1 2017-08-01 Jakub Jelinek <jakub@redhat.com>
4 * optabs.def (vec_extract_optab, vec_init_optab): Change from
5 a direct optab to conversion optab.
6 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
7 with GET_MODE_INNER as last argument instead of optab_handler.
8 * expmed.c (extract_bit_field_1): Likewise. Use vector from
9 vector extraction if possible and optab is available.
10 * expr.c (store_constructor): Use convert_optab_handler instead
11 of optab_handler. Use vector initialization from smaller
12 vectors if possible and optab is available.
13 * tree-vect-stmts.c (vectorizable_load): Likewise.
14 * doc/md.texi (vec_extract, vec_init): Document that the optabs
16 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
17 of vec_init from half-sized vectors with the same element mode.
18 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
19 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
20 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
22 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
23 after mode in gen_vec_extract* calls.
24 (vec_extract<mode>): Renamed to ...
25 (vec_extract<mode><ssescalarmodelower>): ... this.
26 (vec_extract<mode><ssehalfvecmodelower>): New expander.
27 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
28 element mode after mode in gen_vec_init* calls.
29 (VEC_INIT_HALF_MODE): New mode iterator.
30 (vec_init<mode>): Renamed to ...
31 (vec_init<mode><ssescalarmodelower>): ... this.
32 (vec_init<mode><ssehalfvecmodelower>): New expander.
33 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
34 (vec_extractv2sfsf): ... this.
35 (vec_initv2sf): Renamed to ...
36 (vec_initv2sfsf): ... this.
37 (vec_extractv2si): Renamed to ...
38 (vec_extractv2sisi): ... this.
39 (vec_initv2si): Renamed to ...
40 (vec_initv2sisi): ... this.
41 (vec_extractv4hi): Renamed to ...
42 (vec_extractv4hihi): ... this.
43 (vec_initv4hi): Renamed to ...
44 (vec_initv4hihi): ... this.
45 (vec_extractv8qi): Renamed to ...
46 (vec_extractv8qiqi): ... this.
47 (vec_initv8qi): Renamed to ...
48 (vec_initv8qiqi): ... this.
49 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
50 (vec_init<mode>): Renamed to ...
51 (vec_init<mode><VEC_base_l>): ... this.
52 (vec_extract<mode>): Renamed to ...
53 (vec_extract<mode><VEC_base_l>): ... this.
54 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
55 (vec_initv2sfsf): ... this.
56 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
57 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
58 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
59 element mode after mode in gen_vec_init* calls.
60 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
61 (vec_init<mode><Vel>): ... this.
62 (vec_extract<mode>): Renamed to ...
63 (vec_extract<mode><Vel>): ... this.
64 * config/aarch64/iterators.md (Vel): New mode attribute.
65 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
66 Add element mode after mode in gen_vec_extract* calls.
67 * config/s390/vector.md (non_vec_l): New mode attribute.
68 (vec_extract<mode>): Renamed to ...
69 (vec_extract<mode><non_vec_l>): ... this.
70 (vec_init<mode>): Renamed to ...
71 (vec_init<mode><non_vec_l>): ... this.
72 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
73 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
75 * config/arm/iterators.md (V_elem_l): New mode attribute.
76 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
77 (vec_extract<mode><V_elem_l>): ... this.
78 (vec_extractv2di): Renamed to ...
79 (vec_extractv2didi): ... this.
80 (vec_init<mode>): Renamed to ...
81 (vec_init<mode><V_elem_l>): ... this.
82 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
83 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
84 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
85 Add element mode after gen_vec_extract* calls.
86 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
87 (vec_init<mode><unitmode>): ... this.
88 (vec_extract<mode>): Renamed to ...
89 (vec_extract<mode><unitmode>): ... this.
90 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
91 (vec_init<mode><unitmode>): ... this.
92 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
93 (vec_initv2sfsf): ... this.
94 (vec_extractv2sf): Renamed to ...
95 (vec_extractv2sfsf): ... this.
96 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
97 Add element mode after gen_vec_extract* calls.
98 * config/mips/mips.md (unitmode): New mode iterator.
99 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
100 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
101 * config/spu/spu.md (inner_l): New mode attribute.
102 (vec_init<mode>): Renamed to ...
103 (vec_init<mode><inner_l>): ... this.
104 (vec_extract<mode>): Renamed to ...
105 (vec_extract<mode><inner_l>): ... this.
106 * config/sparc/sparc.md (veltmode): New mode iterator.
107 (vec_init<VMALL:mode>): Renamed to ...
108 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
109 * config/ia64/vect.md (vec_initv2si): Renamed to ...
110 (vec_initv2sisi): ... this.
111 (vec_initv2sf): Renamed to ...
112 (vec_initv2sfsf): ... this.
113 (vec_extractv2sf): Renamed to ...
114 (vec_extractv2sfsf): ... this.
115 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
116 (vec_init<mode>): Renamed to ...
117 (vec_init<mode><VEC_base_l>): ... this.
118 (vec_extract<mode>): Renamed to ...
119 (vec_extract<mode><VEC_base_l>): ... this.
120 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
121 (vec_initv2sfsf): ... this.
122 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
123 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
124 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
127 2017-08-01 Richard Biener <rguenther@suse.de>
129 PR tree-optimization/81297
130 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
131 TREE_OVERFLOW from INTEGER_CSTs.
133 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
135 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
137 2017-07-31 Carl Love <cel@us.ibm.com>
139 * config/rs6000/rs6000-c: Add support for built-in functions
140 vector signed char vec_xl_be (signed long long, signed char *);
141 vector unsigned char vec_xl_be (signed long long, unsigned char *);
142 vector signed int vec_xl_be (signed long long, signed int *);
143 vector unsigned int vec_xl_be (signed long long, unsigned int *);
144 vector signed long long vec_xl_be (signed long long, signed long long *);
145 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
146 vector signed short vec_xl_be (signed long long, signed short *);
147 vector unsigned short vec_xl_be (signed long long, unsigned short *);
148 vector double vec_xl_be (signed long long, double *);
149 vector float vec_xl_be (signed long long, float *);
150 * config/rs6000/altivec.h (vec_xl_be): Add #define.
151 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI,
152 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions for the builtins.
153 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
154 (altivec_expand_builtin): Add switch statement to call altivec_expand_xl_be
156 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
157 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di, __builtin_vsx_le_be_v4sf,
158 __builtin_vsx_le_be_v2df, __builtin_vsx_le_be_v16qi.
159 * doc/extend.texi: Update the built-in documentation file for the
160 new built-in functions.
162 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
165 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
167 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
169 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
171 * config.gcc: Add z14.
172 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
173 CPU model numbers for z13s and z14.
174 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
176 * config/s390/s390-opts.h (enum processor_type): Rename
177 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
178 * config/s390/s390.c (processor_table): Add field for CPU name to
179 be passed to Binutils.
180 (s390_asm_output_machine_for_arch): Use the new field in
181 processor_table for Binutils.
182 (s390_expand_builtin): Replace arch12 with z14.
183 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
184 (s390_get_sched_attrmask): Likewise.
185 (s390_get_unit_mask): Likewise.
186 * config/s390/s390.opt: Add z14 to processor_type enum.
188 2017-07-31 Martin Jambor <mjambor@suse.cz>
191 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
192 regardless of optimization level.
194 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
195 Martin Liska <mliska@suse.cz>
197 * predict.def: Remove old comment and adjust probability.
198 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
201 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
204 * config/i386/i386.c (ix86_function_naked): New function.
205 (ix86_can_use_return_insn_p): Return false for naked functions.
206 (ix86_expand_prologue): Skip prologue for naked functions.
207 (ix86_expand_epilogue): Skip epilogue for naked functions
208 and emit trap instruction.
209 (ix86_warn_func_return): New function.
210 (ix86_attribute_table): Add "naked" attribute specification.
211 (TARGET_WARN_FUNC_RETURN): Define.
212 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
214 2017-07-31 Martin Liska <mliska@suse.cz>
216 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
217 (dump_gimple_bb_header): Always dump BB info.
218 (pp_cfg_jump): Do not append info about BB when dumping a jump.
220 2017-07-31 Martin Liska <mliska@suse.cz>
223 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
224 also with current_function_decl non-null equality.
226 2017-07-31 Jakub Jelinek <jakub@redhat.com>
229 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
230 change type to the element type, instead add eltype variable and
231 use it where we are interested in the element type.
233 PR tree-optimization/81603
234 * ipa-polymorphic-call.c
235 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
236 offset arithmetic in offset_int, bail out if the resulting bit offset
237 doesn't fit into shwi.
239 2017-07-31 Martin Liska <mliska@suse.cz>
241 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
242 (gimplify_save_expr): Fix comment.
244 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
247 * config/i386/i386.c (ix86_function_arg): Update arguments for
249 (ix86_compute_frame_layout): Set the initial stack offset to
250 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
251 INCOMING_FRAME_SP_OFFSET.
252 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
253 stack before exception handler returns.
254 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
255 the 'ERROR_CODE' for exception handler.
257 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
259 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
260 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
261 (ASM_OUTPUT_REG_POP): Ditto.
262 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
263 instead of asm_fprintf to output pure string.
265 2017-07-29 Jakub Jelinek <jakub@redhat.com>
267 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
268 to imported_module_or_decl hook.
269 (debug_nothing_tree_tree_tree_bool): Remove.
270 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
271 * debug.c (do_nothing_debug_hooks): Use
272 debug_nothing_tree_tree_tree_bool_bool instead of
273 debug_nothing_tree_tree_tree_bool.
274 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
275 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
276 * sdbout.c (sdb_debug_hooks): Likewise.
277 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
278 (gen_namespace_die): Add DW_AT_export_symbols attribute if
280 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
281 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
282 attribute, don't add anything.
284 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
286 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
287 (fold_build2_stat_loc): Likewise.
288 (fold_build3_stat_loc): Likewise.
289 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
290 (fold_build1_loc): Remove macro.
291 (fold_build2_loc): Likewise.
292 (fold_build3_loc): Likewise.
294 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
296 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
297 (gimple_build_debug_bind_source_stat): Likewise.
298 * gimple.h (gimple_build_debug_bind): Remove macro.
299 (gimple_build_debug_bind_source): Likewise.
301 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
303 * bitmap.c (bitmap_alloc): Adjust.
304 (bitmap_gc_alloc): Likewise.
305 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
307 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
309 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
310 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
311 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
312 (bitmap_gc_alloc_stat): Likewise.
313 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
315 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
317 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
318 * rtl.h (shallow_copy_rtx): Remove macro.
320 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
322 * emit-rtl.c (gen_raw_REG): Adjust.
323 * gengenrtl.c (gendef): Likewise.
324 * rtl.c (rtx_alloc_stat): Remove _stat from name.
325 * rtl.h (rtx_alloc): Remove macro.
327 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
329 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
330 (build_tree_list_stat): Likewise.
331 * tree.h (build_tree_list): Remove macro.
332 (build_tree_list_vec): Likewise.
334 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
336 * tree.c (make_vector_stat): Remove _stat from name.
337 (build_vector_stat): Likewise.
338 * tree.h (make_vector_stat): Remove macro.
339 (build_vector_stat): Likewise.
341 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
343 * tree.h (build_var_debug_value): Remove prototype.
345 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
347 * tree.c (tree_cons_stat): Remove _stat from name.
348 * tree.h (tree_cons): Remove macro.
350 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
352 * tree.c (build_vl_exp_stat): Remove _stat from name.
353 * tree.h (build_vl_exp): Remove macro.
355 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
357 * tree.c (build_decl_stat): Remove _stat from name.
358 * tree.h (build_decl): Remove macro.
360 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
362 * gimple.c (gimple_build_with_ops_stat): Adjust.
363 (gimple_alloc_stat): Remove _stat from name.
364 * gimple.h (gimple_alloc): Remove macro.
366 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
368 * tree.c (make_tree_vec_stat): Remove _stat from name.
369 (grow_tree_vec_stat): Likewise.
370 * tree.h (make_tree_vec_stat): Adjust prototype.
371 (grow_tree_vec_stat): Likewise.
372 (make_tree_vec): Remove macro.
373 (grow_tree_vec): Likewise.
375 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
377 * fold-const.c (fold_build1_stat_loc): Adjust.
378 (fold_build2_stat_loc): Likewise.
379 (fold_build3_stat_loc): Likewise.
380 * tree.c (build0_stat): Remove _stat from name.
381 (build1_stat): Likewise.
382 (build2_stat): Likewise.
383 (build3_stat): Likewise.
384 (build4_stat): Likewise.
385 (build5_stat): Likewise.
386 * tree.h (build1_loc): Remove macro, and rename _stat function
388 (build2_loc): Likewise.
389 (build3_loc): Likewise.
390 (build4_loc): Likewise.
391 (build5_loc): Likewise.
393 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
395 * tree.c (make_int_cst_stat): Remove _stat from name.
396 * tree.h (make_int_cst_stat): Adjust prototype.
397 (make_int_cst): Remove macro.
399 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
401 * tree.c (make_tre_binfo_stat): Remove _stat from name.
402 * tree.h (make_tree_binfo_stat): Adjust prototype.
403 (make_tree_binfo): Remove.
405 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
407 * tree.c (copy_node_stat): Rename to copy_node.
408 (build_distinct_type_copy): Adjust.
409 * tree.h (copy_node_stat): Adjust prototype.
410 (copy_node): Remove macro.
412 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
414 * tree.c (make_node_stat): rename to make_node.
415 (build_tree_list_stat): Adjust.
416 (build0_stat): Likewise.
417 (build2_stat): Likewise.
418 (build3_stat): Likewise.
419 (build4_stat): Likewise.
420 (build5_stat): Likewise.
421 (build_decl_stat): Likewise.
422 * tree.h (make_node_stat): Adjust prototype.
423 (make_node): remove macro.
425 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
427 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
428 (PPC_FEATURE2_SCV): Likewise.
429 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
431 2017-07-28 Tamar Christina <tamar.christina@arm.com>
433 * config/aarch64/aarch64.c
434 (aarch64_internal_mov_immediate): Add new special pattern.
435 * config/aarch64/aarch64.md (*movdi_aarch64):
436 Add reg/32bit const mov case.
438 2017-07-28 Tamar Christina <tamar.christina@arm.com>
439 Richard Sandiford <richard.sandiford@linaro.org>
441 * config/aarch64/aarch64.md (mov<mode>): Generalize.
442 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
443 Add integer and movi cases.
444 (movi-split-hf-df-sf split, fp16): New.
445 (enabled): Added TARGET_FP_F16INST.
446 * config/aarch64/iterators.md (GPF_HF): New.
447 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
449 2017-07-28 Tamar Christina <tamar.christina@arm.com>
451 * config/aarch64/aarch64.c
452 (aarch64_simd_container_mode): Add prototype.
453 (aarch64_expand_mov_immediate): Add HI support.
454 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
455 (aarch64_can_const_movi_rtx_p): New.
456 (aarch64_preferred_reload_class):
457 Remove restrictions of using FP registers for certain SIMD operations.
458 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
459 (aarch64_valid_floating_const): Add integer move validation.
460 (aarch64_simd_imm_scalar_p): Remove.
461 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
462 (aarch64_legitimate_constant_p): Expand list of supported cases.
463 * config/aarch64/aarch64-protos.h
464 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
465 (aarch64_reinterpret_float_as_int): New.
466 (aarch64_simd_imm_scalar_p): Remove.
467 * config/aarch64/constraints.md (Uvi): New.
468 (Dd): Split into Ds and new Dd.
469 * config/aarch64/aarch64.md (*movsi_aarch64):
471 (*movdi_aarch64): Add SIMD mov case.
473 2017-07-28 Bin Cheng <bin.cheng@arm.com>
475 * tree-predcom.c: (struct chain): Handle store-store chain in which
476 stores for elimination only store loop invariant values.
477 (execute_pred_commoning_chain): Ditto.
478 (prepare_initializers_chain_store_elim): Ditto.
479 (prepare_finalizers): Ditto.
480 (is_inv_store_elimination_chain): New function.
481 (initialize_root_vars_store_elim_1): New function.
483 2017-07-28 Bin Cheng <bin.cheng@arm.com>
485 * tree-predcom.c: Revise general description of the pass.
486 (enum chain_type): New enum type for store elimination.
487 (struct chain): New field supporting store elimination.
488 (struct component): Ditto.
489 (dump_chain): Dump store-stores chain.
490 (release_chain): Release resources.
491 (split_data_refs_to_components): Compute and create component
492 contains only stores for elimination.
493 (get_chain_last_ref_at): New function.
494 (make_invariant_chain): Initialization.
495 (make_rooted_chain): Specify chain type in parameter and record it.
496 (add_looparound_copies): Skip for store-stores chain.
497 (determine_roots_comp): Compute type of chain and pass it to
499 (initialize_root_vars_store_elim_2): New function.
500 (finalize_eliminated_stores): New function.
501 (remove_stmt): Handle store for elimination.
502 (execute_pred_commoning_chain): Execute predictive commoning on
504 (determine_unroll_factor): Skip unroll for store-stores chain.
505 (prepare_initializers_chain_store_elim): New function.
506 (prepare_initializers_chain): Hanlde store-store chain.
507 (prepare_finalizers_chain, prepare_finalizers): New function.
508 (tree_predictive_commoning_loop): Return integer value indicating
509 if loop is unrolled or lcssa form is corrupted.
510 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
512 2017-07-28 Bin Cheng <bin.cheng@arm.com>
514 * tree-predcom.c (initialize_root): Delete.
515 (execute_pred_commoning_chain): Initialize root vars and replace
516 reference of non-combined chain directly, rather than call above
519 2017-07-28 Bin Cheng <bin.cheng@arm.com>
521 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
522 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
524 2017-07-28 Bin Cheng <bin.cheng@arm.com>
526 * tree-predcom.c (struct chain): New field init_seq.
527 (release_chain): Release init_seq.
528 (prepare_initializers_chain): Record intialization stmts in above
530 (insert_init_seqs): New function.
531 (tree_predictive_commoning_loop): Call insert_init_seqs.
533 2017-07-28 Bin Cheng <bin.cheng@arm.com>
535 * tree-predcom.c (determine_roots_comp): Skip trivial components.
537 2017-07-28 Richard Biener <rguenther@suse.de>
539 * match.pd: Remove superfluous :c.
540 * genmatch.c (simplify::id): Add member.
541 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
543 (current_id): New global.
544 (dt_node::parent): Move from ...
545 (dt_operand::parent): ... here. Add for_id member.
546 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
547 (decision_tree::find_node): Relax order requirement when
548 merging DT_TRUE nodes to ones inbetween the current simplify
549 and the one we try to merge with. Add diagnostic whenever
550 we need to enforce pattern order by not merging.
551 (decision_tree::insert): Set current_id.
552 (decision_tree::print_node): Dump parent node and for_id.
553 (parser::last_id): Add member.
554 (parser::push_simplify): Assign unique id.
555 (parser::parser): Initialize last_id.
557 2017-07-28 Martin Liska <mliska@suse.cz>
560 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
561 gimple_build_debug_bind.
563 2017-07-28 Richard Biener <rguenther@suse.de>
565 PR tree-optimization/81502
566 * match.pd: Add pattern combining BIT_INSERT_EXPR with
568 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
570 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
571 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
572 for BIT_FIELD_REF args.
573 * fold-const.c (make_bit_field_ref): Likewise.
574 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
576 2017-07-28 Jakub Jelinek <jakub@redhat.com>
579 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
580 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
581 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
582 Or it into SANITIZER_UNDEFINED.
583 * ubsan.c: Include gimple-fold.h and varasm.h.
584 (ubsan_expand_ptr_ifn): New function.
585 (instrument_pointer_overflow): New function.
586 (maybe_instrument_pointer_overflow): New function.
587 (instrument_object_size): Formatting fix.
588 (pass_ubsan::execute): Call instrument_pointer_overflow
589 and maybe_instrument_pointer_overflow.
590 * internal-fn.c (expand_UBSAN_PTR): New function.
591 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
592 * sanitizer.def (__ubsan_handle_pointer_overflow,
593 __ubsan_handle_pointer_overflow_abort): New builtins.
594 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
595 * internal-fn.def (UBSAN_PTR): New internal function.
596 * opts.c (sanitizer_opts): Add pointer-overflow.
597 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
598 * fold-const.c (build_range_check): Compute pointer range check in
599 integral type if pointer arithmetics would be needed. Formatting
602 2017-07-28 Martin Liska <mliska@suse.cz>
605 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
606 parameters that are of a variable-length.
608 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
610 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
612 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
613 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
614 (CRT_CALL_STATIC_FUNCTION): Likewise.
615 (ASM_DEFAULT_SPEC): New define.
616 (ASM_SPEC32): Likewise.
617 (ASM_SPEC64): Likewise.
618 (ASM_SPEC_COMMON): Likewise.
619 (ASM_SPEC): Likewise.
620 (INVALID_64BIT): Likewise.
621 (LINK_OS_DEFAULT_SPEC): Likewise.
622 (LINK_OS_SPEC32): Likewise.
623 (LINK_OS_SPEC64): Likewise.
624 (POWERPC_LINUX): Likewise.
625 (PTRDIFF_TYPE): Likewise.
626 (RESTORE_FP_PREFIX): Likewise.
627 (RESTORE_FP_SUFFIX): Likewise.
628 (SAVE_FP_PREFIX): Likewise.
629 (SAVE_FP_SUFFIX): Likewise.
630 (SIZE_TYPE): Likewise.
631 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
632 (TARGET_64BIT): Likewise.
633 (TARGET_64BIT): Likewise.
634 (TARGET_AIX): Likewise.
635 (WCHAR_TYPE_SIZE): Likewise.
636 (WCHAR_TYPE): Undefine.
637 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
638 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
639 (CPP_OS_RTEMS_SPEC): Delete.
640 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
641 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
643 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
645 2017-07-28 Jakub Jelinek <jakub@redhat.com>
647 PR tree-optimization/81578
648 * tree-parloops.c (build_new_reduction): Bail out if
649 reduction_code isn't one of the standard OpenMP reductions.
650 Move the details printing after that decision.
652 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
654 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
655 related to reload_in_progress.
656 (splat_input_operand): Likewise.
657 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
659 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
661 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
662 (TARGET_INSTANTIATE_DECLS): Likewise.
663 (legitimate_indexed_address_p): Delete reload_in_progress code.
664 (rs6000_debug_legitimate_address_p): Likewise.
665 (rs6000_eliminate_indexed_memrefs): Likewise.
666 (rs6000_emit_le_vsx_store): Likewise.
667 (rs6000_emit_move_si_sf_subreg): Likewise.
668 (rs6000_emit_move): Likewise.
669 (register_to_reg_type): Likewise.
670 (rs6000_pre_atomic_barrier): Likewise.
671 (rs6000_machopic_legitimize_pic_address): Likewise.
672 (rs6000_allocate_stack_temp): Likewise.
673 (rs6000_address_for_fpconvert): Likewise.
674 (rs6000_address_for_altivec): Likewise.
675 (rs6000_secondary_memory_needed_rtx): Delete function.
676 (rs6000_check_sdmode): Likewise.
677 (rs6000_alloc_sdmode_stack_slot): Likewise.
678 (rs6000_instantiate_decls): Likewise.
679 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
680 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
681 Delete reload_in_progress.
682 (*vec_reload_and_plus_<mptrsize>): Likewise.
683 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
684 (vsx_div_v2di): Likewise.
685 (vsx_udiv_v2di): Likewise.
687 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
689 * config/rs6000/rs6000.opt (mlra): Replace with stub.
690 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
691 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
692 (rs6000_debug_reg_global): Delete print of LRA status.
693 (rs6000_option_override_internal): Delete dead LRA related code.
694 (rs6000_lra_p): Delete function.
695 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
697 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
699 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
700 * config/riscv/rtems.h: New file.
702 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
703 Sudakshina Das <sudi.das@arm.com>
705 * config/aarch64/aarch64.md
706 (define_split for and<mode>3nr_compare): Move
707 non aarch64_logical_operand to a register.
708 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
709 register immediate operand to a register.
710 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
712 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
715 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
717 2017-07-27 Richard Biener <rguenther@suse.de>
719 PR tree-optimization/81573
720 PR tree-optimization/81494
721 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
722 multi defuse cycle case.
724 2017-07-27 Richard Biener <rguenther@suse.de>
726 PR tree-optimization/81571
727 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
730 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
732 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
733 earlier and only if MASK_FPU is set. Adjust formatting.
735 2017-07-27 Martin Liska <mliska@suse.cz>
737 * opt-functions.awk: Add validation of value of Init.
738 * optc-gen.awk: Pass new argument.
740 2017-07-27 Martin Liska <mliska@suse.cz>
742 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
745 2017-07-27 Martin Liska <mliska@suse.cz>
747 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
748 BBs and edges seen by autoFDO.
750 2017-07-27 Richard Biener <rguenther@suse.de>
752 PR tree-optimization/81502
753 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
754 with incompatible but same sized type.
755 (execute_update_addresses_taken): Likewise.
757 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
759 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
760 flag_tree_loop_vectorize rather than flag_tree_vectorize.
762 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
765 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
766 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
767 Change s_operand to memory_operand.
769 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
771 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
772 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
773 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
774 Emit instructions rather than returning an expression. Handle TFmode
775 and KFmode by casting to TImode.
776 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
777 (rs6000_emit_le_vsx_store): Likewise.
778 * config/rs6000/vsx.md (VSX_TI): New iterator.
779 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
780 (*vsx_le_undo_permute_<mode>): Likewise.
781 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
782 emit the split sequence.
783 (*vsx_le_perm_store_<mode>): Likewise.
785 2017-07-27 Jakub Jelinek <jakub@redhat.com>
787 PR tree-optimization/81555
788 PR tree-optimization/81556
789 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
790 if true, force CHANGED for the recursive invocation.
791 (reassociate_bb): Remember original length of ops array, pass
792 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
794 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
795 attributes for noipa attribute. For naked attribute use
796 lookup_attribute first before lookup_attribute_spec.
797 * final.c (rest_of_handle_final): Disable IPA RA for functions with
799 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
800 for functions with noipa attribute.
801 (cgraph_externally_visible_p): Return true for functions with noipa
803 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
804 for functions with noipa attribute.
805 * doc/extend.texi: Document noipa function attribute.
806 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
807 also for functions with noipa attribute.
808 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
810 2017-07-26 Andrew Pinski <apinski@cavium.com>
812 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
813 vec_unalign_load_cost and vec_unalign_store_cost.
815 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
817 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
818 -mvsx-small-integer option.
819 (ISA_3_0_MASKS_IEEE): Likewise.
820 (OTHER_VSX_VECTOR_MASKS): Likewise.
821 (POWERPC_MASKS): Likewise.
