042 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
19043 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
19044 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
19045 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
19048 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19050 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
19051 ("vfenez<mode>"): Add missing constraints.
19053 2017-03-08 Martin Sebor <msebor@redhat.com>
19056 * config/nds32/nds32.c (nds32_option_override):
19057 Fix misspelled diagnostic.
19059 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19062 * gimplify.c (gimplify_omp_for): Replace index var in outer
19063 taskloop statement with an artificial variable and add
19064 OMP_CLAUSE_PRIVATE clause for it.
19066 2017-03-08 Richard Biener <rguenther@suse.de>
19068 PR tree-optimization/79955
19069 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
19070 for accesses that are completely outside of the variable.
19072 2017-03-08 Andrew Haley <aph@redhat.com>
19074 PR tree-optimization/79943
19075 * tree-ssa-loop-split.c (compute_new_first_bound): When
19076 calculating the new upper bound, (END-BEG) should be added, not
19079 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19081 * config/avr/avr.md (setmemhi): Make sure match_dup
19082 operand number comes before match_scratch.
19084 2017-03-08 Richard Biener <rguenther@suse.de>
19086 PR tree-optimization/79920
19087 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
19088 with ncopies == 1 to ...
19089 (vect_transform_slp_perm_load): ... here. Properly compute
19090 all element loads by iterating VF times over the group. Do
19091 not handle ncopies (computed in a broken way) in
19092 vect_create_mask_and_perm.
19094 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19097 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
19098 is a uniform vector, use uniform_vector_p return value instead of
19099 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
19101 2017-03-07 Marek Polacek <polacek@redhat.com>
19103 PR middle-end/79809
19104 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
19105 (alloca_call_type): Likewise.
19107 2017-03-07 Martin Liska <mliska@suse.cz>
19109 * gcov.c (process_args): Put comment to correct location.
19111 2017-03-07 Martin Liska <mliska@suse.cz>
19113 PR middle-end/68270
19114 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
19115 Use array_at_struct_end_p instead of DECL_CHAIN (field).
19116 (chkp_narrow_bounds_for_field): Likewise.
19117 (chkp_parse_array_and_component_ref): Pass one more argument to
19120 2017-03-07 Richard Biener <rguenther@suse.de>
19122 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
19125 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
19127 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
19128 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
19130 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19133 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
19134 to end of description.
19135 (PARAM_MAX_STORES_TO_MERGE): Likewise.
19137 2017-03-07 Jakub Jelinek <jakub@redhat.com>
19139 PR rtl-optimization/79901
19140 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
19142 (*avx512f_<code><mode>3<mask_name>): ... this.
19143 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
19144 iterator instead of VI8_AVX2_AVX512BW.
19146 PR rtl-optimization/79901
19147 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
19148 min/max expander, expand it using expand_vec_cond_expr.
19151 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
19154 2017-03-06 Jakub Jelinek <jakub@redhat.com>
19157 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
19158 to void * for PCH reasons.
19159 * dwarf2out.c (output_loc_operands, output_die): Cast
19160 v.val_vec.array to unsigned char *.
19162 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
19165 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
19168 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
19170 PR rtl-optimization/79571
19171 * lra-constraints.c (process_alt_operands): Calculate static
19172 reject and subtract it from overall when only addresses will be
19175 2017-03-06 Julia Koval <julia.koval@intel.com>
19178 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
19179 incoming stack boundary to 128 for 64-bit targets.
19181 2017-03-06 Richard Biener <rguenther@suse.de>
19183 PR tree-optimization/79894
19184 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
19185 to NULL after folding it.
19187 2017-03-06 Richard Biener <rguenther@suse.de>
19189 PR tree-optimization/79824
19190 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
19191 check disabling peeling for gaps.
19193 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
19195 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
19196 attributes): Document gettimeofday.
19198 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
19200 * config/s390/s390.c (s390_option_override_internal): Set
19201 PARAM_MIN_VECT_LOOP_BOUND
19203 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
19205 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
19206 * config/s390/s390.md: Likewise.
19208 2017-03-06 Jakub Jelinek <jakub@redhat.com>
19211 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
19212 (<avx2_avx512>_perm<mode>): Rename to ...
19213 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
19215 (<avx512>_perm<mode>_mask): Rename to ...
19216 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
19218 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
19219 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
19220 instead of VI8F_256_512.
19221 (avx512f_perm<mode>): New define_expand.
19222 (avx512f_perm<mode>_mask): Likewise.
19223 (avx512f_perm<mode>_1<mask_name>): New define_insn.
19224 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
19226 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
19228 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
19229 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
19231 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
19233 2017-03-06 Martin Liska <mliska@suse.cz>
19236 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
19237 when having a SSA NAME w/o VAR_DECL assigned to it.
19239 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
19241 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
19242 msa_dpsub_<su>_d): Fix MODE for vec_select.
19244 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
19246 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
19248 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
19250 2017-03-06 Richard Biener <rguenther@suse.de>
19252 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
19253 * plugin.c (register_plugin_info): Likewise.
19254 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
19256 2017-03-05 Jakub Jelinek <jakub@redhat.com>
19258 * config/i386/sse.md (sse_storehps, sse_storelps,
19259 avx_<castmode><avxsizesuffix>_<castmode>,
19260 avx512f_<castmode><avxsizesuffix>_<castmode>,
19261 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
19262 in condition that at least one operand is not a MEM.
19264 2017-03-03 Jakub Jelinek <jakub@redhat.com>
19266 PR middle-end/79805
19267 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
19268 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
19270 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
19271 gimple_call_nothrow_p flag based on whether original builtin can throw.
19272 If it can, emit following stmts on the fallthrough edge.
19273 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
19274 don't create new bb if inserting just debug stmts on the edge, try to
19275 insert them on the fallthru bb or just reset debug stmts.
19277 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
19280 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
19281 restore recog_data (including the operand rtxes inside it) around
19282 the call to get_insn_template.
19284 2017-03-03 Martin Sebor <msebor@redhat.com>
19286 PR tree-optimization/79699
19287 * context.c (context::~context): Free MPFR caches to avoid
19288 a memory leak on program exit.
19290 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19292 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
19293 Use wide_int::ulow () instead of .elt (0).
19295 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
19297 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
19298 (*pushxf): Limit oF constraint to 32bit targets and add oC
19299 constraint for 64bit targets.
19300 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
19301 (*pushdf): Change rmF constraint to rmC.
19303 2017-03-03 Martin Liska <mliska@suse.cz>
19305 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
19306 Remove unused variable.
19308 2017-03-03 Jakub Jelinek <jakub@redhat.com>
19311 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
19312 is a memory operand, increase num_memory.
19313 (ix86_expand_args_builtin): Likewise.
19315 2017-03-03 Jan Hubicka <jh@suse.cz>
19318 * ipa-devirt.c (maybe_record_node): Properly handle
19319 __cxa_pure_virtual visibility.
19321 2017-03-03 Martin Liska <mliska@suse.cz>
19323 PR tree-optimization/79803
19324 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
19326 (pass_loop_prefetch::execute): Disabled optimization if an
19327 assumption about L1 cache size is not met.
19329 2017-03-03 Martin Liska <mliska@suse.cz>
19331 PR rtl-optimization/79574
19332 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
19333 (hash_scan_set): Likewise.
19334 (dump_hash_table): Likewise.
19335 (hoist_code): Likewise.
19337 2017-03-03 Richard Biener <rguenther@suse.de>
19339 * fixed-value.c (fixed_from_string): Restore use of elt (1)
19340 in place of uhigh ().
19341 (fixed_convert_from_real): Likewise.
19343 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
19346 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
19348 2017-03-03 Richard Biener <rguenther@suse.de>
19350 PR middle-end/79818
19351 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
19352 TYPE_OVERFLOW_UNDEFINED check.
19354 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19356 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
19358 (vector_ae_<mode>_p): Likewise.
19359 (vector_nez_<mode>_p): Likewise.
19360 (vector_ne_v2di_p): Likewise.
19361 (vector_ae_v2di_p): Likewise.
19362 (vector_ne_<mode>_p): Likewise.
19363 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
19365 (vsx_tsqrt<mode>2_fe): Likewise.
19367 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
19370 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
19372 2017-03-02 Jakub Jelinek <jakub@redhat.com>
19374 PR rtl-optimization/79780
19375 * cprop.c (one_cprop_pass): When second and further conditional trap
19376 in a single basic block is turned into an unconditional trap, turn it
19377 into a deleted note to avoid RTL verification failures.
19379 2017-03-02 Richard Biener <rguenther@suse.de>
19381 * fold-const.c (const_binop): Use ulow () instead of elt (0).
19383 2017-03-02 Richard Biener <rguenther@suse.de>
19385 PR tree-optimization/79345
19387 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
19388 param and abort the walk, returning -1 if it is hit.
19389 (walk_aliased_vdefs): Take a limit param and pass it on.
19390 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
19391 defaulting to 0 and return a signed int.
19392 * tree-ssa-uninit.c (struct check_defs_data): New struct.
19393 (check_defs): New helper.
19394 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
19395 about uninitialized memory.
19396 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
19397 bogus uninitialized warning.
19398 (fixed_convert_from_real): Likewise.
19400 2017-03-02 Bin Cheng <bin.cheng@arm.com>
19402 PR tree-optimization/66768
19403 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
19404 iv_use if base object can't be determined.
19406 2017-03-02 Jakub Jelinek <jakub@redhat.com>
19408 PR tree-optimization/79345
19409 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
19410 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
19411 (get_pattern_stats): Initialize it.
19412 * genemit.c (gen_expand): Verify match_scratch numbers come after
19413 match_operand/match_dup numbers.
19414 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
19415 match_scratch numbers.
19416 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
19418 * config/s390/s390.md (trunctdsd2): Likewise.
19420 2017-03-02 Richard Biener <rguenther@suse.de>
19422 * wide-int.h (wide_int_storage::operator=): Implement in terms
19425 2017-03-02 Richard Biener <rguenther@suse.de>
19427 PR tree-optimization/79777
19428 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
19429 the to insert expression to sth existing.
19431 2017-03-01 Martin Sebor <msebor@redhat.com>
19433 PR middle-end/79692
19434 * gimple-ssa-sprintf.c
19435 (directive::known_width_and_precision): New function.
19436 (format_integer): Use it.
19437 (get_mpfr_format_length): Consider the full range of precision
19438 when computing %g output with the # flag. Set the likely byte
19439 count to 3 rather than 1 when precision is indeterminate.
19440 (format_floating): Correct the lower bound of precision.
19442 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19444 * doc/invoke.texi: Document default code model for 64-bit Linux.
19446 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19449 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
19450 udiv rather than div since input pattern is unsigned.
19452 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
19454 * config/i386/i386.c (print_reg): Warn for values of
19455 unsupported size in integer register.
19457 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
19460 * config/rs6000/predicates.md (current_file_function_operand): Do
19461 not allow self calls to be local if the function is replaceable.
19463 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
19466 * config/rs6000/altivec.h (vec_ctz and others): Change the
19467 preprocessor macro that controls conditional compilation from
19468 _ARCH_PWR9 to __POWER9_VECTOR__.
19469 (vec_all_ne): Change parameterization of __altivec_scalar_pred
19470 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
19471 control (instead of _ARCH_PWR9 control) so that template
19472 definition uses power9-specific function.
19473 (vec_any_eq): Likewise.
19474 (vec_all_ne): Change macro definition to use a power9-specific
19475 expansion under #ifdef __POWER9_VECTOR__ control (instead of
19476 _ARCH_PWR9 control).
19477 (vec_any_eq) Likewise.
19478 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
19479 expansion for CMPNEF to remove support for xvcmpnesp instruction.
19480 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
19481 support for xvcmpnedp instruction.
19482 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
19483 macro expansion so that Power9 implementation of vec_all_ne does
19484 not use the AltiVec predicate framework.
19485 (VCMPNEH_P): Likewise.
19486 (VCMPNEW_P): Likewise.
19487 (VCMPNED_P): Likewise.
19488 (VCMPNEFP_P): Likewise.
19489 (VCMPNEDP_P): Likewise.
19490 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
19491 implementation of vec_any_eq to not use AltiVec predicate
19493 (VCMPAEH_P): Likewise.
19494 (VCMPAEW_P): Likewise.
19495 (VCMPAED_P): Likewise.
19496 (VCMPAEFP_P): Likewise.
19497 (VCMPAEDP_P): Likewise.
19498 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
19499 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
19500 not use the AltiVec predicate framework.
19501 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
19502 of vec_any_eq to not use AltiVec predicate framework.
19503 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19504 support for predefined __POWER9_VECTOR__ macro to indicate that
19505 Power9 instruction selection is enabled.
19506 (altivec_overloaded_builtins): Remove extraneous
19507 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
19508 function argument types RS6000_BTI_bool_V16QI and
19509 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
19510 entry for overloaded function argument types RS6000_BTI_bool_V4SI
19511 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
19512 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
19513 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
19514 Power9 for implementations of vec_cmpne. Change the signature for
19515 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
19516 (representing vec_all_ne) to remove the previously described first
19517 argument of type RS6000_BTI_INTSI, as this was an artifact of
19518 reliance on the AltiVec predicate framework, which is no longer
19519 used in the implementation of these functions. Add
19520 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
19521 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
19522 since, unlike the AltiVec predicate framework implementation, we
19523 do not share function descriptors between vec_alle and vec_anyeq.
19524 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
19525 set of modes that receive special treatment even when
19526 TARGET_P9_VECTOR is true. The special treatment emits code that
19527 does not depend on Power9 instructions.
19528 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
19529 define_expand to not rely on AltiVec predicate framework.
19530 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
19532 (vector_ne_v2di_p): Change this define_expand to not rely on
19533 AltiVec predicate framework.
19534 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
19536 (vector_ne_<mode>_p): Change this define_expand to not rely on
19537 AltiVec predicate framework.
19538 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
19540 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
19541 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
19542 define_insn pattern.
19543 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
19544 define_insn pattern because the xvcmpne<VSs>. instruction is not
19546 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
19547 instruction is not supported.
19549 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19551 * config/nvptx/nvptx.c: Include intl.h.
19553 2017-03-01 Martin Jambor <mjambor@suse.cz>
19556 * ipa-prop.h (ipa_bits): Removed field known.
19557 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
19558 to pointers. Adjusted their comments to warn about their sharing.
19559 (ipcp_transformation_summary): Change bits to a vector of pointers.
19560 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
19561 (ipa_get_ipa_bits_for_value): Declare.
19562 * tree-vrp.h (value_range): Mark as GTY((for_user)).
19563 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
19564 (ipa_bits_hash_table): Likewise.
19565 (ipa_vr_ggc_hash_traits): Likewise.
19566 (ipa_vr_hash_table): Likewise.
19567 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
19568 being pointers and vr_known being removed.
19569 (ipa_set_jf_unknown): Likewise.
19570 (ipa_get_ipa_bits_for_value): New function.
19571 (ipa_set_jfunc_bits): Likewise.
19572 (ipa_get_value_range): New overloaded functions.
19573 (ipa_set_jfunc_vr): Likewise.
19574 (ipa_compute_jump_functions_for_edge): Use the above functions to
19575 construct bits and vr parts of jump functions.
19576 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
19577 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
19579 (ipcp_grow_transformations_if_necessary): Also allocate
19580 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
19582 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
19583 them. Fix too long lines.
19584 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
19585 vr_known being removed.
19586 (ipa_read_jump_function): Use new setter functions to construct bits
19587 and vr parts of jump functions or set them to NULL.
19588 (write_ipcp_transformation_info): Adjust for bits being pointers.
19589 (read_ipcp_transformation_info): Likewise.
19590 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
19592 Include gt-ipa-prop.h.
19593 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
19595 (ipcp_store_bits_results): Likewise.
19596 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
19597 Do not write to existing jump functions but use a temporary instead.
19599 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19602 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
19603 attempt to use its first operand as BIT_FIELD_REF base.
19605 2017-03-01 Richard Biener <rguenther@suse.de>
19607 PR middle-end/79721
19608 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
19609 interpolating formula in wrapping arithmetic.
19610 (chrec_apply): Convert chrec_evaluate return value to wanted type.
19612 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19614 PR tree-optimization/79734
19615 * tree-vect-generic.c (expand_vector_condition): Optimize
19616 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
19617 Handle VEC_COND_EXPR where comparison has different inner width from
19618 type's inner width.
19620 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
19622 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
19623 markup, and similar issues. Remove @opindex entries for things
19624 that aren't options. Add missing -mmpy-option entries.
19626 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19628 PR tree-optimization/79737
19629 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
19630 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
19631 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
19632 instead of byte_size. Formatting fix.
19633 (shift_bytes_in_array_right): Formatting fix.
19635 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
19638 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
19639 condition on optimize for the leaf function test.
19641 2017-02-28 Martin Liska <mliska@suse.cz>
19644 * read-rtl-function.c (function_reader::handle_unknown_directive):
19645 Bail out when one uses -flto.
19647 2017-02-28 Martin Liska <mliska@suse.cz>
19649 * common.opt: Replace space with tabular for options of <number>
19651 * config/i386/i386.opt: Show <number> value for
19652 -mlarge-data-threshold.
19653 * opts.c (print_filtered_help): Do not display number in hexadecimal
19656 2017-02-28 Martin Liska <mliska@suse.cz>
19658 * common.opt: Fix --help=option -Q for options which are of
19661 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
19663 * config/i386/i386.c (print_reg): Error out for values
19664 of 8-bit size in invalid integer register.
19666 2017-02-28 Martin Sebor <msebor@redhat.com>
19668 PR tree-optimization/79691
19669 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
19671 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19674 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
19675 gcc_unreachable with output_operand_lossage.
19677 2017-02-28 Richard Biener <rguenther@suse.de>
19679 PR tree-optimization/79740
19680 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
19682 (visit_nary_op): Insert the nary into the hashtable if we
19683 pattern-matched sth.
19684 * tree-ssa-pre.c (eliminate_insert): Robustify.
19686 2017-02-28 Richard Biener <rguenther@suse.de>
19688 PR middle-end/79731
19689 * fold-const.c (decode_field_reference): Reject out-of-bound
19692 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19694 * config/i386/i386.c: Include intl.h.
19695 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
19696 instead of just cond ? "..." : "...".
19697 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
19698 * coverage.c (read_counts_file): Likewise.
19699 * omp-offload.c: Include intl.h.
19700 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
19701 of just cond ? "..." : "...".
19702 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
19703 of just cond ? "..." : "...".
19705 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
19708 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
19710 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
19711 'tune for' CPU name.
19712 * config/arm/arm-cpu-data.h: Regenerated.
19714 2017-02-28 Richard Biener <rguenther@suse.de>
19716 PR tree-optimization/79732
19717 * tree-inline.c (expand_call_inline): Do not shadow var.
19719 2017-02-28 Richard Biener <rguenther@suse.de>
19721 PR tree-optimization/79723
19722 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
19723 address-space properly.
19725 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
19727 * doc/optinfo.texi (Optimization groups): Fix option used for
19729 * doc/invoke.texi (-fopt-info): Document "omp".
19730 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
19731 (OPTGROUP_ALL): Add OPTGROUP_OMP.
19732 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
19733 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
19734 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
19736 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
19738 * dumpfile.c (optgroup_options): Instead of "openmp", associate
19739 OPTGROUP_OMP with "omp".
19741 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
19744 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
19745 for arithmetic shift of unsigned V2DI.
19747 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
19749 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
19750 arc/linux.h headers.
19751 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
19752 (LINK_SPEC): Likewise.
19753 (ARC_TLS_EXTRA_START_SPEC): Likewise.
19754 (EXTRA_SPECS): Likewise.
19755 (STARTFILE_SPEC): Likewise.
19756 (ENDFILE_SPEC): Likewise.
19757 (LIB_SPEC): Likewise.
19758 (TARGET_SDATA_DEFAULT): Likewise.
19759 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
19760 (MULTILIB_DEFAULTS): Likewise.
19761 (DWARF2_UNWIND_INFO): Likewise.
19762 * config/arc/big.h: New file.
19763 * config/arc/elf.h: Likewise.
19764 * config/arc/linux.h: Likewise.
19765 * config/arc/t-uClibc: Remove.