822 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
823 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
824 code, only testing for DImode being allowed in non-VSX floating
826 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
827 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
829 (rs6000_option_override_internal): Delete -mvsx-small-integer.
830 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
831 TARGET_P8_VECTOR test.
832 (rs6000_secondary_reload_simple_move): Likewise.
833 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
834 since TARGET_P9_VECTOR was already tested.
835 (rs6000_opt_masks): Remove -mvsx-small-integer.
836 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
837 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
839 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
840 test for TARGET_VEXTRACTUB was used, and that uses
842 (p9 extract splitter): Likewise.
843 (vsx_extract_<mode>_di_p9): Likewise.
844 (vsx_extract_<mode>_store_p9): Likewise.
845 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
846 for TARGET_P9_VECTOR was used. Delete code that is now dead with
847 the elimination of TARGET_VSX_SMALL_INTEGER.
848 (vsx_extract_<mode>_p8): Likewise.
849 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
850 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
851 (vsx_set_<mode>_p9): Likewise.
852 (vsx_set_v4sf_p9): Likewise.
853 (vsx_set_v4sf_p9_zero): Likewise.
854 (vsx_insert_extract_v4sf_p9): Likewise.
855 (vsx_insert_extract_v4sf_p9_2): Likewise.
856 * config/rs6000/rs6000.md (sign extend splitter): Change
857 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
858 (floatsi<mode>2_lfiwax_mem): Likewise.
859 (floatunssi<mode>2_lfiwzx_mem): Likewise.
860 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
861 since a test for TARGET_P9_VECTOR was used.
862 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
863 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
864 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
865 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
866 TARGET_P8_VECTOR test.
867 (fix_trunc<mode>si2_stfiwx): Likewise.
868 (fix_trunc<mode>si2_internal): Likewise.
869 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
870 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
872 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
873 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
874 TARGET_P8_VECTOR test.
875 (fixuns_trunc<mode>si2_stfiwx): Likewise.
876 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
877 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
879 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
880 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
881 since a test for TARGET_P9_VECTOR was used.
882 (splitter for loading small constants): Likewise.
884 2017-07-26 Andrew Pinski <apinski@cavium.com>
886 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
889 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
892 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
893 (fp_valid_at): Likewise.
895 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
897 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
898 (qdf24xx_addrcost_table): Likewise.
899 (cortexa57_tunings): Update to use generic_branch_cost.
900 (cortexa72_tunings): Likewise.
901 (cortexa73_tunings): Likewise.
902 (qdf24xx_tunings): Likewise.
904 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
906 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
907 (thunderx2t99_branch_cost): Likewise.
908 (cortexa35_tunings): Update to use generic_branch_cost.
909 (cortexa53_tunings): Likewise.
910 (cortexa57_tunings): Likewise.
911 (cortexa72_tunings): Likewise.
912 (cortexa73_tunings): Likewise.
913 (thunderx2t99_tunings): Likewise.
915 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
917 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
918 (sparc_option_override): Honour MASK_FSMULD.
919 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
920 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
921 * config/sparc/sparc.opt (mfsmuld): New option.
922 * doc/invoke.texi (mfsmuld): Document option.
924 2017-07-26 Marek Polacek <polacek@redhat.com>
927 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
929 2017-07-26 Richard Biener <rguenther@suse.de>
931 * gimple-match-head.c (do_valueize): Return OP if valueize
933 (get_def): New helper to get at the def stmt of a SSA name
935 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
936 do_valueize to get at the def stmt.
937 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
939 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
942 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
944 2017-07-26 Martin Liska <mliska@suse.cz>
947 * function.c (expand_function_start): Make expansion of
948 nonlocal_goto_save_area after parm_birth_insn.
950 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
952 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
953 from all CPU target flags enable members.
955 2017-07-26 Richard Biener <rguenther@suse.de>
957 * genmatch.c (dt_simplify::gen): Make iterator vars const.
958 (decision_tree::gen): Make 'type' const.
959 (write_predicate): Likewise.
961 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
963 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
964 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
965 (rs6000_option_override_internal): Likewise.
966 (rs6000_expand_vector_set): Likewise.
967 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
968 (TARGET_UPPER_REGS_SF): Likewise.
969 (TARGET_UPPER_REGS_DI): Likewise.
970 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
971 (TARGET_DIRECT_MOVE_64BIT): Likewise.
972 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
973 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
974 (Splitters for DI constants in Altivec registers): Likewise.
975 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
976 (vsx_set_v4sf_p9): Likewise.
977 (vsx_set_v4sf_p9_zero): Likewise.
978 (vsx_insert_extract_v4sf_p9): Likewise.
979 (vsx_insert_extract_v4sf_p9_2): Likewise.
981 2017-07-25 Carl Love <cel@us.ibm.com>
983 * doc/extend.texi: Update the built-in documentation file for the
984 existing built-in functions
985 vector signed char vec_cnttz (vector signed char);
986 vector unsigned char vec_cnttz (vector unsigned char);
987 vector signed short vec_cnttz (vector signed short);
988 vector unsigned short vec_cnttz (vector unsigned short);
989 vector signed int vec_cnttz (vector signed int);
990 vector unsigned int vec_cnttz (vector unsigned int);
991 vector signed long long vec_cnttz (vector signed long long);
992 vector unsigned long long vec_cnttz (vector unsigned long long);
994 2017-07-25 Andrew Pinski <apinski@cavium.com>
996 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
997 accesses where the use is for the first operand of a BIT_INSERT.
999 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
1002 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
1003 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
1005 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
1007 * config/i386/gstabs.h: Delete.
1008 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
1010 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
1012 * config/i386/i386.c (ix86_decompose_address): Do not check for
1013 register RTX when looking at index_reg or base_reg.
1014 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
1016 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
1018 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
1019 to update EH info here.
1021 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
1023 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
1025 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
1027 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
1029 2017-07-25 Torsten Duwe <duwe@suse.de>
1031 * common.opt: Introduce -fpatchable-function-entry
1032 command line option, and its variables function_entry_patch_area_size
1033 and function_entry_patch_area_start.
1034 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
1035 including a two-value parser.
1036 * target.def (print_patchable_function_entry): New target hook.
1037 * targhooks.h (default_print_patchable_function_entry): New function.
1038 * targhooks.c (default_print_patchable_function_entry): Likewise.
1039 * toplev.c (process_options): Switch off IPA-RA if
1040 patchable function entries are being generated.
1041 * varasm.c (assemble_start_function): Look at the
1042 patchable-function-entry command line switch and current
1043 function attributes and maybe generate NOP instructions by
1044 calling the print_patchable_function_entry hook.
1045 * doc/extend.texi: Document patchable_function_entry attribute.
1046 * doc/invoke.texi: Document -fpatchable_function_entry
1047 command line option.
1048 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
1050 * doc/tm.texi: Re-generate.
1052 2017-07-25 Jakub Jelinek <jakub@redhat.com>
1055 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
1056 TARGET_AVX512DQ rather than TARGET_AVX512BW.
1058 2017-07-25 Tamar Christina <tamar.christina@arm.com>
1060 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
1062 2017-07-25 Richard Biener <rguenther@suse.de>
1064 PR tree-optimization/81455
1065 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
1066 not walk in cycles when looking for guards.
1068 2017-07-25 Richard Biener <rguenther@suse.de>
1070 PR tree-optimization/81529
1071 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
1072 when optimizing backedge uses.
1074 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
1076 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
1078 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
1079 to dl_section_ref. On AIX, append an expression to subtract
1080 the size of the section length to dl_section_ref.
1082 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
1084 * configure.ac: If any of the config.* scripts fail, exit 1.
1085 * configure: Regenerate.
1087 2017-07-25 Richard Biener <rguenther@suse.de>
1090 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
1091 of immediate uses, be more verbose on errors.
1093 2017-07-25 Richard Biener <rguenther@suse.de>
1095 PR tree-optimization/81510
1096 * tree-vect-loop.c (vect_is_simple_reduction): When the
1097 reduction stmt is not inside the loop bail out.
1099 2017-07-25 Richard Biener <rguenther@suse.de>
1101 PR tree-optimization/81303
1102 * tree-vect-loop-manip.c (vect_loop_versioning): Build
1103 profitability check against LOOP_VINFO_NITERSM1.
1105 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
1107 * domwalk.c (cmp_bb_postorder): Simplify.
1108 (sort_bbs_postorder): New function. Use it...
1109 (dom_walker::walk): ...here to optimize common cases.
1111 2017-07-25 Martin Liska <mliska@suse.cz>
1114 * ipa-visibility.c (function_and_variable_visibility): Make the
1115 redirection just on target that supports aliasing.
1116 Fix GNU coding style.
1118 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
1121 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
1123 * config/arm/rtems.h: Likewise.
1124 * config/bfin/rtems.h: Likewise.
1125 * config/i386/rtemself.h: Likewise.
1126 * config/lm32/rtems.h: Likewise.
1127 * config/m32c/rtems.h: Likewise.
1128 * config/m68k/rtemself.h: Likewise.
1129 * config/microblaze/rtems.h: Likewise.
1130 * config/mips/rtems.h: Likewise.
1131 * config/moxie/rtems.h: Likewise.
1132 * config/nios2/rtems.h: Likewise.
1133 * config/powerpcspe/rtems.h: Likewise.
1134 * config/rs6000/rtems.h: Likewise.
1135 * config/rtems.h: Likewise.
1136 * config/sh/rtems.h: Likewise.
1137 * config/sh/rtemself.h: Likewise.
1138 * config/sparc/rtemself.h: Likewise.
1140 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
1143 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
1144 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
1145 * tree-ssa-structalias.c (alias_get_name): Same.
1147 2017-07-25 Bin Cheng <bin.cheng@arm.com>
1150 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
1151 instructions if no du chain is found.
1153 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
1155 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
1157 2017-07-25 Richard Biener <rguenther@suse.de>
1160 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
1163 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
1165 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
1167 (ISA_2_7_MASKS_SERVER): Likewise.
1168 (ISA_3_0_MASKS_IEEE): Likewise.
1169 (OTHER_P8_VECTOR_MASKS): Likewise.
1170 (OTHER_VSX_VECTOR_MASKS): Likewise.
1171 (POWERPC_MASKS): Likewise.
1172 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
1173 duplicate list of options.
1174 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
1175 explicit -mupper-regs options.
1176 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
1177 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
1178 alias for -mupper-regs-df.
1179 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
1180 (rs6000_init_hard_regno_mode_ok): Likewise.
1181 (rs6000_option_override_internal): Likewise.
1182 (rs6000_opt_masks): Likewise.
1183 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
1184 options in terms of whether -mvsx or -mpower8-vector was used.
1185 (TARGET_UPPER_REGS_DI): Likewise.
1186 (TARGET_UPPER_REGS_SF): Likewise.
1187 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
1188 -mupper-regs-* options.
1190 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
1192 * passes.c (emergency_dump_function): Print some empty lines and a
1193 header before the RTL dump.
1195 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
1197 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
1199 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
1202 * config/aarch64/aarch64.c (aarch64_classify_symbol):
1203 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
1205 2017-07-24 Carl Love <cel@us.ibm.com>
1207 * config/rs6000/rs6000-c.c: Add support for built-in functions
1208 vector float vec_extract_fp32_from_shorth (vector unsigned short);
1209 vector float vec_extract_fp32_from_shortl (vector unsigned short);
1210 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
1211 vec_extract_fp_from_shortl): Add defines for the two builtins.
1212 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
1213 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
1215 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
1216 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
1217 * doc/extend.texi: Update the built-in documentation file for the
1218 new built-in function.
1220 2017-07-24 Jakub Jelinek <jakub@redhat.com>
1223 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
1225 * doc/generic.texi: Likewise.
1226 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
1227 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
1229 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
1231 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
1232 (aarch64_mls_elt_merge<mode>): Likewise.
1234 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
1236 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
1237 having __cxa_atexit.
1239 2017-07-23 Michael Collison <michael.collison@arm.com>
1241 * config/arm/arm.c (arm_option_override): Deprecate
1242 use of -mstructure-size-boundary.
1243 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
1244 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
1246 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1249 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
1250 Reduce cost estimate for direct moves.
1252 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
1255 * config/i386/i386.c (ix86_option_override_internal): Disable
1256 BMI, BMI2 and TBM instructions for -m16.
1258 2017-07-21 Carl Love <cel@us.ibm.com>
1260 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1261 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1262 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1263 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1264 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1265 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
1266 VMULOSW): New enum "unspec" values.
1267 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
1268 altivec_vmulosw): New patterns.
1269 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1270 VMULOSW): Add definitions.
1272 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
1274 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
1275 (qdf24xx): Likewise.
1276 * config/aarch64/aarch64-options-extensions.def (rdma); New.
1277 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
1278 (AARCH64_FL_V8_1): Renumber.
1279 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
1280 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
1281 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
1282 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
1283 rdma to feature modifiers list.
1285 2017-07-21 Yury Gribov <tetra2005@gmail.com>
1288 * ipa-visibility (function_and_variable_visibility): Convert
1289 recursive PLT call to direct call if appropriate.
1291 2017-07-21 Andrew Pinski <apinski@cavium.com>
1293 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
1294 operand 1 to see if the types precision matches.
1295 * fold-const.c (operand_equal_p): Likewise.
1297 2017-07-21 Richard Biener <rguenther@suse.de>
1299 PR tree-optimization/81303
1300 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
1301 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
1302 (vect_peeling_hash_get_lowest_cost): Adjust.
1303 (vect_enhance_data_refs_alignment): Likewise. Use
1304 vect_get_peeling_costs_all_drs to compute the penalty for no
1305 peeling to match up costs.
1307 2017-07-21 Richard Biener <rguenther@suse.de>
1309 PR tree-optimization/81500
1310 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
1311 we didn't identify a reduction path.
1313 2017-07-21 Tom de Vries <tom@codesourcery.com>
1314 Cesar Philippidis <cesar@codesourcery.com>
1316 PR gcov-profile/81442
1317 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
1320 2017-07-21 Tom de Vries <tom@codesourcery.com>
1323 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
1325 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
1326 nvptx_override_options_after_change.
1328 2017-07-21 Ulrich Drepper <drepper@redhat.com>
1330 * dwarf2out.c (output_file_names): Avoid double testing for
1333 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
1335 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
1337 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
1339 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
1341 (try_crossjump_to_edge): Do not punt on partitioned functions.
1343 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
1345 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
1346 Put all BBs reachable only via paths crossing cold region to cold
1348 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
1350 2016-07-21 Richard Biener <rguenther@suse.de>
1352 PR tree-optimization/81303
1353 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
1354 into account prologue and epilogue iterations when raising
1355 min_profitable_iters to sth at least covering one vector iteration.
1357 2017-07-21 Tamar Christina <tamar.christina@arm.com>
1359 * config/arm/arm.c (arm_test_cpu_arch_dat):
1362 2017-07-20 Nathan Sidwell <nathan@acm.org>
1364 Remove TYPE_METHODS.
1365 * tree.h (TYPE_METHODS): Delete.
1366 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
1367 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
1368 (dbxout_type_methods): Scan TYPE_FIELDS.
1369 (dbxout_type): Don't check TYPE_METHODS here.
1370 * function.c (use_register_for_decl): Always ignore register for
1371 class types when not optimizing.
1372 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
1373 * tree.c (free_lang_data_in_type): Stitch out member functions and
1374 templates from TYPE_FIELDS.
1375 (build_distinct_type_copy, verify_type_variant,
1376 verify_type): Member fns are on TYPE_FIELDS.
1377 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
1378 * tree-pretty-print.c (dump_generic_node): Likewise.
1380 2017-07-20 Jakub Jelinek <jakub@redhat.com>
1383 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
1384 V2TImode and V4TImode.
1385 (ix86_expand_vector_extract): Likewise.
1386 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
1387 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
1388 (ssescalarmode): Handle V4TImode and V2TImode.
1389 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
1390 (*vec_extractv2ti, *vec_extractv4ti): New insns.
1391 (VEXTRACTI128_MODE): New mode iterator.
1392 (splitter for *vec_extractv?ti first element): New.
1393 (VEC_INIT_MODE): New mode iterator.
1394 (vec_init<mode>): Consolidate 3 expanders into one using
1395 VEC_INIT_MODE mode iterator.
1397 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
1399 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
1400 non_spilled_static_chain_regno_p.
1402 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
1404 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
1406 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
1408 * bb-reorder.c (connect_traces): Allow copying of blocks within
1411 2017-07-20 Richard Biener <rguenther@suse.de>
1413 * gimple.h (gimple_phi_result): Add gphi * overload.
1414 (gimple_phi_result_ptr): Likewise.
1415 (gimple_phi_arg): Likewise. Adjust index assert to only
1416 allow actual argument accesses rather than all slots available
1418 (gimple_phi_arg_def): Add gphi * overload.
1419 * tree-phinodes.c (make_phi_node): Initialize only actual
1421 (resize_phi_node): Clear memory not covered by old node,
1422 do not initialize excess argument slots.
1423 (reserve_phi_args_for_new_edge): Initialize new argument slot
1426 2017-07-20 Bin Cheng <bin.cheng@arm.com>
1428 PR tree-optimization/81388
1430 2016-07-21 Bin Cheng <bin.cheng@arm.com>
1432 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
1433 by removing computation of may_be_zero.
1435 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1436 Tom de Vries <tom@codesourcery.com>
1439 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
1440 when gimple level profile disagrees with what RTL expander did.
1442 2017-07-20 Richard Biener <rguenther@suse.de>
1444 PR tree-optimization/61171
1445 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
1446 (vect_analyze_stmt): Add slp instance parameter.
1447 (vectorizable_reduction): Likewise.
1448 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
1449 (vect_is_simple_reduction): Deal with chains not detected
1450 as SLP reduction chain, specifically not properly associated
1451 chains containing a mix of plus/minus.
1452 (get_reduction_op): Remove.
1453 (get_initial_defs_for_reduction): Simplify, pass in whether
1454 this is a reduction chain, pass in the SLP node for the PHIs.
1455 (vect_create_epilog_for_reduction): Get the SLP instance as
1457 (vectorizable_reduction): Get the SLP instance as arg.
1458 During analysis remember the SLP node with the PHIs in the
1459 instance. Simplify getting at the vectorized reduction PHIs.
1460 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
1461 through SLP instance.
1462 (vect_slp_analyze_operations): Likewise.
1463 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
1464 (vect_transform_stmt): Likewise.
1466 2017-07-20 Tom de Vries <tom@codesourcery.com>
1468 PR tree-optimization/81489
1469 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
1470 read of phi arg location to before loop that modifies phi.
1472 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
1474 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
1477 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1480 * except.c (execute): Fix ordering issue.
1482 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
1484 PR rtl-optimization/81423
1485 * combine.c (make_compound_operation_int): Don't try to optimize
1486 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
1488 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
1490 PR rtl-optimization/81423
1491 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
1492 with a constant that is -1 in the truncated to mode.
1494 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1496 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
1497 (determine_unlikely_bbs): ... here.
1498 * predict.h (propagate_unlikely_bbs_forward): Declare.
1499 * cfgexpand.c (pass_expand::execute): Use it.
1500 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
1502 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
1503 propagate_unlikely_bbs_forward.
1505 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1508 * except.c (maybe_add_nop_after_section_switch): New function.
1511 2017-07-19 Tom de Vries <tom@codesourcery.com>
1513 * gimple.h (gimple_phi_set_arg): Make assert more strict.
1515 2017-07-19 Tom de Vries <tom@codesourcery.com>
1517 * gimple.h (gimple_phi_arg): Make assert more strict.
1519 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
1521 * config.gcc (powerpc*-*-*): Add mmintrin.h.
1522 * config/rs6000/mmintrin.h: New file.
1523 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
1525 2017-07-19 Jakub Jelinek <jakub@redhat.com>
1527 PR tree-optimization/81346
1528 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
1530 2017-07-19 Tom de Vries <tom@codesourcery.com>
1532 * config/nvptx/nvptx.md (VECIM): Add V2DI.
1534 2017-07-19 Tom de Vries <tom@codesourcery.com>
1536 * config/nvptx/nvptx-modes.def: Add V2DImode.
1537 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
1538 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
1539 (nvptx_output_mov_insn): Handle lack of mov.b128.
1540 (nvptx_print_operand): Handle 'H' and 'L' codes.
1541 (nvptx_vector_mode_supported): Allow V2DImode.
1542 (nvptx_preferred_simd_mode): New function.
1543 (nvptx_data_alignment): New function.
1544 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
1545 nvptx_preferred_simd_mode.
1546 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
1548 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
1550 2017-07-19 Tom de Vries <tom@codesourcery.com>
1552 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
1553 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
1554 (nvptx_vector_mode_supported): New function. Allow V2SImode.
1555 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
1556 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
1557 (mov<VECIM>_insn): New define_insn.
1558 (define_expand "mov<VECIM>): New define_expand.
1560 2017-07-19 Tom de Vries <tom@codesourcery.com>
1562 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
1564 2017-07-19 Jakub Jelinek <jakub@redhat.com>
1566 PR tree-optimization/81346
1567 * fold-const.h (fold_div_compare, range_check_type): Declare.
1568 * fold-const.c (range_check_type): New function.
1569 (build_range_check): Use range_check_type.
1570 (fold_div_compare): No longer static, rewritten into
1571 a match.pd helper function.
1572 (fold_comparison): Don't call fold_div_compare here.
1573 * match.pd (X / C1 op C2): New optimization using fold_div_compare
1576 2017-07-19 Nathan Sidwell <nathan@acm.org>
1578 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
1579 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
1580 * tree.c (find_decls_types_r, verify_type): Use
1581 TYPE_{MIN,MAX}_VALUE_RAW.
1582 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
1583 (hash_tree): Likewise.
1584 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
1586 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
1589 2017-07-18 Tom de Vries <tom@codesourcery.com>
1592 * omp-expand.c (expand_omp_for_static_chunk): Handle
1593 equal-argument loop exit phi.
1595 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
1598 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
1599 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
1600 operand 2 predicate.
1601 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
1602 operand 2 predicate.
1603 (ror,rol -> rorx splitters): Use const_int_operand as
1604 operand 2 predicate.
1606 2017-06-18 Richard Biener <rguenther@suse.de>
1608 PR tree-optimization/81410
1609 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
1610 the gap in the ! slp_perm SLP case after each group.
1612 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1615 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
1618 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1621 * predict.c (set_even_probabilities): Cleanup; do not affect
1622 probabilities that are already known.
1623 (combine_predictions_for_bb): Call even when count is set.
1625 2017-07-18 Nathan Sidwell <nathan@acm.org>
1627 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
1630 2017-07-18 Bin Cheng <bin.cheng@arm.com>
1633 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
1634 optimization for loop niter analysis.
1636 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
1639 * config/avr/avr.c (avr_optimize_casesi): Don't use
1640 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
1642 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1644 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
1645 body_cost_vec from _vect_peel_extended_info.
1646 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
1647 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
1650 2017-07-18 Bin Cheng <bin.cheng@arm.com>
1652 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
1654 2017-07-18 Richard Biener <rguenther@suse.de>
1656 PR tree-optimization/80620
1657 PR tree-optimization/81403
1658 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
1659 info when re-using a VN table entry.
1661 2017-07-18 Richard Biener <rguenther@suse.de>
1663 PR tree-optimization/81418
1664 * tree-vect-loop.c (vectorizable_reduction): Properly compute
1665 vectype_in. Verify that with lane-reducing reduction operations
1666 we have a single def-use cycle.
1668 2017-07-17 Carl Love <cel@us.ibm.com>
1670 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
1672 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1673 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1674 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1675 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1676 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1677 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
1678 VMULOSW): New enum "unspec" values.
1679 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
1680 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
1681 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
1682 altivec_vmulosw): New patterns.
1683 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1684 VMULOSW): Add definitions.
1685 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
1687 * config/alpha/alpha.c: Include predict.h.
1689 2017-07-17 Yury Gribov <tetra2005@gmail.com>
1691 * tree-vrp.c (compare_assert_loc): Fix comparison function
1692 to return predictable results.
1694 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1696 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
1699 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
1700 * doc/invoke.texi (mexpand-adddi): Update text.
1702 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1704 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
1705 that also clobbers the CC register. The old expand code is moved
1707 (*arc_clzsi2): ... here.
1708 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
1709 the CC register. The old expand code is moved to ...
1710 (arc_ctzsi2): ... here.
1712 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1714 * config/arc/arc.opt (mindexed-loads): Use initial value
1715 TARGET_INDEXED_LOADS_DEFAULT.
1716 (mauto-modify-reg): Use initial value
1717 TARGET_AUTO_MODIFY_REG_DEFAULT.
1718 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
1719 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
1720 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
1721 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
1723 2017-07-17 Martin Liska <mliska@suse.cz>
1726 * opts.c (finish_options): Do not allow -fgnu-tm
1727 w/ -fsanitize={kernel-,}address. Say sorry.
1729 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1732 * tree-loop-distribution.c (classify_partition): Only assert on
1733 numer of iterations.
1734 (merge_dep_scc_partitions): Delete prameter. Update function call.
1735 (distribute_loop): Remove code handling loop with unknown niters.
1736 (pass_loop_distribution::execute): Skip loop with unknown niters.
1738 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1741 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
1742 function sort_partitions_by_post_order.
1744 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1746 PR tree-optimization/81374
1747 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
1748 the max index of basic blocks, rather than number of basic blocks.
1750 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1752 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
1754 (arc_legitimate_pic_operand_p): Likewise.
1755 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
1757 (arc_needs_pcl_p): Likewise.
1758 (arc_legitimate_pc_offset_p): Likewise.
1759 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
1760 function is also used in constrains.md.
1761 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
1762 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
1763 PLUS. Only return true/false in known cases, otherwise assert.
1764 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
1765 is already called in arc_legitimate_constant_p.
1766 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
1768 (LEGITIMATE_PIC_OPERAND_P): Use
1769 arc_raw_symbolic_reference_mentioned_p function.
1770 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
1775 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1776 Andrew Burgess <andrew.burgess@embecosm.com>
1778 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
1779 (arc_return_address_register): New function.
1780 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
1781 (arc_handle_fndecl_attribute): Add naked attribute.
1782 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
1783 (TARGET_WARN_FUNC_RETURN): Likewise.
1784 (arc_allocate_stack_slots_for_args): New function.
1785 (arc_warn_func_return): Likewise.
1786 (machine_function): Change type fn_type.
1787 (arc_compute_function_type): Consider new naked function type,
1788 change function return type.
1789 (arc_must_save_register): Adapt to handle new
1790 arc_compute_function_type's return type.
1791 (arc_expand_prologue): Likewise.
1792 (arc_expand_epilogue): Likewise.
1793 (arc_return_address_regs): Delete.
1794 (arc_return_address_register): New function.
1795 (arc_epilogue_uses): Use above function.
1796 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
1797 (arc_function_type): Change encoding, add naked type.
1798 (ARC_INTERRUPT_P): Change to handle the new encoding.
1799 (ARC_FAST_INTERRUPT_P): Likewise.
1800 (ARC_NORMAL_P): Define.
1801 (ARC_NAKED_P): Likewise.
1802 (arc_compute_function_type): Delete prototype.
1803 * config/arc/arc.md (in_ret_delay_slot): Use
1804 arc_return_address_register function.
1805 (simple_return): Likewise.
1806 (p_return_i): Likewise.
1808 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1810 PR tree-optimization/81428
1811 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
1812 can't be built for those types.
1814 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1816 Remove stuff dead since r239246.
1818 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
1819 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
1820 (avr_inform_devices): Remove dead stuff.
1822 2017-07-17 Tamar Christina <tamar.christina@arm.com>
1824 * config/arm/arm_neon.h: Fix softp typo.
1826 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1828 PR tree-optimization/81365
1829 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
1830 aggregate moves onto bb predecessor edges, make sure there are no
1831 loads that could alias the lhs in between the start of bb and the
1834 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1837 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
1838 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
1839 [LSHIFTRT, outer_code = TRUNCATE]: Same.
1841 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1843 PR tree-optimization/81396
1844 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
1845 (init_symbolic_number): Initialize it to 1.
1846 (perform_symbolic_merge): Add n_ops from both operands into the new
1848 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
1849 without base_addr as useless if they need more than one operation.
1850 (bswap_replace): Handle !bswap case for NULL base_addr.
1852 2017-07-17 Tom de Vries <tom@codesourcery.com>
1855 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
1858 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
1860 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
1861 conditional builtin define __FIX_LEON3FT_B2BST.
1863 2017-07-17 Daniel Cederman <cederman@gaisler.com>
1865 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
1866 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
1869 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1871 PR rtl-optimization/81424
1872 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
1873 to remove potential trapping from operands if -fnon-call-exceptions.
1875 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1877 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
1878 profile_proability for scalling.
1879 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
1881 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1883 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
1885 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1887 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
1888 fixpoint arithmetics.
1890 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1892 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
1893 fixpoint arithmetics.
1895 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1897 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
1898 fixpoint arithmetics.
1900 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1902 * profile-count.h (profile_probability::from_reg_br_prob_note,
1903 profile_probability::to_reg_br_prob_note): New functions.
1904 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
1905 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
1906 * predict.c (probability_reliable_p): Update.
1907 (edge_probability_reliable_p): Update.
1908 (br_prob_note_reliable_p): Update.
1909 (invert_br_probabilities): Update.
1910 (add_reg_br_prob_note): New function.
1911 (combine_predictions_for_insn): Update.
1912 * asan.c (asan_clear_shadow): Update.
1913 * cfgbuild.c (compute_outgoing_frequencies): Update.
1914 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
1915 (update_br_prob_note): Update.
1916 (rtl_verify_edges): Update.
1917 (purge_dead_edges): Update.
1918 (fixup_reorder_chain): Update.
1919 * emit-rtl.c (try_split): Update.
1920 * ifcvt.c (cond_exec_process_insns): Update.
1921 (cond_exec_process_if_block): Update.
1922 (dead_or_predicable): Update.
1923 * internal-fn.c (expand_addsub_overflow): Update.
1924 (expand_neg_overflow): Update.
1925 (expand_mul_overflow): Update.
1926 * loop-doloop.c (doloop_modify): Update.
1927 * loop-unroll.c (compare_and_jump_seq): Update.
1928 * optabs.c (emit_cmp_and_jump_insn_1): Update.
1929 * predict.h: Update.
1930 * reorg.c (mostly_true_jump): Update.
1932 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
1933 * config/alpha/alpha.c (emit_unlikely_jump): Update.
1934 * config/arc/arc.c: (emit_unlikely_jump): Update.
1935 * config/arm/arm.c: (emit_unlikely_jump): Update.
1936 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
1937 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
1938 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
1939 (ix86_print_operand): Update.
1940 (ix86_split_fp_branch): Update.
1941 (predict_jump): Update.
1942 * config/ia64/ia64.c (ia64_print_operand): Update.
1943 * config/mmix/mmix.c (mmix_print_operand): Update.
1944 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
1945 (rs6000_expand_split_stack_prologue): Update.
1946 * config/rs6000/rs6000.c: Update.
1947 * config/s390/s390.c (s390_expand_vec_strlen): Update.
1948 (s390_expand_vec_movstr): Update.
1949 (s390_expand_cs_tdsi): Update.
1950 (s390_expand_split_stack_prologue): Update.
1951 * config/sh/sh.c (sh_print_operand): Update.
1952 (expand_cbranchsi4): Update.
1953 (expand_cbranchdi4): Update.
1954 * config/sparc/sparc.c (output_v9branch): Update.
1955 * config/spu/spu.c (get_branch_target): Update.
1956 (ea_load_store_inline): Update.
1957 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
1958 * config/tilepro/tilepro.c: Update.
1960 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1962 * gimplify.c (mostly_copy_tree_r): Revert latest change.
1963 (gimplify_save_expr): Likewise.
1965 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1967 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
1969 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1971 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
1973 * timevar.def (TV_IPA_FNSUMMARY): Define.
1975 2017-07-16 Daniel Cederman <cederman@gaisler.com>
1977 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
1978 to back store errata sensitive sequence from being generated.
1979 (sqrtdf2_fix): Likewise.
1981 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1983 * tree-ssa-threadupdate.c (compute_path_counts,
1984 update_joiner_offpath_counts): Use profile_probability.
1986 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
1989 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1991 * config/arm/arm-c.c (arm_cpu_builtins): Define
1992 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
1994 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
1996 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1997 array entries to represent __ieee128 versions of the
1998 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
1999 scalar_extract_sig, and scalar_insert_exp built-in functions.
2000 (altivec_resolve_overloaded_builtin): Add special case handling
2001 for the __builtin_scalar_insert_exp function, as represented by
2002 the P9V_BUILTIN_VEC_VSIEDP constant.
2003 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
2004 exponent support for __ieee128 argument.
2005 (VSESQP): Add scalar extract signature support for __ieee128
2007 (VSTDCNQP): Add scalar test negative support for __ieee128
2009 (VSIEQP): Add scalar insert exponent support for __int128 argument
2010 with __ieee128 result.
2011 (VSIEQPF): Add scalar insert exponent support for __ieee128
2012 argument with __ieee128 result.
2013 (VSTDCQP): Add scalar test data class support for __ieee128
2015 (VSTDCNQP): Add overload support for scalar test negative with
2017 (VSTDCQP): Add overload support for scalar test data class
2019 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
2021 (UNSPEC_VSX_SIEXPQP): New constant.
2022 (xsxexpqp): New insn for VSX scalar extract exponent quad
2024 (xsxsigqp): New insn for VSX scalar extract significand quad
2026 (xsiexpqpf): New insn for VSX scalar insert exponent quad
2027 precision with floating point argument.
2028 (xststdcqp): New expand for VSX scalar test data class quad
2030 (xststdcnegqp): New expand for VSX scalar test negative quad
2032 (xststdcqp): New insn to match expansions for VSX scalar test data
2033 class quad precision and VSX scalar test negative quad precision.
2034 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
2035 special case operand checking to enforce that second operand of
2036 VSX scalar test data class with quad precision argument is a 7-bit
2038 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
2039 prototypes and descriptions of __ieee128 versions of
2040 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
2041 scalar_test_data_class, and scalar_test_neg built-in functions.
2043 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2045 PR tree-optimization/81162
2046 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
2047 replace a negate with an add.
2049 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
2051 * doc/invoke.texi (arm/-mcpu): Document +crypto.
2053 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
2055 * config/arm/arm-c.c (arm_cpu_builtins): Define
2056 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
2058 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
2060 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
2061 (armv8-r): Set ARM Cortex-R52 as default CPU.
2062 * config/arm/arm-tables.opt: Regenerate.
2063 * config/arm/arm-tune.md: Regenerate.
2064 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
2066 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
2067 extension for -mcpu=cortex-r52.
2069 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
2071 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
2072 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
2073 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
2074 (fp-armv8): Define it as FP_ARMv8 only.
2075 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
2076 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
2078 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
2079 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
2080 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
2081 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
2082 than TARGET_FPU_ARMV8.
2083 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
2084 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
2085 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
2087 * config/arm/neon.md (neon_vrint): Likewise.
2088 (neon_vcvt): Likewise.
2089 (neon_<fmaxmin_op><mode>): Likewise.
2090 (<fmaxmin><mode>3): Likewise.
2091 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
2092 * config/arm/predicates.md (arm_cond_move_operator): Check against
2093 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
2095 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
2097 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
2100 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2102 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
2103 loop in comment with memset.
2105 2017-07-14 Martin Liska <mliska@suse.cz>
2107 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
2108 * dwarf2out.c (is_java): Remove the function.
2109 (output_pubname): Remove usage of the function.
2110 (lower_bound_default): Remove usage of DW_LANG_Java.
2111 (gen_compile_unit_die): Likewise.
2112 * gcc.c: Remove compiler defaults for .java and .zip files.
2113 * gimple-expr.c (remove_suffix): Change as there's no longer
2114 extension than 4-letter one.
2115 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
2116 (gimplify_save_expr): Likewise.
2117 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
2118 as it's possible even for other languages than Java.
2119 * langhooks.h (struct lang_hooks): Remove Java from a comment.
2120 * lto-opts.c (lto_write_options): Remove reference to Java.
2121 * opts.c (strip_off_ending): Update file extension handling.
2122 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
2123 * tree-eh.c (lower_resx): Likewise.
2124 * tree.c (free_lang_data_in_type): Remove dead code.
2125 (find_decls_types_r): Likewise.
2126 (build_common_builtin_nodes): Remove Java from a comment.
2127 (verify_type): Remove dead code.
2128 * varasm.c (assemble_external): Remove Java from a comment.
2130 2017-07-14 Martin Liska <mliska@suse.cz>
2132 * opts.c (finish_options): Add quotes.
2133 (common_handle_option): Likewise.
2135 2017-07-14 Martin Liska <mliska@suse.cz>
2137 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
2138 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
2140 * dwarf2out.c (lower_bound_default): Do not handle
2142 (gen_compile_unit_die): Likewise.
2143 * gcc.c: Remove default extension binding for GNU Pascal.
2144 * stmt.c: Remove Pascal language from a comment.
2145 * xcoffout.c: Likewise.
2147 2017-07-13 David Malcolm <dmalcolm@redhat.com>
2150 * diagnostic-show-locus.c (fixit_cmp): New function.
2151 (layout::layout): Sort m_fixit_hints.
2152 (column_range::column_range): Assert that the values are valid.
2153 (struct char_span): New struct.
2154 (correction::overwrite): New method.
2155 (struct source_line): New struct.
2156 (line_corrections::add_hint): Add assertions. Reimplement memcpy
2157 calls in terms of classes source_line and char_span, and
2158 correction::overwrite.
2159 (selftest::test_overlapped_fixit_printing_2): New function.
2160 (selftest::diagnostic_show_locus_c_tests): Call it.
2162 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2164 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
2165 early if there is no lhs.
2167 2017-07-13 Martin Liska <mliska@suse.cz>
2169 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
2170 (gen_reference_type_die): Likewise.
2171 * stor-layout.c: Remove Pascal-related comment.
2173 2017-07-13 Martin Liska <mliska@suse.cz>
2175 * opts.c (finish_options): Add quotes to error messages.
2176 (parse_sanitizer_options): Likewise.
2178 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
2180 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
2182 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
2184 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
2186 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
2188 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
2190 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
2192 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
2195 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
2196 provides the hardware capability bits, define the macro
2197 __BUILTIN_CPU_SUPPORTS__.
2198 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
2199 if GLIBC does not provide the hardware capability bits. Add a
2200 gcc_unreachable call if the built-in cpu function is neither
2201 __builtin_cpu_is nor __builtin_cpu_supports.
2202 (rs6000_get_function_versions_dispatcher): Change the warning
2203 that an old GLIBC is used which does not export the capability
2204 bits to be an error.
2205 * doc/extend.texi (target_clones attribute): Document the
2206 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
2207 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
2208 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
2209 the macros defined by GCC if the newer GLIBC is available.
2211 2017-07-12 Jeff Law <law@redhat.com>
2213 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
2214 remaining includes slightly.
2215 * config/riscv/riscv-builtins.c: Include profile-count.h.
2217 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
2220 * config/avr/avr.c (avr_set_current_function): In diagnostic
2221 messages: Quote keywords and (parts of) identifiers.
2222 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
2225 2017-07-12 Carl Love <cel@us.ibm.com>
2227 * config/rs6000/rs6000-c.c: Add support for built-in functions
2228 vector bool char vec_revb (vector bool char);
2229 vector bool short vec_revb (vector short char);
2230 vector bool int vec_revb (vector bool int);
2231 vector bool long long vec_revb (vector bool long long);
2232 * doc/extend.texi: Update the built-in documentation file for the
2233 new built-in functions.
2235 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2237 * config/s390/s390.md: Remove movcc splitter.
2239 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2241 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
2242 load/store on condition.
2244 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
2247 * config/avr/avr.c (avr_encode_section_info)
2248 [progmem && !TREE_READONLY]: Error if progmem object needs
2251 2017-07-11 Michael Collison <michael.collison@arm.com>
2253 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
2256 2017-07-11 Carl Love <cel@us.ibm.com>
2258 * config/rs6000/rs6000-c.c: Add support for builtins
2259 vector unsigned int vec_parity_lsbb (vector signed int);
2260 vector unsigned int vec_parity_lsbb (vector unsigned int);
2261 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
2262 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
2263 vector unsigned long long vec_parity_lsbb (vector signed long long);
2264 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
2265 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
2266 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
2267 * doc/extend.texi: Update the built-in documentation file for the
2268 new built-in functions.
2270 2017-07-11 David Malcolm <dmalcolm@redhat.com>
2272 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
2273 (layout::m_primary_loc): New field.
2274 (layout::layout): Initialize new field. Move location filtering
2275 logic from here to...
2276 (layout::maybe_add_location_range): ...this new method. Add
2277 support for filtering to just the lines already specified by other
2279 (layout::will_show_line_p): New method.
2280 (gcc_rich_location::add_location_if_nearby): New method.
2281 (selftest::test_add_location_if_nearby): New test function.
2282 (selftest::diagnostic_show_locus_c_tests): Call it.
2283 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
2286 2017-07-11 Tom de Vries <tom@codesourcery.com>
2288 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
2289 (bb_first_real_insn): New function.
2290 (nvptx_single): Add extra initialization of broadcasted condition
2293 2017-07-11 Nathan Sidwell <nathan@acm.org>
2295 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
2297 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
2299 * doc/extend.texi (AVR Function Attributes): Remove weblink to
2300 Binutils doc as TEXI will mess them up.
2301 * doc/invoke.texi (AVR Options): Same here.
2303 2017-07-11 Daniel Cederman <cederman@gaisler.com>
2305 * config/sparc/sparc.opt (mfix-ut700): New option.
2306 (mfix-gr712rc): Likewise.
2307 (sparc_fix_b2bst): New variable.
2308 * doc/invoke.texi (SPARC options): Document them.
2309 (ARM options): Fix warnings.
2310 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
2311 instructions to prevent sequences that can trigger the store-store
2312 errata for certain LEON3FT processors.
2313 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
2314 (sparc_option_override): Set sparc_fix_b2bst appropriately.
2315 * config/sparc/sparc.md (fix_b2bst): New attribute.
2316 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
2318 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
2321 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
2323 (*rsqrtsf2_sse): Ditto.
2325 (div<mode>3): Macroize insn from divdf3 and divsf3
2326 using MODEF mode iterator.
2328 2017-07-10 Martin Sebor <msebor@redhat.com>
2330 PR tree-optimization/80397
2331 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
2332 instead of testing for equality to INTEGER_TYPE.
2334 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
2336 * config.gcc: Remove uclibc from arc target spec.
2338 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
2340 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
2342 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
2345 * lto-wrapper.c (remove_option): New function.
2346 (merge_and_complain): Merge PIC/PIE options more realistically.
2348 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
2350 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
2354 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
2355 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
2356 * config.in: Regenerate.
2357 * configure: Regenerate.
2358 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
2359 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
2360 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
2361 (TARGET_GASISR_PROLOGUES): ...target mask.
2362 * common/config/avr/avr-common.c
2363 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
2364 Set -mgas-isr-prologues.
2365 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
2366 INSERT_PASS_BEFORE for it.
2367 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
2368 * config/avr/avr.c (avr_option_override)
2369 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
2370 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
2371 (avr_attribute_table) <no_gccisr>: Add new function attribute.
2372 (avr_set_current_function) <is_no_gccisr>: Init machine field.
2373 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
2375 (make_avr_pass_pre_proep): New function.
2376 (emit_push_sfr) <treg>: Add argument to function and use it
2378 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
2379 and set machine->gasisr.yes.
2380 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
2381 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
2382 __gcc_isr.n_pushed to .L__stack_usage.
2383 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
2384 (avr_asm_final_postscan_insn): ...this new static function.
2385 * config/avr/avr.h (machine_function)
2386 <is_no_gccisr, use_L__stack_usage>: New fields.
2387 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
2388 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
2389 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
2390 (gasisr, *gasisr): New expander and insn.
2391 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
2392 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
2393 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
2395 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
2397 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
2400 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
2402 Move jump-tables out of .text again.
2405 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
2406 (ASM_OUTPUT_ADDR_VEC): New function.
2407 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
2408 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
2409 INSN_ADDRESSes as asm comment.
2410 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
2411 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
2412 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
2413 * config/avr/avr.md (*tablejump): Adjust comment.
2414 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
2415 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
2417 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
2418 (avr_output_addr_vec): New proto.
2419 (avr_log_t) <insn_addresses>: New field.
2421 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
2424 * config/i386/i386.c (ix86_function_arg_advance): Set
2425 outgoing_args_on_stack to true if there are outgoing arguments
2427 (ix86_function_arg): Likewise.
2428 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
2429 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
2430 * config/i386/i386.h (machine_function): Add
2431 outgoing_args_on_stack.
2433 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
2435 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
2437 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
2439 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
2441 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
2442 (REAL_H): Remove $(MACHMODE_H).
2443 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
2445 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
2446 $(MACHMODE_H) and double-int.h.
2447 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
2449 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
2452 2017-07-07 Andrew Pinski <apinski@cavium.com>
2454 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
2455 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
2457 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
2459 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
2460 Add warning if GCC was not configured to link against a GLIBC that
2461 exports the hardware capability bits.
2462 (make_resolver_func): Make resolver function private and not a
2463 COMDAT function. Create the name with clone_function_name instead
2464 of make_unique_name.
2467 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
2468 correct operand in doing the split.
2470 2017-07-07 Carl Love <cel@us.ibm.com>
2472 * config/rs6000/rs6000-c: Add support for built-in function
2473 vector unsigned short vec_pack_to_short_fp32 (vector float,
2475 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
2476 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
2477 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
2478 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
2479 (convert_4f32_8i16): Add define_expand.
2480 * doc/extend.texi: Update the built-in documentation file for the
2481 new built-in function.
2483 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2485 * config/sparc/m8.md: New file.
2486 * config/sparc/sparc.md: Include m8.md.
2488 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2490 * config/sparc/sparc.opt: New option -mvis4b.
2491 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
2492 (sparc_option_override): Handle VIS4B.
2493 (enum sparc_builtins): Define
2494 SPARC_BUILTIN_DICTUNPACK{8,16,32},
2495 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
2496 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
2497 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
2498 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
2499 (check_constant_argument): New function.
2500 (sparc_vis_init_builtins): Define builtins
2501 __builtin_vis_dictunpack{8,16,32},
2502 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
2503 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
2504 __builtin_vis_fpcmpde{8,16,32}shl and
2505 __builtin_vis_fpcmpur{8,16,32}shl.
2506 (sparc_expand_builtin): Check that the constant operands to
2507 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
2508 constant and in range.
2509 * config/sparc/sparc-c.c (sparc_target_macros): Handle
2511 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
2512 (SPARC_IMM5_P): Likewise.
2513 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
2514 (enabled): Handle vis4b.
2515 (UNSPEC_DICTUNPACK): New unspec.
2516 (UNSPEC_FPCMPSHL): Likewise.
2517 (UNSPEC_FPUCMPSHL): Likewise.
2518 (UNSPEC_FPCMPDESHL): Likewise.
2519 (UNSPEC_FPCMPURSHL): Likewise.
2520 (cpu_feature): New CPU feature `vis4b'.
2521 (dictunpack{8,16,32}): New insns.
2522 (FPCSMODE): New mode iterator.
2523 (fpcscond): New code iterator.
2524 (fpcsucond): Likewise.
2525 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
2526 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
2527 (fpcmpde{8,16,32}{si,di}shl): Likewise.
2528 (fpcmpur{8,16,32}{si,di}shl): Likewise.
2529 * config/sparc/constraints.md: Define constraints `q' for unsigned
2530 2-bit integer constants and `t' for unsigned 5-bit integer
2532 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
2534 (imm5_operand_dictunpack16): Likewise.
2535 (imm5_operand_dictunpack32): Likewise.
2536 (imm2_operand): Likewise.
2537 * doc/invoke.texi (SPARC Options): Document -mvis4b.
2538 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
2539 ditunpack* and fpcmp*shl builtins.
2541 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2543 * config.gcc: Handle m8 in --with-{cpu,tune} options.
2544 * config.in: Add HAVE_AS_SPARC6 define.
2545 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
2547 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
2549 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
2550 (CPP_CPU_SPEC): Handle m8.
2551 (ASM_CPU_SPEC): Likewise.
2552 * config/sparc/sparc-opts.h (enum processor_type): Add
2554 * config/sparc/sparc.c (m8_costs): New struct.
2555 (sparc_option_override): Handle TARGET_CPU_m8.
2556 (sparc32_initialize_trampoline): Likewise.
2557 (sparc64_initialize_trampoline): Likewise.
2558 (sparc_issue_rate): Likewise.
2559 (sparc_register_move_cost): Likewise.
2560 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
2561 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
2562 (ASM_CPU64_DEFAULT_SPEC): Likewise.
2563 (CPP_CPU_SPEC): Handle M8.
2564 (ASM_CPU_SPEC): Likewise.
2565 (AS_M8_FLAG): Define.
2566 * config/sparc/sparc.md: Add m8 to the cpu attribute.
2567 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
2568 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
2570 * configure: Regenerate.
2571 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
2574 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2576 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
2578 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
2579 ("*movdi_insn_sp32"): Do not set v3pipe.
2580 ("*movsi_insn"): Likewise.
2581 ("*movdi_insn_sp64"): Likewise.
2582 ("*movsf_insn"): Likewise.
2583 ("*movdf_insn_sp32"): Likewise.
2584 ("*movdf_insn_sp64"): Likewise.
2585 ("*zero_extendsidi2_insn_sp64"): Likewise.
2586 ("*sign_extendsidi2_insn"): Likewise.
2587 ("*mov<VM32:mode>_insn"): Likewise.
2588 ("*mov<VM64:mode>_insn_sp64"): Likewise.
2589 ("*mov<VM64:mode>_insn_sp32"): Likewise.
2590 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
2591 ("<vlop:code><VL:mode>3"): Likewise.
2592 ("*not_<vlop:code><VL:mode>3"): Likewise.
2593 ("*nand<VL:mode>_vis"): Likewise.
2594 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
2595 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
2596 ("one_cmpl<VL:mode>2"): Likewise.
2597 ("faligndata<VM64:mode>_vis"): Likewise.
2598 ("alignaddrsi_vis"): Likewise.
2599 ("alignaddrdi_vis"): Likweise.
2600 ("alignaddrlsi_vis"): Likewise.
2601 ("alignaddrldi_vis"): Likewise.
2602 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
2603 ("bmaskdi_vis"): Likewise.
2604 ("bmasksi_vis"): Likewise.
2605 ("bshuffle<VM64:mode>_vis"): Likewise.
2606 ("cmask8<P:mode>_vis"): Likewise.
2607 ("cmask16<P:mode>_vis"): Likewise.
2608 ("cmask32<P:mode>_vis"): Likewise.
2609 ("pdistn<P:mode>_vis"): Likewise.
2610 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
2612 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2614 * config/sparc/sparc.md ("subtype"): New insn attribute.
2615 ("*wrgsr_sp64"): Set insn subtype.
2616 ("*rdgsr_sp64"): Likewise.
2617 ("alignaddrsi_vis"): Likewise.
2618 ("alignaddrdi_vis"): Likewise.
2619 ("alignaddrlsi_vis"): Likewise.
2620 ("alignaddrldi_vis"): Likewise.
2621 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
2622 ("fexpand_vis"): Likewise.
2623 ("fpmerge_vis"): Likewise.
2624 ("faligndata<VM64:mode>_vis"): Likewise.
2625 ("bshuffle<VM64:mode>_vis"): Likewise.
2626 ("cmask8<P:mode>_vis"): Likewise.
2627 ("cmask16<P:mode>_vis"): Likewise.
2628 ("cmask32<P:mode>_vis"): Likewise.
2629 ("fchksm16_vis"): Likewise.
2630 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
2631 ("fmean16_vis"): Likewise.
2632 ("fp<plusminus_insn>64_vis"): Likewise.
2633 ("<plusminus_insn>v8qi3"): Likewise.
2634 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
2635 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
2636 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
2637 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
2638 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
2639 ("*movqi_insn"): Likewise.
2640 ("*movhi_insn"): Likewise.
2641 ("*movsi_insn"): Likewise.
2642 ("movsi_pic_gotdata_op"): Likewise.
2643 ("*movdi_insn_sp32"): Likewise.
2644 ("*movdi_insn_sp64"): Likewise.
2645 ("movdi_pic_gotdata_op"): Likewise.
2646 ("*movsf_insn"): Likewise.
2647 ("*movdf_insn_sp32"): Likewise.
2648 ("*movdf_insn_sp64"): Likewise.
2649 ("*zero_extendhisi2_insn"): Likewise.
2650 ("*zero_extendqihi2_insn"): Likewise.
2651 ("*zero_extendqisi2_insn"): Likewise.
2652 ("*zero_extendqidi2_insn"): Likewise.
2653 ("*zero_extendhidi2_insn"): Likewise.
2654 ("*zero_extendsidi2_insn_sp64"): Likewise.
2655 ("ldfsr"): Likewise.
2656 ("prefetch_64"): Likewise.
2657 ("prefetch_32"): Likewise.
2658 ("tie_ld32"): Likewise.
2659 ("tie_ld64"): Likewise.
2660 ("*tldo_ldub_sp32"): Likewise.
2661 ("*tldo_ldub1_sp32"): Likewise.