19767 2017-02-27 Bin Cheng <bin.cheng@arm.com>
19769 PR tree-optimization/77536
19770 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
19771 (tree_transform_and_unroll_loop): Use above function to compute the
19772 estimated niter of unrolled loop and use it when scaling profile.
19773 Also use count info rather than frequency if it's non-zero.
19774 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
19775 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
19776 (vect_transform_loop): Call above function.
19778 2017-02-27 Richard Biener <rguenther@suse.de>
19780 PR tree-optimization/45397
19781 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
19782 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
19783 (visit_nary_op): Add pattern matching for CSEing sign-changed
19784 or truncated operations with wider ones.
19786 2017-02-27 Richard Biener <rguenther@suse.de>
19788 PR tree-optimization/79690
19789 * tree-vect-stmts.c (vectorizable_store): Use vector type
19790 built from the DR with address-space.
19792 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
19794 * doc/invoke.texi (Optimize Options): Refine the description
19795 of asan-use-after-return.
19797 2017-02-25 Alan Modra <amodra@gmail.com>
19799 PR rtl-optimization/79584
19800 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
19801 base, not ad->base_term, the reg within base. Remove assertion
19802 that ad->base == ad->base_term. Replace gen_int_mode using
19803 bogus mode with const0_rtx.
19805 2017-02-25 Jakub Jelinek <jakub@redhat.com>
19807 PR middle-end/79396
19808 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
19809 FMA_EXPR like tcc_binary or tcc_unary.
19811 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
19814 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
19816 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
19817 (output_loc_operands): Handle DW_OP_call_ref and
19818 DW_OP_GNU_variable_value.
19819 (struct variable_value_struct): New type.
19820 (struct variable_value_hasher): Likewise.
19821 (variable_value_hash): New variable.
19822 (string_types): Remove.
19823 (copy_loc_descr): New function.
19824 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
19825 (prepend_loc_descr_to_each): New function.
19826 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
19827 instead of add_loc_descr_to_each if the first argument is single
19828 location list and the second has multiple.
19829 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
19830 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
19831 when looking for variable value which doesn't have other location info.
19832 (loc_list_from_tree): Formatting fix.
19833 (gen_array_type_die): Simplify DW_AT_string_length handling.
19834 (adjust_string_types): Remove.
19835 (gen_subprogram_die): Don't call adjust_string_types nor test/set
19836 string_types. Call resolve_variable_values.
19837 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
19838 (resolve_addr_in_expr): Likewise. Add A argument.
19839 (copy_deref_exprloc): Remove deref argument. Adjust for the
19840 original expression being DW_OP_GNU_variable_value with optionally
19841 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
19842 optionally after it.
19843 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
19844 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
19845 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
19846 (variable_value_hasher::hash, variable_value_hasher::equal): New
19848 (resolve_variable_value_in_expr, resolve_variable_value,
19849 resolve_variable_values, note_variable_value_in_expr,
19850 note_variable_value): New functions.
19851 (dwarf2out_early_finish): Call note_variable_value on all toplevel
19854 2017-02-24 Jakub Jelinek <jakub@redhat.com>
19857 * opts.h (handle_generated_option): Add GENERATED_P argument.
19858 * opts-common.c (handle_option): Adjust function comment.
19859 (handle_generated_option): Add GENERATED_P argument, pass it to
19861 (control_warning_option): Pass false to handle_generated_option
19863 * opts.c (maybe_default_option): Pass true to handle_generated_option
19865 * optc-gen.awk: Likewise.
19867 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19869 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
19870 a REG, look at the REG it is a SUBREG of.
19871 (splitter for cmpeqsi_t): Ditto.
19873 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19875 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
19876 the special USEs with the pattern of the insn, not the insn itself.
19878 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
19881 * doc/invoke.texi: Document -mload-store-pairs.
19883 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19884 Sandra Loosemore <sandra@codesourcery.com>
19886 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
19887 argument isn't a CONST_INT.
19888 (nios2_alternate_compare_const): Assert op is a CONST_INT.
19889 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
19890 (nios2_validate_compare): Bypass alternate compare logic if *op2
19891 is not a CONST_INT.
19892 (ldstwm_operation_p): Return false if first_base is not a REG or
19893 if first_offset is not a CONST_INT.
19895 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19897 * config/cris/cris.md: Use correct operand in a define_peephole2.
19899 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19901 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
19903 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19905 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
19906 this_insn if it is an INSN or JUMP_INSN.
19907 (force_offsettable): Look at base, not at addr.
19908 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
19909 on things that aren't necessarily CONST_INTs.
19911 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
19913 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
19914 -mfpmath=sse is the default also for x86-32 targets with SSE2
19915 instruction set when @option{-ffast-math} is enabled
19917 2017-02-24 Jeff Law <law@redhat.com>
19919 PR rtl-optimizatoin/79286
19920 * ira.c (update_equiv_regs): Drop may_trap_p exception to
19923 2017-02-24 Richard Biener <rguenther@suse.de>
19925 PR tree-optimization/79389
19926 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
19929 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
19931 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
19932 function comment to reflect reality.
19933 (loop_exits_before_overflow): Fix typo in function description.
19935 2017-02-24 Richard Biener <rguenther@suse.de>
19937 PR tree-optimization/79389
19938 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
19939 properly that a threading opportunity exists. Detect conditional
19940 copy/constant propagation opportunities.
19942 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
19944 * config/visium/visium.md (type): Add trap.
19945 (b): New mode attribute.
19946 (*btst): Rename into...
19947 (*btst<mode>): ...this and adjust.
19948 (*cbranchsi4_btst_insn): Rename into...
19949 (*cbranch<mode>4_btst_insn): ...this and adjust.
19950 (trap): New define_insn.
19952 2017-02-23 Jakub Jelinek <jakub@redhat.com>
19954 PR tree-optimization/79389
19955 * ifcvt.c (struct noce_if_info): Add rev_cond field.
19956 (noce_reversed_cond_code): New function.
19957 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
19958 reversed_comparison_code. Formatting fix.
19959 (noce_try_store_flag): Test rev_cond != NULL in addition to
19960 reversed_comparison_code.
19961 (noce_try_store_flag_constants): Likewise.
19962 (noce_try_store_flag_mask): Likewise.
19963 (noce_try_addcc): Use rev_cond if non-NULL instead of
19964 reversed_comparison_code.
19965 (noce_try_cmove_arith): Likewise. Formatting fixes.
19966 (noce_try_minmax, noce_try_abs): Clear rev_cond.
19967 (noce_find_if_block): Initialize rev_cond.
19968 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
19969 instead of false as last argument never attempt to reverse it
19972 2017-02-23 Bin Cheng <bin.cheng@arm.com>
19974 PR tree-optimization/79663
19975 * tree-predcom.c (combine_chains): Process refs in reverse order
19976 only for ZERO length chains, and add explaining comment.
19978 2017-02-23 Jeff Law <law@redhat.com>
19980 PR tree-optimization/79578
19981 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
19982 in call to operand_equal_p.
19984 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
19987 * config/i386/cpuid.h: Fix another undefined behavior.
19989 2017-02-23 Richard Biener <rguenther@suse.de>
19991 PR tree-optimization/79683
19992 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
19993 vector types for data-refs.
19995 2017-02-23 Martin Liska <mliska@suse.cz>
19997 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
19999 2017-02-23 Jakub Jelinek <jakub@redhat.com>
20001 PR middle-end/79665
20002 * internal-fn.c (get_range_pos_neg): Moved to ...
20003 * tree.c (get_range_pos_neg): ... here. No longer static.
20004 * tree.h (get_range_pos_neg): New prototype.
20005 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
20006 are known to be in between 0 and signed maximum inclusive, try to
20007 expand both unsigned and signed divmod and use the cheaper one from
20010 2017-02-22 Jeff Law <law@redhat.com>
20012 PR tree-optimization/79578
20013 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
20014 to compare base operands.
20016 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
20019 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
20020 gpc_reg_operand instead of fpr_reg_operand.
20022 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
20024 * config/mips/mips.c (mips_return_in_memory): Force FP
20025 vector types to be returned in memory for o32 ABI.
20027 2017-02-22 Jakub Jelinek <jakub@redhat.com>
20029 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
20030 instead of DW_TAG_member for static data member declarations and don't
20031 set no_linkage_name for static inline data members.
20032 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
20035 2017-02-22 Martin Liska <mliska@suse.cz>
20037 * doc/invoke.texi: Replace inequality signs with square brackets
20040 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20042 PR tree-optimization/68644
20043 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
20045 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
20048 * lra-constraints.c (simplify_operand_subreg): Handle
20049 WORD_REGISTER_OPERATIONS targets.
20051 2017-02-22 Jakub Jelinek <jakub@redhat.com>
20054 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
20055 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
20056 elimination by swapping fld*.
20058 2017-02-22 Richard Biener <rguenther@suse.de>
20060 PR tree-optimization/79673
20061 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
20062 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
20063 irrelevant address-space qualifiers and avoiding a
20064 ADDR_SPACE_CONVERT_EXPR from fold_convert.
20066 2017-02-22 Richard Biener <rguenther@suse.de>
20068 PR tree-optimization/79666
20069 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
20070 to not symbolically negate if that may introduce undefined
20073 2017-02-22 Martin Liska <mliska@suse.cz>
20076 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
20077 * data-streamer-out.c (streamer_write_gcov_count_stream):
20079 * value-prof.c (stream_out_histogram_value): Make assert more
20080 precise based on type of counter.
20082 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
20085 * config/i386/i386.md (standard_x87sse_constant_load splitter):
20086 Use nonimmediate_operand instead of memory_operand for operand 1.
20087 (float-extend standard_x87sse_constant_load splitter): Ditto.
20089 2017-02-21 Jeff Law <law@redhat.com>
20091 PR tree-optimization/79621
20092 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
20093 blocks with edges to themselves.
20095 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20098 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
20099 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
20100 Use gimple_call_builtin_p.
20103 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
20104 on temporarily removed DEBUG_INSNs.
20106 PR tree-optimization/79649
20107 * tree-loop-distribution.c (classify_partition): Give up on
20108 non-generic address space loads/stores.
20110 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
20112 * doc/loop.texi (Loop manipulation): Remove nonexistent
20113 tree_ssa_loop_version from the documentation.
20114 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
20116 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20119 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
20120 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
20121 * config/rs6000/rs6000.c: Include except.h.
20122 (rs6000_expand_split_stack_prologue): Call
20123 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
20125 2017-02-21 Martin Jambor <mjambor@suse.cz>
20128 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
20129 have been analyzed.
20131 2017-02-21 Martin Jambor <mjambor@suse.cz>
20133 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
20134 for backward compatibility only.
20135 * doc/invoke.texi (Option Summary): Remove all references to
20136 -fipa-cp-alignment.
20138 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
20142 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20144 * lra-constraints.c (curr_insn_transform): Handle
20145 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
20147 2017-02-21 Martin Liska <mliska@suse.cz>
20149 * config/i386/i386.opt: Replace -masm-dialect with -masm.
20151 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
20153 PR translation/79638
20154 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
20156 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
20159 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
20160 (arm_function_ok_for_sibcall): Return false for an indirect call by
20161 descriptor if all the argument registers are used.
20162 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
20163 alignment of the function.
20165 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20167 PR tree-optimization/61441
20168 * simplify-rtx.c (simplify_const_unary_operation): For
20169 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
20170 the sNaN unmodified.
20172 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
20174 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20175 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20176 instead of SYSTEM_HEADER_DIR.
20178 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
20179 Martin Liška <mliska@suse.cz>
20181 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
20182 Fix typos and grammar, use active voice, and clarify.
20184 2017-02-20 Marek Polacek <polacek@redhat.com>
20186 PR middle-end/79537
20187 * gimplify.c (gimplify_expr): Handle unused *&&L;.
20190 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
20192 2017-02-20 Jakub Jelinek <jakub@redhat.com>
20195 * config/i386/i386.c (ix86_expand_builtin): Handle
20196 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
20197 ix86_builtins_isa[fcode].isa as a requirement of those
20198 flags and any other flag in the bitmask.
20199 (ix86_init_mmx_sse_builtins): Use 0 instead of
20200 ~OPTION_MASK_ISA_64BIT as mask.
20201 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
20202 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
20203 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
20204 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
20206 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20209 * lra-constraints.c (split_reg): Check requested split mode
20210 is supported by the register.
20212 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20214 * lra-constraints.c (simplify_operand_subreg): Remove early
20217 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20220 * lra-constraints.c (curr_insn_transform): Tighten condition
20221 for converting SUBREG reloads from OP_OUT to OP_INOUT.
20223 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20226 * lra-constraints.c (curr_insn_transform): Handle
20227 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
20229 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
20232 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
20234 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
20236 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
20239 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
20242 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20244 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
20247 2017-02-18 Jakub Jelinek <jakub@redhat.com>
20249 * final.c (last_columnnum, override_columnnum): New variables.
20250 (final_start_function): Set last_columnnum, pass it to begin_prologue
20251 hook and pass 0 to dwarf2out_begin_prologue.
20252 (final_scan_insn): Update override_columnnum. Pass last_columnnum
20253 to source_line debug hook.
20254 (notice_source_line): Compute last_columnnum and for debug_column_info
20255 return true on column changes.
20256 * debug.h (struct gcc_debug_hooks): Add column argument to
20257 source_line and begin_prologue hooks.
20258 (debug_nothing_int_charstar_int_bool): Remove prototype.
20259 (debug_nothing_int_int_charstar,
20260 debug_nothing_int_int_charstar_int_bool): New prototypes.
20261 (dwarf2out_begin_prologue): Add column argument.
20262 * debug.c (do_nothing_debug_hooks): Adjust source_line and
20263 begin_prologue hooks.
20264 (debug_nothing_int_charstar_int_bool): Remove.
20265 (debug_nothing_int_int_charstar,
20266 debug_nothing_int_int_charstar_int_bool): New functions.
20267 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
20268 through to dwarf2out_source_line.
20269 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
20270 (dwarf2out_source_line): Add column argument, emit it if requested.
20271 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
20273 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
20274 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
20275 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
20276 through to dwarf2out_begin_prologue.
20277 (vmsdbgout_source_line): Add column argument, pass it through to
20278 dwarf2out_source_line.
20279 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
20280 dbxout_source_line caller.
20281 (dbxout_source_line): Add column argument.
20283 * common.opt (gno-column-info, gcolumn-info): New options.
20284 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
20285 (check_die): Also test for multiple DW_AT_decl_column attributes.
20286 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
20287 DW_AT_decl_column if requested.
20288 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
20290 (gen_variable_die): Likewise.
20291 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
20292 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
20295 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
20296 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
20297 (ix86_handle_option): Handle OPT_m3dnowa.
20298 * doc/invoke.texi (-m3dnowa): Document.
20299 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
20300 -m3dnowa instead of -m3dnow -march=athlon.
20303 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
20304 instead of gcc_assert for K, r and R code checks. Formatting fixes.
20306 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20309 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
20310 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
20311 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
20312 generator for vsx_xxpermdi_<mode>_be.
20313 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
20314 force big-endian semantics.
20315 (vsx_xxpermdi_<mode>_be): New define_expand with same
20316 implementation as previous version of vsx_xxpermdi_<mode>.
20318 2017-02-17 Jakub Jelinek <jakub@redhat.com>
20320 PR tree-optimization/79327
20321 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
20322 variable, its initialization and use.
20324 2017-02-17 Julia Koval <julia.koval@intel.com>
20326 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
20327 (OPTION_MASK_ISA_PKU_UNSET): New.
20328 (ix86_handle_option): Handle -mrdpid.
20329 * config/i386/cpuid.h (bit_RDPID): New.
20330 * config/i386/driver-i386.c (host_detect_local_cpu):
20331 Detect RDPID feature.
20332 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
20333 * config/i386/i386-c.c (ix86_target_macros_internal):
20335 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
20336 (ix86_valid_target_attribute_inner_p): Add "rdpid".
20337 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
20338 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
20339 * config/i386/i386.md (define_insn "rdpid"): New.
20340 * config/i386/i386.opt Add -mrdpid.
20341 * config/i386/immintrin.h (_rdpid_u32): New.
20343 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
20345 PR rtl-optimization/79541
20346 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
20347 instead of transforming it into USE.
20349 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
20351 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
20352 If HONOR_SNANS (SFmode) force the input to a register.
20353 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
20354 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
20355 an frsp or similar insn.
20357 2017-02-17 Martin Liska <mliska@suse.cz>
20359 PR rtl-optimization/79577
20360 * params.def (selsched-max-sched-times): Increase minimum to 1.
20362 2017-02-17 Martin Liska <mliska@suse.cz>
20364 PR rtl-optimization/79574
20365 * gcse.c (want_to_gcse_p): Prevent integer overflow.
20367 2017-02-17 Martin Liska <mliska@suse.cz>
20369 PR tree-optimization/79529
20370 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
20371 ssa_defined_default_def_p to handle cases which are implicitly
20373 * tree-ssa.c (ssa_defined_default_def_p): New function.
20374 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
20375 which are implicitly defined.
20376 * tree-ssa.h (ssa_defined_default_def_p): Declare.
20378 2017-02-17 Richard Biener <rguenther@suse.de>
20380 PR middle-end/79576
20381 * params.def (max-ssa-name-query-depth): Limit to 10.
20383 2017-02-17 Richard Biener <rguenther@suse.de>
20385 PR tree-optimization/79552
20386 * tree-ssa-structalias.c (visit_loadstore): Properly verify
20389 2017-02-17 Richard Biener <rguenther@suse.de>
20392 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
20394 2017-02-17 Marek Polacek <polacek@redhat.com>
20396 PR middle-end/79536
20397 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
20398 (fold_negate_expr): New wrapper.
20400 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
20402 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
20403 Correct terminology and de-emphasize pre-standard behavior.
20405 2017-02-16 Alan Modra <amodra@gmail.com>
20407 PR rtl-optimization/79286
20408 * ira.c (def_dominates_uses): New function.
20409 (update_equiv_regs): Don't create an equivalence for insns that
20410 may trap where the register def does not dominate the use.
20412 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
20414 PR rtl-optimization/78127
20415 * lra.c (lra): Call lra_eliminate before finish the loop after
20418 2017-02-16 Richard Biener <rguenther@suse.de>
20420 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
20422 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
20423 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
20424 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
20425 (isl_val_int_from_wi): New function.
20426 (extract_affine_gmp): Rename to ...
20427 (extract_affine_wi): ... this, take a widest_int.
20428 (extract_affine_int): Just wrap extract_affine_wi.
20429 (add_param_constraints): Use isl_val_int_from_wi.
20430 (add_loop_constraints): Likewise, and extract_affine_wi.
20432 2017-02-15 Jeff Law <law@redhat.com>
20434 PR middle-end/79521
20435 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
20436 ira_init_register_move_cost_if_necessary.
20438 2017-02-15 Martin Sebor <msebor@redhat.com>
20440 PR middle-end/32003
20441 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
20442 removed in a prior commit.
20444 2017-02-15 Bin Cheng <bin.cheng@arm.com>
20446 PR tree-optimization/79347
20447 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
20448 counters during peeling.
20450 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
20452 * Makefile.in (site.exp): Remove "set ISLVER".
20454 2017-02-15 Jakub Jelinek <jakub@redhat.com>
20457 * real.c (real_from_integer): Call real_convert even for decimal.
20459 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
20462 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
20464 2017-02-14 Andrew Pinski <apinski@cavium.com>
20466 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
20467 cores and change the partno/implementer to be correct.
20468 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
20469 the 'B" as the implementer.
20470 * config/aarch64/aarch64-tune.md: Regenerate.
20472 2017-02-14 Carl Love <cel@us.ibm.com>
20474 * config/rs6000/rs6000.c: Add case statement entry to make the
20475 xvcvuxdsp built-in argument unsigned.
20476 * config/rs6000/vsx.md: Fix the source and return operand types so they
20477 match the instruction definitions from the ISA document. Fix typo
20478 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
20481 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
20484 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
20485 member early_clobber_alts.
20486 * lra-lives.c (reg_early_clobber_p): New.
20487 (process_bb_lives): Use it.
20488 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
20489 (debug_operand_data): Initialize early_clobber_alts.
20490 (setup_operand_alternative): Set up early_clobber_alts.