2662 ("*tldo_ldub2_sp32"): Likewise.
2663 ("*tldo_ldub_sp64"): Likewise.
2664 ("*tldo_ldub1_sp64"): Likewise.
2665 ("*tldo_ldub2_sp64"): Likewise.
2666 ("*tldo_ldub3_sp64"): Likewise.
2667 ("*tldo_lduh_sp32"): Likewise.
2668 ("*tldo_lduh1_sp32"): Likewise.
2669 ("*tldo_lduh_sp64"): Likewise.
2670 ("*tldo_lduh1_sp64"): Likewise.
2671 ("*tldo_lduh2_sp64"): Likewise.
2672 ("*tldo_lduw_sp32"): Likewise.
2673 ("*tldo_lduw_sp64"): Likewise.
2674 ("*tldo_lduw1_sp64"): Likewise.
2675 ("*tldo_ldx_sp64"): Likewise.
2676 ("*mov<VM32:mode>_insn"): Likewise.
2677 ("*mov<VM64:mode>_insn_sp64"): Likewise.
2678 ("*mov<VM64:mode>_insn_sp32"): Likewise.
2680 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2682 * config/sparc/sparc.md ("type"): New insn type viscmp.
2683 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
2685 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
2686 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
2687 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
2688 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
2690 ("n7_vis_logical_11cycle"): Likewise.
2691 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
2692 * config/sparc/niagara2.md ("niag3_vis": Likewise.
2693 * config/sparc/niagara.md ("niag_vis"): Likewise.
2694 * config/sparc/ultra3.md ("us3_fga"): Likewise.
2695 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
2697 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2699 * config/sparc/sparc.md: New instruction type `bmask'.
2700 (bmaskdi_vis): Use the `bmask' type.
2701 (bmasksi_vis): Likewise.
2702 * config/sparc/ultra3.md (us3_array): Likewise.
2703 * config/sparc/niagara7.md (n7_array): Likewise.
2704 * config/sparc/niagara4.md (n4_array): Likewise.
2705 * config/sparc/niagara2.md (niag2_vis): Likewise.
2706 (niag3_vis): Likewise.
2707 * config/sparc/niagara.md (niag_vis): Likewise.
2709 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2711 * ipa-comdats.c: Remove optimize check from gate.
2712 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
2713 for functions not optimized.
2714 (ipa_fn_summary_read): Skip optimize check.
2715 (ipa_fn_summary_write): Likewise.
2716 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
2718 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
2720 (can_inline_edge_p): Check flag_pcc_struct_return for match.
2721 (check_callers): Give up on caller which is not optimized.
2722 (inline_small_functions): Likewise.
2723 (ipa_inline): Do not give up when not optimizing.
2724 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
2725 away unoptimizes cdtors.
2726 (whole_program_function_and_variable_visibility): Do
2727 ipa_discover_readonly_nonaddressable_vars in LTO mode.
2728 * ipa.c (process_references): Do not check optimize.
2729 (symbol_table::remove_unreachable_nodes): Update optimize check.
2730 (set_writeonly_bit): Update optimize check.
2731 (pass_ipa_cdtor_merge::gate): Do not check optimize.
2732 (pass_ipa_single_use::gate): Remove.
2734 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2736 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
2737 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
2738 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
2739 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
2740 permute_load, permute_store, adjust_extract, adjust_splat,
2741 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
2742 replace_swap_with_copy, dump_swap_insn_table,
2743 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
2744 recombine_lvx_pattern, recombine_stvx_pattern,
2745 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
2746 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
2747 to file rs6000-p8swap.c.
2748 * config/rs6000/rs6000-p8swap.c: New file.
2749 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
2750 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
2751 and rs6000*-*-* targets.
2753 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2755 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
2757 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2759 * lto-wrapper.c (merge_and_complain): Do not merge
2760 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
2761 fsigned_zeros, ftrapping_math, fwrapv.
2762 (append_compiler_options): Do not track these options.
2763 (append_linker_options): Likewie
2765 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2767 * cgraphunit.c (cgraph_node::finalize_function): When
2768 !flag_toplevel_reorde set no_reorder flag.
2769 (varpool_node::finalize_decl): Likewise.
2770 (symbol_table::compile): Drop no toplevel reorder path.
2772 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2774 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
2775 edges; zero probability is not better than uninitialized.
2777 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2779 * asan.h (asan_sanitize_allocas_p): Declare.
2780 * asan.c (asan_sanitize_allocas_p): New function.
2781 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
2782 (handle_builtin_alloca): Likewise.
2783 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
2784 if !asan_sanitize_allocas_p.
2785 * params.def (asan-instrument-allocas): Add new option.
2786 * params.h (ASAN_PROTECT_ALLOCAS): Define.
2787 * opts.c (common_handle_option): Disable allocas sanitization for
2790 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2792 * asan.c: Include gimple-fold.h.
2793 (get_last_alloca_addr): New function.
2794 (handle_builtin_stackrestore): Likewise.
2795 (handle_builtin_alloca): Likewise.
2796 (asan_emit_allocas_unpoison): Likewise.
2797 (get_mem_refs_of_builtin_call): Add new parameter, remove const
2798 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
2799 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
2800 (instrument_builtin_call): Pass gimple iterator to
2801 get_mem_refs_of_builtin_call.
2802 (last_alloca_addr): New global.
2803 * asan.h (asan_emit_allocas_unpoison): Declare.
2804 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
2805 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
2806 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
2807 if function calls alloca.
2808 * gimple-fold.c (replace_call_with_value): Remove static keyword.
2809 * gimple-fold.h (replace_call_with_value): Declare.
2810 * internal-fn.c: Include asan.h.
2811 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
2812 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
2814 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2816 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
2817 (C_SELFTEST_FLAGS): New.
2818 (CPP_SELFTEST_FLAGS): New.
2819 (SELFTEST_DEPS): New, from deps of s-selftest.
2820 (C_SELFTEST_DEPS): New, from deps of s-selftest.
2821 (CPP_SELFTEST_DEPS): New.
2822 (selftest): Add dependency on s-selftest-c++.
2823 (s-selftest): Rename to...
2824 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
2825 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
2826 than SELFTEST_FLAGS.
2827 (selftest-gdb): Rename to...
2828 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
2830 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
2831 (selftest-valgrind): Rename to...
2832 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
2834 (selftest-valgrind): Reintroduce as an alias for
2835 selftest-c-valgrind.
2836 (s-selftest-c++): New.
2837 (selftest-c++-gdb): New.
2838 (selftest-c++-valgrind): New.
2840 2017-07-06 Olivier Hainque <hainque@adacore.com>
2842 * gcc.c (process_command): When deciding if undefined variables
2843 should be ignored when processing specs, accept "gcc -v" as well.
2845 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2847 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
2848 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
2850 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
2852 * config/arm/arm-cpus.in (armv8-r): Add new entry.
2853 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
2854 * config/arm/arm-tables.opt: Regenerate.
2855 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
2857 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
2859 2017-07-06 Carl Love <cel@us.ibm.com>
2861 * ChangeLog: Clean up from mid air collision
2863 2017-07-06 Carl Love <cel@us.ibm.com>
2865 * config/rs6000/rs6000-c.c: Add support for built-in functions
2866 vector signed int vec_subc (vector signed int, vector signed int);
2867 vector signed __int128 vec_subc (vector signed __int128,
2868 vector signed __int128);
2869 vector unsigned __int128 vec_subc (vector unsigned __int128,
2870 vector unsigned __int128);
2871 vector signed int vec_sube (vector signed int, vector signed int,
2873 vector unsigned int vec_sube (vector unsigned int,
2874 vector unsigned int,
2875 vector unsigned int);
2876 vector signed __int128 vec_sube (vector signed __int128,
2877 vector signed __int128,
2878 vector signed__int128);
2879 vector unsigned __int128 vec_sube (vector unsigned __int128,
2880 vector unsigned __int128,
2881 vector unsigned __int128);
2882 vector signed int vec_subec (vector signed int, vector signed int,
2884 vector unsigned int vec_subec (vector unsigned int,
2885 vector unsigned int,
2886 vector unsigned int);
2887 vector signed __int128 vec_subec (vector signed __int128,
2888 vector signed __int128,
2889 vector signed__int128);
2890 vector unsigned __int128 vec_subec (vector unsigned __int128,
2891 vector unsigned __int128,
2892 vector unsigned __int128);
2893 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
2894 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
2895 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
2896 BU_ALTIVEC_OVERLOAD_X definitions.
2897 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
2898 * doc/extend.texi: Update the built-in documentation file for the new
2901 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2904 * diagnostic-show-locus.c (layout::layout): Use start and finish
2905 spelling location for the start and finish of each range.
2906 * genmatch.c (linemap_client_expand_location_to_spelling_point):
2907 Add unused aspect param.
2908 * input.c (expand_location_1): Add "aspect" param, and use it
2909 to access the correct part of the location.
2910 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
2912 (expand_location_to_spelling_point): Likewise.
2913 (linemap_client_expand_location_to_spelling_point): Add "aspect"
2914 param, and pass it to expand_location_1.
2916 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
2918 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
2919 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
2920 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
2921 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
2922 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
2923 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
2924 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
2925 _mm_maskz_getmant_ss): New intrinsics.
2926 (__builtin_ia32_getexpss128_mask): Changed to ...
2927 __builtin_ia32_getexpss128_round ... this.
2928 (__builtin_ia32_getexpsd128_mask): Changed to ...
2929 __builtin_ia32_getexpsd128_round ... this.
2930 * config/i386/i386-builtin-types.def
2931 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
2932 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
2933 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
2934 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
2935 __builtin_ia32_getmantss_mask_round): New builtins.
2936 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
2937 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
2938 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
2939 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
2940 * config/i386/sse.md
2941 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
2942 avx512f_sgetexp<mode><mask_scalar_name>
2943 <round_saeonly_scalar_name> ... this.
2944 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
2945 %0, %1, %2<round_saeonly_op3>}): Changed to ...
2946 vgetexp<ssescalarmodesuffix>
2947 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2948 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
2949 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
2950 avx512f_vgetmant<mode><mask_scalar_name>
2951 <round_saeonly_scalar_name> ... this.
2952 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
2953 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
2954 vgetmant<ssescalarmodesuffix>
2955 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
2956 %0<mask_scalar_operand4>, %1, %2
2957 <round_saeonly_scalar_mask_op4>, %3} ... this.
2958 * config/i386/subst.md (mask_scalar_operand4,
2959 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
2960 round_saeonly_scalar_nimm_predicate): New subst attributes.
2962 2017-07-06 Julia Koval <julia.koval@intel.com>
2964 * config/i386/i386.c (ix86_erase_embedded_rounding):
2965 Remove code for old rounding pattern.
2967 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
2969 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
2971 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
2973 * doc/sourcebuild.texi (Test Directives, Variants of
2974 dg-require-support): Add documentation for dg-require-stack-check.
2976 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
2978 * config/i386/subst.md (mask_scalar, round_scalar,
2979 round_saeonly_scalar): New meta-templates.
2980 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
2981 round_scalar_mask_operand3, round_scalar_mask_op3,
2982 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
2983 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
2984 round_saeonly_scalar_constraint,
2985 round_saeonly_scalar_prefix): New subst attribute.
2986 * config/i386/sse.md
2987 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
2988 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
2989 <round_scalar_name> ... this.
2990 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
2991 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
2992 <round_scalar_name> ... this.
2993 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
2994 <sse>_vm<code><mode>3<mask_scalar_name>
2995 <round_saeonly_scalar_name> ... this.
2996 (v<plusminus_mnemonic><ssescalarmodesuffix>
2997 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
2998 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
2999 v<plusminus_mnemonic><ssescalarmodesuffix>
3000 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
3001 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
3002 (v<multdiv_mnemonic><ssescalarmodesuffix>
3003 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
3004 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
3005 v<multdiv_mnemonic><ssescalarmodesuffix>
3006 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
3007 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
3008 (v<maxmin_float><ssescalarmodesuffix>
3009 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
3010 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
3011 v<maxmin_float><ssescalarmodesuffix>
3012 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
3013 %0<mask_scalar_operand3>, %1, %<iptr>2
3014 <round_saeonly_scalar_mask_op3>} ... this.
3016 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
3018 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
3019 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
3021 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
3022 Alan Hayward <alan.hayward@arm.com>
3023 David Sherwood <david.sherwood@arm.com>
3025 * combine.c (simplify_if_then_else): Remove "enum" before
3027 * compare-elim.c (can_eliminate_compare): Likewise.
3028 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
3030 (aarch64_lookup_simd_builtin_type): Likewise.
3031 (aarch64_simd_builtin_type): Likewise.
3032 (aarch64_init_simd_builtin_types): Likewise.
3033 (aarch64_simd_expand_args): Likewise.
3034 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
3036 (aarch64_reverse_mask): Likewise.
3037 (aarch64_simd_emit_reg_reg_move): Likewise.
3038 (aarch64_gen_adjusted_ldpstp): Likewise.
3039 (aarch64_ccmp_mode_to_code): Likewise.
3040 (aarch64_operands_ok_for_ldpstp): Likewise.
3041 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3042 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
3044 (aarch64_min_divisions_for_recip_mul): Likewise.
3045 (aarch64_reassociation_width): Likewise.
3046 (aarch64_get_condition_code_1): Likewise.
3047 (aarch64_simd_emit_reg_reg_move): Likewise.
3048 (aarch64_simd_attr_length_rglist): Likewise.
3049 (aarch64_reverse_mask): Likewise.
3050 (aarch64_operands_ok_for_ldpstp): Likewise.
3051 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3052 (aarch64_gen_adjusted_ldpstp): Likewise.
3053 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
3055 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
3056 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
3057 (arm_lookup_simd_builtin_type): Likewise.
3058 (arm_simd_builtin_type): Likewise.
3059 (arm_init_simd_builtin_types): Likewise.
3060 (arm_expand_builtin_args): Likewise.
3061 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
3062 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
3063 (ft32_setup_incoming_varargs): Likewise.
3064 (ft32_function_arg): Likewise.
3065 (ft32_function_arg_advance): Likewise.
3066 (ft32_pass_by_reference): Likewise.
3067 (ft32_arg_partial_bytes): Likewise.
3068 (ft32_valid_pointer_mode): Likewise.
3069 (ft32_addr_space_pointer_mode): Likewise.
3070 (ft32_addr_space_legitimate_address_p): Likewise.
3071 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
3073 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
3074 (ix86_emit_outlined_ms2sysv_restore): Likewise.
3075 (iamcu_alignment): Likewise.
3076 (canonicalize_vector_int_perm): Likewise.
3077 (ix86_noce_conversion_profitable_p): Likewise.
3078 (ix86_mpx_bound_mode): Likewise.
3079 (ix86_operands_ok_for_move_multiple): Likewise.
3080 * config/microblaze/microblaze-protos.h
3081 (microblaze_expand_conditional_branch_reg): Likewise.
3082 * config/microblaze/microblaze.c
3083 (microblaze_expand_conditional_branch_reg): Likewise.
3084 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
3086 (rs6000_reassociation_width): Likewise.
3087 (rs6000_invalid_binary_op): Likewise.
3088 (fusion_p9_p): Likewise.
3089 (emit_fusion_p9_load): Likewise.
3090 (emit_fusion_p9_store): Likewise.
3091 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
3093 (riscv_hard_regno_mode_ok_p): Likewise.
3094 (riscv_address_insns): Likewise.
3095 (riscv_split_symbol): Likewise.
3096 (riscv_legitimize_move): Likewise.
3097 (riscv_function_value): Likewise.
3098 (riscv_hard_regno_nregs): Likewise.
3099 (riscv_expand_builtin): Likewise.
3100 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
3101 (riscv_build_integer): Likewise.
3102 (riscv_split_integer): Likewise.
3103 (riscv_legitimate_constant_p): Likewise.
3104 (riscv_cannot_force_const_mem): Likewise.
3105 (riscv_regno_mode_ok_for_base_p): Likewise.
3106 (riscv_valid_base_register_p): Likewise.
3107 (riscv_valid_offset_p): Likewise.
3108 (riscv_valid_lo_sum_p): Likewise.
3109 (riscv_classify_address): Likewise.
3110 (riscv_legitimate_address_p): Likewise.
3111 (riscv_address_insns): Likewise.
3112 (riscv_load_store_insns): Likewise.
3113 (riscv_force_binary): Likewise.
3114 (riscv_split_symbol): Likewise.
3115 (riscv_force_address): Likewise.
3116 (riscv_legitimize_address): Likewise.
3117 (riscv_move_integer): Likewise.
3118 (riscv_legitimize_const_move): Likewise.
3119 (riscv_legitimize_move): Likewise.
3120 (riscv_address_cost): Likewise.
3121 (riscv_subword): Likewise.
3122 (riscv_output_move): Likewise.
3123 (riscv_canonicalize_int_order_test): Likewise.
3124 (riscv_emit_int_order_test): Likewise.
3125 (riscv_function_arg_boundary): Likewise.
3126 (riscv_pass_mode_in_fpr_p): Likewise.
3127 (riscv_pass_fpr_single): Likewise.
3128 (riscv_pass_fpr_pair): Likewise.
3129 (riscv_get_arg_info): Likewise.
3130 (riscv_function_arg): Likewise.
3131 (riscv_function_arg_advance): Likewise.
3132 (riscv_arg_partial_bytes): Likewise.
3133 (riscv_function_value): Likewise.
3134 (riscv_pass_by_reference): Likewise.
3135 (riscv_setup_incoming_varargs): Likewise.
3136 (riscv_print_operand): Likewise.
3137 (riscv_elf_select_rtx_section): Likewise.
3138 (riscv_save_restore_reg): Likewise.
3139 (riscv_for_each_saved_reg): Likewise.
3140 (riscv_register_move_cost): Likewise.
3141 (riscv_hard_regno_mode_ok_p): Likewise.
3142 (riscv_hard_regno_nregs): Likewise.
3143 (riscv_class_max_nregs): Likewise.
3144 (riscv_memory_move_cost): Likewise.
3145 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
3146 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
3147 (rl78_addr_space_address_mode): Likewise.
3148 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3150 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3151 (rs6000_reassociation_width): Likewise.
3152 (rs6000_invalid_binary_op): Likewise.
3153 (fusion_p9_p): Likewise.
3154 (emit_fusion_p9_load): Likewise.
3155 (emit_fusion_p9_store): Likewise.
3156 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
3157 (ok_for_simple_move_operands): Likewise.
3158 (ok_for_simple_move_strict_operands): Likewise.
3159 (ok_for_simple_arith_logic_operands): Likewise.
3160 (visium_legitimize_reload_address): Likewise.
3161 (visium_select_cc_mode): Likewise.
3162 (output_cbranch): Likewise.
3163 (visium_split_double_move): Likewise.
3164 (visium_expand_copysign): Likewise.
3165 (visium_expand_int_cstore): Likewise.
3166 (visium_expand_fp_cstore): Likewise.
3167 * config/visium/visium.c (visium_pass_by_reference): Likewise.
3168 (visium_function_arg): Likewise.
3169 (visium_function_arg_advance): Likewise.
3170 (visium_libcall_value): Likewise.
3171 (visium_setup_incoming_varargs): Likewise.
3172 (visium_legitimate_constant_p): Likewise.
3173 (visium_legitimate_address_p): Likewise.
3174 (visium_legitimize_address): Likewise.
3175 (visium_secondary_reload): Likewise.
3176 (visium_register_move_cost): Likewise.
3177 (visium_memory_move_cost): Likewise.
3178 (prepare_move_operands): Likewise.
3179 (ok_for_simple_move_operands): Likewise.
3180 (ok_for_simple_move_strict_operands): Likewise.
3181 (ok_for_simple_arith_logic_operands): Likewise.
3182 (visium_function_value_1): Likewise.
3183 (rtx_ok_for_offset_p): Likewise.
3184 (visium_legitimize_reload_address): Likewise.
3185 (visium_split_double_move): Likewise.
3186 (visium_expand_copysign): Likewise.
3187 (visium_expand_int_cstore): Likewise.
3188 (visium_expand_fp_cstore): Likewise.
3189 (visium_split_cstore): Likewise.
3190 (visium_select_cc_mode): Likewise.
3191 (visium_split_cbranch): Likewise.
3192 (output_cbranch): Likewise.
3193 (visium_print_operand_address): Likewise.
3194 * expmed.c (flip_storage_order): Likewise.
3195 * expmed.h (emit_cstore): Likewise.
3196 (flip_storage_order): Likewise.
3197 * genrecog.c (validate_pattern): Likewise.
3198 * hsa-gen.c (gen_hsa_addr): Likewise.
3199 * internal-fn.c (expand_arith_overflow): Likewise.
3200 * ira-color.c (allocno_copy_cost_saving): Likewise.
3201 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3202 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
3203 (process_invariant_for_inheritance): Likewise.
3204 * lra-eliminations.c (move_plus_up): Likewise.
3205 * omp-low.c (lower_oacc_reductions): Likewise.
3206 * simplify-rtx.c (simplify_subreg): Likewise.
3207 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
3208 (TARGET_CHKP_BOUND_MODE): Likewise..
3209 * targhooks.c (default_chkp_bound_mode): Likewise.
3210 (default_setup_incoming_vararg_bounds): Likewise.
3211 * targhooks.h (default_chkp_bound_mode): Likewise.
3212 (default_setup_incoming_vararg_bounds): Likewise.
3213 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
3214 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
3215 (have_whole_vector_shift): Likewise.
3216 * tree-vect-stmts.c (vectorizable_load): Likewise.
3217 * doc/tm.texi: Regenerate.
3219 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
3221 Graceful degrade if Binutils PR21472 is not available.
3224 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
3225 .rodata in flash test fails.
3226 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
3227 * confgure: Regenerate.
3228 * config.in: Regenerate.
3229 * config/avr/avr.c (avr_asm_named_section)
3230 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
3231 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
3232 (avr_asm_init_sections): Same.
3234 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3236 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
3237 (fma<VH:mode>4_intrinsic): Likewise.
3238 (*fmsub<VCVTF:mode>4): Likewise.
3239 (*fmsub<VH:mode>4_intrinsic): Likewise.
3241 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
3244 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
3245 Don't depend on "optimize > 0".
3246 (out_movhi_r_mr, out_movqi_mr_r): Same.
3247 (out_movhi_mr_r, out_movqi_r_mr): Same.
3248 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
3249 io_address_operand on "optimize > 0".
3251 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3253 * tree-loop-distribution.c: Add general explanantion on the pass.
3254 (generate_loops_for_partition): Mark distributed loop.
3255 (pg_add_dependence_edges): New parameter. Handle alias data
3256 dependence specially and record it in the parameter if asked.
3257 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
3258 (init_partition_graph_vertices, add_partition_graph_edge): New.
3259 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
3260 (free_partition_graph_vdata, build_partition_graph): New.
3261 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
3262 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
3263 (data_ref_segment_size, latch_dominated_by_data_ref): New.
3264 (compute_alias_check_pairs, version_loop_by_alias_check): New.
3265 (version_for_distribution_p, finalize_partitions): New.
3266 (distribute_loop): Handle alias data dependence specially. Factor
3267 out loop fusion code as functions and call these functions.
3269 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3271 * tree-loop-distribution.c (classify_partition): New parameter and
3272 better handle reduction statement.
3273 (rdg_build_partitions): Revise comment.
3274 (distribute_loop): Compute statements in all partitions and pass it
3275 to classify_partition.
3277 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3279 * tree-loop-distribution.c (enum partition_type): New.
3280 (struct partition): New field type.
3281 (partition_merge_into): Add parameter. Update partition type.
3282 (data_dep_in_cycle_p, update_type_for_merge): New functions.
3283 (build_rdg_partition_for_vertex): Compute partition type.
3284 (rdg_build_partitions): Dump partition type.
3285 (distribute_loop): Update calls to partition_merge_into.
3287 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3289 * tree-loop-distribution.c (struct ddr_hasher): New.
3290 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
3292 (classify_partition): Call get_data_dependence.
3293 (pg_add_dependence_edges): Ditto.
3294 (distribute_loop): Release data dependence hash table.
3296 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3298 * tree-loop-distribution.c (ref_base_address): Delete.
3299 (similar_memory_accesses): Rename ...
3300 (share_memory_accesses): ... to this. Check if partitions access
3301 the same memory reference.
3302 (distribute_loop): Call share_memory_accesses.
3304 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3306 * tree-loop-distribution.c (struct partition): New field recording
3308 (partition_alloc, partition_free): Init and release data refs.
3309 (partition_merge_into): Merge data refs.
3310 (build_rdg_partition_for_vertex): Collect data refs for partition.
3311 (pg_add_dependence_edges): Change parameters from vector to bitmap.
3313 (distribute_loop): Remve data refs from vertice data of partition
3316 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3318 * tree-loop-distribution.c (params.h): Include header file.
3319 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
3320 (datarefs_vec): New global var.
3321 (create_rdg_vertices): Use datarefs_vec directly.
3322 (free_rdg): Don't free data references.
3323 (build_rdg): Update use. Don't free data references.
3324 (distribute_loop): Compute global variable for data references.
3325 Bail out if there are too many data references.
3327 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3329 * tree-loop-distribution.c (loop_nest): New global var.
3330 (build_rdg): Use loop directly, rather than loop nest.
3331 (pg_add_dependence_edges): Remove loop nest parameter. Use global
3333 (distribute_loop): Compute global variable loop nest. Update use.
3335 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3337 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
3338 (partition_merge_into): New parameter. Dump reason for fusion.
3339 (distribute_loop): Update use of partition_merge_into.
3341 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3343 * tree-loop-distribution.c (bb_top_order_index): New.
3344 (bb_top_order_index_size, bb_top_order_cmp): New.
3345 (stmts_from_loop): Use topological order.
3346 (pass_loop_distribution::execute): Compute and release topological
3347 order for basic blocks.
3349 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3351 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
3354 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3356 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
3357 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
3358 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
3359 * internal-fn.def (LOOP_DIST_ALIAS): New.
3360 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
3361 (fold_loop_internal_call): ... this.
3362 (vect_loop_dist_alias_call): New function.
3363 (set_uid_loop_bbs): Call fold_loop_internal_call.
3364 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
3367 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
3370 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
3371 Require dead FLAGS_REG at the beginning of a peephole.
3373 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
3376 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
3377 arguments in the call to __builtin_ia32_sbb_u32.
3378 (_subborrow_u64): Swap _X and _Y arguments in the call to
3379 __builtin_ia32_sbb_u64.
3381 2017-07-04 Jakub Jelinek <jakub@redhat.com>
3384 * tree-vrp.c (compare_assert_loc): Turn into a function template
3385 with stable template parameter. Only test if a->e is NULL,
3386 !a->e == !b->e has been verified already. Use e == NULL or
3387 e != NULL instead of e or ! e tests. If stable is true, don't use
3388 iterative_hash_expr, on the other side allow a or b or both NULL
3389 and sort the NULLs last.
3390 (process_assert_insertions): Sort using compare_assert_loc<false>
3391 instead of compare_assert_loc, later sort using
3392 compare_assert_loc<true> before calling process_assert_insertions_for
3393 in a loop. Use break instead of continue once seen NULL pointer.
3395 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
3397 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
3398 Cortex-R7 and Cortex-R8 processors.
3400 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3402 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
3403 uninitialized while src is not.
3405 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
3407 * common/config/arm/arm-common.c: Adjust include path for
3409 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
3410 (arm-cpu.h): Create in build directory. Adjust dependency rules.
3411 (arm-cpu-data.h): Likewise.
3412 (arm-cpu-cdata.h): Likewise.
3413 * config/arm/arm-cpu.h: Delete.
3414 * config/arm/arm-cpu-cdata.h: Delete.
3415 * config/arm/arm-cpu-data.h: Delete.
3417 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
3419 * config/arm/arm-cpus.in (cortex-a55): New.
3420 (cortex-a75): Likewise.
3421 (cortex-a75.cortex-a55): Likewise.
3422 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
3424 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
3425 * config/arm/arm-cpu-cdata.h: Regenerate.
3426 * config/arm/arm-cpu-data.h: Regenerate.
3427 * config/arm/arm-cpu.h: Regenerate.
3428 * config/arm/arm-tables.opt: Regenerate.
3429 * config/arm/arm-tune.md: Regenerate.
3431 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3433 * haifa-sched.c (sched_create_recovery_edges): Update profile.
3435 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3437 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
3440 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
3442 PR tree-optimization/81292
3443 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
3444 full_string_p, also call adjust_related_strinfos if the adjustment
3445 is simple, otherwise invalidate related strinfos.
3447 2017-07-04 Martin Liska <mliska@suse.cz>
3450 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
3451 newly created variable as DECL_IGNORED_P.
3453 2017-07-04 Martin Liska <mliska@suse.cz>
3456 * ipa-inline.c (inline_small_functions):
3457 Use xstrdup_for_dump.
3459 2017-07-04 Tom de Vries <tom@codesourcery.com>
3461 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
3463 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
3466 * config/darwin.c (darwin_function_switched_text_sections):
3469 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
3471 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
3473 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
3475 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
3477 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3479 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
3480 min_profitable_iters, and th as inclusive lower bounds.
3481 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
3482 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
3483 for min_profitable_iters and min_profitable_estimate.
3484 (vect_transform_loop): Treat th as an inclusive lower bound.
3485 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
3487 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
3490 * config/darwin.c (darwin_function_switched_text_sections):
3491 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
3492 in two pieces, and suppress the use of buf.
3494 2017-07-03 Nathan Sidwell <nathan@acm.org>
3496 * hash-table.h (hash_table_mod1): Fix indentation.
3498 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3501 * predict.c (force_edge_cold): Be more careful about propagation
3503 * profile-count.h (profile_probability::guessed,
3504 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
3506 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
3508 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
3510 * doc/invoke.texi (rcpc architecture extension): Document it.
3512 2017-07-03 Richard Biener <rguenther@suse.de>
3514 PR tree-optimization/60510
3515 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
3516 the scalar reduction PHI and use it.
3517 (vectorizable_reduction): Properly guard the single_defuse_cycle
3518 path for non-SLP reduction chains where we cannot use it.
3519 Rework reduc_def/index and vector type deduction. Rework
3520 vector operand gathering during reduction op code-gen.
3521 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
3522 chains dissolve the chain and leave it to non-SLP reduction
3525 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3527 * tree-data-ref.h (dr_alignment): Declare.
3528 * tree-data-ref.c (dr_alignment): New function.
3529 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
3530 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
3532 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
3534 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3536 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
3537 and base_misalignment fields.
3538 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
3539 * tree-data-ref.c: Include builtins.h.
3540 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
3541 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
3542 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
3543 * tree-vect-data-refs.c: Include tree-cfg.h.
3544 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
3545 fields instead of calculating an alignment here.
3546 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
3547 innermost_loop_behavior fields.
3549 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3551 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
3553 (DR_STEP_ALIGNMENT): New macro.
3554 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
3555 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
3556 (create_data_ref): Print it.
3557 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
3558 to tell whether the step preserves vector (mis)alignment.
3559 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
3560 Move the check for an integer step and generalise to all INTEGER_CST.
3561 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
3562 Print the outer step alignment.
3564 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3566 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
3567 with offset_alignment.
3568 (DR_ALIGNED_TO): Delete.
3569 (DR_OFFSET_ALIGNMENT): New macro.
3570 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
3571 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
3572 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
3573 (create_data_ref): Likewise.
3574 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
3575 (vect_analyze_data_refs): Likewise.
3576 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
3577 creating dummy innermost behavior.
3579 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3581 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
3582 with a "innermost_loop_behavior *" and refeence tree.
3583 * tree-data-ref.c (dr_analyze_innermost): Likewise.
3584 (create_data_ref): Update call accordingly.
3585 * tree-predcom.c (find_looparound_phi): Likewise.
3587 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3589 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
3590 fields with dr_wrt_vec_loop.
3591 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
3592 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
3593 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
3594 (vect_dr_behavior): New function.
3595 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
3596 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
3597 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
3598 track whether the step preserves the misalignment.
3599 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
3600 Use vect_dr_behavior.
3601 (vect_setup_realignment): Update call accordingly.
3602 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
3603 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
3604 call to vect_create_addr_base_for_vector_ref.
3605 (vect_create_cond_for_align_checks): Likewise.
3606 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
3607 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
3608 (vect_recog_mask_conversion_pattern): Likewise.
3609 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
3610 (new_stmt_vec_info): Remove redundant zeroing.
3612 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
3614 * common/config/arm/arm-common.c (arm_be8_option): New function.
3615 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
3616 (ISA_ARMv6): Add isa_bit_be8.
3617 * config/arm/arm.h (arm_be8_option): Add prototype.
3618 (BE8_SPEC_FUNCTION): New define.
3619 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
3620 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
3621 (mlittle-endian): Similarly.
3622 (mbe8, mbe32): New options.
3623 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
3624 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
3626 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3628 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
3630 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3632 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
3633 (cleanup_tree_cfg_bb): Use it.
3634 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
3636 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
3638 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3641 * loop-doloop.c (add_test): Update profile.
3643 2017-07-03 Martin Liska <mliska@suse.cz>
3646 * sanopt.c (rewrite_usage_of_param): New function.
3647 (sanitize_rewrite_addressable_params): Likewise.
3648 (pass_sanopt::execute): Call rewrite_usage_of_param.
3650 2017-07-03 Richard Biener <rguenther@suse.de>
3652 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
3653 back to using VIEW_CONVERT_EXPR.
3655 2017-07-03 Martin Liska <mliska@suse.cz>
3658 * doc/extend.texi: Document when a resolver function is
3659 generated for target_clones.
3661 2017-07-03 Martin Liska <mliska@suse.cz>
3663 * asan.c (asan_emit_stack_protection): Unpoison just red zones
3664 and shadow memory of auto variables which are subject of
3665 use-after-scope sanitization.
3666 (asan_expand_mark_ifn): Add do set only when is_poison.
3668 2016-07-03 Richard Biener <rguenther@suse.de>
3670 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
3672 (vect_force_simple_reduction): Record reduction def -> phi mapping.
3673 (vectorizable_reduction): Perform reduction PHI creation when
3674 visiting a reduction PHI and adjust and simplify code generation
3675 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
3676 (vect_transform_loop): Visit reduction PHIs.
3677 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
3678 defs into the SLP tree.
3679 (vect_build_slp_tree): Reduction defs terminate the recursion.
3680 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
3682 (vect_get_vec_defs_for_stmt_copy): Export.
3683 (vect_get_vec_defs): Likewise.
3684 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
3686 (vect_get_vec_defs_for_stmt_copy): Declare.
3687 (vect_get_vec_defs): Likewise.
3689 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3691 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
3692 parameter with a "loop" parameter and use it instead of the
3693 loop containing DR_STMT. Don't check simple_iv when doing
3694 BB analysis. Describe the two analysis modes in the comment.
3696 2017-07-03 Tom de Vries <tom@codesourcery.com>
3698 PR tree-optimization/69468
3699 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
3700 (find_same_succ_bb): Handle ignore_edge_flags.
3702 2017-07-03 Tom de Vries <tom@codesourcery.com>
3704 PR tree-optimization/81192
3705 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
3707 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
3709 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
3711 2017-07-03 Tom de Vries <tom@codesourcery.com>
3713 PR tree-optimization/81192
3714 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
3715 BB_SAME_SUCC (bb) == NULL.
3717 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3719 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
3722 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3724 * dumpfile.c: Include profile-count.h
3725 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
3727 (insert_cond_bb): Update profile.
3728 * tree-cfg.h (insert_cond_bb): Update prototype.
3729 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
3730 * tree-dump.c: Do not include tree-cfg.
3732 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3734 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
3736 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3738 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
3741 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3743 * tree-complex.c (expand_complex_div_wide): update profile.
3745 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3746 Alan Hayward <alan.hayward@arm.com>
3747 David Sherwood <david.sherwood@arm.com>
3749 * Makefile.in (MACHMODE_H): Remove insn-modes.h
3750 (CORETYPES_H): New define.
3751 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
3752 (insn-modes-inline.h, s-modes-inline-h): New rules.
3753 (generated_files): Add insn-modes-inline.h.
3754 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
3755 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
3756 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
3757 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
3758 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
3759 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
3760 (build/gencodes.o, build/genconditions.o): Likewise.
3761 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
3762 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
3763 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
3764 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
3765 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
3766 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
3767 * coretypes.h: Include everything up to real.h for generators.
3768 Include insn-modes.h first. Include wide-int-print.h after
3769 wide-int.h. Include insn-modes-inline.h and then machmode.h.
3770 * machmode.h: Don't include insn-modes.h here.
3771 * function-tests.c: Remove includes of signop.h, machmode.h,
3772 double-int.h and wide-int.h.
3774 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
3776 * optc-save-gen.awk: Likewise.
3777 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
3778 * godump.c: Remove include of wide-int-print.h.
3779 * pretty-print.h: Likewise.
3780 * wide-int-print.cc: Likewise.
3781 * wide-int.cc: Likewise.
3782 * hash-map-tests.c: Remove include of signop.h.
3783 * hash-set-tests.c: Likewise.
3784 * rtl-tests.c: Likewise.
3785 * mkconfig.sh: Remove include of machmode.h.
3786 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
3788 (emit_insn_modes_inline_h): ...this new function. Emit the code
3789 into an insn-modes-inline.h header file, adding appropriate
3790 include guards and end comments.
3791 (emit_insn_modes_c_header): Remove include of machmode.h.
3792 (emit_min_insn_modes_c_header): Include coretypes.h rather than
3794 (main): Handle -i flag and call emit_insn_modes_inline_h when
3797 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3799 * tree-ssa-strlen.c (strinfo): Rename the length field to
3800 nonzero_chars. Add a full_string_p field.
3801 (compare_nonzero_chars, zero_length_string_p): New functions.
3802 (get_addr_stridx): Add an offset_out parameter.
3803 Use compare_nonzero_chars.
3804 (get_stridx): Update accordingly. Use compare_nonzero_chars.
3805 (new_strinfo): Update after above changes to strinfo.
3806 (set_endptr_and_length): Set full_string_p.
3807 (get_string_length): Update after above changes to strinfo.
3808 (unshare_strinfo): Update call to new_strinfo.
3809 (maybe_invalidate): Likewise.
3810 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
3811 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
3812 as a uhwi instead of an shwi. Update after above changes to
3813 strinfo and new_strinfo.
3814 (zero_length_string): Assert that chainsi contains full strings.
3815 Use zero_length_string_p. Update call to new_strinfo.
3816 (adjust_related_strinfos): Update after above changes to strinfo.
3817 Copy full_string_p from origsi.
3818 (adjust_last_stmt): Use zero_length_string_p.
3819 (handle_builtin_strlen): Update after above changes to strinfo and
3820 new_strinfo. Install the lhs as the string length if the previous
3821 entry didn't describe a full string.
3822 (handle_builtin_strchr): Update after above changes to strinfo
3824 (handle_builtin_strcpy): Likewise.
3825 (handle_builtin_strcat): Likewise.
3826 (handle_builtin_malloc): Likewise.
3827 (handle_pointer_plus): Likewise.
3828 (handle_builtin_memcpy): Likewise. Track nonzero characters
3829 that aren't necessarily followed by a nul terminator.
3830 (handle_char_store): Likewise.
3832 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3834 PR tree-optimization/80769
3835 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
3836 for malloc and calloc. Document the new invariant that all related
3837 strinfos have delayed lengths or none do.
3838 (verify_related_strinfos): Move earlier in file.
3839 (set_endptr_and_length): New function, split out from...
3840 (get_string_length): ...here. Also set the lengths of related
3842 (zero_length_string): Assert that chainsi has known (rather than
3844 (adjust_related_strinfos): Likewise.
3846 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3848 PR tree-optimization/81136
3849 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
3850 assert that two references with the same misalignment have the same
3851 compile-time misalignment if those compile-time misalignments
3854 2017-07-01 Andi Kleen <ak@linux.intel.com>
3856 * print-tree.c (print_node): Print all attributes.
3858 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3860 * cfg.c (scale_bbs_frequencies): New function.
3861 * cfg.h (scale_bbs_frequencies): Declare it.
3862 * cfgloopanal.c (single_likely_exit): Cleanup.
3863 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
3865 (scale_loop_profile): Likewise.
3866 (loop_version): Likewise.
3867 (create_empty_loop_on_edge): Update.
3868 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
3869 scale_loop_frequencies, scale_loop_profile, loopify,
3870 loop_version): Update prototypes.
3871 * modulo-sched.c (sms_schedule): Update.
3872 * predict.c (unlikely_executed_edge_p): Also check probability.
3873 (probably_never_executed_edge_p): Fix typo.
3874 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3875 * tree-parloops.c (gen_parallel_loop): Update.
3876 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
3877 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3878 * tree-ssa-loop-split.c (split_loop): Update.
3879 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3880 * tree-vect-loop-manip.c (vect_do_peeling): Update.
3881 (vect_loop_versioning): Update.
3882 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3884 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3886 * trans-mem.c (split_bb_make_tm_edge): Update profile.
3888 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3890 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
3891 to keep profile consistent.
3893 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3895 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
3896 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
3897 * profile-count.h (max_safe_multiplier): Make unsigned.
3898 (profile_count::guessed_zero): New.
3900 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3902 * bb-reorder.c (fix_up_crossing_landing_pad,
3903 fix_crossing_conditional_branches): Use make_single_succ_edge
3904 to keep profile consistent.
3906 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3908 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
3911 2017-07-01 Jakub Jelinek <jakub@redhat.com>
3914 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
3915 the right scopes, make sure cond_jump isn't preserved between multiple
3916 iterations. Search for fallthru edge whenever there are 3+ edges and
3917 use find_fallthru_edge for it.
3919 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3921 Patch by Alexander Monakov <amonakov@ispras.ru>
3922 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
3923 probabilities consistently.
3925 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3927 * pa.c (pa_expand_compare_and_swap_loop): Update call of
3928 emit_cmp_and_jump_insns.
3930 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3933 * tree-inline.c (expand_call_inline): Combine profile statuses.
3935 2017-06-30 Andrew Pinski <apinski@cavium.com>
3937 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
3938 fold_stmt returned true.
3940 2017-06-30 Nathan Sidwell <nathan@acm.org>
3942 * ggc.h (empty_string): Delete.
3943 * cfgexpand.c (expand_asm_stmt): Use plain "".
3944 * optabs.c (expand_asm_memory_barrier): Likewise.
3945 * stringpool.c (empty_string): Delete.
3946 (digit_vector, digit_string): Delete.
3947 (ggc_alloc_string): Use plain "", don't optimize single digit
3948 strings. Use ggc_alloc_atomic.
3950 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
3952 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
3953 comparison set and one other set, use the cost of the non-comparison
3956 2017-06-30 Nathan Sidwell <nathan@acm.org>
3958 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
3961 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
3963 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
3964 loops. Remove now unneeded calls to gimple_switch_set_label() that
3965 just set removed labels to NULL_TREE.
3967 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
3969 * tree-ssanames.c (set_range_info_raw): Abstract from ...
3970 (set_range_info): ...here. Only call set_range_info_raw if domain
3972 (set_nonzero_bits): Call set_range_info_raw.
3973 * tree-ssanames.h (set_range_info_raw): New.
3975 2017-06-30 Jakub Jelinek <jakub@redhat.com>
3978 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
3979 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
3980 of nonimmediate_operand and <store_mask_constraint> instead of m
3981 for the input operand. For V8FI iterator, always split if input
3982 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
3983 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
3984 <store_mask_predicate> instead of register_operand and
3985 <store_mask_constraint> instead of v for the input operand. Make
3986 sure both operands aren't MEMs for if not <mask_applied>.
3988 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
3990 * lto-wrapper.c (copy_file) Close both file descriptors before
3993 2017-06-30 Martin Liska <mliska@suse.cz>
3996 * multiple_target.c (create_dispatcher_calls): Make ifunc
3997 also for function that don't have calls or are not referenced.
3999 2017-06-30 Richard Biener <rguenther@suse.de>
4001 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
4002 analyze the first scalar stmt. Move vector type computation
4003 for the BB case here from ...
4004 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
4005 live operation processing in the SLP case properly.
4007 2017-06-30 Richard Biener <rguenther@suse.de>
4009 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
4011 2017-06-30 Martin Liska <mliska@suse.cz>
4014 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
4015 before BUILT_IN_UNWIND_RESUME when ASAN is used.
4017 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
4019 * doc/invoke.texi (AArch64): Add missing options and remove redundant
4022 2017-06-30 Richard Biener <rguenther@suse.de>
4024 PR tree-optimization/81249
4025 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
4026 condition reduction result to original scalar type.
4028 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4030 * profile-count.h (enum profile_quality): Fix typos and whitespace
4033 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4035 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
4036 type for branch probabilities.
4038 2017-06-29 Julian Brown <julian@codesourcery.com>
4039 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4041 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
4042 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
4043 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
4044 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
4046 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4048 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
4049 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
4050 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
4051 CC usage from generic code to here.
4052 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
4053 CC usage into the target macros.
4055 2017-06-29 Maya Rashish <coypu@sdf.org>
4057 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
4060 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
4062 * arm/arm-builtins.c: Include profile-count.h
4063 * except.c (sjlj_emit_function_enter): Use
4064 profile_probability::unlikely.
4066 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4068 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
4069 and tocrel_offset be pointer args rather than implicitly using
4071 (legitimate_constant_pool_address_p, rs6000_emit_move,
4072 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
4073 tocrel_offset and use in toc_relative_expr_p call.
4074 (print_operand, print_operand_address): Use static tocrel_base_oac
4075 and tocrel_offset_oac.
4076 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
4079 2017-06-29 Maya Rashish <coypu@sdf.org>
4081 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
4083 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
4085 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
4086 objects, take into account only the alignment of 'op0' and 'mode1' if
4089 2017-06-29 Steve Ellcey <sellcey@cavium.com>
4091 * ccmp.c (ccmp_tree_comparison_p): New function.
4092 (ccmp_candidate_p): Update to use above function.
4093 (get_compare_parts): New function.
4094 (expand_ccmp_next): Update to use new functions.
4095 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
4097 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
4098 take mode as argument.
4099 * ccmp.h (expand_ccmp_expr): Add mode as argument.
4100 * expr.c (expand_expr_real_1): Pass mode as argument.
4102 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
4104 * combine.c (combine_instructions): Print insns to dump_file, together
4107 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
4109 * asan.c (asan_emit_stack_protection): Update.
4110 (create_cond_insert_point): Update.
4111 * auto-profile.c (afdo_propagate_circuit): Update.
4112 * basic-block.h (struct edge_def): Turn probability to
4113 profile_probability.
4114 (EDGE_FREQUENCY): Update.
4115 * bb-reorder.c (find_traces_1_round): Update.
4116 (better_edge_p): Update.
4117 (sanitize_hot_paths): Update.
4118 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
4119 (make_single_succ_edge): Update.
4120 (check_bb_profile): Update.
4121 (dump_edge_info): Update.
4122 (update_bb_profile_for_threading): Update.
4123 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
4124 probabilitycount to 0.
4125 * cfgbuild.c (compute_outgoing_frequencies): Update.
4126 * cfgcleanup.c (try_forward_edges): Update.
4127 (outgoing_edges_match): Update.
4128 (try_crossjump_to_edge): Update.
4129 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
4130 (expand_gimple_tailcall): Update.
4131 (construct_init_block): Use make_single_succ_edge.
4132 (construct_exit_block): Use make_single_succ_edge.
4133 * cfghooks.c (verify_flow_info): Update.
4134 (redirect_edge_succ_nodup): Update.
4135 (split_edge): Update.
4136 (account_profile_record): Update.
4137 * cfgloopanal.c (single_likely_exit): Update.
4138 * cfgloopmanip.c (scale_loop_profile): Update.
4139 (set_zero_probability): Remove.
4140 (duplicate_loop_to_header_edge): Update.
4141 * cfgloopmanip.h (loop_version): Update prototype.
4142 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4143 (force_nonfallthru_and_redirect): Update.
4144 (update_br_prob_note): Update.
4145 (rtl_verify_edges): Update.
4146 (purge_dead_edges): Update.
4147 (rtl_lv_add_condition_to_bb): Update.
4148 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
4149 * cgraphunit.c (init_lowered_empty_function): Update.
4150 (cgraph_node::expand_thunk): Update.
4151 * cilk-common.c: Include profile-count.h
4152 * dojump.c (inv): Remove.
4153 (jumpifnot): Update.
4154 (jumpifnot_1): Update.
4155 (do_jump_1): Update.
4157 (do_jump_by_parts_greater_rtx): Update.
4158 (do_compare_rtx_and_jump): Update.
4159 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
4160 do_jump_1. do_compare_rtx_and_jump): Update prototype.
4161 * dwarf2cfi.c: Include profile-count.h
4162 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
4163 (sjlj_emit_dispatch_table): Likewise.
4164 * explow.c: Include profile-count.h
4165 * expmed.c (emit_store_flag_force): Update.
4166 (do_cmp_and_jump): Update.
4167 * expr.c (compare_by_pieces_d::generate): Update.
4168 (compare_by_pieces_d::finish_mode): Update.
4169 (emit_block_move_via_loop): Update.
4170 (store_expr_with_bounds): Update.
4171 (store_constructor): Update.
4172 (expand_expr_real_2): Update.
4173 (expand_expr_real_1): Update.
4174 * expr.h (try_casesi, try_tablejump): Update prototypes.
4175 * gimple-pretty-print.c (dump_probability): Update.
4176 (dump_profile): New.
4177 (dump_gimple_label): Update.
4178 (dump_gimple_bb_header): Update.
4179 * graph.c (draw_cfg_node_succ_edges): Update.
4180 * hsa-gen.c (convert_switch_statements): Update.
4181 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
4182 (find_if_case_1): Update.
4183 (find_if_case_2): Update.
4184 * internal-fn.c (expand_arith_overflow_result_store): Update.
4185 (expand_addsub_overflow): Update.
4186 (expand_neg_overflow): Update.
4187 (expand_mul_overflow): Update.
4188 (expand_vector_ubsan_overflow): Update.
4189 * ipa-cp.c (good_cloning_opportunity_p): Update.
4190 * ipa-split.c (split_function): Use make_single_succ_edge.
4191 * ipa-utils.c (ipa_merge_profiles): Update.
4192 * loop-doloop.c (add_test): Update.
4193 (doloop_modify): Update.
4194 * loop-unroll.c (compare_and_jump_seq): Update.
4195 (unroll_loop_runtime_iterations): Update.
4196 * lra-constraints.c (lra_inheritance): Update.
4197 * lto-streamer-in.c (input_cfg): Update.
4198 * lto-streamer-out.c (output_cfg): Update.
4199 * mcf.c (adjust_cfg_counts): Update.
4200 * modulo-sched.c (sms_schedule): Update.
4201 * omp-expand.c (expand_omp_for_init_counts): Update.
4202 (extract_omp_for_update_vars): Update.
4203 (expand_omp_ordered_sink): Update.
4204 (expand_omp_for_ordered_loops): Update.
4205 (expand_omp_for_generic): Update.
4206 (expand_omp_for_static_nochunk): Update.
4207 (expand_omp_for_static_chunk): Update.
4208 (expand_cilk_for): Update.
4209 (expand_omp_simd): Update.
4210 (expand_omp_taskloop_for_outer): Update.
4211 (expand_omp_taskloop_for_inner): Update.
4212 * omp-simd-clone.c (simd_clone_adjust): Update.
4213 * optabs.c (expand_doubleword_shift): Update.
4214 (expand_abs): Update.
4215 (emit_cmp_and_jump_insn_1): Update.
4216 (expand_compare_and_swap_loop): Update.
4217 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
4218 * predict.c (predictable_edge_p): Update.
4219 (edge_probability_reliable_p): Update.
4220 (set_even_probabilities): Update.
4221 (combine_predictions_for_insn): Update.
4222 (combine_predictions_for_bb): Update.
4223 (propagate_freq): Update.
4224 (estimate_bb_frequencies): Update.
4225 (force_edge_cold): Update.
4226 * profile-count.c (profile_count::dump): Add missing space into dump.
4227 (profile_count::debug): Add newline.
4228 (profile_count::differs_from_p): Explicitly convert to unsigned.
4229 (profile_count::stream_in): Update.
4230 (profile_probability::dump): New member function.
4231 (profile_probability::debug): New member function.
4232 (profile_probability::differs_from_p): New member function.
4233 (profile_probability::differs_lot_from_p): New member function.
4234 (profile_probability::stream_in): New member function.
4235 (profile_probability::stream_out): New member function.
4236 * profile-count.h (profile_count_quality): Rename to ...
4237 (profile_quality): ... this one.
4238 (profile_probability): New.
4239 (profile_count): Update.
4240 * profile.c (compute_branch_probabilities): Update.
4241 * recog.c (peep2_attempt): Update.
4242 * sched-ebb.c (schedule_ebbs): Update.
4243 * sched-rgn.c (find_single_block_region): Update.
4244 (compute_dom_prob_ps): Update.
4245 (schedule_region): Update.
4246 * sel-sched-ir.c (compute_succs_info): Update.
4247 * stmt.c (struct case_node): Update.
4248 (do_jump_if_equal): Update.
4249 (get_outgoing_edge_probs): Update.
4250 (conditional_probability): Update.
4251 (emit_case_dispatch_table): Update.
4252 (expand_case): Update.