20491 (collect_non_operand_hard_regs): Ditto. Pass early clobber
20492 alternatives to new_insn_reg.
20493 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
20495 (lra_update_insn_regno_info): Pass the new arg.
20497 2017-02-14 Jakub Jelinek <jakub@redhat.com>
20499 PR middle-end/79505
20500 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
20501 (new_oacc_loop_raw): Don't clear already cleared fields.
20504 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
20505 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
20506 _mm512_prefetch_i64gather_ps): New inline functions and macros.
20508 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
20511 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
20513 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
20516 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
20517 the extra instruction to the right place to store 128-bit constant
20520 2017-02-14 Martin Sebor <msebor@redhat.com>
20522 PR middle-end/79448
20523 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
20524 warning for strings of unknown length.
20526 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
20528 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
20530 2017-02-14 Jeff Law <law@redhat.com>
20533 * ira-costs.c (scan_one_insn): Initialize register move costs
20534 for pseudos seen in USE/CLOBBER insns.
20536 PR tree-optimization/79095
20537 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
20538 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
20539 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
20540 if the operands are known to be not equal, then the resulting range
20542 (intersect_ranges): If the new range is ~[0,0] and the old range is
20543 wide, then prefer ~[0,0].
20544 * tree-vrp.c (overflow_comparison_p_1): New function.
20545 (overflow_comparison_p): New function.
20546 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
20547 if NAME is used in an overflow test.
20548 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
20549 overflow check that can be expressed as an equality test, then adjust
20550 ops to be that equality test.
20552 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20554 * config/s390/s390-builtin-types.def: Remove flags argument.
20555 * config/s390/s390.c (s390_init_builtins): Likewise.
20557 2017-02-14 Martin Liska <mliska@suse.cz>
20559 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
20560 vector. Fix trailing white spaces.
20562 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
20564 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
20567 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20569 PR rtl-optimization/68664
20570 * config/arm/arm.c (arm_sched_can_speculate_insn):
20571 New function. Declare prototype.
20572 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
20574 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20576 PR rtl-optimization/68664
20577 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
20579 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
20581 2017-02-14 Amit Pawar <amit.pawar@amd.com>
20583 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
20584 max skip bytes for function, loop and jump.
20586 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20588 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
20589 ABS_EXPR for gimple dump.
20591 2017-02-14 Jakub Jelinek <jakub@redhat.com>
20594 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
20596 PR tree-optimization/79408
20597 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
20598 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
20599 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
20600 also if rhs1 is INTEGER_CST.
20602 2017-02-14 Richard Biener <rguenther@suse.de>
20604 PR middle-end/79432
20605 * tree-into-ssa.c (insert_phi_nodes): When the function can
20606 have abnormal edges rewrite SSA names with broken use-def
20607 dominance out of SSA and register them for PHI insertion.
20609 2017-02-13 Martin Sebor <msebor@redhat.com>
20611 PR middle-end/79496
20612 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
20613 clearing info.nowrite flag when snprintf size argument is a range.
20615 2017-02-13 Jakub Jelinek <jakub@redhat.com>
20617 * cprop.c (cprop_jump): Add missing space in string literal.
20618 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
20619 (get_constraint_for_component_ref): Likewise.
20620 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
20621 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
20622 * lra-constraints.c (process_alt_operands): Likewise.
20623 * ipa-inline.c (inline_small_functions): Likewise.
20624 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
20625 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
20626 * trans-mem.c (diagnose_tm_1_op): Likewise.
20627 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
20628 (grid_parallel_clauses_gridifiable): Likewise.
20630 * config/nvptx/mkoffload.c (process): Add space in between
20633 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
20634 "MOD4_SSE_REGS" and "ALL_REGS".
20636 * spellcheck.c (test_data): Add , in between "foo" and "food".
20638 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20641 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
20642 boundary crossing check and subsequent code generation agree.
20644 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20646 * config/aarch64/aarch64.c (has_memory_op): Delete.
20647 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
20650 2017-02-13 Jakub Jelinek <jakub@redhat.com>
20652 PR rtl-optimization/79388
20653 PR rtl-optimization/79450
20654 * combine.c (distribute_notes): When removing TEM_INSN for which
20655 corresponding dest has last value recorded, invalidate that last
20658 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20660 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
20661 of explicit '@'. Add missing assembly comment marker on branch costs
20664 2017-02-13 Nathan Sidwell <nathan@acm.org>
20666 * gengtype-lex.l (<in_struct>): Add '/'.
20668 2017-02-13 Martin Liska <mliska@suse.cz>
20671 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
20673 2017-02-13 Richard Biener <rguenther@suse.de>
20675 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
20677 * configure: Re-generate.
20678 * config.in: Likewise.
20679 * graphite-dependences.c: Simplify as if
20680 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
20681 * graphite-isl-ast-to-gimple.c: Likewise.
20682 * graphite-optimize-isl.c: Likewise.
20683 * graphite-poly.c: Likewise.
20684 * graphite-sese-to-poly.c: Likewise.
20685 * graphite.h: Likewise.
20686 * toplev.c: Include isl/version.h and use isl_version () for
20687 printing the ISL version.
20688 * doc/install.texi: Update ISL requirement.
20690 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
20692 * doc/standards.texi (Standards): Update reference to
20695 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
20697 * doc/extend.texi (Named Address Spaces): sourceware.org now
20699 * doc/install.texi (Binaries): Ditto.
20702 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
20704 * doc/cpp.texi: Replace "stringify"/"stringification" with C
20705 standard terminology "stringize"/"stringizing" throughout.
20706 * doc/cppinternals.texi: Likewise.
20708 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
20710 * doc/extend.texi: Fix some spelling mistakes and typos.
20711 * doc/invoke.texi: Likewise.
20713 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
20716 * params.def (inline-min-speedup) Change from 10 to 8.
20718 2017-02-11 Jakub Jelinek <jakub@redhat.com>
20720 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
20723 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
20726 * ipa-inline-analysis.c (get_minimal_bb): New function.
20727 (record_modified): Use it.
20728 (remap_edge_change_prob): Handle also ancestor functions.
20730 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
20732 * doc/contrib.texi (Contributors): Remove broken link into
20733 the Mauve CVS repository.
20735 2017-02-11 Jakub Jelinek <jakub@redhat.com>
20737 PR middle-end/79454
20738 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
20739 result computation whenever lhs doesn't have vector mode, not
20740 just when it has BLKmode.
20742 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
20744 * doc/makefile.texi (profiledbootstrap): Refer to the
20745 installation instructions only in textual form.
20747 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20750 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
20752 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
20754 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
20755 (Specific): Update mingw-w64 reference.
20757 (Specific): Remove broken link to Renesas RX processor.
20759 2017-02-10 Richard Biener <rguenther@suse.de>
20761 * toplev.c (process_options): Do not mention obsolete graphite
20762 options when printing sorry message about missing graphite support.
20763 Mention -floop-nest-optimize.
20765 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
20767 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
20768 (vtst_p16): Likewise.
20769 (vtstq_p8): Likewise.
20770 (vtstq_p16): Likewise.
20772 (vtstq_p64): Likewise.
20773 * config/arm/arm_neon.h (vgetq_lane_p64): New.
20774 (vset_lane_p64): New.
20775 (vsetq_lane_p64): New.
20777 2017-02-10 Jakub Jelinek <jakub@redhat.com>
20779 PR tree-optimization/79411
20780 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
20781 stmt operands are SSA_NAMEs used in abnormal phis.
20782 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
20785 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
20788 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
20791 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
20793 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
20795 2017-02-09 Jakub Jelinek <jakub@redhat.com>
20797 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
20801 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
20802 not arbitrary TREE_CONSTANT.
20805 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
20806 "omp declare target link" attribute unless is_global_var.
20807 * omp-offload.c (find_link_var_op): Likewise.
20809 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
20810 Chung-Lin Tang <cltang@codesourcery.com>
20812 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
20814 (gimplify_adjust_omp_clauses): Don't delete TILE.
20815 (gimplify_omp_for): Deal with TILE.
20816 * internal-fn.c (expand_GOACC_TILE): New function.
20817 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
20819 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
20820 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
20822 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
20823 avoid DIV for outermost collapse var.
20824 (expand_oacc_for): Insert tile element loop as needed. Adjust.
20825 Remove out of date comments, fix whitespace.
20826 * omp-general.c (omp_extract_for_data): Deal with tiling.
20827 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
20828 adjust OLF_DIM_BASE value.
20829 (struct omp_for_data): Add tiling field.
20830 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
20831 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
20832 for auto loops. Remove default auto determining, moved to
20833 oacc_loop_fixed_partitions.
20834 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
20835 stmts, add e_mask field.
20836 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
20837 (oacc_thread_numbers): Use oacc_dim_call.
20838 (oacc_xform_tile): New.
20839 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
20840 (finish_oacc_loop): Adjust for ifns vector.
20841 (oacc_loop_discover_walk): Append loop abstraction sites to list,
20842 add case for GOACC_TILE fns.
20843 (oacc_loop_xform_loop): Delete.
20844 (oacc_loop_process): Iterate over call list directly, and add
20845 handling for GOACC_TILE fns.
20846 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
20848 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
20849 vector partitioning to outer loops. Assign 2 partitions to loops
20850 when available. Add TILE handling.
20851 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
20852 (execite_oacc_device_lower): Process GOACC_TILE fns,
20853 ignore unknown specs.
20854 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
20855 * tree.c (omp_clause_num_ops): Adjust TILE ops.
20856 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
20858 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
20860 * configure.ac (ACX_BUGURL): Update.
20861 * configure: Regenerate.
20863 2017-02-09 Richard Biener <rguenther@suse.de>
20865 PR tree-optimization/69823
20866 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
20867 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
20869 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
20871 * config/arc/arc-c.def: Add __NPS400__ definition.
20872 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
20873 (TARGET_NPS400): Define.
20875 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
20877 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
20879 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
20880 pointer, arch_info.
20881 (arc_cpu_types): Fill the arch_info field with a pointer into the
20882 arc_arch_types table.
20883 (arc_selected_cpu): Declare.
20884 * config/arc/arc.c (arc_selected_cpu): Make global.
20885 (arc_selected_arch): Delete.
20886 (arc_base_cpu): Delete.
20887 (arc_override_options): Remove references to deleted variables,
20888 update access to arch information.
20889 (ARC_OPT): Update access to arch information.
20890 (ARC_OPTX): Likewise.
20891 * config/arc/arc.h (arc_base_cpu): Remove declaration.
20892 (TARGET_ARC600): Update access to arch information.
20893 (TARGET_ARC601): Likewise.
20894 (TARGET_ARC700): Likewise.
20895 (TARGET_EM): Likewise.
20896 (TARGET_HS): Likewise.
20897 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
20900 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
20903 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
20904 condition/operands for integer GE/LE/GEU/LEU operations.
20906 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
20908 PR translation/79397
20909 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
20912 2017-02-08 Martin Jambor <mjambor@suse.cz>
20915 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
20916 whether allocation happened.
20917 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
20918 nothing was allocated.
20920 2017-02-08 Jakub Jelinek <jakub@redhat.com>
20922 PR tree-optimization/79408
20923 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
20924 constant, but SSA_NAME with a known integer range, use the minimum
20925 of that range instead of op1 to determine if modulo can be replaced
20926 with its first operand.
20928 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20930 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
20932 2017-02-08 Richard Biener <rguenther@suse.de>
20934 PR tree-optimization/71824
20935 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
20936 Check all loops contained in the merged region.
20938 2017-02-07 Andrew Pinski <apinski@cavium.com>
20940 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
20942 2017-02-07 Andrew Pinski <apinski@cavium.com>
20944 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
20945 (thunderxt88): Likewise.
20946 (thunderxt81): Disable LSE and change v8.1 to v8.
20947 (thunderxt83): Likewise.
20949 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20950 Richard Biener <rguenther@suse.de>
20952 PR middle-end/79399
20953 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
20954 type from int to size_t.
20955 * ira-costs.c (struct_costs_size): Change type from int to size_t.
20957 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20959 PR rtl-optimization/79386
20960 * cprop.c (bypass_conditional_jumps): Initialize
20961 bypass_last_basic_block already before splitting bbs after
20962 unconditional traps...
20963 (bypass_conditional_jumps): ... rather than here.
20966 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
20967 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
20968 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
20969 fix -masm=intel patterns.
20971 2017-02-07 Richard Biener <rguenther@suse.de>
20973 PR tree-optimization/79256
20974 PR middle-end/79278
20975 * builtins.c (get_object_alignment_2): Use min_align_of_type
20976 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
20977 and ADJUST_FIELD_ALIGN.
20979 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
20981 * doc/tm.texi: Regenerate.
20982 * stor-layout.c (layout_decl): Adjust.
20983 (update_alignment_for_field): Likewise.
20984 (place_field): Likewise.
20985 (min_align_of_type): Likewise.
20986 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
20987 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
20988 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
20989 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
20990 * config/frv/frv.c (frv_adjust_field_align): Likewise.
20991 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
20992 * config/i386/i386.c (x86_field_alignment): Likewise.
20993 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
20994 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
20995 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
20996 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
20997 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
20998 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
21002 2017-01-30 Richard Biener <rguenther@suse.de>
21004 PR tree-optimization/79256
21005 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
21006 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
21009 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
21011 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
21012 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
21013 builtins to SImode and emit a zero-extend, if necessary.
21015 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21017 * docs/invoke.texi (RISC-V Options): Alphabetize.
21019 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21021 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
21024 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21026 * config/riscv/riscv.c: New file.
21027 * common/config/riscv/riscv-common.c: Likewise.
21028 * config.gcc: Likewise.
21029 * config/riscv/constraints.md: Likewise.
21030 * config/riscv/elf.h: Likewise.
21031 * config/riscv/generic.md: Likewise.
21032 * config/riscv/linux.h: Likewise.
21033 * config/riscv/multilib-generator: Likewise.
21034 * config/riscv/peephole.md: Likewise.
21035 * config/riscv/pic.md: Likewise.
21036 * config/riscv/predicates.md: Likewise.
21037 * config/riscv/riscv-builtins.c: Likewise.
21038 * config/riscv/riscv-c.c: Likewise.
21039 * config/riscv/riscv-ftypes.def: Likewise.
21040 * config/riscv/riscv-modes.def: Likewise.
21041 * config/riscv/riscv-opts.h: Likewise.
21042 * config/riscv/riscv-protos.h: Likewise.
21043 * config/riscv/riscv.h: Likewise.
21044 * config/riscv/riscv.md: Likewise.
21045 * config/riscv/riscv.opt: Likewise.
21046 * config/riscv/sync.md: Likewise.
21047 * config/riscv/t-elf-multilib: Likewise.
21048 * config/riscv/t-linux: Likewise.
21049 * config/riscv/t-linux-multilib: Likewise.
21050 * config/riscv/t-riscv: Likewise.
21051 * configure.ac: Likewise.
21052 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
21053 Waterman as RISC-V maintainers.
21054 * doc/install.texi: Add RISC-V entries.
21055 * doc/invoke.texi: Add RISC-V options section.
21056 * doc/md.texi: Add RISC-V constraints section.
21057 * configure: Regenerated.
21059 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
21062 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
21063 false values to be constant vectors with all 0 or all 1 bits set.
21064 (vcondu<mode><mode>): Likewise.
21065 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
21067 (fpmask_comparison_operator): Update comment.
21068 (vecint_comparison_operator): New predicate.
21069 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
21070 vector conditionals when the true and false values are constant
21071 vectors with all 0 bits or all 1 bits set.
21073 2017-02-06 Martin Sebor <msebor@redhat.com>
21075 PR tree-optimization/79376
21076 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
21078 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
21080 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
21081 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
21082 to simplify split condition.
21084 2017-02-06 Jakub Jelinek <jakub@redhat.com>
21086 * omp-expand.c (oxpand_omp_atomic_fetch_op,
21087 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
21090 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
21092 PR rtl-optimization/68664
21093 * target.def (can_speculate_insn): New hook.
21094 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
21095 * doc/tm.texi: Regenerate.
21096 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
21097 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
21098 (rs6000_sched_can_speculate_insn): New function.
21100 2017-02-06 Jakub Jelinek <jakub@redhat.com>
21102 PR tree-optimization/79284
21103 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
21104 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
21105 vectorizable_mask_load_store, vectorizable_operation,
21106 vect_is_simple_cond, get_same_sized_vectype): Use it instead
21107 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
21108 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
21109 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
21110 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
21111 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
21112 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
21113 is_gimple_assign (stmt). Replace another such test with
21114 is_gimple_assign (stmt).
21116 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
21119 * config/avr/avr.c (rtl-iter.h): Include it.
21120 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
21121 (avr_legitimate_combined_insn): ...and implementation.
21123 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
21125 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
21126 * config/s390/s390.c (s390_const_operand_ok)
21127 (s390_canonicalize_comparison, s390_extract_part)
21128 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
21129 (s390_contiguous_bitmask_p, s390_rtx_costs)
21130 (legitimize_pic_address): Likewise.
21131 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
21132 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
21133 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
21134 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
21135 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
21137 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
21139 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
21140 REGNO($0) == REGNO($1).
21142 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21144 * config/s390/linux.h(SIZE_TYPE): Add comment.
21146 2017-02-06 Julian Brown <julian@codesourcery.com>
21147 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21148 Virendra Pathak <virendra.pathak@broadcom.com>
21150 * config/aarch64/aarch64-cores.def: Change the scheduler
21152 * config/aarch64/aarch64.md: Include thunderx2t99.md.
21153 * config/aarch64/thunderx2t99.md: New file.
21155 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
21157 * doc/standards.texi (Go Language): Update link to language
21160 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
21162 * tree-eh.c (lower_resx): Sanitize profile.
21163 (cleanup_empty_eh_move_lp): Likewise.
21165 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
21168 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
21170 * cfgloopmanip.h (loop_version): Update prototype.
21171 * modulo-sched.c (sms_schedule): Update call of loop_version.
21172 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
21173 * tree-parloops.c (gen_parallel_loop): Likewise.
21174 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
21175 * tree-ssa-loop-split.c (split_loop): Likewise.
21176 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
21177 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
21179 2017-02-05 Martin Liska <mliska@suse.cz>
21182 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
21184 (print_operand_address): Initialize a struct to zero.
21186 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
21188 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
21189 garbage collector only in textual form.
21191 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
21193 * doc/extend.texi (x86 specific memory model extensions for
21194 transactional memory): Simplify a phrase.
21196 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
21199 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
21200 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
21201 (atomic_storedi_1): Likewise.
21203 2017-02-04 Jakub Jelinek <jakub@redhat.com>
21205 PR tree-optimization/79338
21206 * tree-parloops.c (gather_scalar_reductions): Don't call
21207 vect_analyze_loop_form for loop->inner before destroying loop's
21210 2017-02-03 Martin Sebor <msebor@redhat.com>
21212 PR tree-optimization/79327
21213 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
21214 when precision has resulted in leading zeros.
21215 (format_integer): Adjust the likely counter to assume an unknown
21216 argument that may be zero is non-zero.
21218 2017-02-03 Jason Merrill <jason@redhat.com>
21221 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
21222 avoid copying non-taken branch.
21224 2017-02-03 Jakub Jelinek <jakub@redhat.com>
21226 PR tree-optimization/79340
21227 * tree-vect-loop.c (vectorizable_reduction): Release
21228 vec_defs elements after safe_splicing them into other vectors.
21231 PR tree-optimization/79327
21232 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
21233 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
21235 (format_integer): Use wide_int_to_tree instead of build_int_cst
21236 + to_?hwi. If argmin is NULL, just set argmin and argmax to
21237 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
21238 of shortest and longest sequence.
21240 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
21242 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
21243 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
21245 2017-02-03 Walter Lee <walt@tilera.com>
21248 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
21249 after initial stackframe link reg save.
21250 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
21252 2017-02-03 Jakub Jelinek <jakub@redhat.com>
21255 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
21256 wu for stxssp alternative.
21258 2017-02-03 Martin Sebor <msebor@redhat.com>
21260 PR tree-optimization/79352
21261 * gimple-fold.c (get_range_strlen): Add argument.
21262 (get_range_strlen): Change return type to bool.
21263 (get_maxval_strlen): Pass in a dummy argument.
21264 * gimple-fold.h (get_range_strlen): Change return type to bool.
21265 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
21266 * tree.h (array_at_struct_end_p): Add argument.
21267 * tree.c (array_at_struct_end_p): Handle it.
21269 2017-02-03 Martin Liska <mliska@suse.cz>
21272 * multiple_target.c (create_dispatcher_calls): Redirect edge
21273 from a caller of a dispatcher.
21274 (expand_target_clones): Make the clones local.
21275 (ipa_target_clone): Do both target clones and resolvers.
21276 (ipa_dispatcher_calls): Remove the pass.
21277 (pass_dispatcher_calls::gate): Likewise.
21278 (make_pass_dispatcher_calls): Likewise.
21279 * passes.def (pass_target_clone): Put as very first IPA early
21282 2017-02-03 Martin Liska <mliska@suse.cz>
21284 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
21285 in case of a function with ifunc attribute.
21287 2017-02-03 Martin Liska <mliska@suse.cz>
21289 * cgraph.c (cgraph_node::dump): Dump function version info.
21290 * symtab.c (symtab_node::dump_base): Add missing new line.
21292 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
21294 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
21295 (ifcombine_ifandif): Use it.
21297 2017-02-03 Martin Liska <mliska@suse.cz>
21299 * doc/invoke.texi: Document default value for
21300 use-after-scope-direct-emission-threshold.
21302 2017-02-03 Martin Liska <mliska@suse.cz>
21304 PR tree-optimization/79339
21305 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
21306 (format_floating): Likewise.
21308 2017-02-03 Martin Liska <mliska@suse.cz>
21311 * ipa-prop.c (ipa_node_params_t::insert): Remove current
21313 (ipa_node_params_t::remove): Likewise.
21314 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
21315 initialization from removed ipa_node_params_t::insert.
21316 (ipa_node_params::~ipa_node_params): Move from removed
21317 ipa_node_params_t::release.
21318 * symbol-summary.h (symbol_summary::m_released): New member.
21319 Do not release a summary twice. Do not allow to call finalizer
21320 for types of a summary that live in GGC memory.
21322 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21324 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
21327 2017-02-02 Martin Sebor <msebor@redhat.com>
21329 PR middle-end/79275
21330 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
21331 (format_string): Tighten up the range of output for non-constant
21332 strings and correct the expected range for wide non-constant strings.
21334 2017-02-02 Martin Sebor <msebor@redhat.com>
21336 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
21338 PR middle-end/32003
21339 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
21341 (-fdump-tree-@var): Add to index and document how to come up
21342 with pass-specific option and dump file names.
21343 (-fdump-passes): Clarify where to look for output.
21345 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
21347 PR middle-end/77445
21348 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
21349 statistics of the analyzed path; allow threading for speed when
21350 any of BBs along the path are optimized for speed.
21352 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
21354 PR middle-end/78468
21355 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
21356 settings of the virtual registers.
21359 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
21361 * explow.c (get_dynamic_stack_size): Take known alignment of stack
21362 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
21365 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21367 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
21368 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
21370 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21372 * config/s390/s390.md: Add missing comments with the expanded
21374 * config/s390/vector.md: Likewise.
21375 * config/s390/vx-builtins.md: Likewise.
21377 2017-02-02 Jakub Jelinek <jakub@redhat.com>
21380 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
21381 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
21382 conditions on a single line.
21384 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21386 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
21387 __S390_VX__ to __VX__.
21389 2017-02-01 Andrew Pinski <apinski@cavium.com>
21391 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
21392 stmt_info to record_stmt_cost.
21393 (vect_get_known_peeling_cost): Pass stmt_info if known to
21395 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
21396 cpu_vector_cost field into
21397 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
21398 field into vec_int_stmt_cost and vec_fp_stmt_cost.
21399 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
21400 splitting of scalar_stmt_cost and vec_stmt_cost.
21401 (thunderx_vector_cost): Likewise.
21402 (cortexa57_vector_cost): LIkewise.
21403 (exynosm1_vector_cost): Likewise.
21404 (xgene1_vector_cost): Likewise.
21405 (thunderx2t99_vector_cost): Improve after the splitting of the two
21407 (aarch64_builtin_vectorization_cost): Update for the splitting of
21408 scalar_stmt_cost and vec_stmt_cost.
21410 2017-02-01 Torvald Riegel <triegel@redhat.com>
21411 Richard Henderson <rth@redhat.com>
21413 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
21414 conditional on existance of a fast atomic load.
21415 * optabs-query.c (can_atomic_load_p): New function.
21416 * optabs-query.h (can_atomic_load_p): Declare it.
21417 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
21418 no fast atomic load is available for the particular size of access.
21419 (expand_atomic_compare_and_swap): Likewise.
21420 (expand_atomic_load): Likewise.
21421 (expand_atomic_store): Likewise.
21422 (expand_atomic_fetch_op): Likewise.
21423 * testsuite/lib/target-supports.exp
21424 (check_effective_target_sync_int_128): Remove x86 because it provides
21425 no fast atomic load.
21426 (check_effective_target_sync_int_128_runtime): Likewise.
21428 2017-02-01 Richard Biener <rguenther@suse.de>
21430 * graphite.c: Include tree-vectorizer.h for find_loop_location.
21431 (graphite_transform_loops): Provide opt-info for optimized nests.
21432 * tree-parloop.c (parallelize_loops): Provide opt-info for
21433 parallelized loops.
21435 2017-02-01 Richard Biener <rguenther@suse.de>
21437 PR middle-end/79315
21438 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
21439 was not set before.
21441 2017-02-01 Richard Biener <rguenther@suse.de>
21443 PR tree-optimization/71824
21444 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
21445 Verify the loops are valid in the merged SESE region.
21446 (scop_detection::can_represent_loop_1): Check analyzing the
21447 evolution of the number of iterations in the region succeeds.
21449 2017-01-31 Ian Lance Taylor <iant@golang.org>
21451 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
21452 REG_ARGS_SIZE note to 32-bit push insns and call insn.
21454 2017-01-31 David Malcolm <dmalcolm@redhat.com>
21456 PR preprocessor/79210
21457 * input.c (get_substring_ranges_for_loc): Replace line_width
21458 assertion with error-handling.
21460 2017-01-31 Richard Biener <rguenther@suse.de>
21462 PR tree-optimization/77318
21463 * graphite-sese-to-poly.c (extract_affine): Fix assert.
21464 (create_pw_aff_from_tree): Take loop parameter.
21465 (add_condition_to_pbb): Pass loop of the condition to
21466 create_pw_aff_from_tree.
21468 2017-01-31 Jakub Jelinek <jakub@redhat.com>
21470 * config/s390/s390.c (s390_asan_shadow_offset): New function.
21471 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
21473 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
21477 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
21479 (convert_int_to_float128): Likewise.
21480 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
21481 (convert_int_to_float128): Likewise.
21482 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
21483 (UNSPEC_IEEE128_CONVERT): Likewise.
21484 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
21485 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
21486 Use local variables for IBM extended format.
21487 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
21488 (fix_trunc<mode>si2_fprs): Likewise.
21489 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
21490 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
21491 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
21492 to know that we can now have integers of all sizes in vector
21494 (fix<uns>_<mode>di2_hw): Likewise.
21495 (float<uns>_<mode>si2_hw): Likewise.
21496 (fix_<mode>si2_hw): Likewise.
21497 (fixuns_<mode>si2_hw): Likewise.
21498 (float<uns>_<mode>di2_hw): Likewise.
21499 (float_<mode>di2_hw): Likewise.
21500 (float_<mode>si2_hw): Likewise.
21501 (floatuns_<mode>di2_hw): Likewise.
21502 (floatuns_<mode>si2_hw): Likewise.
21503 (xscvqp<su>wz_<mode>): Delete, no longer used.
21504 (xscvqp<su>dz_<mode>): Likewise.
21505 (xscv<su>dqp_<mode>): Likewise.
21506 (ieee128_mfvsrd_64bit): Likewise.
21507 (ieee128_mfvsrd_32bit): Likewise.
21508 (ieee128_mfvsrwz): Likewise.
21509 (ieee128_mtvsrw): Likewise.
21510 (ieee128_mtvsrd_64bit): Likewise.
21511 (ieee128_mtvsrd_32bit): Likewise.
21513 2017-01-31 Martin Liska <mliska@suse.cz>
21516 * ipa-prop.c (ipa_free_all_node_params): Call release method
21517 instead of ~sumbol_summary to not to trigger double times
21520 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
21522 PR tree-optimization/71691
21523 * bitmap.h (class auto_bitmap): New.
21524 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
21525 is_maybe_undefined instead of ssa_undefined_value_p.
21527 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21529 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
21530 __S390_ARCH_LEVEL__ to __ARCH__.
21532 2017-01-31 Jakub Jelinek <jakub@redhat.com>
21534 PR tree-optimization/79267
21535 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
21536 if should_remove_lhs_p is true.
21538 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
21541 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
21542 (add_alignment_attribute): New.
21543 (base_type_die): Add alignment attribute.
21544 (subrange_type_die): Likewise.
21545 (modified_type_die): Likewise.
21546 (gen_array_type_die): Likewise.
21547 (gen_descr_array_type_die: Likewise.
21548 (gen_enumeration_type_die): Likewise.
21549 (gen_subprogram_die): Likewise.
21550 (gen_variable_die): Likewise.
21551 (gen_field_die): Likewise.
21552 (gen_ptr_to_mbr_type_die): Likewise.
21553 (gen_struct_or_union_type_die): Likewise.
21554 (gen_subroutine_type_die): Likewise.
21555 (gen_typedef_die): Likewise.
21556 (base_type_cmp): Compare alignment attribute.
21558 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21561 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
21562 (setb_unsigned) New pattern for setb with CCUNS.
21563 * config/rs6000/rs6000.c (expand_block_compare): Use a different
21564 subfc./subfe sequence to avoid overflow problems. Generate a
21565 shorter sequence with cmpld/setb for power9.
21566 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
21567 for generating subfc. instruction.
21568 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
21569 now uses this instruction.
21571 2017-01-30 Ian Lance Taylor <iant@google.com>
21574 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
21575 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
21577 2017-01-30 Martin Sebor <msebor@redhat.com>
21579 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
21580 Move constant to the right of a relational operator.
21581 (get_mpfr_format_length, format_character, format_string): Ditto.
21582 (should_warn_p, maybe_warn): Same.
21584 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
21586 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
21589 * asan.c (get_translation_unit_decl): Remove function.
21590 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
21592 2017-01-30 Martin Liska <mliska@suse.cz>
21594 PR gcov-profile/79259
21595 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
21596 -fprofile-generate.
21598 2017-01-30 Martin Liska <mliska@suse.cz>
21601 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
21602 Initialize variables with NULL value.
21604 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
21607 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
21609 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
21611 2017-01-30 Richard Biener <rguenther@suse.de>
21613 PR tree-optimization/79276
21614 * tree-vrp.c (process_assert_insertions): Properly adjust common
21615 when removing a duplicate.
21617 * gcc.dg/torture/pr79276.c: New testcase.
21619 2017-01-30 Richard Biener <rguenther@suse.de>
21621 PR tree-optimization/79256
21622 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
21623 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
21625 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
21627 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
21630 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
21631 ("*r<noxa>sbg_<mode>_sll_bitmask")
21632 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
21633 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
21634 Use contiguous_bitmask_nowrap_operand.
21636 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21639 * config/rs6000/altivec.h (vec_xl): Revise #define.
21640 (vec_xst): Likewise.
21642 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
21644 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
21646 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
21648 PR rtl-optimization/79194
21649 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
21650 traps before call to bypass_conditional_jumps.
21652 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
21654 PR tree-optimization/71374
21655 * lra-constraints.c (check_conflict_input_operands): New.
21656 (match_reload): Use it.
21658 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
21661 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
21662 account to calculate conflict_set.
21664 2017-01-27 Bin Cheng <bin.cheng@arm.com>
21666 PR rtl-optimization/78559
21667 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
21668 other_insn in combine.
21670 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21672 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
21673 uint16_type_node for BT_UINT16.
21675 2017-01-27 David Malcolm <dmalcolm@redhat.com>
21677 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
21678 "RTL Tests" to menu.
21679 (GIMPLE Tests): New node.
21680 (RTL Tests): New node.
21682 2017-01-27 Richard Biener <rguenther@suse.de>
21684 PR tree-optimization/79245
21685 * tree-loop-distribution.c (distribute_loop): Apply cost
21686 modeling also to detected patterns.
21688 2017-01-27 Richard Biener <rguenther@suse.de>
21690 PR tree-optimization/71433
21691 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
21692 (compare_assert_loc): New function.
21693 (process_assert_insertions): Sort and optimize assert locations
21694 to remove duplicates and push down identical assertions on
21695 edges to their destination block.
21697 2017-01-27 Richard Biener <rguenther@suse.de>
21699 PR tree-optimization/79244
21700 * tree-vrp.c (remove_range_assertions): Forcefully propagate
21701 out SSA names even if abnormal.
21703 2017-01-27 Jakub Jelinek <jakub@redhat.com>
21705 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
21706 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
21707 instead of MPFR_RNDN.
21709 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
21712 * arm.c (arm_option_override): Don't call build_target_option_node
21713 until after doing all option overrides.
21714 (arm_valid_target_attribute_tree): Likewise.
21716 2017-01-27 Martin Liska <mliska@suse.cz>
21718 * doc/invoke.texi (-fprofile-arcs): Document profiling support
21719 for {cd}tors and C++ {cd}tors.
21721 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
21723 * config/s390/s390.md ("*setmem_long_and")
21724 ("*setmem_long_and_31z"): Use zero_extend instead of and.
21726 2017-01-26 Martin Sebor <msebor@redhat.com>
21728 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
21731 2017-01-26 Martin Sebor <msebor@redhat.com>
21733 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
21734 HAVE_DFmode before using XFmode or DFmode.
21735 (parse_directive): Avoid using the z length modifier to avoid
21736 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
21738 PR middle-end/78703
21739 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
21740 to accept adjustment as an array.
21741 (get_int_range): New function.
21742 (struct directive): Make width and prec arrays.
21743 (directive::set_width, directive::set_precision): Call get_int_range.
21744 (format_integer, format_floating): Handle width and precision ranges.
21745 (format_string, parse_directive): Same.
21747 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21750 * dwarf2out.c (generate_skeleton_bottom_up): For children with
21751 comdat_type_p set, just clone them, but keep the children in the
21755 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
21756 which have direct callers with -fvar-tracking-assignments enabled
21758 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
21759 inside of type units.
21761 2017-01-26 Martin Sebor <msebor@redhat.com>
21763 PR middle-end/78703
21764 * gimple-ssa-sprintf.c (struct result_range): Add likely and
21766 (struct format_result): Replace number_chars, number_chars_min,
21767 and number_chars_max with a single member of struct result_range.
21769 (format_result::operator+=): Adjust.
21770 (struct fmtresult): Remove bounded. Handle likely and unlikely
21772 (fmtresult::adjust_for_width_or_precision): New function.
21773 (fmtresult:type_max_digits): New function.
21774 (bytes_remaining): Handle likely and unlikely counters.
21775 (min_bytes_remaining): Remove.
21776 (format_percent): Simplify.
21777 (format_integer, format_floating): Set likely and unlikely counters.
21778 (get_string_length, format_character, format_string): Same.
21779 (format_plain, should_warn_p): New function.
21780 (maybe_warn): Call should_warn_p. Update diagnostic messages
21781 and handle those for all directives, including plain strings.
21782 (format_directive): Handle likely and unlikely counters.
21783 Remove unnecessary quoting from diagnostics. Add an informational
21785 (add_bytes): Remove.
21786 (pass_sprintf_length::compute_format_length): Simplify.
21787 (try_substitute_return_value): Handle likely and unlikely counters.
21789 2017-01-26 Carl Love <cel@us.ibm.com>
21791 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
21792 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
21794 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
21797 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
21798 endianess for subregs into account.
21799 * lra-constraints.c (lra_constraints): Do risky transformations
21800 always on the first iteration.
21801 * lra-lives.c (check_pseudos_live_through_calls): Add arg
21802 last_call_used_reg_set.
21803 (process_bb_lives): Define and use last_call_used_reg_set.
21804 * lra.c (lra): Always continue after lra_constraints on the first
21807 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
21809 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
21811 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
21813 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21815 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
21816 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
21817 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
21818 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
21819 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
21820 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
21821 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
21822 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
21823 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
21825 2017-01-26 Marek Polacek <polacek@redhat.com>
21828 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
21829 for the third operand.
21831 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21833 PR middle-end/79236
21834 * omp-low.c (struct omp_context): Add simt_stmt field.
21835 (scan_omp_for): Return omp_context *.
21836 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
21837 context to the _simt_ SIMD stmt.
21838 (lower_omp_for): For combined SIMD with sibling _simt_
21839 SIMD, make sure to use the same decls in _looptemp_
21840 clauses as in the sibling.
21842 2017-01-26 David Sherwood <david.sherwood@arm.com>
21844 PR middle-end/79212
21845 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
21848 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21851 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
21852 emit fld b; fld a; if possible.
21854 * brig-builtins.def: Update copyright years.
21855 * config/arm/arm_acle_builtins.def: Update copyright years.
21857 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
21860 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
21861 constraint instead of o for the stxsd instruction.
21863 2017-01-25 Carl Love <cel@us.ibm.com>
21865 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
21866 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
21868 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
21870 * doc/invoke.texi (C++ Dialect Options): Fix typo.
21872 2017-01-25 Richard Biener <rguenther@suse.de>
21874 PR tree-optimization/69264
21875 * target.def (vector_alignment_reachable): Improve documentation.
21876 * doc/tm.texi: Regenerate.
21877 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
21879 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
21880 earlier changes with respect to TYPE_USER_ALIGN.
21881 (vector_alignment_reachable_p): Likewise. Improve dumping.
21883 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21886 * config/arm/arm.md (xordi3): Force constant operand into a register
21889 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21891 * doc/invoke.texi (-fstore-merging): Correct default optimization
21892 levels at which it is enabled.
21893 (-O): Move -fstore-merging from list to...
21896 2017-01-25 Richard Biener <rguenther@suse.de>
21899 * omp-expand.c: Include debug.h.
21900 (expand_omp_taskreg): Make sure to generate early debug before
21901 outlining anything from a function.
21902 (expand_omp_target): Likewise.
21903 (grid_expand_target_grid_body): Likewise.
21905 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
21908 * asan.c (get_translation_unit_decl): New function.
21909 (asan_add_global): Extract modules file name from globals
21910 TRANSLATION_UNIT_DECL name.
21912 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
21915 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
21916 for long calls with APCS frame and VFP.
21918 2017-01-24 David Malcolm <dmalcolm@redhat.com>
21920 * cfg.c (original_copy_tables_initialized_p): New function.
21921 * cfg.h (original_copy_tables_initialized_p): New decl.
21922 * cfgrtl.c (relink_block_chain): Guard the call to
21923 free_original_copy_tables with a call to
21924 original_copy_tables_initialized_p.
21925 * cgraph.h (symtab_node::native_rtl_p): New decl.
21926 * cgraphunit.c (symtab_node::native_rtl_p): New function.
21927 (symtab_node::needed_p): Don't assert for early assembly output
21928 for __RTL functions.
21929 (cgraph_node::finalize_function): Set "force_output" for __RTL
21931 (cgraph_node::analyze): Bail out early for __RTL functions.
21932 (analyze_functions): Update assertion to support __RTL functions.
21933 (cgraph_node::expand): Bail out early for __RTL functions.
21934 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
21936 * function.h (struct function): Update comment for field
21938 * gimple-expr.c: Include "tree-pass.h".
21939 (gimple_has_body_p): Return false for __RTL functions.
21940 * Makefile.in (OBJS): Add run-rtl-passes.o.
21941 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
21943 (gcc::pass_manager::get_clean_slate): New accessor.
21944 * passes.c: Include "insn-addr.h".
21945 (should_skip_pass_p): Add logging. Update logic for running
21946 "expand" to be compatible with both __GIMPLE and __RTL. Guard
21947 property-provider override so it is only done for gimple passes.
21949 (skip_pass): New function.
21950 (execute_one_pass): Call skip_pass when skipping passes.