4253 (expand_sjlj_dispatch_table): Update.
4254 (emit_case_nodes): Update.
4255 * targhooks.c: Update.
4256 * tracer.c (better_p): Update.
4257 (find_best_successor): Update.
4258 * trans-mem.c (expand_transaction): Update.
4259 * tree-call-cdce.c: Update.
4260 * tree-cfg.c (gimple_split_edge): Upate.
4261 (move_sese_region_to_fn): Upate.
4262 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
4263 * tree-eh.c (lower_resx): Upate.
4264 (cleanup_empty_eh_move_lp): Upate.
4265 * tree-if-conv.c (version_loop_for_if_conversion): Update.
4266 * tree-inline.c (copy_edges_for_bb): Update.
4267 (copy_cfg_body): Update.
4268 * tree-parloops.c (gen_parallel_loop): Update.
4269 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
4270 (gimple_gen_time_profiler): Update.
4271 * tree-ssa-dce.c (remove_dead_stmt): Update.
4272 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4273 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4274 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4275 (unloop_loops): Update.
4276 (try_peel_loop): Update.
4277 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4278 * tree-ssa-loop-split.c (connect_loops): Update.
4279 (split_loop): Update.
4280 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
4281 (hoist_guard): Update.
4282 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4283 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4284 (value_replacement): Update.
4285 * tree-ssa-reassoc.c (branch_fixup): Update.
4286 * tree-ssa-tail-merge.c (replace_block_by): Update.
4287 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4288 (create_edge_and_update_destination_phis): Update.
4289 (compute_path_counts): Update.
4290 (recompute_probabilities): Update.
4291 (update_joiner_offpath_counts): Update.
4292 (freqs_to_counts_path): Update.
4293 (duplicate_thread_path): Update.
4294 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4295 (struct switch_conv_info): Update.
4296 (gen_inbound_check): Update.
4297 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4298 (vect_do_peeling): Update.
4299 (vect_loop_versioning): Update.
4300 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4301 (optimize_mask_stores): Update.
4302 * ubsan.c (ubsan_expand_null_ifn): Update.
4303 * value-prof.c (gimple_divmod_fixed_value): Update.
4304 (gimple_divmod_fixed_value_transform): Update.
4305 (gimple_mod_pow2): Update.
4306 (gimple_mod_pow2_value_transform): Update.
4307 (gimple_mod_subtract): Update.
4308 (gimple_mod_subtract_transform): Update.
4309 (gimple_ic): Update.
4310 (gimple_stringop_fixed_value): Update.
4311 (gimple_stringops_transform): Update.
4312 * value-prof.h: Update.
4314 2017-06-29 Carl Love <cel@us.ibm.com>
4316 * config/rs6000/rs6000-c.c: Add support for built-in functions
4317 vector signed int vec_signed (vector float);
4318 vector signed long long vec_signed (vector double);
4319 vector signed int vec_signed2 (vector double, vector double);
4320 vector signed int vec_signede (vector double);
4321 vector signed int vec_signedo (vector double);
4322 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
4323 instruction generator.
4324 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
4325 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
4326 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
4328 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
4329 vunsignede_v2df): Add define_expands.
4330 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
4331 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
4332 VEC_UNSIGNEDO): Add definitions.
4333 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
4334 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
4335 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
4336 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
4337 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
4338 * config/rs6000/altivec.h (vec_signed, vec_signed2,
4339 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
4340 vec_unsignede, vec_unsignedo): Add builtin defines.
4341 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
4343 * doc/extend.texi: Update the built-in documentation file for the
4344 new built-in functions.
4346 2017-06-29 Richard Biener <rguenther@suse.de>
4348 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
4349 reduction chains to LOOP_VINFO_REDUCTIONS.
4350 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
4351 SLP reductions after processing reduction chains.
4353 2017-06-29 Nathan Sidwell <nathan@acm.org>
4355 * builtins.c (fold_builtin_FUNCTION): Use
4356 lang_hooks.decl_printable_name.
4358 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
4361 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
4362 with only one label.
4363 * stmt.c (expand_case): Assert NCASES is greater than one.
4365 2017-06-29 Richard Biener <rguenther@suse.de>
4367 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
4369 (group_case_labels): Likewise.
4370 (find_taken_edge): Push sanity checking on val to workers...
4371 (find_taken_edge_cond_expr): ... here
4372 (find_taken_edge_switch_expr): ... and here, handle cases
4373 with just a default label.
4374 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
4375 (group_case_labels): Likewise.
4376 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
4377 group_case_labels does anything cleanup the CFG again.
4379 2017-06-29 Bin Cheng <bin.cheng@arm.com>
4381 PR tree-optimization/81196
4382 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
4383 exit condition comparing two IVs.
4385 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
4387 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
4388 profile to the dummy entry at the end of the list of architectures.
4389 * config/arm/arm-cpu-cdata.h: Regenerated.
4391 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4392 Michael Collison <michael.collison@arm.com>
4395 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
4397 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
4398 (*aarch64_reg_<mode>3_minus_mask): New pattern.
4399 (*aarch64_<optab>_reg_di3_mask2): New pattern.
4400 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
4401 of shift when the shift amount is masked with constant equal to
4402 the size of the mode.
4403 * config/aarch64/predicates.md (subreg_lowpart_operator): New
4406 2017-06-29 Martin Liska <mliska@suse.cz>
4408 * config/i386/i386.opt: Change range from [1,5] to [0,5].
4410 2017-06-29 Yury Gribov <tetra2005@gmail.com>
4413 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
4416 (edge_growth_cache_entry::edge_growth_cache_entry): New
4418 (reset_edge_growth_cache): Update to use constructor.
4420 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4422 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
4423 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
4424 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
4426 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
4428 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
4429 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
4431 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
4433 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
4434 (*-linux-uclibc*): Add t-uclibc tmake_file.
4435 * config/t-musl: New.
4436 * config/t-uclibc: New.
4438 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
4440 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
4442 (gen_comm_data): Emit architectural setting of arch_prof.
4443 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
4445 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
4446 (armv8-m.base, armv8-m.main): Likewise.
4447 * arm-protos.h (arm_build_target): Add profile field.
4448 (arch_option): Likewise.
4449 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
4451 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
4452 arm_active_target.profile.
4454 2017-06-28 Richard Biener <rguenther@suse.de>
4457 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
4458 TYPE_OVERFLOW_WRAPS.
4459 * match.pd (negate_expr_p): Likewise.
4460 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
4461 fold_build2, not fold_binary.
4463 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
4465 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
4466 Convert memory address to Pmode.
4467 (aarch64_print_operand): Assert MEM operands are always Pmode.
4469 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
4472 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
4473 Remove redundant if.
4474 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
4475 * config/arm/aarch-common-protos.h
4476 (aarch_forward_to_shift_is_not_shifted_re): Remove.
4477 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
4479 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
4482 * multiple_target.c (create_dispatcher_calls): Set the default
4483 clone to be static, not public.
4485 2017-06-28 Richard Biener <rguenther@suse.de>
4487 * tree-vect-loop.c (vectorizable_reduction): Move special
4488 cond reduction IV var creation ...
4489 (vect_create_epilog_for_reduction): ... here. Remove induction_index
4490 parameter. Use STMT_VINFO_VECTYPE.
4491 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
4494 2017-06-28 Martin Liska <mliska@suse.cz>
4497 * ipa-visibility.c (non_local_p): Handle visibility.
4499 2017-06-28 Martin Liska <mliska@suse.cz>
4502 * common.opt: Add IntegerRange to various options.
4503 * opt-functions.awk (integer_range_info): New function.
4504 * optc-gen.awk: Add integer_range_info to cl_options struct.
4505 * opts-common.c (decode_cmdline_option): Handle
4506 CL_ERR_INT_RANGE_ARG.
4507 (cmdline_handle_error): Likewise.
4508 * opts.c (print_filtered_help): Show valid interval in
4509 when --help is provided.
4510 * opts.h (struct cl_option): Add range_min and range_max fields.
4511 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
4513 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
4515 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
4516 (x * C EQ/NE y * C): New transformation.
4518 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
4520 * genmultilib (combination_space): Accept '+' in option names.
4522 2017-06-28 Martin Liska <mliska@suse.cz>
4525 * asan.c (instrument_derefs): Bail out inner references
4526 that are hard register variables.
4528 2017-06-28 Jakub Jelinek <jakub@redhat.com>
4531 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
4532 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
4534 2017-06-28 Richard Biener <rguenther@suse.de>
4536 * tree-vectorizer.h (vect_get_vec_defs): Remove.
4537 (vect_get_slp_defs): Adjust.
4538 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
4540 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
4542 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4543 get_initial_defs_for_reduction instead of vect_get_vec_defs.
4544 (vectorizable_reduction): Adjust.
4545 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
4547 (vect_get_slp_defs): Likewise.
4548 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
4549 (vectorizable_bswap): Adjust.
4550 (vectorizable_call): Likewise.
4551 (vectorizable_conversion): Likewise.
4552 (vectorizable_assignment): Likewise.
4553 (vectorizable_shift): Likewise.
4554 (vectorizable_operation): Likewise.
4555 (vectorizable_store): Likewise.
4556 (vectorizable_condition): Likewise.
4557 (vectorizable_comparison): Likewise.
4559 2017-06-28 Michael Collison <michael.collison@arm.com>
4562 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
4564 (arm_gen_movmemqi): Removed unused variable 'i'.
4565 Convert 'for' loop into 'while' loop.
4566 (arm_expand_prologue): Remove last unnecessary set of insn.
4567 (thumb_pop): Remove unused variable 'pushed_words'.
4568 (thumb_exit): Remove last unnecessary set of regs_to_pop.
4570 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4572 * config/s390/predicates.md: Use s390_rel_address_ok_p.
4573 * config/s390/s390-protos.h: Add prototype of
4574 s390_rel_address_ok_p.
4575 * config/s390/s390.c (s390_got_symbol): New function.
4576 (s390_rel_address_ok_p): New function.
4577 (legitimize_pic_address): Use s390_rel_address_ok_p.
4578 (s390_load_got): Use s390_got_symbol.
4579 (s390_option_override): Issue error if
4580 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
4581 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
4583 * config/s390/s390.opt: New option mpic-data-is-text-relative.
4585 2017-06-27 Andrew Pinski <apinski@cavium.com>
4587 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
4588 (X * copysign (1.0, X)): New pattern.
4589 (X * copysign (1.0, -X)): New pattern.
4590 (copysign (-1.0, CST)): New pattern.
4592 2017-06-27 Joseph Myers <joseph@codesourcery.com>
4594 * genmultilib (combination_space): Remove variable.
4595 Validate reuse rules against regular expression for any sequence
4596 of multilib options in any order.
4598 2017-06-27 Michael Collison <michael.collison@arm.com>
4600 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
4601 call aarch64_split_simd_combine.
4602 * (aarch64_combine_internal<mode>): Delete pattern.
4603 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
4604 Allow register and subreg operands.
4606 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4608 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
4609 specific need, just fallback on defaults.
4610 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
4612 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4613 Olivier Hainque <hainque@adacore.com>
4615 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
4617 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
4618 targets. Pick a default if no particular attempt applied.
4619 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
4622 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4624 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
4625 (x86_64-wrs-vxworks7): Likewise.
4627 2017-06-27 Marek Polacek <polacek@redhat.com>
4630 * ubsan.c (instrument_null): Check get_base_address's result for null.
4632 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
4634 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
4636 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
4638 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
4639 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
4640 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
4641 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
4643 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
4644 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
4645 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
4646 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
4647 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
4648 BUILT_IN_FEUPDATEENV): New builtins.
4649 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
4650 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
4651 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
4652 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
4654 (builtin_structptr_types): Adjust size.
4655 * tree.c (builtin_structptr_types): Add four entries.
4657 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4658 Olivier Hainque <hainque@adacore.com>
4660 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
4661 (TLS_SYM): New local macro, forcing reference to __tls__ on
4662 link command lines for VxWorks 7 RTPs, triggering initialization
4664 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
4665 OS features TLS support, true for RTPs on VxWorks 7.
4666 * config/vxworks.c (vxworks_override_options): Setup emutls
4669 2017-06-27 Jakub Jelinek <jakub@redhat.com>
4671 * predict.c (test_prediction_value_range): Use -1U instead of -1
4672 to avoid narrowing conversion warning.
4673 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
4674 to avoid narrowing conversion warning.
4675 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
4677 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
4679 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4681 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
4682 64bit configurations.
4683 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
4684 (SIZE_TYPE): Likewise.
4685 * config/vxworks.c (vxworks_emutls_var_fields): Use
4686 long_unsigned_type_node instead of unsigned_type_node as the offset
4687 field type, which is "pointer" mode in emutls.c.
4689 2017-06-27 Jakub Jelinek <jakub@redhat.com>
4692 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
4695 * gimple-fold.c (replace_call_with_call_and_fold): Handle
4696 gimple_vuse copying separately from gimple_vdef copying.
4698 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4700 * value-prof.c (free_hist): Remove call to memset and the enclosing if
4703 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
4704 Olivier Hainque <hainque@adacore.com>
4706 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
4707 for all vxworks7 targets.
4708 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
4709 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
4710 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
4711 variations for VX6/VX7 and 32/64bits later on in ...
4712 (VXWORKS_LIB_SPEC): Leverage new macros.
4713 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
4714 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
4716 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
4718 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
4719 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
4721 2017-06-26 Carl Love <cel@us.ibm.com>
4723 * config/rs6000/rs6000-c.c: Add support for built-in functions
4724 vector bool char vec_reve (vector bool char);
4725 vector signed char vec_reve (vector signed char);
4726 vector unsigned char vec_reve (vector unsigned char);
4727 vector bool int vec_reve (vector bool int);
4728 vector signed int vec_reve (vector signed int);
4729 vector unsigned int vec_reve (vector unsigned int);
4730 vector bool long long vec_reve (vector bool long long);
4731 vector signed long long vec_reve (vector signed long long);
4732 vector unsigned long long vec_reve (vector unsigned long long);
4733 vector bool short vec_reve (vector bool short);
4734 vector signed short vec_reve (vector signed short);
4735 vector double vec_reve (vector double);
4736 vector float vec_reve (vector float);
4737 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
4738 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
4739 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
4740 (altivec_vreve): New pattern.
4741 * config/rs6000/altivec.h (vec_reve): New define.
4742 * doc/extend.texi (vec_rev): Update the built-in documentation file
4743 for the new built-in functions.
4745 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4747 PR tree-optimization/71815
4748 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
4750 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
4752 (slsr_process_phi): Likewise.
4753 (replace_uncond_cands_and_profitable_phis): Don't replace a
4754 multiply candidate with a stride of 1 (copy or cast).
4755 (phi_incr_cost): Call uses_consumed_by_stmt rather than
4757 (lowest_cost_path): Likewise.
4758 (total_savings): Likewise.
4760 2017-06-26 Richard Biener <rguenther@suse.de>
4763 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
4764 Use def_builtin_pure for all gather builtins.
4766 2017-06-26 Richard Biener <rguenther@suse.de>
4768 PR tree-optimization/81203
4769 * tree-tailcall.c (find_tail_calls): Do not move stmts into
4772 2017-06-26 Marek Polacek <polacek@redhat.com>
4775 * doc/invoke.texi: Document -Wmultistatement-macros.
4777 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
4779 * doc/sourcebuild.texi (ARM-specific attributes): Document new
4780 arm_neon_ok_no_float_abi effective target.
4782 2017-06-26 Richard Biener <rguenther@suse.de>
4784 PR tree-optimization/80928
4785 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
4786 (copy_bbs): Set BB_DUPLICATED flag early.
4787 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
4789 (execute_on_shrinking_pred): Likewise.
4790 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
4791 BB_DUPLICATED blocks.
4792 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
4793 iterate over all PHIs considering removal of *gsi.
4795 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
4797 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
4800 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4802 * config/rs6000/rs6000-string.c: (expand_block_clear,
4803 do_load_for_compare, select_block_compare_mode,
4804 compute_current_alignment, expand_block_compare,
4805 expand_strncmp_align_check, expand_strn_compare,
4806 expand_block_move, rs6000_output_load_multiple)
4807 Move functions related to string/block move/compare
4809 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
4810 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
4811 for this function which is now used in two files.
4812 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
4813 * config.gcc: Add rs6000-string.o to extra_objs for
4814 targets powerpc*-*-* and rs6000*-*-*.
4816 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4819 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
4820 32-bit, since indexed is not valid for DImode.
4821 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
4822 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
4823 (define_peephole2 for Altivec d-form load): Add 32-bit support.
4824 (define_peephole2 for Altivec d-form store): Likewise.
4827 * multiple_target.c (create_dispatcher_calls): Only create the
4828 dispatcher call if the function is the default clone of a
4831 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
4834 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
4835 a call, force the call to not be a tail call.
4837 2017-06-23 Jeff Law <law@redhat.com>
4839 * doc/contrib.texi: Add entry for Steven Pemberton's work on
4842 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
4844 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
4845 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
4846 handling for early expansion of vector shifts (sl,sr,sra,rl).
4847 (builtin_function_type): Add vector shift right instructions
4848 to the unsigned argument list.
4850 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4852 rtl-optimizatoin/79286
4853 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
4854 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
4855 trap. PIC register plus a const unspec without offset can never trap.
4857 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
4859 * tree.h (builtin_structptr_type): New type.
4860 (builtin_structptr_types): Declare new array.
4861 * tree.c (builtin_structptr_types): New array.
4862 (free_lang_data, build_common_tree_nodes): Use it.
4864 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
4867 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
4870 2017-06-22 Matt Turner <mattst88@gmail.com>
4872 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
4873 Lake models to skylake case. Assume skylake for unknown
4874 models with clflushopt.
4876 2017-06-22 Jeff Law <law@redhat.com>
4878 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
4879 frame sizes that do not satisfy aarch64_uimm12_shift.
4881 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
4883 * profile-count.h (apply_probability,
4884 apply_scale, probability_in): Fix checks for zero.
4886 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4888 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
4889 * doc/cppdiropts.texi (-I @var{dir}): Document it.
4891 2016-06-22 Richard Biener <rguenther@suse.de>
4893 * tree-vect-loop.c (vect_model_reduction_cost): Handle
4894 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
4895 REDUC_MAX_EXPR support.
4896 (vectorizable_reduction): Likewise.
4897 (vect_create_epilog_for_reduction): Likewise.
4899 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
4901 * match.pd (A / (1 << B) -> A >> B): New.
4902 * generic-match-head.c: Include optabs-tree.h.
4903 * gimple-match-head.c: Likewise.
4904 * optabs-tree.h (target_supports_op_p): New.
4905 * optabs-tree.c (target_supports_op_p): New.
4907 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4909 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
4910 $gcc_cv_ld --help output.
4911 (gcc_cv_ld_demangle): Likewise.
4912 (gcc_cv_ld_eh_frame_hdr): Likewise.
4913 (gcc_cv_ld_pie): Likewise.
4914 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
4915 (gcc_cv_ld_buildid): Likewise.
4916 (gcc_cv_ld_sysroot): Likewise.
4917 (ld_bndplt_support): Likewise.
4918 (ld_pushpopstate_support): Likewise.
4919 * configure: Regenerate.
4920 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
4922 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4925 * config/i386/sse.md (round<mode>2): Renumber match_dup and
4926 operands indexes to avoid gap between operands and match_dups.
4928 2017-06-21 Andrew Pinski <apinski@cavium.com>
4930 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
4931 Increment Arith_shift and Arith_shift_reg by 1.
4932 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
4934 * config/aarch64/aarch64.c (thunderx_tunings): Enable
4935 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
4936 (aarch64_strip_extend): Add new argument and test for it.
4937 (aarch64_cheap_mult_shift_p): New function.
4938 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
4939 add a cost if it is true.
4940 Update calls to aarch64_strip_extend.
4941 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
4943 2017-06-21 Andrew Pinski <apinski@cavium.com>
4945 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
4947 (thunderxt88): Likewise.
4948 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
4949 (thunderx_prefetch_tune): New variable.
4950 (thunderx2t99_prefetch_tune): Update for the correct values.
4951 (thunderxt88_tunings): New variable.
4952 (thunderx_tunings): Use thunderx_prefetch_tune instead of
4953 generic_prefetch_tune.
4954 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
4956 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4958 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
4959 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
4960 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
4961 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
4962 (aarch64_atomic_cas<mode>, GPI): Likewise.
4964 2017-06-21 Martin Liska <mliska@suse.cz>
4966 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
4967 statements on cold and hot labels.
4968 * predict.c (tree_estimate_probability_bb): Remove the
4969 prediction from this place.
4971 2017-06-21 Martin Liska <mliska@suse.cz>
4973 PR tree-optimization/79489
4974 * gimplify.c (maybe_add_early_return_predict_stmt): New
4976 (gimplify_return_expr): Call the function.
4977 * predict.c (tree_estimate_probability_bb): Remove handling
4979 * predict.def: Update comment about early return predictor.
4980 * gimple-predict.h (is_gimple_predict): New function.
4981 * predict.def: Change default value of early return to 66.
4982 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
4984 * passes.def: Put pass_strip_predict_hints to the beginning of
4987 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
4989 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
4990 FUNCTION_DECL declarations.
4991 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
4993 (dwaf2out_decl): Likewise.
4994 * godump.c (go_early_global_decl): Skip call to the real debug hook
4995 for FUNCTION_DECL declarations.
4996 * passes.c (rest_of_decl_compilation): Skip call to the
4997 early_global_decl debug hook for FUNCTION_DECL declarations, unless
4998 -fdump-go-spec is passed.
5000 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
5002 * config/i386/i386.c (struct builtin_isa): New field pure_p.
5003 Reorder for compactness.
5004 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
5005 (def_builtin_pure, def_builtin_pure2): New functions.
5006 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
5008 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
5010 * match.pd (nop_convert): New predicate.
5011 ((A +- CST1) +- CST2): Allow some NOP conversions.
5013 2017-06-21 Jakub Jelinek <jakub@redhat.com>
5016 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
5017 with ctors/dtors if GOVD_SHARED is set.
5019 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
5021 * config/aarch64/aarch64.md (movti_aarch64):
5022 Emit mov rather than orr.
5023 (movtf_aarch64): Likewise.
5024 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
5025 Emit mov rather than orr.
5027 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
5029 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
5030 Swap alternatives, make integer dup more expensive.
5032 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
5034 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
5035 Return true for non-tls symbols.
5037 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
5039 * config/aarch64/aarch64-cores.def (cortex-a55): New.
5040 (cortex-a75): Likewise.
5041 (cortex-a75.cortex-a55): Likewise.
5042 * config/aarch64/aarch64-tune.md: Regenerate.
5043 * doc/invoke.texi (-mtune): Document new values for -mtune.
5045 2017-06-21 Tom de Vries <tom@codesourcery.com>
5047 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
5049 (Effective-Target Keywords, Other attributes): Suggest using
5050 dg-add-options stack_size feature to get stack limit in stack_size
5051 effective target documentation.
5053 2017-06-21 Julian Brown <julian@codesourcery.com>
5054 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5056 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
5057 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
5058 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
5060 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
5061 attribute type list for neon_multiply.
5062 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
5063 attribute type list for neon_multiply.
5064 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
5065 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
5066 attribute type list for neon_multiply.
5067 * config/arm/types.md (crypto_pmull): Add.
5068 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
5069 attribute type list.
5071 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
5073 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
5076 2017-06-20 Jakub Jelinek <jakub@redhat.com>
5078 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
5079 to make sure not to dereference a NULL cost_classes_ptr pointer.
5081 2017-06-20 Carl Love <cel@us.ibm.com>
5083 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5084 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5085 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5086 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5087 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5088 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5089 VMULOSW): New enum "unspec" values.
5090 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
5091 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
5092 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5093 altivec_vmulosw): New patterns.
5094 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5095 VMULOSW): Add definitions.
5097 2017-06-20 Julia Koval <julia.koval@intel.com>
5099 * config/i386/i386.c: Fix rounding expand for new pattern.
5100 * config/i386/subst.md: Fix pattern (parallel -> unspec).
5102 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
5104 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
5105 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
5107 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
5109 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
5112 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
5114 * config/aarch64/aarch64-cores.def: Rearrange to sort by
5115 architecture, then by implementer ID.
5116 * config/aarch64/aarch64-tune.md: Regenerate.
5118 2017-06-20 Richard Biener <rguenther@suse.de>
5121 * fold-const.c (split_tree): Fold to type before negating.
5123 2017-06-20 David Malcolm <dmalcolm@redhat.com>
5125 * diagnostic-show-locus.c
5126 (selftest::test_fixit_deletion_affecting_newline): New function.
5127 (selftest::diagnostic_show_locus_c_tests): Call it.
5129 2017-06-20 Andreas Schwab <schwab@suse.de>
5132 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
5135 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
5137 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
5138 __ARM_FEATURE_COPROC according to support.
5140 2017-06-20 Jakub Jelinek <jakub@redhat.com>
5142 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
5143 Rewritten to avoid overflow for > 32-bit pointers.
5146 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
5147 by removing enum keyword.
5148 (ubsan_type_descriptor): Likewise. Formatting fix.
5151 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
5152 splitter): Require TARGET_SSE2 in the condition.
5154 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
5157 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
5158 for doing vector set of SFmode on ISA 3.0.
5159 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
5160 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
5162 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
5163 SFmode value into a V4SF variable that was extracted from another
5164 V4SF variable without converting the element to double precision
5165 and back to single precision vector format.
5166 (vsx_insert_extract_v4sf_p9_2): Likewise.
5168 2017-06-19 Jakub Jelinek <jakub@redhat.com>
5170 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
5171 in UWHI to avoid undefined overflow.
5174 * ubsan.h (enum ubsan_encode_value_phase): New.
5175 (ubsan_encode_value): Change second argument to
5176 enum ubsan_encode_value_phase with default value of
5177 UBSAN_ENCODE_VALUE_GENERIC.
5178 * ubsan.c (ubsan_encode_value): Change second argument to
5179 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
5180 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
5181 create_tmp_var_raw instead of create_tmp_var and use a
5183 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
5184 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
5185 ubsan_encode_value callers.
5188 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
5189 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
5190 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
5192 2017-06-19 Richard Biener <rguenther@suse.de>
5195 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
5196 estimates if we changed anything.
5198 2017-06-19 Richard Biener <rguenther@suse.de>
5200 PR tree-optimization/80887
5201 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
5202 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
5203 simplified lookups, then reset mprts_hook.
5204 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
5206 (try_to_simplify): Likewise.
5208 2017-06-19 Martin Liska <mliska@suse.cz>
5211 * gimplify.c (gimplify_switch_expr):
5212 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
5214 2017-06-19 Martin Liska <mliska@suse.cz>
5216 * doc/install.texi: Document that PGO runs in 4 stages.
5218 2017-06-19 Martin Liska <mliska@suse.cz>
5221 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
5222 to dispatcher function name.
5223 * multiple_target.c (replace_function_decl): New function.
5224 (create_dispatcher_calls): Redirect both edges and references.
5226 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
5228 * profile-count.c (profile_count::dump): Dump quality.
5229 (profile_count::differs_from_p): Update for unsigned val.
5230 * profile-count.h (profile_count_quality): New enum.
5231 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
5233 2017-06-19 Richard Biener <rguenther@suse.de>
5235 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
5236 struct function as arg.
5237 (estimate_numbers_of_iterations): Export overload with loop arg.
5238 (free_numbers_of_iterations_estimates_loop): Use an overload of
5239 free_numbers_of_iterations_estimates instead.
5240 * tree-cfg.c (remove_bb): Adjust.
5241 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
5242 * tree-parloops.c (gen_parallel_loop): Likewise.
5243 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
5245 (tree_unroll_loops_completely): Likewise.
5246 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
5247 Use an overload instead and export.
5248 (estimated_loop_iterations): Adjust.
5249 (max_loop_iterations): Likewise.
5250 (likely_max_loop_iterations): Likewise.
5251 (estimate_numbers_of_iterations): Take struct function as arg
5253 (loop_exits_before_overflow): Adjust.
5254 (free_numbers_of_iterations_estimates_loop): Use an overload.
5255 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
5256 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
5258 2017-06-19 Richard Biener <rguenther@suse.de>
5261 * ipa-prop.c (find_constructor_constant_at_offset): Handle
5262 RANGE_EXPR conservatively.
5264 2017-06-16 Carl Love <cel@us.ibm.com>
5266 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5267 definitions for vec_float, vec_float2, vec_floato,
5268 vec_floate built-ins.
5269 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
5270 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
5272 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
5273 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
5274 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
5275 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
5276 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
5277 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
5278 vec_floato): Add builtin defines.
5279 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
5280 Update the built-in documentation file for the new built-in
5283 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5285 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
5286 (mthumb): Mark as the negative of -marm.
5288 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5290 * doc/invoke.texi (ARM Options, -mcpu): Document supported
5292 (ARM Options, -mtune): Document that this accepts the same
5293 extension options as -mcpu.
5294 (ARM Options, -mfpu): Document addition of -mfpu=auto.
5296 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5298 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
5299 permitted extensions.
5301 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5303 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
5304 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
5305 (armv8-m.main): Add option +nodsp.
5306 * config/arm/arm-cpu-cdata.h: Regenerated.
5308 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5310 * config/arm/t-fuchsia: New file.
5311 * config.gcc (arm*-*-fuchsia*): Use it.
5313 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5315 * config/arm/t-symbian: Rewrite for new option infrastructure.
5317 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5319 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
5320 (MULTILIB_REQUIRED): Likewise.
5322 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5324 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
5325 (MULTILIB_RESUE): Likewise.
5326 (MULTILIB_MATCHES): Likewise.
5327 (MULTLIB_REQUIRED): Likewise.
5329 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5331 * config/arm/t-rtems: Rewrite for new option framework.
5333 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5335 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
5336 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
5337 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
5338 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
5339 * config/arm/t-multilib: ... here.
5340 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
5341 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
5342 armv7-a and armv8*-a when A-profile libraries have not been built.
5343 * config/arm/t-rmprofile: Rewrite.
5345 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5347 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
5348 with a backslash. Remove the backslash after substituting unescaped
5350 * doc/fragments.texi (MULTILIB_REUSE): Document it.
5352 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5354 * config.gcc: (arm*-*-*): When building a-profile libraries, force
5355 the driver to pass through the default setting of -mfloat-abi.
5356 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
5358 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
5359 (all_feat_combs): New rule.
5360 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
5362 * config/arm/t-aprofile: Rewrite.
5364 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5366 * config/arm/arm.h (FPUTYPE_AUTO): Define.
5367 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
5368 fpu is not specified by the user/command-line.
5369 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
5370 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
5371 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
5372 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
5373 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
5374 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
5376 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5378 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
5379 * config/arm/t-arm-elf: Rewritten.
5381 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5383 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
5384 have some floating-point instructions.
5385 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
5386 (TARGET_MAYBE_HARD_FLOAT): New macro.
5387 * config/arm/arm-builtins.c (arm_init_builtins): Use
5388 TARGET_MAYBE_HARD_FLOAT.
5389 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
5391 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5393 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
5394 (configargs.h): Include it.
5395 (arm_print_hint_for_fpu_option): New function.
5396 (arm_parse_fpu_option): New function.
5397 (candidate_extension): New class.
5398 (arm_canon_for_multilib): New function.
5399 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
5400 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
5401 (ARCH_CANONICAL_SPECS): New macro.
5402 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
5404 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5406 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
5407 are set after handling multilib fragments. Set target_cpu_default2
5410 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5412 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
5414 (arm*-*-*): Set target_cpu_default2 to a quoted string.
5415 * config/arm/parsecpu.awk (check_cpu): Validate any extension
5417 (check_arch): Likewise.
5418 * config/arm/arm.c (arm_configure_build_target): Handle
5419 TARGET_CPU_DEFAULT being a string constant. Scan any feature
5420 options in the default.
5422 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5424 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
5425 when an option is an alias of another.
5426 * config/arm/parsecpu.awk (optalias): New parser token.
5427 (gen_comm_data): Mark non-alias options as such. Emit entries
5428 for extension aliases.
5429 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
5430 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
5431 (armv6kz, armv6zk, armv6t2): Likewise.
5432 (armv7): Make vfpv3-d16 an alias.
5433 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
5435 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
5436 Sort in canonical order.
5437 (armv8-a): Sort in canonical order.
5438 (armv8.1-a, armv8.2-a): Likewise.
5439 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
5441 (cortex-a9): Sort in canonical order.
5442 * config/arm/arm.c (selftests.h): Include it.
5443 (arm_test_cpu_arch_data): New function.
5444 (arm_run_self_tests): New function.
5445 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
5446 (targetm): Move declaration to the end of the file.
5447 * arm-cpu-cdata.h: Regenerated.
5449 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5451 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
5452 call to target_mode_check describing the type of option passed.
5453 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
5454 (arm_target_thumb_only): Use arm_parse_arch_option_name or
5455 arm_parse_cpu_option_name to match parameters against list of
5457 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
5458 arm_arch_core_flags data structure.
5459 * config/arm/arm-cpu_cdata.h: Regenerated.
5461 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5463 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
5465 (arm_print_hint_for_cpu_option): Likewise.
5466 (arm_print_hint_for_arch_option): Likewise.
5467 (arm_parse_cpu_option_name): Likewise.
5468 (arm_parse_arch_option_name): Likewise.
5469 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
5470 of entries in the all_fpus list.
5471 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
5472 (arm_parse_cpu_option_name): Declare.
5473 (arm_parse_arch_option_name): Declare.
5474 (arm_parse_option_features): Declare.
5475 (arm_intialize_isa): Declare.
5476 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
5478 (gen_comm_data): ... here. Make definitions non-static.
5479 * config/arm/arm-cpu-data.h: Regenerated.
5480 * config/arm/arm-cpu-cdata.h: Regenerated.
5482 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5484 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
5485 (cpu_arch_extension): New structure.
5486 (cpu_arch_option, arch_option, cpu_option): New structures.
5487 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
5489 (gen_data): Generate new format data tables.
5490 * config/arm/arm.c (cpu_tune): New structure.
5491 (cpu_option, processors): Delete.
5492 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
5493 (arm_print_hint_for_cpu_option): ... this and ...
5494 (arm_print_hint_for_arch_option): ... this.
5495 (arm_parse_arch_cpu_name): Delete. Replace with ...
5496 (arm_parse_cpu_option_name): ... this and ...
5497 (arm_parse_arch_option_name): ... this.
5498 (arm_unrecognized_feature): Change type of target parameter to
5500 (arm_parse_arch_cpu_features): Delete. Replace with ...
5501 (arm_parse_option_features): ... this.
5502 (arm_configure_build_target): Rework to use new configuration data
5504 (arm_print_tune_info): Rework for new configuration data tables.
5505 * config/arm/arm-cpu-data.h: Regenerated.
5506 * config/arm/arm-cpu.h: Regenerated.
5508 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5510 * Makefile.in (OBJS): Move sbitmap.o from here ...
5511 (OBJS-libcommon): ... to here.
5513 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5515 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
5516 (ISA_ALL_CRYPTO): New macro.
5517 (ISA_ALL_SIMD): New macro
5518 (ISA_ALL_FP): New macro.
5519 * config/arm/arm.c (fpu_bitlist): Update initializer.
5520 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
5522 (arm9e): Add fpu. Add option for nofp
5523 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
5524 (arm926ej-s, arm1026ej-s): Likewise.
5525 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
5526 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
5527 neon-fp16, neon-vfpv4, nofp and nosimd.
5528 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
5529 (cortex-a8): Add fpu. Add option for nofp.
5530 (cortex-a9): Add fpu. Add options for nosimd and nofp.
5531 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
5532 (cortex-r4f): Add fpu.
5533 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
5534 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
5536 (cortex-r8): Likewise.
5537 (cortex-m4): Add fpu. Add option for nofp.
5538 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
5539 (cortex-a17.cortex-a7): Likewise.
5540 (cortex-a32): Add fpu. Add options for crypto and nofp.
5541 (cortex-a35, cortex-a53): Likewise.
5542 (cortex-a57): Add fpu. Add option for crypto.
5543 (cortex-a72, cortex-a73): Likewise.
5544 (exynos-m1): Likewise.
5545 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
5546 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
5547 (cortex-m33): Add fpu. Add option for nofp.
5548 * config/arm/arm-cpu-cdata.h: Regenerated
5549 * config/arm/arm-cpu-data.h: Regenerated.
5551 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5553 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
5554 (armv5te, armv5tej): Likewise.
5555 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
5556 (armv7): Add options fp and vfpv3-d16.
5557 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
5558 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
5560 (armv7ve): Likewise.
5561 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
5562 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
5563 (armv8-a): Add nocrypto option.
5564 (armv8.1-a, armv8.2-a): Likewise.
5565 (armv8-m.main): add options fp, fp.dp and nofp.
5567 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5569 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
5571 (armv8-a+crc): Delete.
5572 (armv8.1-a): Add options simd, crypto and nofp.
5573 (armv8.2-a): Add options fp16, simd, crypto and nofp.
5574 (armv8.2-a+fp16): Delete.
5575 (armv8-m.main): Add option dsp.
5576 (armv8-m.main+dsp): Delete.
5577 (cortex-a8): Add fpu. Add nofp option.
5578 (cortex-a9): Add fpu. Add nofp and nosimd options.
5579 * config/arm/parsecpu.awk (gen_data): Generate option tables and
5580 link to main cpu and architecture data structures.
5581 (gen_comm_data): Only put isa attributes from the main architecture
5583 (option): New statement for architecture and CPU entries.
5584 * arm.c (struct cpu_option): New structure.
5585 (struct processors): Add entry for options.
5586 (arm_unrecognized_feature): New function.
5587 (arm_parse_arch_cpu_name): Ignore any characters after the first
5589 (arm_parse_arch_cpu_feature): New function.
5590 (arm_configure_build_target): Separate out any CPU and architecture
5591 features and parse separately. Don't error out if -mfpu=auto is
5592 used with only an architecture string.
5593 (arm_print_asm_arch_directives): New function.
5594 (arm_file_start): Call it.
5595 * config/arm/arm-cpu-cdata.h: Regenerated.
5596 * config/arm/arm-cpu-data.h: Likewise.
5597 * config/arm/arm-tables.opt: Likewise.
5599 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5601 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
5602 assembler when it is not -mfpu=auto.
5604 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5606 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
5607 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
5608 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
5609 (ASM_CPU_SPEC): Rewrite.
5610 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
5611 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
5612 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
5613 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
5614 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
5615 copied string is NUL-terminated. Also strip any characters prefixed
5617 (arm_rewrite_selected_arch): New function.
5618 (arm_rewrite_march): New function.
5620 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5622 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
5623 (x_arm_cpu_string, x_arm_tune_string): Likewise.
5624 (march, mcpu, mtune): Convert to string-based options.
5625 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
5626 (arm_parse_arch_cpu_name): New function.
5627 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
5628 identify selected architecture or CPU.
5629 (arm_option_save): New function.
5630 (TARGET_OPTION_SAVE): Redefine.
5631 (arm_option_restore): Restore string options.
5632 (arm_option_print): Print string options.
5634 2017-06-16 Martin Sebor <msebor@redhat.com>
5636 PR tree-optimization/80933
5637 PR tree-optimization/80934
5638 * builtins.c (fold_builtin_3): Do not handle bcmp here.
5639 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
5640 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
5641 (gimple_fold_builtin): Call them.
5643 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5645 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
5646 as unlikely; update profile.
5648 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5650 * predict.c (force_edge_cold): Handle declaring edges impossible
5653 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5655 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
5657 (try_unroll_loop_completely): Fix reporting.
5659 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5661 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
5663 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
5666 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
5667 if given a non-constant argument for an intrinsic which requires a
5670 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5672 * profile.c (compare_freqs): New function.
5673 (branch_prob): Sort edge list.
5674 (find_spanning_tree): Assume that the list is priority sorted.
5676 2017-06-16 Richard Biener <rguenther@suse.de>
5678 PR tree-optimization/81090
5679 * passes.def (pass_record_bounds): Remove.
5680 * tree-pass.h (make_pass_record_bounds): Likewise.
5681 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
5682 make_pass_record_bounds): Likewise.
5683 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
5684 not free niter estimates at the beginning but at the end.
5685 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
5687 2017-06-16 Richard Biener <rguenther@suse.de>
5689 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
5690 initializer to workaround ICE in host GCC 4.8.
5692 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5694 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
5696 (clone_inlined_nodes): Update.
5698 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5700 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
5701 prefetch settings, and enable prefetching by default at -O3.
5703 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5705 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5706 Set flag_prefetch_loop_arrays according to tuning data.
5708 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5710 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
5712 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
5713 [Unrelated to main purpose of the patch] Place the pointer field last
5714 to enable type checking errors when tune structure are wrongly merged.
5715 * config/aarch64/aarch64.c (generic_prefetch_tune,)
5716 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
5717 (thunderx2t99_prefetch_tune): New tune constants.
5718 (tune_params *_tunings): Update all tunings (no functional change).
5719 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
5720 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
5721 from tunings structures.
5723 2017-06-16 Jakub Jelinek <jakub@redhat.com>
5726 * ubsan.c (instrument_null): Add T argument, use it instead
5727 of computing it based on IS_LHS.
5728 (instrument_object_size): Likewise.
5729 (pass_ubsan::execute): Adjust instrument_null and
5730 instrument_object_size callers to pass gimple_get_lhs or
5731 gimple_assign_rhs1 result to it. Use instrument_null instead of
5732 calling get_base_address and instrument_mem_ref. Handle
5733 aggregate call arguments for object-size sanitization.
5735 2017-06-16 Yury Gribov <tetra2005@gmail.com>
5737 PR tree-optimization/81089
5738 * tree-vrp.c (is_masked_range_test): Validate operands of
5741 2017-06-15 Martin Sebor <msebor@redhat.com>
5744 * dumpfile.c (dump_register): Avoid calling memset to initialize
5745 a class with a default ctor.
5746 * gcc.c (struct compiler): Remove const qualification.
5747 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
5748 * hash-table.h: Ditto.
5749 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
5751 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
5752 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
5754 * params.h (struct param_info): Make struct members non-const.
5755 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
5756 with default initialization.
5757 * vec.h (vec_copy_construct, vec_default_construct): New helper
5759 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
5760 with vec_copy_construct.
5761 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
5762 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
5763 * doc/invoke.texi (-Wclass-memaccess): Document.
5765 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5767 * emit-rtl.h (is_leaf): Update comment about local
5770 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5773 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
5774 for a variable to have a section before checking if the section has a
5776 Set section to.persistent if persistent attribute is set.
5777 Warn if .persistent attribute is used on an automatic variable.
5779 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
5781 PR rtl-optimization/80474
5782 * reorg.c (update_block): Do not ignore instructions in a delay slot.
5784 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
5786 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
5789 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
5791 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
5792 (casesi): Emit bounds checking as RTL.
5793 (casesi_internal_mips16_<mode>): Remove bounds checking.
5795 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
5797 * config/xtensa/xtensa.c (xtensa_option_override): Append
5798 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
5799 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
5800 xtensa_doloop_hooks): Define unconditionally.
5801 (xtensa_reorg_loops): Only call reorg_loops in the presence of
5803 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
5804 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
5805 for it in xtensa_option_override.
5806 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
5807 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
5809 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
5811 * doc/cppopts.texi: Document '-' special value to -MF.
5813 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
5815 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
5816 (cortex_a53_fconst): Likewise.
5817 (cortex_a53_fpmul): Likewise.
5818 (cortex_a53_f_load_64): Likewise.
5819 (cortex_a53_f_load_many): Likewise.
5820 (cortex_a53_advsimd_alu): Likewise.
5821 (cortex_a53_advsimd_alu_q): Likewise.
5822 (cortex_a53_advsimd_mul): Likewise.
5823 (cortex_a53_advsimd_mul_q): Likewise.
5824 (fpmac bypass): Add new bypass for fpmac-fpmac case.
5825 Add missing fmul, r2f_cvt and fconst cases.
5827 2017-06-14 Richard Biener <rguenther@suse.de>
5830 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
5832 (fold_binary_loc): When associating do not treat pre-existing
5833 TREE_OVERFLOW on literal constants as a reason to allow
5834 TREE_OVERFLOW on associated literal constants.
5836 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5838 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
5839 (MASK_FEATURES): New macro.
5840 * config/sparc/sparc.c (sparc_option_override): Remove the special
5841 handling of -mfpu and generalize it to all MASK_FEATURES switches.
5843 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5845 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
5846 a division of 0 if non-call exceptions are enabled.
5848 2017-06-14 Andrew Pinski <apinski@cavium.com>
5849 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5852 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
5853 Improve vector initialization code gen for only variable case.
5855 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5857 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
5859 2017-06-14 Richard Biener <rguenther@suse.de>
5861 PR tree-optimization/81083
5862 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
5865 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5867 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
5868 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
5869 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
5870 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
5871 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
5872 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
5874 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5876 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
5877 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
5879 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5881 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
5883 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5885 * config/rs6000/t-rtems: Don't handle SPE.
5887 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5889 * config/rs6000/t-linux: Don't handle SPE.
5891 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5893 * config/rs6000/eabispe.h: Delete file.
5895 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5897 * config/rs6000/t-spe: Delete file.
5899 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5901 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
5902 (rs6000_legitimate_offset_address_p): Return false for anything in
5903 V2SImode or V2SFmode.
5905 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5907 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
5908 except V2SF and V2SI. Rearrange the vector modes, and add comments.
5909 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
5911 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
5912 (rs6000_legitimate_offset_address_p): Ditto.
5913 (rs6000_emit_move): Ditto.
5914 (rs6000_init_builtins): Remove V4HI_type_node.
5916 2017-06-13 Martin Liska <mliska@suse.cz>
5919 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
5920 (gate_asan): Likewise.
5921 * asan.h (asan_no_sanitize_address_p): Remove the function.
5922 (sanitize_flags_p): New function.
5923 * builtins.def: Fix coding style.
5924 * common.opt: Use renamed enum value.
5925 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
5926 * doc/extend.texi: Document no_sanitize attribute.
5927 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
5928 to SANITIZE_UNDEFINED_NONDEFAULT.
5929 * gcc.c (sanitize_spec_function): Use the renamed enum value.
5930 * gimple-fold.c (optimize_atomic_compare_exchange_p):
5931 Use sanitize_flags_p.
5932 * gimplify.c (gimplify_function_tree): Likewise.
5933 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
5934 * opts.c (parse_no_sanitize_attribute): New function.
5935 (common_handle_option): Use renamed enum value.
5936 * opts.h (parse_no_sanitize_attribute): Declare.
5937 * tree.c (sanitize_flags_p): New function.
5938 * tree.h: Declared here.
5939 * tsan.c: Use sanitize_flags_p.
5940 * ubsan.c (ubsan_expand_null_ifn): Likewise.
5941 (instrument_mem_ref): Likewise.
5942 (instrument_bool_enum_load): Likewise.
5943 (do_ubsan_in_current_function): Remove the function.
5944 (pass_ubsan::execute): Use sanitize_flags_p.
5945 * ubsan.h: Remove do_ubsan_in_current_function
5946 * tree-cfg.c (print_no_sanitize_attr_value): New function.
5947 (dump_function_to_file): Use it here.
5949 2017-06-13 Martin Jambor <mjambor@suse.cz>
5951 PR tree-optimization/80803
5952 PR tree-optimization/81063
5953 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
5954 (propagate_subaccesses_across_link): Enqueue subtree whenever
5955 necessary instead of relying on the caller.
5957 2017-06-13 Martin Jambor <mjambor@suse.cz>
5959 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
5960 that have a first_link.
5961 (sort_and_splice_var_accesses): Do not check first_link before
5963 (subtree_mark_written_and_enqueue): Likewise.
5964 (propagate_all_subaccesses): Likewise and do not stop at first
5965 parent with a first_link.
5967 2017-06-13 Martin Jambor <mjambor@suse.cz>
5969 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
5972 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5974 * match.pd: New pattern.
5976 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5978 * tree-vrp.c (is_masked_range_test): New function.
5979 (register_edge_assert_for): Determine ranges for
5982 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5984 PR tree-optimization/67328
5985 * fold-const.c (maskable_range_p): New function.
5986 (build_range_check): Generate bittests if possible.
5988 2017-06-13 Martin Liska <mliska@suse.cz>
5990 * gimple-pretty-print.c (dump_probability): Add new argument.
5991 (dump_edge_probability): Dump both probability and count.
5992 (dump_gimple_label): Likewise.
5993 (dump_gimple_bb_header): Likewise.
5995 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
5998 * config/avr/avr-devices.c: Fix indentation.
5999 * config/avr/gen-avr-mmcu-specs.c: Dito.
6001 2017-06-13 Richard Biener <rguenther@suse.de>
6003 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
6004 instead get vector type from stmt_info.
6005 (vectorizable_reduction): Adjust. Remove dead code.
6007 2017-06-13 Richard Biener <rguenther@suse.de>
6010 * fold-const.c (extract_muldiv_1): Remove bogus distribution
6011 case of C * (x * C2 + C3).
6012 (fold_addr_of_array_ref_difference): Properly fold index difference.
6014 2017-06-12 David S. Miller <davem@davemloft.net>
6017 * config/sparc/sparc.md (return expander): Emit frame blockage if
6018 function uses alloca.
6020 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
6022 * combine.c (make_field_assignment): Check len rather than the mode
6023 precision when calling force_to_mode.
6025 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
6027 Support multilibs and devices that see flash in RAM address range.
6030 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
6031 (avr_mcu_t) <flash_pm_offset>: New field.
6032 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
6033 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
6034 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
6035 (AVR_TINY_PM_OFFSET): Remove macro.
6036 * config/avr/avr.opt (-mshort-calls): New option.
6037 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
6038 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
6039 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
6040 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
6041 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
6042 instead of avr_arch->have_jmp_call.
6043 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
6044 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
6045 avr_arch->flash_pm_offset to define.
6046 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
6047 new field flash_pm_offset. Add entry for avrxmega3.
6048 (avr_texinfo): Add entry for avrxmega3.
6049 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
6050 attiny212, attiny214,
6051 attiny412, attiny414, attiny416, attiny417,
6052 attiny814, attiny816, attiny817,
6053 attiny1614, attiny1616, attiny1617,
6054 attiny3214, attiny3216, attiny3217.
6055 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
6056 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
6057 (avr_print_operand_address) [AVR_TINY]: Same.
6058 (avr_asm_init_sections) <readonly_data_section>: Only patch
6059 callback if avr_arch->flash_pm_offset = 0.
6060 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
6061 for rodata if avr_arch->flash_pm_offset != 0.
6062 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
6063 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
6064 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
6065 (m_options): Append opt_rcall.
6066 (m_dirnames): Append dir_rcall.
6067 * config/avr/t-multilib: Regenerate.
6069 * configure.ac [target=avr]: Check whether avrxmega3 default
6070 linker description file works as needed.
6071 * configure: Regenerate.
6072 * doc/avr-mmcu.texi: Regenerate.
6073 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
6074 <__AVR_ARCH__>: Document avrxmega3 and 103.
6075 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
6076 <__AVR_SHORT_CALLS__>: Document it.
6077 <__AVR_PM_BASE_ADDRESS__>: Document it.
6078 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
6079 (AVR Variable Attributes) <progmem>: Document this is
6080 not needed for avrxmega3.
6081 (AVR Named Address Spaces) <__flash>: Dito.
6083 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
6085 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
6087 2017-06-12 Doug Rupp <rupp@adacore.com>
6089 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
6090 Append vxworks-stdint.h to the tm_file list.
6091 * config/vxworks-stdint.h: New file.
6093 2017-06-12 Martin Liska <mliska@suse.cz>
6095 PR tree-optimization/81041
6096 * tree-profile.c (gimple_gen_ic_func_profiler):
6097 Create an extra BB in profile-generate
6098 (gimple_gen_time_profiler): Likewise.
6100 2017-06-12 Jakub Jelinek <jakub@redhat.com>
6102 PR tree-optimization/81003
6103 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
6104 (update_range_test): Use it instead of force_gimple_operand_gsi.
6106 2017-06-12 Richard Biener <rguenther@suse.de>
6108 PR tree-optimization/81053
6109 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
6110 with backedge value not defined in loop. Simplify def stmt
6113 2017-06-11 Tom de Vries <tom@codesourcery.com>
6116 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
6118 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
6119 nvptx_cannot_force_const_mem.
6121 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
6123 * opts.c (finish_options): Move test for flag_split_stack after
6124 it has been initialized.
6126 2017-06-11 Jason Merrill <jason@redhat.com>
6128 * tree.h (id_equal): New.
6129 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
6130 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
6131 instead of strcmp of IDENTIFIER_POINTER.
6133 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
6135 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
6136 (mark_all_inlined_calls_cdtor): Fix formating.
6137 (inline_transform): Rescale profile before inlining.
6139 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
6141 * cgraph.h (cgraph_edge::clone): Update prototype.
6142 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
6143 (cgraph_node::create_clone): Update.
6144 (cgraph_node::create_version_clone): Update.
6145 * tree-inline.c (copy_bb): Update.
6146 (expand_call_inline): Update.
6148 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
6150 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
6151 factored out from ...
6152 (rs6000_emit_prologue): ... here.
6154 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
6156 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
6157 factored out from ...
6158 (rs6000_emit_prologue): ... here.