21951 * read-md.c (md_reader::read_char): Support filtering
21952 the input to a subset of line numbers.
21953 (md_reader::md_reader): Initialize fields
21954 m_first_line and m_last_line.
21955 (md_reader::read_file_fragment): New function.
21956 * read-md.h (md_reader::read_file_fragment): New decl.
21957 (md_reader::m_first_line): New field.
21958 (md_reader::m_last_line): New field.
21959 * read-rtl-function.c (function_reader::create_function): Only
21960 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
21961 curr_properties. Set DECL_INITIAL to a dummy block.
21962 (read_rtl_function_body_from_file_range): New function.
21963 * read-rtl-function.h (read_rtl_function_body_from_file_range):
21965 * run-rtl-passes.c: New file.
21966 * run-rtl-passes.h: New file.
21968 2017-01-24 Jeff Law <law@redhat.com>
21970 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
21973 2017-01-24 Bin Cheng <bin.cheng@arm.com>
21975 PR tree-optimization/79159
21976 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
21977 (record_nonwrapping_iv): Improve boundary using above function if no
21978 value range information.
21980 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
21981 Martin Jambor <mjambor@suse.cz>
21983 * brig-builtins.def: New file.
21984 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
21985 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
21986 (DEF_HSAIL_SAT_BUILTIN): Likewise.
21987 (DEF_HSAIL_INTR_BUILTIN): Likewise.
21988 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
21989 * builtin-types.def (BT_INT8): New.
21990 (BT_INT16): Likewise.
21991 (BT_UINT8): Likewise.
21992 (BT_UINT16): Likewise.
21993 (BT_FN_ULONG): Likewise.
21994 (BT_FN_UINT_INT): Likewise.
21995 (BT_FN_UINT_ULONG): Likewise.
21996 (BT_FN_UINT_LONG): Likewise.
21997 (BT_FN_UINT_PTR): Likewise.
21998 (BT_FN_ULONG_PTR): Likewise.
21999 (BT_FN_INT8_FLOAT): Likewise.
22000 (BT_FN_INT16_FLOAT): Likewise.
22001 (BT_FN_UINT32_FLOAT): Likewise.
22002 (BT_FN_UINT16_FLOAT): Likewise.
22003 (BT_FN_UINT8_FLOAT): Likewise.
22004 (BT_FN_UINT64_FLOAT): Likewise.
22005 (BT_FN_UINT16_UINT32): Likewise.
22006 (BT_FN_UINT32_UINT16): Likewise.
22007 (BT_FN_UINT16_UINT16_UINT16): Likewise.
22008 (BT_FN_INT_PTR_INT): Likewise.
22009 (BT_FN_UINT_PTR_UINT): Likewise.
22010 (BT_FN_LONG_PTR_LONG): Likewise.
22011 (BT_FN_ULONG_PTR_ULONG): Likewise.
22012 (BT_FN_VOID_UINT64_UINT64): Likewise.
22013 (BT_FN_UINT8_UINT8_UINT8): Likewise.
22014 (BT_FN_INT8_INT8_INT8): Likewise.
22015 (BT_FN_INT16_INT16_INT16): Likewise.
22016 (BT_FN_INT_INT_INT): Likewise.
22017 (BT_FN_UINT_FLOAT_UINT): Likewise.
22018 (BT_FN_FLOAT_UINT_UINT): Likewise.
22019 (BT_FN_ULONG_UINT_UINT): Likewise.
22020 (BT_FN_ULONG_UINT_PTR): Likewise.
22021 (BT_FN_ULONG_ULONG_ULONG): Likewise.
22022 (BT_FN_UINT_UINT_UINT): Likewise.
22023 (BT_FN_VOID_UINT_PTR): Likewise.
22024 (BT_FN_UINT_UINT_PTR: Likewise.
22025 (BT_FN_UINT32_UINT64_PTR): Likewise.
22026 (BT_FN_INT_INT_UINT_UINT): Likewise.
22027 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
22028 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
22029 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
22030 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
22031 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
22032 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
22033 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
22034 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
22035 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
22036 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
22037 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
22038 * doc/frontends.texi: List BRIG FE.
22039 * doc/install.texi (Testing): Add BRIG tesring requirements.
22040 * doc/invoke.texi (Overall Options): Mention BRIG.
22041 * doc/standards.texi (Standards): Doucment BRIG HSA version.
22043 2017-01-24 Richard Biener <rguenther@suse.de>
22045 PR translation/79208
22046 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
22048 2017-01-24 Martin Jambor <mjambor@suse.cz>
22051 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
22052 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
22053 and known_contexts.
22055 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
22057 PR middle-end/79123
22058 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
22059 casts from signed to unsigned really don't have a range.
22061 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
22063 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
22064 GMP_RNDx for compatiblity.
22066 2017-01-24 Martin Liska <mliska@suse.cz>
22069 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
22070 that would prevent us to call alloca with -1 as argument.
22072 2017-01-24 Jakub Jelinek <jakub@redhat.com>
22074 * dwarf2out.c (output_compilation_unit_header, output_file_names):
22075 Avoid -Wformat-security warning.
22077 2017-01-23 Andrew Pinski <apinski@cavium.com>
22079 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
22082 2017-01-23 Martin Sebor <msebor@redhat.com>
22084 PR middle-end/78703
22085 * gimple-ssa-sprintf.c (warn_level): New global.
22086 (format_integer): Use it here and throughout the rest of the file.
22087 Use the same switch to compute sign as base.
22088 (maybe_warn): New function.
22089 (format_directive): Factor out warnings into maybe_warn.
22090 Add debugging output. Use warn_level.
22091 (add_bytes): Use warn_level.
22092 (pass_sprintf_length::compute_format_length): Add debugging output.
22093 (try_substitute_return_value): Same.
22094 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
22096 PR middle-end/78703
22097 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
22098 (struct fmtresult, format_integer, format_floating): Adjust.
22099 (fmtresult::fmtresult): Set max correctly in two argument ctor.
22100 (get_string_length, format_string,format_directive): Same.
22101 (pass_sprintf_length::compute_format_length): Same.
22102 (try_substitute_return_value): Simplify slightly.
22104 PR middle-end/78703
22105 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
22106 (fmtresult::operator+=): Outlined.
22107 (struct fmtresult): Add ctors.
22108 (struct conversion_spec): Rename...
22109 (struct directive): ...to this. Add and remove data members.
22110 (directive::set_width, directive::set_precision): New functions.
22111 (format_percent): Use fmtresult ctor.
22112 (get_width_and_precision): Remove.
22113 (format_integer): Make naming changes. Avoid computing width and
22115 (format_floating): Same. Adjust indentation.
22116 (format_character, format_none): New functions.
22117 (format_string): Moved character handling to format_character.
22118 (format_directive): Remove arguments, change return type.
22119 (parse_directive): New function.
22120 (pass_sprintf_length::compute_format_length): Move directive
22121 parsing to parse_directive.
22123 2017-01-23 Jakub Jelinek <jakub@redhat.com>
22125 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
22126 (assign_assembler_name_if_needed): ... this.
22127 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
22128 (assign_assembler_name_if_needed): ... this.
22129 (free_lang_data_in_cgraph): Adjust callers.
22130 * cgraphunit.c (cgraph_node::analyze): Likewise.
22131 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
22134 2017-01-23 Richard Biener <rguenther@suse.de>
22136 PR tree-optimization/79088
22137 PR tree-optimization/79188
22138 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
22139 resetting loop bounds after last path deletion. Reset loop
22140 bounds of the target loop, make code match the comments.
22141 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
22142 Make sure loops need no fixups.
22144 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
22146 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
22147 exponent support with double type for first argument.
22148 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
22149 type returned by __builtin_vec_extract_sig,
22150 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
22151 functions from "vector int" to "vector unsigned int" or from
22152 "vector long long int" to "vector unsigned long long int".
22153 Changed type returned by __builtin_vec_extract_exp,
22154 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
22155 functions from "vector int" to "vector unsigned int" or from
22156 "vector long long int" to "vector unsigned long long int".
22157 Changed return type of __builtin_vec_test_data_class,
22158 __builtin_vec_test_data_class_sp, and
22159 __builtin_vec_test_data_class_dp from "vector int" to
22160 "vector bool int" or from "vector long long int" to "vector bool
22161 long long int" and changed second argument type from "unsigned
22162 int" to "int". Added new overloaded function forms "vector float
22163 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
22164 "vector float __builtin_vec_insert_exp_sp (vector float, vector
22165 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
22166 double, vector unsigned long long int)" and "vector double
22167 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
22168 long int)". Changed return type of
22169 __builtin_scalar_test_data_class and
22170 __builtin_scalar_test_data_class_sp and
22171 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
22172 int" and changed second argument from "unsigned int" to "int".
22173 Changed type returned by __builtin_scalar_test_neg,
22174 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
22175 from "int" to "bool int". Added new overloaded function form
22176 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
22177 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
22178 exponent double-precision with floating point first argument.
22179 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
22180 documentation of scalar_test_data_class, scalar_test_neg,
22181 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
22182 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
22183 vec_test_data_class built-in functions to reflect refinements in
22184 their type signatures.
22186 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
22188 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
22190 (aarch64_elf_asm_destructor): Likewise.
22192 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
22194 PR rtl-optimization/78634
22195 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
22196 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
22197 * ifcvt.c (noce_try_cmove): Add missing cost check.
22199 PR rtl-optimization/71724
22200 * combine.c (if_then_else_cond): Look for situations where it is
22201 beneficial to undo the work of one of the recursive calls.
22203 2017-01-23 Bin Cheng <bin.cheng@arm.com>
22205 PR tree-optimization/70754
22206 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
22207 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
22208 combined stmt before it if not NULL.
22209 (combine_chains): Process refs reversely and compute dominance point
22212 2017-01-23 Martin Liska <mliska@suse.cz>
22214 PR tree-optimization/79196
22215 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
22216 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
22218 (strlen_optimize_stmt): Call the renamed function.
22220 2017-01-23 Michael Matz <matz@suse.de>
22222 PR tree-optimization/78384
22223 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
22225 2017-01-23 Richard Biener <rguenther@suse.de>
22227 PR tree-optimization/79186
22228 * tree-vrp.c (register_new_assert_for): Make sure we've seen
22229 both incoming edges before moving an assert.
22231 2017-01-23 Martin Jambor <mjambor@suse.cz>
22233 * ipa-prop.c (load_from_param_1): Removed.
22234 (load_from_unmodified_param): Bits from load_from_param_1 put back
22236 (load_from_param): Removed.
22237 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
22238 with stmt. Reverted back to use of load_from_unmodified_param.
22240 2017-01-23 Martin Jambor <mjambor@suse.cz>
22243 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
22244 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
22245 field a pointer to garbage collected vector, mark lattices and
22246 ipcp_orig_node with GTY((skip)).
22247 (ipa_get_param_count): Adjust to descriptors being a pointer.
22248 (ipa_get_param): Likewise.
22249 (ipa_get_type): Likewise.
22250 (ipa_get_param_move_cost): Likewise.
22251 (ipa_set_param_used): Likewise.
22252 (ipa_get_controlled_uses): Likewise.
22253 (ipa_set_controlled_uses): Likewise.
22254 (ipa_is_param_used): Likewise.
22255 (ipa_node_params_t): Move into garbage collector. New methods insert
22257 (ipa_node_params_sum): Annotate wth GTY(()).
22258 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
22260 (ipa_load_from_parm_agg): Adjust declaration.
22261 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
22262 * ipa-profile.c (ipa_profile): Likewise.
22263 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
22264 (ipa_populate_param_decls): Make descriptors parameter garbage
22266 (ipa_dump_param): Adjust to descriptors being a pointer.
22267 (ipa_alloc_node_params): Likewise.
22268 (ipa_initialize_node_params): Likewise.
22269 (load_from_param_1): Make descriptors parameter garbage collected.
22270 (load_from_unmodified_param): Likewise.
22271 (load_from_param): Likewise.
22272 (ipa_load_from_parm_agg): Likewise.
22273 (ipa_node_params::~ipa_node_params): Removed.
22274 (ipa_free_all_node_params): Remove call to delete operator.
22275 (ipa_node_params_t::insert): New.
22276 (ipa_node_params_t::remove): Likewise.
22277 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
22278 copy known_csts and known_contexts vectors.
22279 (ipa_read_node_info): Adjust to descriptors being a pointer.
22280 (ipcp_modif_dom_walker): Make m_descriptors field garbage
22282 (ipcp_transform_function): Make descriptors variable garbage
22285 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
22287 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
22288 * config/i386/avx512dqintrin.h: Ditto.
22289 * config/i386/avx512fintrin.h: Ditto.
22290 * config/i386/i386.c: Handle new builtins.
22291 * config/i386/i386-builtin.def: Add new builtins.
22292 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
22293 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
22295 2017-01-23 Jakub Jelinek <jakub@redhat.com>
22296 Martin Liska <mliska@suse.cz>
22298 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
22299 * asan.c (asan_expand_poison_ifn): Support stores and use
22300 appropriate ASAN report function.
22301 * internal-fn.c (expand_ASAN_POISON_USE): New function.
22302 * internal-fn.def (ASAN_POISON_USE): Declare.
22303 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
22304 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
22305 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
22306 ASAN_POISON calls w/o LHS.
22307 * tree-ssa.c (execute_update_addresses_taken): Create clobber
22308 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
22309 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
22310 * gimplify.c (asan_poison_variables): Add attribute
22311 use_after_scope_memory to variables that really needs to live
22313 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
22314 having the attribute.
22316 2017-01-23 Martin Liska <mliska@suse.cz>
22318 * asan.c (create_asan_shadow_var): New function.
22319 (asan_expand_poison_ifn): Likewise.
22320 * asan.h (asan_expand_poison_ifn): New declaration.
22321 * internal-fn.c (expand_ASAN_POISON): Likewise.
22322 * internal-fn.def (ASAN_POISON): New builtin.
22323 * sanopt.c (pass_sanopt::execute): Expand
22324 asan_expand_poison_ifn.
22325 * tree-inline.c (copy_decl_for_dup_finish): Make function
22327 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
22328 * tree-ssa.c (is_asan_mark_p): New function.
22329 (execute_update_addresses_taken): Rewrite local variables
22330 (identified just by use-after-scope as addressable) into SSA.
22332 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
22334 * doc/install.texi (Specific): opensource.apple.com uses https
22335 now. Remove trailing slash.
22337 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
22339 * README.Portability: Remove note on an Irix compatibility issue.
22341 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
22343 * gcov.c (INCLUDE_ALGORITHM): Define.
22344 (INCLUDE_VECTOR): Define.
22345 No longer include <vector> and <algorithm> directly.
22347 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
22349 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
22351 * doc/invoke.texi (Code Gen Options): Ditto.
22353 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
22356 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
22358 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
22360 rtl-optimization/79125
22361 * cprop.c (local_cprop_pass): Handle cases where we make an
22362 unconditional trap.
22364 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
22368 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
22369 read from, for big endian.
22371 2017-01-20 Jiong Wang <jiong.wang@arm.com>
22373 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
22374 register pauth builtins for LP64 only.
22376 2017-01-20 Marek Polacek <polacek@redhat.com>
22379 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
22382 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
22384 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
22386 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
22387 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
22388 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
22390 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22393 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
22394 in big-endian mode when they are not a single duplicated value.
22396 2017-01-20 Richard Biener <rguenther@suse.de>
22398 * BASE-VER: Bump to 7.0.1.
22400 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
22402 * omp-low.c (omplow_simd_context): New struct. Use it...
22403 (lower_rec_simd_input_clauses): ...here and...
22404 (lower_rec_input_clauses): ...here to hold common data. Adjust all
22405 references to idx, lane, max_vf, is_simt.
22407 2017-01-20 Graham Markall <graham.markall@embecosm.com>
22409 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
22412 2017-01-20 Martin Jambor <mjambor@suse.cz>
22414 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
22415 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
22417 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
22418 (GTFILES): Rename hsa.c to hsa-common.c.
22419 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
22420 * hsa-dump.c: Likewise.
22421 * hsa-gen.c: Likewise.
22422 * hsa-regalloc.c: Likewise.
22423 * ipa-hsa.c: Likewise.
22424 * omp-expand.c: Likewise.
22425 * omp-low.c: Likewise.
22426 * toplev.c: Likewise.
22428 2017-01-20 Marek Polacek <polacek@redhat.com>
22431 * doc/invoke.texi: Document -Wduplicated-branches.
22432 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
22433 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
22434 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
22435 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
22436 return 0 only when not OEP_LEXICOGRAPHIC.
22437 (fold_build_cleanup_point_expr): Use the expression
22438 location when building CLEANUP_POINT_EXPR.
22439 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
22440 * tree.c (add_expr): Handle error_mark_node.
22442 2017-01-20 Martin Liska <mliska@suse.cz>
22445 * tree-profile.c (init_ic_make_global_vars): Do not call
22447 (gimple_init_gcov_profiler): Likewise.
22449 2017-01-20 Martin Liska <mliska@suse.cz>
22452 * cgraph.h (maybe_create_reference): Remove argument and
22454 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
22456 * ipa-cp.c (create_specialized_node): Likewise.
22457 * symtab.c (symtab_node::maybe_create_reference): Handle
22458 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
22460 2017-01-20 Martin Liska <mliska@suse.cz>
22462 * read-rtl-function.c (function_reader::create_function): Use
22463 build_decl instread of build_decl_stat.
22465 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
22467 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
22468 * config/i386/avx512dqintrin.h: Ditto.
22469 * config/i386/avx512fintrin.h: Ditto.
22470 * config/i386/i386-builtin-types.def: Add new types.
22471 * config/i386/i386.c: Handle new types.
22472 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
22473 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
22474 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
22475 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
22476 (__builtin_ia32_kshiftridi): New.
22477 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
22479 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
22483 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
22484 define to rs6000_init_stack_protect_guard.
22485 (rs6000_init_stack_protect_guard): New function.
22487 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
22488 Yunqiang Su <yunqiang.su@imgtec.com>
22490 * config.gcc (supported_defaults): Add madd4.
22491 (with_madd4): Add validation.
22492 (all_defaults): Add madd4.
22493 * config/mips/mips.opt (mmadd4): New option.
22494 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
22496 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
22498 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
22499 (ISA_HAS_FUSED_MADD4): Likewise.
22500 * doc/invoke.texi (-mmadd4): Document the new option.
22501 * doc/install.texi (--with-madd4): Document the new option.
22503 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22505 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
22506 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
22507 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
22508 (aarch64_init_pauth_hint_builtins): New.
22509 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
22510 (aarch64_expand_builtin): Expand new builtins.
22512 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22514 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
22515 * combine-stack-adj.c (no_unhandled_cfa): Handle
22516 REG_CFA_TOGGLE_RA_MANGLE.
22517 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
22518 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
22519 info for return address signing.
22520 (aarch64_expand_epilogue): Likewise.
22522 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22524 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
22525 * config/aarch64/aarch64-protos.h
22526 (aarch64_return_address_signing_enabled): New declaration.
22527 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
22529 (aarch64_expand_prologue): Sign return address before it's pushed onto
22531 (aarch64_expand_epilogue): Authenticate return address fetched from
22533 (aarch64_override_options): Sanity check for ILP32 and ISA level.
22534 (aarch64_attributes): New function attributes for "sign-return-address".
22535 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
22536 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
22537 ("*do_return"): Generate combined instructions according to key index.
22538 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
22539 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
22541 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
22542 * config/aarch64/aarch64.opt (msign-return-address=): New.
22543 * doc/extend.texi (AArch64 Function Attributes): Documents
22544 "sign-return-address=".
22545 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
22547 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
22549 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
22550 overall option summary.
22552 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22554 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
22555 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
22556 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
22557 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
22559 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
22561 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
22562 -mpower9-minmax by default for -mcpu=power9.
22563 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
22564 128-bit floating point.
22566 2017-01-20 Alan Modra <amodra@gmail.com>
22568 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
22569 optimizing for size.
22571 2017-01-20 Alan Modra <amodra@gmail.com>
22574 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
22575 for strcmp and strncmp from corresponding builtin decl.
22577 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
22579 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
22580 instead of i386/rtems-64.h.
22581 * config/i386/rtems-64.h: Remove.
22583 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
22587 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
22589 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
22591 2017-01-19 Tamar Christina <tamar.christina@arm.com>
22593 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
22594 Change int to HOST_WIDE_INT.
22595 * config/aarch64/aarch64-protos.h
22596 (aarch64_simd_gen_const_vector_dup): Likewise.
22597 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
22599 2017-01-19 David Malcolm <dmalcolm@redhat.com>
22601 * langhooks-def.h (lhd_type_for_size): New decl.
22602 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
22603 * langhooks.c (lhd_type_for_size): New function, taken from
22606 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
22608 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
22609 define_bypass for CR latency.
22610 (power9-cracked-alu): Update bypass latency and remove power9-branch.
22611 (power9-alu2): Add define_bypass for CR latency.
22613 (power9-mul): Update insn latency.
22614 (power9-mul-compare): Update insn latency, bypass latency and remove
22617 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22619 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
22621 * config/aarch64/aarch64.md
22622 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
22623 aarch64_nopcrelative_literal_loads.
22624 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
22626 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
22628 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
22629 TARGET_LOONGSON_3A.
22630 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
22632 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
22635 * config.gcc (supported_defaults): Add lxc1-sxc1.
22636 (with_lxc1_sxc1): Add validation.
22637 (all_defaults): Add lxc1-sxc1.
22638 * config/mips/mips.opt (mlxc1-sxc1): New option.
22639 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
22641 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
22642 __mips_no_lxc1_sxc1.
22643 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
22644 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
22645 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
22647 2017-01-19 Richard Biener <rguenther@suse.de>
22649 PR tree-optimization/72488
22650 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
22651 sure to restore SSA info.
22652 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
22654 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
22656 PR rtl-optimization/79121
22657 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
22658 of the inner type when shifting an extended value.
22660 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
22663 * symtab.c (symtab_node::equal_address_to): Fix comparing of
22664 interposable aliases.
22666 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
22669 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
22670 Use the evmergelohi instruction.
22671 (mov_si<mode>_e500_subreg4_2_le): Likewise.
22672 (mov_sitf_e500_subreg8_2_be): Likewise.
22673 (mov_sitf_e500_subreg12_2_le): Likewise.
22674 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
22675 (mov_si<mode>_e500_subreg4_2_be): Likewise.
22676 (mov_sitf_e500_subreg8_2_le): Likewise.
22677 (mov_sitf_e500_subreg12_2_be): Likewise.
22679 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22681 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
22682 attribute from vecsimple to vecperm.
22683 (altivec_vbpermq2): Likewise.
22685 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22688 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
22690 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22691 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
22692 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
22693 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
22694 case where N arg is SIZE_MAX.
22695 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
22696 (cmpstrsi): Add pattern.
22698 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22700 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22701 __builtin_vec_revb builtins.
22702 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
22703 built-in functions to support generation of the ISA 3.0 XXBR<x>
22704 vector byte reverse instructions.
22705 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
22706 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
22707 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
22708 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
22709 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
22710 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
22711 (P9V_BUILTIN_VEC_REVB): Likewise.
22712 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
22713 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
22714 (p9_xxbrq_v16qi): Likewise.
22715 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
22716 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
22717 (p9_xxbrh_v8hi): Likewise.
22718 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
22719 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
22720 vec_revb built-in functions.
22722 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
22724 PR rtl-optimization/78952
22725 * config/i386/i386.md (any_extract): New code iterator.
22726 (*insvqi_2): Use any_extract for source operand.
22727 (*insvqi_3): Use any_shiftrt for source operand.
22729 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
22731 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
22733 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
22735 2017-01-18 Matthias Klose <doko@ubuntu.com>
22737 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
22739 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22741 * config/rs6000/altivec.h (vec_bperm): Change #define.
22742 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
22743 (altivec_vbpermq2): New define_insn.
22744 (altivec_vbpermd): Likewise.
22745 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
22746 function interface.
22747 (VBPERMD): Likewise.
22748 (VBPERM): New polymorphic function interface.
22749 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
22750 Add entries for P9V_BUILTIN_VEC_VBPERM.
22751 * doc/extend.texi: Add interfaces for vec_bperm.
22753 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22755 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
22756 first letter of error messages.
22757 (s390_resolve_overloaded_builtin): Likewise.
22758 * config/s390/s390.c (s390_expand_builtin): Likewise.
22759 (s390_invalid_arg_for_unprototyped_fn): Likewise.
22760 (s390_valid_target_attribute_inner_p): Likewise.
22761 * config/s390/s390.md ("tabort"): Likewise.
22763 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
22765 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
22766 (ISA_AVOID_DIV_HILO): New macro.
22767 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
22768 (ISA_HAS_DDIV): Likewise.
22770 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22772 * doc/invoke.texi (fabi-version): Correct number of occurrences.
22774 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22776 * doc/invoke.texi (fabi-version): Spelling fix.
22778 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22781 * doc/invoke.texi (fabi-version): Mention mangling fix for
22784 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22787 * doc/invoke.texi (fabi-version): Document discriminator mangling.
22789 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
22792 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
22793 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
22795 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
22797 (stack_protect_test): Ditto.
22798 * config/rs6000/rs6000.opt (mstack-protector-guard=,
22799 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
22801 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
22802 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
22803 -mstack-protector-guard-offset=.
22804 (RS/6000 and PowerPC Options): Ditto.
22806 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
22808 * config/i386/i386.h (MASK_CLASS_P): New define.
22809 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
22810 there are no registers from different register sets also when
22811 mask registers are used. Update function comment.
22812 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
22813 to (*k/*r) and (*k/*km) alternatives.
22815 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
22817 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
22818 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
22819 (EH_RETURN_HANDLER_RTX): New define.
22820 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
22821 Force frame pointer in EH return functions.
22822 (aarch64_expand_epilogue): Add barrier for eh_return.
22823 (aarch64_final_eh_return_addr): Remove.
22824 (aarch64_eh_return_handler_rtx): New function.
22825 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
22827 (aarch64_eh_return_handler_rtx): New prototype.
22829 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22831 * config/rs6000/altivec.h (vec_rlmi): New #define.
22832 (vec_vrlnm): Likewise.
22833 (vec_rlnm): Likewise.
22834 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
22835 (UNSPEC_VRLNM): Likewise.
22836 (VIlong): New mode iterator.
22837 (altivec_vrl<VI_char>mi): New define_insn.
22838 (altivec_vrl<VI_char>nm): Likewise.
22839 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
22841 (VRLDNM): Likewise.
22842 (RLNM): New polymorphic function entry.
22843 (VRLWMI): New monomorphic function entry.
22844 (VRLDMI): Likewise.
22845 (RLMI): New polymorphic function entry.
22846 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
22847 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
22848 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
22851 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22854 * dwarf2out.c (field_byte_offset): Restore the
22855 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
22856 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
22857 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
22860 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
22863 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
22865 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22868 * dwarf2out.c (add_data_member_location_attribute): For constant
22869 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
22870 instead of DW_AT_data_member_location, DW_AT_bit_offset and
22871 DW_AT_byte_size attributes.
22873 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
22875 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
22876 after forcing to constant memory when the code model is medium.
22878 2017-01-17 Julia Koval <julia.koval@intel.com>
22881 * config/i386/avx512fintrin.h
22882 (_mm512_i32gather_ps): Change __addr type to void const*.
22883 (_mm512_mask_i32gather_ps): Ditto.
22884 (_mm512_i32gather_pd): Ditto.
22885 (_mm512_mask_i32gather_pd): Ditto.
22886 (_mm512_i64gather_ps): Ditto.
22887 (_mm512_mask_i64gather_ps): Ditto.
22888 (_mm512_i64gather_pd): Ditto.
22889 (_mm512_mask_i64gather_pd): Ditto.
22890 (_mm512_i32gather_epi32): Ditto.
22891 (_mm512_mask_i32gather_epi32): Ditto.
22892 (_mm512_i32gather_epi64): Ditto.
22893 (_mm512_mask_i32gather_epi64): Ditto.
22894 (_mm512_i64gather_epi32): Ditto.
22895 (_mm512_mask_i64gather_epi32): Ditto.
22896 (_mm512_i64gather_epi64): Ditto.
22897 (_mm512_mask_i64gather_epi64): Ditto.
22898 (_mm512_i32scatter_ps): Change __addr type to void*.
22899 (_mm512_mask_i32scatter_ps): Ditto.
22900 (_mm512_i32scatter_pd): Ditto.
22901 (_mm512_mask_i32scatter_pd): Ditto.
22902 (_mm512_i64scatter_ps): Ditto.
22903 (_mm512_mask_i64scatter_ps): Ditto.
22904 (_mm512_i64scatter_pd): Ditto.
22905 (_mm512_mask_i64scatter_pd): Ditto.
22906 (_mm512_i32scatter_epi32): Ditto.
22907 (_mm512_mask_i32scatter_epi32): Ditto.
22908 (_mm512_i32scatter_epi64): Ditto.
22909 (_mm512_mask_i32scatter_epi64): Ditto.
22910 (_mm512_i64scatter_epi32): Ditto.
22911 (_mm512_mask_i64scatter_epi32): Ditto.
22912 (_mm512_i64scatter_epi64): Ditto.
22913 (_mm512_mask_i64scatter_epi64): Ditto.
22914 * config/i386/avx512pfintrin.h
22915 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
22916 (_mm512_mask_prefetch_i32gather_ps): Ditto.
22917 (_mm512_mask_prefetch_i64gather_pd): Ditto.
22918 (_mm512_mask_prefetch_i64gather_ps): Ditto.
22919 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
22920 (_mm512_prefetch_i32scatter_ps): Ditto.
22921 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
22922 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
22923 (_mm512_prefetch_i64scatter_pd): Ditto.
22924 (_mm512_prefetch_i64scatter_ps): Ditto.
22925 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
22926 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
22927 * config/i386/avx512vlintrin.h
22928 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
22929 (_mm_mmask_i32gather_ps): Ditto.
22930 (_mm256_mmask_i32gather_pd): Ditto.
22931 (_mm_mmask_i32gather_pd): Ditto.
22932 (_mm256_mmask_i64gather_ps): Ditto.
22933 (_mm_mmask_i64gather_ps): Ditto.
22934 (_mm256_mmask_i64gather_pd): Ditto.
22935 (_mm_mmask_i64gather_pd): Ditto.
22936 (_mm256_mmask_i32gather_epi32): Ditto.
22937 (_mm_mmask_i32gather_epi32): Ditto.
22938 (_mm256_mmask_i32gather_epi64): Ditto.
22939 (_mm_mmask_i32gather_epi64): Ditto.
22940 (_mm256_mmask_i64gather_epi32): Ditto.
22941 (_mm_mmask_i64gather_epi32): Ditto.
22942 (_mm256_mmask_i64gather_epi64): Ditto.
22943 (_mm_mmask_i64gather_epi64): Ditto.
22944 (_mm256_i32scatter_ps): Change __addr type to void*.
22945 (_mm256_mask_i32scatter_ps): Ditto.
22946 (_mm_i32scatter_ps): Ditto.
22947 (_mm_mask_i32scatter_ps): Ditto.
22948 (_mm256_i32scatter_pd): Ditto.
22949 (_mm256_mask_i32scatter_pd): Ditto.
22950 (_mm_i32scatter_pd): Ditto.
22951 (_mm_mask_i32scatter_pd): Ditto.
22952 (_mm256_i64scatter_ps): Ditto.
22953 (_mm256_mask_i64scatter_ps): Ditto.
22954 (_mm_i64scatter_ps): Ditto.
22955 (_mm_mask_i64scatter_ps): Ditto.
22956 (_mm256_i64scatter_pd): Ditto.
22957 (_mm256_mask_i64scatter_pd): Ditto.
22958 (_mm_i64scatter_pd): Ditto.
22959 (_mm_mask_i64scatter_pd): Ditto.
22960 (_mm256_i32scatter_epi32): Ditto.
22961 (_mm256_mask_i32scatter_epi32): Ditto.
22962 (_mm_i32scatter_epi32): Ditto.
22963 (_mm_mask_i32scatter_epi32): Ditto.
22964 (_mm256_i32scatter_epi64): Ditto.
22965 (_mm256_mask_i32scatter_epi64): Ditto.
22966 (_mm_i32scatter_epi64): Ditto.
22967 (_mm_mask_i32scatter_epi64): Ditto.
22968 (_mm256_i64scatter_epi32): Ditto.
22969 (_mm256_mask_i64scatter_epi32): Ditto.
22970 (_mm_i64scatter_epi32): Ditto.
22971 (_mm_mask_i64scatter_epi32): Ditto.
22972 (_mm256_i64scatter_epi64): Ditto.
22973 (_mm256_mask_i64scatter_epi64): Ditto.
22974 (_mm_i64scatter_epi64): Ditto.
22975 (_mm_mask_i64scatter_epi64): Ditto.
22976 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
22977 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
22978 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
22979 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
22980 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
22981 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
22982 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
22983 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
22984 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
22985 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
22986 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
22987 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
22988 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
22989 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
22990 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
22991 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
22992 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
22993 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
22994 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
22995 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
22996 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
22997 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
22998 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
22999 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
23000 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
23001 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
23002 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
23003 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
23004 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
23005 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
23006 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
23007 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
23008 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
23009 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
23010 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
23011 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
23012 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
23013 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
23014 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
23015 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
23016 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
23017 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
23018 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
23019 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
23020 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
23021 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
23022 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
23023 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
23024 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
23025 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
23026 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
23027 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
23028 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
23029 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
23030 definitions accordingly.
23032 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
23033 Kuan-Lin Chen <kuanlinchentw@gmail.com>
23036 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
23039 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
23042 * ira-conflicts.c (ira_build_conflicts): Update total conflict
23043 hard regs for inner regno.
23045 2017-01-17 Martin Liska <mliska@suse.cz>
23048 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
23049 assumption and add comment.
23051 2017-01-17 Nathan Sidwell <nathan@acm.org>
23053 * ipa-visibility.c (localize_node): New function, broken out of ...
23054 (function_and_variable_visibility): ... here. Call it.
23056 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
23058 PR middle-end/77445
23059 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
23060 correctly set frequency of oudgoing edge.
23061 (duplicate_thread_path): Fix profile updating.
23063 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23066 * configure.ac: Add GCC_BASE_VER.
23067 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
23068 version from BASE-VER file.
23069 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
23070 (gcc.o): Depend on $(BASEVER).
23071 * common.opt (dumpfullversion): New option.
23072 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
23073 * doc/invoke.texi: Document -dumpfullversion.
23074 * doc/install.texi: Document --with-gcc-major-version-only.
23075 * configure: Regenerated.
23077 2017-01-17 Richard Biener <rguenther@suse.de>
23079 PR tree-optimization/71433
23080 * tree-vrp.c (register_new_assert_for): Merge same asserts
23081 on all incoming edges.
23082 (process_assert_insertions_for): Handle insertions at the
23085 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
23087 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
23088 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
23090 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
23093 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
23096 2017-01-17 Alan Modra <amodra@gmail.com>
23099 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
23100 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
23101 symbolic stack limit when pic.
23103 2017-01-16 Martin Sebor <msebor@redhat.com>
23105 PR tree-optimization/78608
23106 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
23108 2017-01-16 Jeff Law <law@redhat.com>
23111 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
23112 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
23113 for several include directories that may be relative to sysroot.
23114 * config/i386/x-mingw32 (gplus_includedir): Define.
23115 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
23116 (native_system_includedir): Likewise.
23117 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
23118 override if TARGET_SYSTEM_ROOT is defined.
23119 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
23121 PR tree-optimization/79090
23122 PR tree-optimization/33562
23123 PR tree-optimization/61912
23124 PR tree-optimization/77485
23125 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
23126 and computed trims into the dump file.
23128 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
23130 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
23132 2017-01-16 Jakub Jelinek <jakub@redhat.com>
23135 * gimplify.c (gimplify_init_constructor): If want_value and
23136 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
23140 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
23141 sequence. Formatting fixes.
23142 (doloop_optimize): Formatting fixes.
23145 * gcc.c (debug_level_greater_than_spec_func): New function.
23146 (static_spec_functions): Add debug-level-gt spec function.
23147 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
23149 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
23150 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
23151 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
23152 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
23153 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
23154 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
23156 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
23158 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
23159 QImode fixups to general and mask registers only.
23161 2017-01-16 Carl Love <cel@us.ibm.com>
23163 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
23164 for built-in functions
23165 vector signed char vec_nabs (vector signed char)
23166 vector signed short vec_nabs (vector signed short)
23167 vector signed int vec_nabs (vector signed int)
23168 vector signed long long vec_nabs (vector signed long long)
23169 vector float vec_nabs (vector float)
23170 vector double vec_nabs (vector double)
23171 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
23173 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
23174 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
23175 * doc/extend.texi: Update the documentation file for the new built-in
23178 2017-01-16 Martin Sebor <msebor@redhat.com>
23180 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
23183 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23185 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
23186 UNSPEC_VSX__XXSPLTD to require special splat handling.
23188 2017-01-16 David Malcolm <dmalcolm@redhat.com>
23191 * system.h: Poison strndup.
23193 2017-01-16 Alan Modra <amodra@gmail.com>
23196 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
23199 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
23201 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
23203 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
23205 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
23206 call recog here. Assert that INSN_CODE (insn) is non-negative.
23208 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
23211 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
23213 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
23214 in the currently scheduled RTL fragment.
23216 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
23218 PR rtl-optimization/78751
23219 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
23222 2017-01-14 Jeff Law <law@redhat.com>
23224 PR tree-optimization/79090
23225 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
23226 variable length stores.
23227 (compute_trims): Delete dead assignment to *trim_tail.
23228 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
23231 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
23233 PR rtl-optimization/78626
23234 PR rtl-optimization/78727
23235 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
23236 of a block, and split such blocks after everything else is finished.
23238 2017-01-14 Alan Modra <amodra@gmail.com>
23241 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
23242 target legitimate_combined_insn.
23243 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
23244 (rs6000_legitimate_combined_insn): New function.
23245 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
23247 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
23248 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
23249 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
23251 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
23253 * doc/frontends.texi (G++ and GCC): Remove references to Java.
23255 2017-01-13 Jeff Law <law@redhat.com>
23257 PR tree-optimization/33562
23258 PR tree-optimization/61912
23259 PR tree-optimization/77485
23260 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
23262 (delete_dead_assignment): Likewise.
23263 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
23264 statement to delete_dead_call and delete_dead_assignment.
23266 2017-01-13 David Malcolm <dmalcolm@redhat.com>
23269 * substring-locations.c (format_warning_va): Strengthen case 1 so
23270 that both endpoints of the substring must be within the format
23271 range for just the substring to be printed.
23273 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
23275 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
23276 * config/i386/i386.c (ix86_target_string): Add missing options
23277 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
23278 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
23279 flags_other and ix86_target_other to flags2_other. Display unknown
23281 (ix86_valid_target_attribute_inner_p): Add missing options and
23282 reorder options by implied ISAs, as in ix86_target_string.
23284 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
23286 * hash-table.h (hash_table::too_empty_p): New function.
23287 (hash_table::expand): Use it.
23288 (hash_table::traverse): Likewise.
23289 (hash_table::empty_slot): Use sizeof (value_type) instead of
23290 sizeof (PTR) to convert bytes to elements. Shrink the table
23291 if the current size is excessive for the current number of
23294 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
23296 * ira-costs.c (record_reg_classes): Break from the inner loop
23297 early once alt_fail is known to be true. Update outer loop
23298 handling accordingly.
23300 2017-01-13 Jeff Law <law@redhat.com>
23302 * tree-ssa-dse.c (decrement_count): New function.
23303 (increment_start_addr, maybe_trim_memstar_call): Likewise.
23304 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
23305 when we know the partially dead statement is a mem* function.
23307 PR tree-optimization/61912
23308 PR tree-optimization/77485
23309 * tree-ssa-dse.c: Include expr.h.
23310 (maybe_trim_constructor_store): New function.
23311 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
23313 PR tree-optimization/33562
23314 PR tree-optimization/61912
23315 PR tree-optimization/77485
23316 * doc/invoke.texi: Document new dse-max-object-size param.
23317 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
23318 * tree-ssa-dse.c: Include params.h.
23319 (dse_store_status): New enum.
23320 (initialize_ao_ref_for_dse): New, partially extracted from
23322 (valid_ao_ref_for_dse, normalize_ref): New.
23323 (setup_live_bytes_from_ref, compute_trims): Likewise.
23324 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
23325 (maybe_trim_partially_dead_store): Likewise.
23326 (maybe_trim_complex_store): Likewise.
23327 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
23328 Track what bytes live from the original store. Return tri-state
23329 for dead, partially dead or live.
23330 (dse_dom_walker): Add constructor, destructor and new private members.
23331 (delete_dead_call, delete_dead_assignment): New extracted from
23333 (dse_optimize_stmt): Make a member of dse_dom_walker.
23334 Use initialize_ao_ref_for_dse.
23336 PR tree-optimization/33562
23337 PR tree-optimization/61912
23338 PR tree-optimization/77485
23339 * sbitmap.h (bitmap_count_bits): Prototype.
23340 (bitmap_clear_range, bitmap_set_range): Likewise.
23341 * sbitmap.c (bitmap_clear_range): New function.
23342 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
23344 2017-01-13 Martin Liska <mliska@suse.cz>
23347 * function.c (set_cfun): Add new argument force.
23348 * function.h (set_cfun): Likewise.
23349 * ipa-inline-transform.c (inline_call): Use the function when
23350 strict alising from is dropped for function we inline to.
23352 2017-01-13 Richard Biener <rguenther@suse.de>
23354 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
23355 for dumping GIMPLE INTEGER_CSTs.
23357 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23359 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
23360 to 201112L since C++17.
23362 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
23365 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
23366 if -fsanitize=kernel-address is present.
23368 2017-01-13 Richard Biener <rguenther@suse.de>
23370 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
23371 as _Literal ( type ) number in case usual suffixes do not
23372 preserve all information.
23374 2017-01-13 Richard Biener <rguenther@suse.de>
23376 PR tree-optimization/77283
23377 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
23378 and ssa-iterators.h.
23379 (is_feasible_trace): Implement a cost model based on joiner
23382 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
23385 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
23386 char or short to __float128/_Float128 directly.
23388 2017-01-12 Martin Sebor <msebor@redhat.com>
23390 to -Wformat-overflow.
23391 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
23392 (min_bytes_remaining): Same.
23393 (get_string_length): Same.
23394 (format_string): Same.
23395 (format_directive): Same.
23397 (pass_sprintf_length::handle_gimple_call): Same.
23399 2017-01-12 Jakub Jelinek <jakub@redhat.com>
23401 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
23402 info.nowrite calls with no lhs that can't throw. Return bool
23403 whether gsi_remove has been called or not.
23404 (pass_sprintf_length::handle_gimple_call): Return bool whether
23405 try_substitute_return_value called gsi_remove. Formatting fix.
23406 (pass_sprintf_length::execute): Don't use gsi_remove if
23407 handle_gimple_call returned true.
23410 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
23411 be removed due to side-effects, don't remove following barrier nor
23412 turn the successor edge into fallthru edge.
23414 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23417 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
23418 element-reversing loads and stores as not swappable.
23420 2017-01-12 Nathan Sidwell <nathan@acm.org>
23421 Nicolai Stange <nicstange@gmail.com>
23423 * combine.c (try_combine): Don't ignore result of overlap checking
23424 loop. Combine overlap & asm check into single loop.
23426 2017-01-12 Richard Biener <rguenther@suse.de>
23428 * tree-pretty-print.c (dump_generic_node): Provide -gimple
23429 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
23431 2017-01-12 Richard Biener <rguenther@suse.de>
23433 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
23434 and TS_TARGET_OPTION directly derive from TS_BASE.
23435 * tree-core.h (tree_optimization_option): Derive from tree_base.
23436 (tree_target_option): Likewise.
23438 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
23440 * config/i386/i386.c (memory_address_length): Increase len
23441 only when rip_relative_addr_p returns false.
23443 2017-01-11 Julia Koval <julia.koval@intel.com>
23445 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
23446 (OPTION_MASK_ISA_SGX_SET): New.
23447 (ix86_handle_option): Handle OPT_msgx.
23448 * config.gcc: Added sgxintrin.h.
23449 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
23450 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
23451 * config/i386/i386.c (ix86_target_string): Add -msgx.
23453 (ix86_option_override_internal): Handle new options.
23454 (ix86_valid_target_attribute_inner_p): Add sgx.
23455 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
23456 * config/i386/i386.opt: Add msgx.
23457 * config/i386/sgxintrin.h: New file.
23458 * config/i386/x86intrin.h: Add sgxintrin.h.
23460 2017-01-11 Jakub Jelinek <jakub@redhat.com>
23463 * fold-const.c (maybe_nonzero_address): Return 1 for function
23465 (tree_single_nonzero_warnv_p): Don't handle function local objects
23469 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
23472 2017-01-11 David Malcolm <dmalcolm@redhat.com>
23475 * opts.c: Include "spellcheck.h"
23476 (struct string_fragment): New struct.
23477 (struct edit_distance_traits<const string_fragment &>): New
23479 (get_closest_sanitizer_option): New function.
23480 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
23482 2017-01-11 Jakub Jelinek <jakub@redhat.com>
23484 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
23486 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
23487 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
23488 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
23489 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
23490 for initial die_offset if dwarf_split_debug_info.
23491 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
23492 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
23494 (output_skeleton_debug_sections): Formatting fix. Use
23495 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
23496 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
23498 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
23500 * config/arm/cortex-a53.md: Add bypasses for
23501 cortex_a53_r2f_cvt.
23502 (cortex_a53_r2f): Only use for transfers.
23503 (cortex_a53_f2r): Likewise.
23504 (cortex_a53_r2f_cvt): Add reservation for conversions.
23505 (cortex_a53_f2r_cvt): Likewise.
23507 2017-01-11 Tamar Christina <tamar.christina@arm.com>
23509 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
23510 to all inlined functions, change static to extern.
23512 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
23515 * config/arm/arm.c (legitimize_pic_address): Handle reference to
23517 (arm_assemble_integer): Likewise.
23519 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
23521 * config.gcc: Use new awk script to check CPU, FPU and architecture
23522 parameters for --with-... options.
23523 * config/arm/parsecpu.awk: New file
23524 * config/arm/arm-cpus.in: New file.
23525 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
23527 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
23529 * config/arm/t-arm: Update dependency rules.
23530 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
23531 of processing .def files.
23532 * config/arm/genopt.sh: Deleted.
23533 * config/arm/gentune.sh: Deleted.
23534 * config/arm/arm-cores.def: Deleted.
23535 * config/arm/arm-arches.def: Deleted.
23536 * config/arm/arm-fpus.def: Deleted.
23537 * config/arm/arm-tune.md: Regenerated.
23538 * config/arm/arm-tables.opt: Regenerated.
23539 * config/arm/arm-cpu.h: New generated file.
23540 * config/arm/arm-cpu-data.h: New generated file.
23541 * config/arm/arm-cpu-cdata.h: New generated file.
23543 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
23546 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
23548 (input_varpool_node): Unpack dynamically_initialized bit.
23550 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
23552 PR rtl-optimization/79032
23553 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
23554 the alignment of the adjusted memory reference against that of MODE,
23555 instead of the alignment of the original memory reference.
23557 2017-01-11 Martin Jambor <mjambor@suse.cz>
23559 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
23561 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
23562 decorated functions.
23564 2017-01-11 Richard Biener <rguenther@suse.de>
23566 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
23567 set range/nonnull info for PHI results. Do not set it on
23568 stmts marked for removal.
23570 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
23572 * expr.c (store_field): In the bitfield case, fetch the return value
23573 from the registers before applying a single big-endian adjustment.
23574 Always do a final load for a BLKmode value not larger than a word.
23576 2017-01-10 David Malcolm <dmalcolm@redhat.com>
23579 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
23580 that we correctly handle column numbers greater than
23581 LINE_MAP_MAX_COLUMN_NUMBER.
23583 2017-01-10 Martin Sebor <msebor@redhat.com>
23585 PR middle-end/78245
23586 * gimple-ssa-sprintf.c (get_destination_size): Call
23587 {init,fini}object_sizes.
23588 * tree-object-size.c (addr_object_size): Adjust.
23589 (pass_through_call): Adjust.
23590 (pass_object_sizes::execute): Adjust.
23591 * tree-object-size.h (fini_object_sizes): Declare.
23593 2017-01-10 Martin Sebor <msebor@redhat.com>
23595 PR tree-optimization/78775
23596 * builtins.c (get_size_range): Move...
23597 * calls.c: ...to here.
23598 (alloc_max_size): Accept zero argument.
23599 (operand_signed_p): Remove.
23600 (maybe_warn_alloc_args_overflow): Call get_size_range.
23601 * calls.h (get_size_range): Declare.
23603 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
23605 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
23606 from TI's devices.csv file as of September 2016.
23607 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
23609 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
23611 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
23612 * doc/invoke.texi: Likewise.
23613 * doc/md.texi: Likewise.
23614 * doc/objc.texi: Likewise.
23616 2017-01-10 Joshua Conner <joshconner@google.com>
23618 * config/arm/fuchsia-elf.h: New file.
23619 * config/fuchsia.h: New file.
23620 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
23621 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
23623 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
23625 2016-01-10 Richard Biener <rguenther@suse.de>
23627 PR tree-optimization/79034
23628 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
23629 Propagate out degenerate PHIs in the joiner.
23631 2017-01-10 Martin Liska <mliska@suse.cz>
23633 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
23634 (sort_congruence_classes_by_decl_uid): Likewise.
23635 (sort_congruence_class_groups_by_decl_uid): Likewise.
23636 (sem_item_optimizer::merge_classes): Sort class, groups in these
23637 classes and members in the groups by DECL_UID of declarations.
23638 This would make merge operations stable.
23640 2017-01-10 Martin Liska <mliska@suse.cz>
23642 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
23643 usage of m_classes_vec.
23644 (sem_item_optimizer::~sem_item_optimizer): Likewise.
23645 (sem_item_optimizer::get_group_by_hash): Likewise.
23646 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
23647 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
23648 (sem_item_optimizer::verify_classes): Likewise.
23649 (sem_item_optimizer::process_cong_reduction): Likewise.
23650 (sem_item_optimizer::dump_cong_classes): Likewise.
23651 (sem_item_optimizer::merge_classes): Likewise.
23652 * ipa-icf.h (congruence_class_hash): Rename from
23653 congruence_class_group_hash. Remove declaration of m_classes_vec.
23655 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
23657 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
23658 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
23659 * config.gcc: Add avx512vpopcntdqintrin.h.
23660 * config/i386/avx512vpopcntdqintrin.h: New.
23661 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
23662 * config/i386/i386-builtin-types.def: Add new types.
23663 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
23664 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
23665 __builtin_ia32_vpopcountq_v8di_mask): New.
23666 * config/i386/i386-c.c (ix86_target_macros_internal): Define
23667 __AVX512VPOPCNTDQ__.
23668 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
23669 (PTA_AVX512VPOPCNTDQ): Define.
23670 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
23671 TARGET_AVX512VPOPCNTDQ_P): Define.
23672 * config/i386/i386.opt: Add mavx512vpopcntdq.
23673 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
23674 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
23676 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23678 PR middle-end/77484
23679 * predict.def (PRED_CALL): Set to 67.
23681 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
23683 * expr.c (store_field): In the bitfield case, if the value comes from
23684 a function call and is of an aggregate type returned in registers, do
23685 not modify the field mode; extract the value in all cases if the mode
23686 is BLKmode and the size is not larger than a word.
23688 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
23691 * config/i386/cpuid.h: Fix undefined behavior.
23693 2017-01-04 Jeff Law <law@redhat.com>
23695 PR tree-optimization/79007
23696 PR tree-optimization/67955
23697 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
23698 conservative for pt.null when flag_non_call_exceptions is on.
23700 2017-01-09 Jakub Jelinek <jakub@redhat.com>
23702 PR translation/79019
23703 PR translation/79020
23704 * params.def (PARAM_INLINE_MIN_SPEEDUP,
23705 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
23706 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
23708 * config/avr/avr.opt (maccumulate-args): Likewise.
23709 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
23710 * common.opt (freport-bug): Likewise.
23711 * cif-code.def (CIF_FINAL_ERROR): Likewise.
23712 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
23713 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
23714 translatable string.
23715 * config/i386/i386.c (function_value_32): Likewise.
23716 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
23717 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
23719 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
23720 * common/config/msp430/msp430-common.c (msp430_handle_option):
23722 * symtab.c (symtab_node::verify_base): Likewise.
23723 * opts.c (set_debug_level): Likewise.
23724 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
23725 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
23726 missing whitespace to translatable strings.
23727 * config/avr/avr.md (bswapsi2): Fix typo in comment.
23728 * config/sh/superh.h: Likewise.
23729 * config/i386/xopintrin.h: Likewise.
23730 * config/i386/znver1.md: Likewise.
23731 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
23732 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
23733 * double-int.h (struct double_int): Likewise.
23734 * double-int.c (div_and_round_double): Likewise.
23735 * wide-int.cc: Likewise.
23736 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
23737 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
23738 * cfgcleanup.c (crossjumps_occured): Renamed to ...
23739 (crossjumps_occurred): ... this.
23740 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
23743 PR tree-optimization/78899
23744 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
23745 returning bool return struct loop *, NULL for failure and the new
23747 (versionable_outer_loop_p): Don't version outer loop if it has
23748 dont_vectorized bit set.
23749 (tree_if_conversion): When versioning outer loop, ensure
23750 tree_if_conversion is performed also on the inner loop of the
23751 non-vectorizable outer loop copy.
23752 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
23753 LOOP_VECTORIZED in inner loop of the scalar outer loop and
23754 prevent vectorization of it.
23755 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
23756 the outer loop vectorization of the non-scalar version is attempted
23757 before vectorization of the inner loop in scalar version. If
23758 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
23759 vectorization of its inner loop.
23760 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
23761 has 2 inner loops, rename also on edges from bb whose single pred
23762 is outer_loop->header. Fix typo in function comment.
23764 2017-01-09 Martin Sebor <msebor@redhat.com>
23767 * asan.c (asan_emit_stack_protection): Increase local buffer size
23768 to avoid snprintf truncation warning.
23770 2017-01-09 Andrew Pinski <apinski@cavium.com>
23772 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
23773 to reference thunderx2t99 for the tuning structure
23774 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
23776 (thunderx2t99_extra_costs): This.
23777 * config/aarch64/aarch64-tune.md: Regenerate.
23778 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
23779 (vulcan_addrcost_table): This.
23780 (vulcan_regmove_cost): Rename to ...
23781 (thunderx2t99_regmove_cost): This.
23782 (vulcan_vector_cost): Rename to ...
23783 (thunderx2t99_vector_cost): this.
23784 (vulcan_branch_cost): Rename to ...
23785 (thunderx2t99_branch_cost): This.
23786 (vulcan_tunings): Rename to ...
23787 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
23788 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
23790 2017-01-09 Martin Jambor <mjambor@suse.cz>
23794 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
23795 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
23796 (propagate_vr_accross_jump_function): Use the above function for all
23797 value range computations for pass-through jump functions and type
23798 converasion from explicit value range values.
23799 (ipcp_propagate_stage): Do not attempt to deduce types of formal
23800 parameters from TYPE_ARG_TYPES.
23801 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
23802 (ipa_write_node_info): Stream type of the actual argument.
23803 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
23805 2017-01-09 Martin Liska <mliska@suse.cz>
23808 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
23809 (lookup_compiler): Do not show error message with have_E.
23811 2017-01-09 Jakub Jelinek <jakub@redhat.com>
23813 PR tree-optimization/78938
23814 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
23815 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
23816 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
23817 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
23820 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23822 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
23825 2017-01-09 Richard Biener <rguenther@suse.de>
23827 PR tree-optimization/78997
23828 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
23829 name condition properly.
23831 2017-01-09 Richard Biener <rguenther@suse.de>
23834 * dwarf2out.c (is_cxx): New overload with context.
23835 (is_naming_typedef_decl): Use it.
23837 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23839 * invoke.texi (Option Summary): Correct spacing in option lists
23840 and add line breaks to fix over-long lines.
23842 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23844 PR middle-end/17660
23846 * extend.texi (Common Variable Attributes): Add xref to GCC
23847 Internals manual to explain mode attribute keywords.
23849 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23852 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
23853 and Preprocessor Options.
23854 (Options for Linking): Document -pthread here....
23855 (RS/6000 and PowerPC Options): ...not here.
23856 (Solaris 2 Options): ...or here.
23857 * doc/cppopts.texi: Document -pthread.
23859 2017-01-08 Martin Sebor <msebor@redhat.com>
23861 PR middle-end/77708
23862 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
23863 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
23864 New member functions.
23865 (format_directive): Used them.
23867 (pass_sprintf_length::handle_gimple_call): Same.
23868 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
23869 to avoid truncation for any argument.
23870 (extract_affine_mul): Same.
23871 * tree.c (get_file_function_name): Same.
23873 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23875 PR middle-end/77484
23876 * predict.def (PRED_INDIR_CALL): Set to 86.
23878 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23880 PR preprocessor/54124
23881 * doc/cppopts.texi: Reformat -d subtable to list the full name
23882 of the options. Add cross-reference to the docs for the general
23883 compiler -d options.
23884 * doc/invoke.texi (Developer Options): Add cross-reference to the
23885 preprocessor-specific -d option documentation.
23887 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23889 PR preprocessor/13498
23890 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
23891 redudant material, and reflect new command-line options.
23892 (System Headers): Likewise.
23894 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23896 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
23897 -isystem, and -idirafter. Copy-edit.
23898 * doc/cppopts.texi: Copy-edit. Remove contradiction about
23899 default for -ftrack-macro-expansion. Delete obsolete and
23900 badly-formatted implementation details about -fdebug-cpp output.
23901 * doc/cppwarnopts.texi: Copy-edit.
23903 2017-01-07 David Malcolm <dmalcolm@redhat.com>
23906 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
23907 that the transition from a max line width >= 1<<10 to narrower
23908 lines works correctly.
23910 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
23912 * doc/options.texi (PerFunction): New.
23913 * opt-functions.awk (switch_flags): Map both Optimization and
23914 PerFunction to CL_OPTIMIZATION.
23915 * opth-gen.awk: Test for PerFunction flag along with
23917 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
23918 it only when the latter is present. Skip those that don't in
23919 the hash function generator.
23920 * common.opt (fvar-tracking): Mark as PerFunction instead of
23922 (fvar-tracking-assignments): Likewise.
23923 (fvar-tracking-assignments-toggle): Likewise.
23924 (fvar-tracking-uninit): Likewise.
23926 2017-01-07 Jakub Jelinek <jakub@redhat.com>
23928 PR translation/79018
23929 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
23932 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
23935 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
23936 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
23937 TARGET_LONG_JUMP_TABLE_OFFSETS.
23938 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
23939 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
23940 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
23941 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
23942 * config/m68k/m68k.md (tablejump expander): Likewise.
23943 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
23944 TARGET_LONG_JUMP_TABLE_OFFSETS.
23945 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
23946 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
23948 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
23949 David Holsgrove <david.holsgrove@xilinx.com>
23951 * common/config/microblaze/microblaze-common.c
23952 (TARGET_EXCEPT_UNWIND_INFO): Remove.
23953 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
23955 * config/microblaze/microblaze.c (microblaze_must_save_register)
23956 (microblaze_expand_epilogue, microblaze_return_addr): Handle
23958 (microblaze_eh_return): New function.
23959 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
23960 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
23961 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
23962 * config/microblaze/microblaze.md (eh_return): New pattern.
23964 2017-01-06 Jakub Jelinek <jakub@redhat.com>
23966 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
23967 GCC_DIAGNOSTIC_STRINGIFY): Define.
23969 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
23971 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23973 * config/arm/arm.md (<mcrr>): New.
23975 * config/arm/arm.c (arm_arch5te): New.
23976 (arm_option_override): Set arm_arch5te.
23977 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
23979 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
23980 (arm_mcrr_qualifiers): ... this. New.
23981 (MRRC_QUALIFIERS): Define to...
23982 (arm_mrrc_qualifiers): ... this. New.
23983 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
23985 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
23986 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
23987 (MRRCI, mrrc, MRRC): New.
23988 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
23989 VUNSPEC_MRRC2): New.
23991 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23993 * config/arm/arm.md (<mcr>): New.
23995 * config/arm/arm.c (arm_coproc_builtin_available): Add
23996 support for mcr, mrc, mcr2 and mrc2.
23997 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
23998 (arm_mcr_qualifiers): ... this. New.
23999 (MRC_QUALIFIERS): Define to ...
24000 (arm_mrc_qualifiers): ... this. New.
24001 (MCR_QUALIFIERS): Define to ...
24002 (arm_mcr_qualifiers): ... this. New.
24003 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
24005 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
24006 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
24007 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
24008 VUNSPEC_MRC2): New.
24010 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24012 * config/arm/arm.md (*ldc): New.
24016 * config/arm/arm.c (arm_coproc_builtin_available): Add
24017 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
24018 (arm_coproc_ldc_stc_legitimate_address): New.
24019 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
24020 'qualifier_const_pointer'.
24021 (LDC_QUALIFIERS): Define to...
24022 (arm_ldc_qualifiers): ... this. New.
24023 (STC_QUALIFIERS): Define to...
24024 (arm_stc_qualifiers): ... this. New.
24025 * config/arm/arm-protos.h
24026 (arm_coproc_ldc_stc_legitimate_address): New.
24027 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
24028 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
24029 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
24030 stc2, stcl, stc2l): New.
24031 * config/arm/constraints.md (Uz): New.
24032 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
24033 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
24034 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
24035 VUNSPEC_STC2L): New.
24037 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24039 * config/arm/arm.md (<cdp>): New.
24040 * config/arm/arm.c (neon_const_bounds): Rename this ...
24041 (arm_const_bounds): ... this.
24042 (arm_coproc_builtin_available): New.
24043 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
24044 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
24045 (CDP_QUALIFIERS): Define to...
24046 (arm_cdp_qualifiers): ... this. New.
24048 (arm_expand_builtin_args): Add case for 6 arguments.
24049 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
24050 (arm_const_bounds): ... this.
24051 (arm_coproc_builtin_available): New.
24052 * config/arm/arm_acle.h (__arm_cdp): New.
24054 * config/arm/arm_acle_builtins.def (cdp): New.
24056 * config/arm/iterators.md (CDPI,CDP,cdp): New.
24057 * config/arm/neon.md: Rename all 'neon_const_bounds' to
24058 'arm_const_bounds'.
24059 * config/arm/types.md (coproc): New.
24060 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
24061 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
24062 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
24063 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
24065 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24067 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
24068 (UBINOP_QUALIFIERS): New.
24070 (acle_builtin_data): New. Change comment.
24071 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
24072 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
24073 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
24074 arm_acle_builtins.def.
24075 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
24076 (arm_init_acle_builtins): New.
24077 (CRC32_BUILTIN): Remove.
24078 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
24079 crc32cb, crc32ch and crc32cw.
24080 (arm_init_crc32_builtins): Remove.
24081 (arm_init_builtins): Use arm_init_acle_builtins rather
24082 than arm_init_crc32_builtins.
24083 (arm_expand_acle_builtin): New.
24084 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
24085 * config/arm/arm_acle_builtins.def: New.
24087 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24089 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
24090 (arm_builtin_datum): ... this.
24091 (arm_init_neon_builtin): Rename to ...
24092 (arm_init_builtin): ... this. Add a new parameters PREFIX
24093 and USE_SIG_IN_NAME.
24094 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
24095 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
24096 'arm_builtin_datum'.
24097 (arm_init_vfp_builtins): Likewise.
24098 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
24099 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
24100 (arm_expand_neon_args): Rename to ...
24101 (arm_expand_builtin_args): ... this. Rename builtin_arg
24102 enum values and differentiate between ARG_BUILTIN_MEMORY
24103 and ARG_BUILTIN_NEON_MEMORY.
24104 (arm_expand_neon_builtin_1): Rename to ...
24105 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
24106 values, arm_expand_builtin_args and add bool parameter NEON.
24107 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
24108 (arm_expand_vfp_builtin): Likewise.
24109 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
24111 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24113 PR middle-end/77484
24114 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
24115 * predict.c (tree_estimate_probability_bb): Reverse direction of
24116 polymorphic call predictor.
24118 2017-01-06 David Malcolm <dmalcolm@redhat.com>
24120 * passes.c (execute_one_pass): Split out pass-skipping logic into...
24121 (determine_pass_name_match): ...this new function and...
24122 (should_skip_pass_p): ...this new function.
24124 2017-01-06 Nathan Sidwell <nathan@acm.org>
24126 * ipa-visibility.c (function_and_variable_visibility): Reformat
24127 comments and long lines. Remove extrneous if.
24128 * symtab.c (symtab_node::make_decl_local): Fix code format.
24129 (symtab_node::set_section_for_node): Fix comment typo.
24131 2017-01-06 Martin Liska <mliska@suse.cz>
24134 * lra-constraints.c: Rename invariant to lra_invariant.
24135 * predict.c (set_even_probabilities): Initialize e to NULL.
24137 2017-01-05 Martin Sebor <msebor@redhat.com>
24139 PR tree-optimization/78910
24140 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
24141 (format_integer): Correct off-by-one error in the handling
24142 of precision with negative numbers in signed conversions..
24144 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
24146 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
24148 2017-01-05 Jakub Jelinek <jakub@redhat.com>
24150 PR tree-optimization/71016
24151 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
24152 factor_out_conditional_conversion. Formatting fix.
24153 (factor_out_conditional_conversion): Add cond_stmt argument.
24154 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
24155 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
24158 2017-01-05 David Malcolm <dmalcolm@redhat.com>
24160 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
24161 read-rtl-function.o, and selftest-rtl.o.
24162 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
24163 (selftest::aarch64_test_loading_full_dump): New function.
24164 (selftest::aarch64_run_selftests): New function.
24165 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
24166 selftest::aarch64_run_selftests.
24167 * config/i386/i386.c
24168 (selftest::ix86_test_loading_dump_fragment_1): New function.
24169 (selftest::ix86_test_loading_call_insn): New function.
24170 (selftest::ix86_test_loading_full_dump): New function.
24171 (selftest::ix86_test_loading_unspec): New function.
24172 (selftest::ix86_run_selftests): Call the new functions.
24173 * emit-rtl.c (maybe_set_max_label_num): New function.
24174 * emit-rtl.h (maybe_set_max_label_num): New decl.
24175 * function.c (instantiate_decls): Guard call to
24176 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
24177 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
24179 * gensupport.c (gen_reader::gen_reader): Pass "false"
24180 for new "compact" param of rtx_reader.
24181 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
24182 rather than an empty string for NULL strings.
24183 * read-md.c: Potentially include config.h rather than bconfig.h.
24184 Wrap include of errors.h with #ifdef GENERATOR_FILE.
24185 (have_error): New global, copied from errors.c.
24186 (md_reader::read_name): Rename to...
24187 (md_reader::read_name_1): ...this, adding "out_loc" param,
24188 and converting "missing name or number" to returning false, rather
24190 (md_reader::read_name): Reimplement in terms of read_name_1.
24191 (md_reader::read_name_or_nil): New function.
24192 (md_reader::read_string): Handle "(nil)" by returning NULL.
24193 (md_reader::md_reader): Add new param "compact".
24194 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
24195 (md_reader::read_file): New method.
24196 * read-md.h (md_reader::md_reader): Add new param "compact".
24197 (md_reader::read_file): New method.
24198 (md_reader::is_compact): New accessor.
24199 (md_reader::read_name): Convert return type from void to file_location.
24200 (md_reader::read_name_or_nil): New decl.
24201 (md_reader::read_name_1): New decl.
24202 (md_reader::m_compact): New field.
24203 (noop_reader::noop_reader): Pass "false" for new "compact" param
24205 (rtx_reader::rtx_reader): Add new "compact" param.
24206 (rtx_reader::read_rtx_operand): Make virtual and convert return
24207 type from void to rtx.
24208 (rtx_reader::read_until): New decl.
24209 (rtx_reader::handle_any_trailing_information): New virtual function.
24210 (rtx_reader::postprocess): New virtual function.
24211 (rtx_reader::finalize_string): New virtual function.
24212 (rtx_reader::m_in_call_function_usage): New field.
24213 (rtx_reader::m_reuse_rtx_by_id): New field.
24214 * read-rtl-function.c: New file.
24215 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
24216 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
24217 (selftest::verify_three_block_rtl_cfg): New decl.
24218 * read-rtl-function.h: New file.
24219 * read-rtl.c: Potentially include config.h rather than bconfig.h.
24220 For host, include function.h, memmodel.h, and emit-rtl.h.
24221 (one_time_initialization): New function.
24222 (struct compact_insn_name): New struct.
24223 (compact_insn_names): New array.
24224 (find_code): Handle insn codes in compact dumps.
24225 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
24226 (bind_subst_iter_and_attr): Likewise.
24227 (add_condition_to_string): Likewise.
24228 (add_condition_to_rtx): Likewise.
24229 (apply_attribute_uses): Likewise.
24230 (add_current_iterators): Likewise.
24231 (apply_iterators): Likewise.
24232 (initialize_iterators): Guard usage of apply_subst_iterator with
24233 #ifdef GENERATOR_FILE.
24234 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
24235 (md_reader::read_mapping): Likewise.
24236 (add_define_attr_for_define_subst): Likewise.
24237 (add_define_subst_attr): Likewise.
24238 (read_subst_mapping): Likewise.
24239 (check_code_iterator): Likewise.
24240 (rtx_reader::read_rtx): Likewise. Move one-time initialization
24242 (one_time_initialization): New function.
24243 (rtx_reader::read_until): New method.
24244 (read_flags): New function.
24245 (parse_reg_note_name): New function.
24246 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
24247 Handle reuse_rtx ids.
24248 Wrap iterator lookup within #ifdef GENERATOR_FILE.
24249 Add parsing support for RTL dumps, mirroring the special-cases in
24250 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
24251 values, and calling handle_any_trailing_information.
24252 (rtx_reader::read_rtx_operand): Convert return type from void
24253 to rtx, returning return_rtx. Handle case 'e'. Call
24254 finalize_string on XSTR and XTMPL fields.
24255 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
24256 "(nil)" values were omitted. Call the postprocess vfunc on the
24258 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
24259 class ctor. Initialize m_in_call_function_usage. Call
24260 one_time_initialization.
24261 * rtl-tests.c (selftest::test_uncond_jump): Call
24262 set_new_first_and_last_insn.
24263 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
24264 * selftest-rtl.c: New file.
24265 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
24266 (selftest::get_insn_by_uid): New decl.
24267 * selftest-run-tests.c (selftest::run_tests): Call
24268 read_rtl_function_c_tests.
24269 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
24270 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
24273 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
24275 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
24276 operands in a special way. Assert that pos+len <= mode precision.
24278 2017-01-05 Jakub Jelinek <jakub@redhat.com>
24280 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
24281 3 argument Alias with unlimited for the negative form.
24282 (fno-vect-cost-model): Removed.
24284 2017-01-05 Martin Liska <mliska@suse.cz>
24286 * hsa-gen.c (gen_hsa_divmod): New function.
24287 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
24289 2017-01-05 Martin Liska <mliska@suse.cz>
24292 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
24295 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24297 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
24298 small constant length operands.
24300 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24302 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
24303 between loop iterations.
24305 2017-01-05 Martin Liska <mliska@suse.cz>
24308 * gimplify.c (gimplify_decl_expr): Compare to
24309 asan_poisoned_variables instread of checking flags.
24310 (gimplify_target_expr): Likewise.
24311 (gimplify_expr): Likewise.
24312 (gimplify_function_tree): Conditionally initialize
24313 asan_poisoned_variables.
24315 2017-01-04 Jeff Law <law@redhat.com>
24317 PR tree-optimizatin/78812
24318 * rtl.h (contains_mem_rtx_p): Prototype.
24319 * ifcvt.c (containts_mem_rtx_p): Move from here to...
24320 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
24321 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
24322 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
24323 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
24325 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24327 * input.c (assert_char_at_range): Default-initialize actual_range.
24329 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24331 * df-scan.c (df_ref_create_structure): Make regno unsigned,
24332 to match the caller.
24334 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24336 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
24337 insns after final jump in test to emit dummy move.
24339 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24341 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
24342 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
24344 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
24346 * multiple_target.c (create_dispatcher_calls): Init e_next.
24347 * tree-ssa-loop-split.c (split_loop): Init border.
24348 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
24351 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
24356 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
24357 (altivec_register_operand): Do not return true if the operand
24358 contains a SUBREG mixing SImode and SFmode.
24359 (vsx_register_operand): Likewise.
24360 (vsx_reg_sfsubreg_ok): New predicate.
24361 (vfloat_operand): Do not return true if the operand contains a
24362 SUBREG mixing SImode and SFmode.
24363 (vint_operand): Likewise.
24364 (vlogical_operand): Likewise.
24365 (gpc_reg_operand): Likewise.
24366 (int_reg_operand): Likewise.
24367 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
24368 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
24369 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
24371 (rs6000_emit_move_si_sf_subreg): New helper function.
24372 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
24373 fixup SUBREGs involving SImode and SFmode.
24374 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
24375 numbers for the new peephole2 optimization.
24376 (peephole2 for SFmode unions): New peephole2 to optimize cases in
24377 the GLIBC math library that do AND/IOR/XOR operations on single
24378 precision floating point.
24379 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
24380 target macros to say whether we need to avoid SUBREGs mixing
24382 (TARGET_ALLOW_SF_SUBREG): Likewise.
24383 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
24384 (UNSPEC_SI_FROM_SF): Likewise.
24385 (iorxor): Change spacing.
24386 (and_ior_xor): New iterator for AND, IOR, and XOR.
24387 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
24388 (movdi_from_sf_zero_ext): Likewise.
24389 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
24390 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
24391 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
24392 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
24393 (fms<mode>4): Likewise.
24394 (fnma<mode>4): Likewise.
24395 (fnms<mode>4): Likewise.
24396 (nfma<mode>4): Likewise.
24397 (nfms<mode>4): Likewise.
24399 2017-01-04 Marek Polacek <polacek@redhat.com>
24402 * doc/invoke.texi: Document -Wpointer-compare.
24404 2017-01-04 Jakub Jelinek <jakub@redhat.com>
24406 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
24409 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
24410 descriptions for -gdwarf-5 and emit them as uleb128 instead of
24413 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
24416 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
24417 documentation of the powerpc_popcntb_ok attribute.
24418 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24419 code to issue warning messages if a requested CPU configuration is
24420 not supported by the binary (assembler and loader) toolchain.
24421 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
24422 made to define a built-in function that has been disabled.
24423 (paired_init_builtins): Add assertion to prevent ICE if attempt is
24424 made to define a built-in function that has been disabled.
24425 (altivec_init_builtins): Add comment explaining why definition
24426 of the DST built-in functions is not preceded by an assertion
24427 check. Add assertions to prevent ICE if attempts are made to
24428 define an altivec predicate or an abs* built-in function that has
24430 (htm_init_builtins): Add comment explaining why definition of the
24431 htm built-in functions is not preceded by an assertion check.
24433 2017-01-04 Jeff Law <law@redhat.com>
24435 PR tree-optimizatin/67955
24436 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
24437 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
24438 the points-to solution does not include pt_null. Use DECL_PT_UID
24441 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
24443 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
24444 Use gen_int_mode instead of gen_lopwart for const_int operands.
24446 2017-01-04 Jakub Jelinek <jakub@redhat.com>
24448 PR tree-optimization/71563
24449 * match.pd: Simplify X << Y into X if Y is known to be 0 or
24450 out of range value - has low bits known to be zero.
24452 2017-01-04 Alan Modra <amodra@gmail.com>
24454 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
24455 * configure: Regenerate.
24456 * config.in: Regenerate.
24458 2017-01-04 Jakub Jelinek <jakub@redhat.com>
24461 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
24462 a substring of the message, but strcmp with the whole message. Ifdef
24463 ENABLE_NLS, translate the message first using dgettext.
24465 2017-01-03 Jeff Law <law@redhat.com>
24467 PR tree-optimizatin/78856
24468 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
24469 (mark_threaded_blocks): Remove code to truncate thread paths that
24470 cross multiple loop headers. Instead invalidate the cached loop
24471 iteration information and handle case of a thread path walking
24472 into an irreducible region.
24474 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
24477 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
24478 assertions. Add support for doing the signbit if the IEEE 128-bit
24479 floating point value is in a GPR.
24480 * config/rs6000/rs6000.md (Fsignbit): Delete.
24481 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
24482 Update the length attribute if the value is in a GPR.
24483 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
24484 the sign or zero extension instruction, since the value is always 0/1.
24485 (signbit<mode>2_dm2): Delete using <Fsignbit>.
24488 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
24489 extracting SImode to a GPR register so that we can generate a
24490 store, limit the vector to be in a traditional Altivec register
24491 for the vextuwrx instruction.
24493 2017-01-03 Ian Lance Taylor <iant@google.com>
24495 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
24497 2017-01-03 Martin Sebor <msebor@redhat.com>
24499 PR tree-optimization/78696
24500 * gimple-ssa-sprintf.c (format_floating): Correct handling of
24501 precision. Use MPFR for %f for greater fidelity. Correct handling
24503 (pass_sprintf_length::compute_format_length): Set width and precision
24504 specified by asrerisk to void_node for vararg functions.
24505 (try_substitute_return_value): Adjust dump output.
24507 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
24509 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
24511 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
24513 * doc/invoke.texi (SPARC options): Document -mlra as the default.
24514 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
24515 -mlra/-mno-lra was passed to the compiler.
24517 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
24519 PR rtl-optimization/65618
24520 * emit-rtl.c (try_split): Move initialization of "before" and
24521 "after" to just before the call to emit_insn_after_setloc.
24523 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
24525 * doc/md.texi (Standard Names): Remove reference to Java frontend.
24527 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
24529 * dwarf2out.c (gen_enumeration_type_die): When
24530 -gno-strict-dwarf, add a DW_AT_encoding attribute.
24532 2017-01-03 Jakub Jelinek <jakub@redhat.com>
24534 PR tree-optimization/78965
24535 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
24536 Change first argument from const call_info & to call_info &. For %n
24537 set info.nowrite to false.
24539 PR middle-end/78901
24540 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
24541 possibly throwing calls.
24543 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
24544 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
24545 and fns handling, rather than in a separate case SSA_NAME.
24547 2017-01-02 Jeff Law <law@redhat.com>
24549 * config/darwin-driver.c (darwin_driver_init): Const-correctness
24550 fixes for first_period and second_period variables.
24552 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
24555 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
24556 (*insvqi_1): New insn pattern.
24557 (*insvqi_1_mem_rex64): Ditto.
24558 (*insvqi_2): Ditto.
24559 (*insvqi_3): Rename from *insvqi.
24561 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
24563 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
24565 * doc/cfg.texi (Edges): Remove reference to Java.
24566 (Maintaining the CFG): Ditto.
24568 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24570 PR middle-end/77674
24571 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
24572 transparent aliases.
24574 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24576 PR middle-end/77484
24577 * predict.def (PRED_CALL): Update hitrate.
24578 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
24579 * predict.c (tree_estimate_probability_bb): Split CALL predictor
24580 into direct/indirect/polymorphic variants.
24582 2017-01-01 Jakub Jelinek <jakub@redhat.com>
24584 Update copyright years.
24586 * gcc.c (process_command): Update copyright notice dates.
24587 * gcov-dump.c (print_version): Ditto.
24588 * gcov.c (print_version): Ditto.
24589 * gcov-tool.c (print_version): Ditto.
24590 * gengtype.c (create_file): Ditto.
24591 * doc/cpp.texi: Bump @copying's copyright year.
24592 * doc/cppinternals.texi: Ditto.
24593 * doc/gcc.texi: Ditto.
24594 * doc/gccint.texi: Ditto.
24595 * doc/gcov.texi: Ditto.
24596 * doc/install.texi: Ditto.
24597 * doc/invoke.texi: Ditto.
24599 Copyright (C) 2017 Free Software Foundation, Inc.
24601 Copying and distribution of this file, with or without modification,
24602 are permitted in any medium without royalty provided the copyright
24603 notice and this notice are preserved.