6160 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
6162 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
6164 (handle_missing_profiles): Fix computation of tp_first_run.
6165 (counts_to_freqs): Do not touch freqs when count is 0.
6167 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
6169 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
6172 2017-06-10 Tom de Vries <tom@codesourcery.com>
6174 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
6175 attributes): Document signal effective target.
6177 2017-06-10 Tom de Vries <tom@codesourcery.com>
6179 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
6180 Document effective target stack_size.
6182 2017-06-09 David Malcolm <dmalcolm@redhat.com>
6184 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
6185 to the edit_context if they can be auto-applied.
6187 2017-06-9 Ian Lance Taylor <iant@golang.org>
6189 * opts.c (finish_options): If -fsplit-stack, disable implicit
6190 -forder-blocks-and-partition.
6191 * doc/invoke.texi (Optimize Options): Document that when using
6192 -fsplit-stack -forder-blocks-and-partition is not implicitly
6195 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
6197 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
6198 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
6199 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
6200 * builtins.def (abort, trap, unreachable): Declare cold.
6201 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
6202 * tree-core.h (ECF_COLD): New.
6203 * tree.c (set_call_expr_flags): Handle ECF_COLD.
6204 (build_common_builtin_nodes): Mark unreachable and abort as cold.
6206 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
6208 * predict.c (unlikely_executed_stmt_p): Cleanup.
6210 2017-06-09 Richard Biener <rguenther@suse.de>
6212 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
6213 model if the ref is always written to.
6215 2017-06-09 Tamar Christina <tamar.christina@arm.com>
6217 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
6219 2017-06-09 Tamar Christina <tamar.christina@arm.com>
6221 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
6224 2017-06-09 Tom de Vries <tom@codesourcery.com>
6227 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
6228 "target cannot support label values" when encountering LABEL_REF.
6230 2017-06-09 Martin Liska <mliska@suse.cz>
6232 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
6233 (gimple_gen_ic_func_profiler): Emit direct comparison
6234 of __gcov_indirect_call_callee with NULL.
6235 (gimple_gen_time_profiler): Change probability from
6236 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
6238 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
6240 * profile.c (edge_gcov_counts): Turn to pointer.
6241 (compute_branch_probabilities, compute_branch_probabilities): Update.
6242 (branch_prob): Do not clear edge_gcov_count.
6243 * profile.h (edge_gcov_counts): Turn to pointer.
6244 (edge_gcov_count): Update.
6246 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
6248 * gimple.h (gimple_check_failed): Mark cold.
6250 2017-06-09 Richard Biener <rguenther@suse.de>
6252 PR tree-optimization/66623
6253 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
6254 refactor check_reduction into two parts, properly computing
6255 whether we have to check reduction validity for outer loop
6258 2017-06-09 Richard Biener <rguenther@suse.de>
6260 PR tree-optimization/79483
6261 * graphite-scop-detection.c (order): New global.
6262 (get_order): Compute bb to order mapping that satisfies code
6263 generation constraints.
6264 (cmp_pbbs): New helper.
6265 (build_scops): Start domwalk at entry block, sort generated
6268 2017-06-09 Richard Biener <rguenther@suse.de>
6271 * ipa-polymorphic-call.c
6272 (ipa_polymorphic_call_context::restrict_to_inner_class):
6273 Skip FIELD_DECLs with error_mark_node type.
6274 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
6277 2017-06-09 Martin Liska <mliska@suse.cz>
6279 * predict.c (struct branch_predictor): New struct.
6280 (test_prediction_value_range): New test.
6281 (predict_c_tests): New function.
6282 * selftest-run-tests.c (selftest::run_tests): Run the function.
6283 * selftest.h: Declare new tests.
6285 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
6288 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
6289 gen_add3_insn did not fail.
6290 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
6291 r0, construct that number in a temporary reg and add that reg to r0.
6292 If asked to put the result in r0 as well, fail.
6294 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
6296 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6297 for early expansion of vec_eqv.
6299 2017-06-08 Jakub Jelinek <jakub@redhat.com>
6302 * ubsan.c (instrument_null): Avoid pointless code temporary.
6303 (pass_ubsan::execute): Instrument aggregate arguments of calls.
6305 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
6309 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
6312 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
6315 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6317 * predict.c (unlikely_executed_edge_p): Move ahead.
6318 (probably_never_executed_edge_p): Use it.
6320 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
6323 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
6324 gimple_call_builtin_p call.
6326 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6328 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
6329 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
6330 rtl_check_failed_type2, rtl_check_failed_code1,
6331 rtl_check_failed_code2, rtl_check_failed_code_mode,
6332 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
6333 rtvec_check_failed_bounds, rtl_check_failed_flag,
6334 _fatal_insn_not_found, _fatal_insn): Likewise.
6335 * tree.h (tree_contains_struct_check_failed,
6336 tree_check_failed, tree_not_check_failed,
6337 tree_class_check_failed, tree_range_check_failed,
6338 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
6339 tree_vec_elt_check_failed, phi_node_elt_check_failed,
6340 tree_operand_check_failed, omp_clause_check_failed,
6341 omp_clause_operand_check_failed, omp_clause_range_check_failed):
6344 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6346 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
6347 flag_branch_probabilities.
6348 * ipa-inline.c (edge_badness): Likewise.
6349 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
6350 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
6351 * predict.c (maybe_hot_frequency_p): Likewise.
6352 (probably_never_executed): Likewise.
6353 * sched-ebb.c (schedule_ebbs): Likewise.
6354 * sched-rgn.c (find_single_block_region): Likewise.
6355 * tracer.c (tail_duplicate): Likewise.
6357 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6359 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
6360 longer requires x_flag_profile_use.
6362 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6364 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
6365 instead of flag_reorder_blocks_and_partition.
6366 * dbxout.c (dbxout_function_end): Likewise.
6367 * dwarf2out.c (gen_subprogram_die): Likewise.
6368 * haifa-sched.c (sched_create_recovery_edges): Likewise.
6369 * hw-doloop.c (reorg_loops): Likewise.
6370 * varasm.c (assemble_start_function,
6371 assemble_end_function): Likewise.
6372 (decide_function_section): Do not check for
6373 flag_reorder_blocks_and_partition.
6375 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
6377 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
6379 (chkp_get_hard_register_fake_addr_expr): Ditto.
6380 (chkp_build_addr_expr): Add check for hard reg case.
6381 (chkp_parse_array_and_component_ref): Ditto.
6382 (chkp_find_bounds_1): Ditto.
6383 (chkp_process_stmt): Don't generate bounds store for
6386 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6388 * predict.c (maybe_hot_bb_p): Do not check profile status.
6389 (maybe_hot_edge_p): Likewise.
6390 (probably_never_executed): Check for zero counts even if profile
6392 (unlikely_executed_edge_p): New function.
6393 (unlikely_executed_stmt_p): New function.
6394 (unlikely_executed_bb_p): New function.
6395 (set_even_probabilities): Use unlikely predicates.
6396 (combine_predictions_for_bb): Likewise.
6397 (predict_paths_for_bb): Likewise.
6398 (predict_paths_leading_to_edge): Likewise.
6399 (determine_unlikely_bbs): New function.
6400 (estimate_bb_frequencies): Use it.
6401 (compute_function_frequency): Use zero counts even if profile is
6403 * profile-count.h: Fix typo.
6405 2017-08-08 Julia Koval <julia.koval@intel.com>
6407 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
6408 _mm512_mask_cvtsepi16_storeu_epi8,
6409 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
6410 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
6411 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
6412 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
6413 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
6414 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
6415 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
6416 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
6417 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
6418 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
6419 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
6420 __builtin_ia32_pmovuswb256mem_mask,
6421 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
6422 __builtin_ia32_pmovwb512mem_mask): New builtins.
6424 2017-08-08 Julia Koval <julia.koval@intel.com>
6426 PR target/73350,80862
6427 * config/i386/subst.md (round): Fix round pattern.
6428 * config/i386/i386.c (ix86_erase_embedded_rounding):
6429 Fix erasing rounding for the fixed pattern.
6431 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6433 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
6435 2017-06-08 Martin Liska <mliska@suse.cz>
6437 PR gcov-profile/80911
6438 * gcov.c (block_info::block_info): New constructor.
6440 2017-06-07 Carl Love <cel@us.ibm.com>
6442 * config/rs6000/rs6000-c: The return type of the following
6443 built-in functions was implemented as int not long long. Fix sign
6444 of return value for the unsigned version of vec_mulo and vec_mule.
6445 vector unsigned long long vec_bperm (vector unsigned long long,
6446 vector unsigned char)
6447 vector signed long long vec_mule (vector signed int,
6449 vector unsigned long long vec_mule (vector unsigned int,
6450 vector unsigned int)
6451 vector signed long long vec_mulo (vector signed int,
6453 vector unsigned long long vec_mulo (vector unsigned int,
6454 vector unsigned int)
6455 * doc/extend.texi: Fix the documentation for the built-in
6458 2017-06-07 Carl Love <cel@us.ibm.com>
6461 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
6464 2017-06-07 Carl Love <cel@us.ibm.com>
6466 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
6467 support, Generate doublehv for signed int/float for BE case only.
6469 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
6471 * doc/invoke.texi (mcx16): Rewrite.
6473 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6475 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
6476 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
6477 *mov<mode>_softfloat, and an anonymous splitter): Use
6478 nonimmediate_operand instead of rs6000_nonimmediate_operand.
6480 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6482 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
6484 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
6485 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
6486 (rs6000_debug_reg_global): Adjust.
6487 (rs6000_init_hard_regno_mode_ok): Adjust.
6488 (rs6000_dbx_register_number): Adjust.
6489 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
6490 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
6491 Remove SPE_ACC and SPEFSCR.
6492 (REG_ALLOC_ORDER): Ditto.
6493 (FRAME_POINTER_REGNUM): Change to 111.
6494 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
6495 (REG_CLASS_NAMES): Ditto.
6496 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
6497 (REGISTER_NAMES): Ditto.
6498 (ADDITIONAL_REG_NAMES): Ditto.
6499 (rs6000_reg_names): Ditto.
6500 * config/rs6000/rs6000.md: Renumber some register number
6503 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6505 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
6507 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
6508 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
6510 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
6511 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
6512 Delete the SPE high registers.
6513 (REG_ALLOC_ORDER): Ditto.
6514 (enum reg_class): Remove SPE_HIGH_REGS.
6515 (REG_CLASS_NAMES): Ditto.
6516 (REG_CLASS_CONTENTS): Delete the SPE high registers.
6517 (REGISTER_NAMES): Ditto.
6518 (rs6000_reg_names): Ditto.
6519 * doc/tm.texi.in: Remove SPE as example.
6520 * doc/tm.texi: Regenerate.
6522 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6524 * config/rs6000/8540.md (ppc8540_brinc): Delete.
6525 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
6526 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
6527 * config/rs6000/rs6000.md (type): Remove "brinc".
6529 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6531 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
6532 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
6533 * config/rs6000/linuxspe.h: Delete file.
6534 * config/rs6000/rs6000.md: Don't include spe.md.
6535 * config/rs6000/spe.h: Delete file.
6536 * config/rs6000/spe.md: Delete file.
6537 * config/rs6000/t-rs6000: Remove spe.md.
6539 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6541 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
6542 (reg_or_none500mem_operand): Delete.
6543 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
6544 instead of reg_or_none500mem_operand.
6546 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6548 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
6549 handling of SPE flags.
6550 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
6552 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6554 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
6556 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
6557 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
6558 paired_divv2sf3): Similar.
6559 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
6560 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
6561 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
6563 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
6564 Rename the paired_* instruction patterns.
6565 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
6567 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
6568 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
6569 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
6570 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
6571 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
6573 (struct machine_function): Delete field spe_insn_chain_scanned_p.
6574 (spe_func_has_64bit_regs_p): Delete.
6575 (spe_expand_predicate_builtin): Delete.
6576 (spe_expand_evsel_builtin): Delete.
6577 (TARGET_DWARF_REGISTER_SPAN): Do not define.
6578 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
6579 (invalid_e500_subreg): Delete.
6580 (rs6000_legitimize_address): Always force_reg op2 as well, for
6581 paired single memory accesses.
6582 (rs6000_member_type_forces_blk): Delete.
6583 (rs6000_spe_function_arg): Delete.
6584 (rs6000_expand_unop_builtin): Delete SPE handling.
6585 (rs6000_expand_binop_builtin): Ditto.
6586 (spe_expand_stv_builtin): Delete.
6587 (bdesc_2arg_spe): Delete.
6588 (spe_expand_builtin): Delete.
6589 (spe_expand_predicate_builtin): Delete.
6590 (spe_expand_evsel_builtin): Delete.
6591 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
6592 (spe_init_builtins): Delete.
6593 (spe_func_has_64bit_regs_p): Delete.
6594 (savres_routine_name): Delete "info" parameter. Adjust callers.
6595 (rs6000_emit_stack_reset): Ditto.
6596 (rs6000_dwarf_register_span): Delete.
6597 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
6598 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
6599 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
6601 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
6603 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
6604 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
6605 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
6606 mulv2sf3, divv2sf3): Delete expanders.
6608 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6610 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
6612 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6614 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
6615 * config/rs6000/rs6000.c: Ditto.
6617 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6619 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
6620 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
6621 comparison_operator.
6623 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6625 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
6626 * config/rs6000/rs6000.opt: Ditto.
6627 * config/rs6000/t-rtems: Ditto.
6629 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6631 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
6632 TARGET_E500_SINGLE by 0, simplify.
6633 * config/rs6000/rs6000.c: Ditto.
6634 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
6635 (spe_build_register_parallel): Delete.
6636 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
6637 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
6638 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
6639 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
6640 (E500_CONVERT): Delete.
6641 * config/rs6000/spe.md: Remove many patterns and all define_constants.
6643 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6645 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
6646 * config/rs6000/dfp.md: Ditto.
6647 (negdd2, *negdd2_fpr): Merge.
6648 (absdd2, *absdd2_fpr): Merge.
6649 (negtd2, *negtd2_fpr): Merge.
6650 (abstd2, *abstd2_fpr): Merge.
6651 * config/rs6000/e500.h: Delete file.
6652 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
6653 TARGET_FPRS by 1 and simplify.
6654 * config/rs6000/rs6000-c.c: Ditto.
6655 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
6657 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
6659 * config/rs6000/rs6000.md: Ditto.
6660 (floatdidf2, *floatdidf2_fpr): Merge.
6661 (move_from_CR_gt_bit): Delete.
6662 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
6663 (E500_CR_IOR_COMPARE): Delete.
6664 (All patterns that require !TARGET_FPRS): Delete.
6665 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
6667 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6669 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
6671 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6673 * graphds.c (add_edge): Intitialize edge's attached data.
6674 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
6675 pointer parameter. Call pointed function on each edge during
6676 graph traversing. Skip traversing the edge when the function
6678 (graphds_dfs, graphds_scc): Ditto.
6679 (for_each_edge): New parameter. Pass the new parameter to callback
6681 * graphds.h (skip_edge_callback): New function pointer type.
6682 (graphds_dfs, graphds_scc): New function pointer parameter.
6683 (graphds_edge_callback, for_each_edge): New parameter.
6685 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6687 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
6688 out code checking if runtime alias check is possible to below ...
6689 Call the new function.
6690 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
6691 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
6693 2017-06-07 Marek Polacek <polacek@redhat.com>
6696 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
6697 TYPE_OVERFLOW_WRAPS check.
6699 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6701 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
6702 if versioning is required.
6703 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
6704 peeling with the check for versioning.
6706 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6708 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
6709 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
6710 Set true to new parameter if new ssa variable is defined.
6711 (vect_gen_vector_loop_niters): Refactor. Set range information
6712 for the new vector loop bound variable.
6713 (vect_do_peeling): Ditto.
6715 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6717 * tree-affine.c (ssa.h): Include header file.
6718 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
6719 has wrapping overflow behavior.
6721 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6723 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
6725 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6727 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
6728 (tree_to_aff_combination): ... here.
6730 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6732 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
6733 reg_pressure model function.
6734 (ivopts_global_cost_for_size): Delete.
6735 (determine_set_costs, iv_ca_recount_cost): Call new model function
6736 ivopts_estimate_reg_pressure.
6738 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6740 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
6741 expensive than udiv. Remove floating point cases from mod.
6743 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6745 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
6748 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6750 * config/aarch64/aarch64.md
6751 (copysignsf3): Fix mask generation.
6753 2017-06-07 Jakub Jelinek <jakub@redhat.com>
6755 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
6757 (class dump_manager): Add register_dumps method.
6758 * dumpfile.c: Include langhooks.h.
6759 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
6760 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
6761 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
6762 (dump_manager::dump_register): Start with 512 entries instead of 32.
6763 (dump_manager::register_dumps): New method.
6764 * toplev.c (general_init): Instead of invoking register_dumps
6765 langhook, invoke register_dumps method on the dump manager.
6766 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
6769 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
6771 * doc/md.texi: Clarify the restrictions on a define_insn condition.
6772 Say that # requires an associated define_split to exist, and that
6773 the define_split must be suitable for use after register allocation.
6775 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6777 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
6778 (compute_outgoing_frequencies): Also initialize zero counts.
6779 (find_many_sub_basic_blocks): Do not produce uninitialized profile
6780 around loops; preserve more of profile when nothing changes.
6782 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
6784 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
6786 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
6787 * config/arm/arm-cpu-cdata.h: Regenerate.
6788 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
6789 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
6790 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
6791 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
6792 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
6794 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
6796 * config/arm/t-rmprofile: Likewise.
6797 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
6799 2017-06-06 David S. Miller <davem@davemloft.net>
6802 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
6803 blockage if function uses alloca.
6805 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6807 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
6808 New "uid" fields to hold pretty-print IDs of group and ref.
6809 Memory references are now identified as <group_id>:<ref_id>
6810 instead of using [random] addresses.
6811 (dump_mem_details): Simplify, no functional change.
6812 (dump_mem_ref): Simplify and make output more concise.
6813 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
6814 (find_or_create_group): Initialize group uid.
6815 (record_ref): Initialize ref uid. Improve debug output.
6816 (prune_group_by_reuse, should_issue_prefetch_p,)
6817 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
6818 (mark_nontemporal_store, determine_loop_nest_reuse):
6819 Improve debug output.
6821 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6823 * dbgcnt.def (prefetch): New debug counter.
6824 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
6825 (schedule_prefetches): Stop issueing prefetches if debug counter
6828 2017-06-06 Tom de Vries <tom@codesourcery.com>
6830 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
6831 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
6833 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6835 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
6836 Use aarch64_reg_or_zero predicate for operand 4.
6837 (aarch64_compare_and_swap<mode> define_insn_and_split):
6838 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
6839 (aarch64_store_exclusive<mode>): Likewise for operand 2.
6841 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6843 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
6844 (arm_compute_save_core_reg_mask): This.
6845 (thumb1_compute_save_reg_mask): Rename into ...
6846 (thumb1_compute_save_core_reg_mask): This.
6847 (arm_compute_save_reg0_reg12_mask): Adapt comment.
6848 (arm_compute_frame_layout): Likewise.
6850 2017-06-06 Richard Biener <rguenther@suse.de>
6852 PR tree-optimization/80974
6853 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
6854 keep or clear leaders SSA info.
6856 2017-06-06 Tom de Vries <tom@codesourcery.com>
6858 * config/nvptx/nvptx.c (split_mode_p): New function.
6859 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
6861 2017-06-06 Tom de Vries <tom@codesourcery.com>
6863 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
6865 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6868 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
6871 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6873 * shrink-wrap.c (handle_simple_exit): Update profile.
6874 (try_shrink_wrapping): Upate profile.
6876 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6878 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
6879 (tree_guess_outgoing_edge_probabilities): New.
6880 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
6881 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
6883 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6885 * ipa-split.c (split_function): Initialize return bb profile.
6887 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6889 * profile.c (compute_branch_probabilities): Also initialize
6892 2017-06-06 Richard Biener <rguenther@suse.de>
6894 PR tree-optimization/80928
6895 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
6896 (vect_analyze_loop_operations): Properly guard analysis for
6898 (vect_transform_loop): Likewise.
6899 (vect_analyze_loop_2): Also reset SLP type on PHIs.
6900 (vect_model_induction_cost): Do not cost for pure SLP.
6901 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
6902 of induction in inner loop vectorization.
6903 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
6904 (vect_get_and_check_slp_defs): Handle vect_induction_def.
6905 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
6907 (vect_analyze_slp_cost_1): Cost induction.
6908 (vect_detect_hybrid_slp_stmts): Handle PHIs.
6909 (vect_get_slp_vect_defs): Likewise.
6910 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
6911 (vect_transform_stmt): Handle SLP reductions.
6912 * tree-vectorizer.h (vectorizable_induction): Adjust.
6914 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6916 * config/rs6000/rs6000.c (make_resolver_func): Update
6917 init_lowered_empty_function call.
6919 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
6921 * doc/invoke.texi: Document the -fprofile-abs-path option.
6922 * common.opt (fprofile-abs-path): New option.
6923 * gcov-io.h (gcov_write_filename): Declare.
6924 * gcov-io.c (gcov_write_filename): New function.
6925 * coverage.c (coverage_begin_function): Use gcov_write_filename.
6926 * profile.c (output_location): Likewise.
6928 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6930 * shring-wrap.c: Revert accidental commit.
6932 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
6934 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
6936 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6938 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
6940 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
6941 profile in callgraph edge.
6942 * profile-count.h (apply_probability): If THIS is 0, then result is 0
6943 (apply_scale): Likewise.
6944 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
6945 Also scale profile when inlining function with zero profile.
6946 (initialize_cfun): Update exit block profile even when it is zero.
6947 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
6948 when profile is read.
6950 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6952 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
6953 (CLONE_*): New constants to define the processors we can generate
6954 code for with the target_clone attribute.
6955 (rs6000_clone_map): New array to identify which clone processors
6956 the current program is running on.
6957 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
6958 target_clone attribute.
6959 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
6960 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
6961 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
6962 (cpu_expand_builtin): Add support for target_clone attribute.
6963 (rs6000_valid_attribute_p): Allow "default" attribute.
6964 (get_decl_name): New debug function to simplify printing the
6965 current function name in debugging statements.
6966 (rs6000_clone_priority): New functions to support the target_clone
6967 attribute, and be able to generate code to switch between ISA 2.05
6968 through ISA 3.0 (power6 through power9).
6969 (rs6000_compare_version_priority): Likewise.
6970 (rs6000_get_function_versions_dispatcher): Likewise.
6971 (make_resolver_func): Likewise.
6972 (add_condition_to_bb): Likewise.
6973 (dispatch_function_versions): Likewise.
6974 (rs6000_generate_version_dispatcher_body): Likewise.
6975 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
6976 (fusion_gpr_load_p): Fix a spacing issue.
6977 * doc/extend.texi (Common Function Attributes): Document that the
6978 PowerPC supports the target_clone attribute.
6980 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
6982 * config/arm/arm.h: explain F symbol found in description of ARM
6983 register allocation in its legend.
6985 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6987 * config/mips/frame-header-opt.c: Include profile-count.h.
6988 * config/riscv/riscv.c: Include profile-count.h
6990 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6992 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
6994 (sm_set_flag_if_changed): Add bbs field.
6995 (execute_sm_if_changed_flag_set): Pass BBS.
6996 (execute_sm): Update.
6998 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7000 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7003 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7005 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
7006 (peephole2): New peephole2 to emit the above.
7007 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
7009 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7011 * config/aarch64/aarch64.c (define_peephole2 above
7012 *sub_<shift>_<mode>): New peephole.
7014 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7016 * config/i386/i386.c (make_resolver_func): Update.
7017 * Makefile.in: Add profile-count.h and profile-count.o
7018 * auto-profile.c (afdo_indirect_call): Update to new API.
7019 (afdo_set_bb_count): Update.
7020 (afdo_propagate_edge): Update.
7021 (afdo_propagate_circuit): Update.
7022 (afdo_calculate_branch_prob): Update.
7023 (afdo_annotate_cfg): Update.
7024 * basic-block.h: Include profile-count.h
7025 (struct edge_def): Turn count to profile_count.
7026 (struct basic_block_def): Likewie.
7027 (REG_BR_PROB_BASE): Move to profile-count.h
7028 (RDIV): Move to profile-count.h
7029 * bb-reorder.c (max_entry_count): Turn to profile_count.
7030 (find_traces): Update.
7031 (rotate_loop):Update.
7032 (connect_traces):Update.
7033 (sanitize_hot_paths):Update.
7034 * bt-load.c (migrate_btr_defs): Update.
7035 * cfg.c (RDIV): Remove.
7036 (init_flow): Use alloc_block.
7037 (alloc_block): Uninitialize count.
7038 (unchecked_make_edge): Uninitialize count.
7039 (check_bb_profile): Update.
7040 (dump_edge_info): Update.
7041 (dump_bb_info): Update.
7042 (update_bb_profile_for_threading): Update.
7043 (scale_bbs_frequencies_int): Update.
7044 (scale_bbs_frequencies_gcov_type): Update.
7045 (scale_bbs_frequencies_profile_count): New.
7046 * cfg.h (update_bb_profile_for_threading): Update.
7047 (scale_bbs_frequencies_profile_count): Declare.
7048 * cfgbuild.c (compute_outgoing_frequencies): Update.
7049 (find_many_sub_basic_blocks): Update.
7050 * cfgcleanup.c (try_forward_edges): Update.
7051 (try_crossjump_to_edge): Update.
7052 * cfgexpand.c (expand_gimple_tailcall): Update.
7053 (construct_exit_block): Update.
7054 * cfghooks.c (verify_flow_info): Update.
7055 (dump_bb_for_graph): Update.
7056 (split_edge): Update.
7057 (make_forwarder_block): Update.
7058 (duplicate_block): Update.
7059 (account_profile_record): Update.
7060 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
7061 (get_estimated_loop_iterations): Update.
7062 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
7063 (single_likely_exit): Update.
7064 * cfgloopmanip.c (scale_loop_profile): Update.
7066 (set_zero_probability): Update.
7067 (lv_adjust_loop_entry_edge): Update.
7068 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
7069 (purge_dead_edges): Update.
7070 (rtl_account_profile_record): Update.
7071 * cgraph.c (cgraph_node::create): Uninitialize count.
7072 (symbol_table::create_edge): Uninitialize count.
7073 (cgraph_update_edges_for_call_stmt_node): Update.
7074 (cgraph_edge::dump_edge_flags): Update.
7075 (cgraph_node::dump): Update.
7076 (cgraph_edge::maybe_hot_p): Update.
7077 * cgraph.h: Include profile-count.h
7078 (create_clone), create_edge, create_indirect_edge): Update.
7079 (cgraph_node): Turn count to profile_count.
7080 (cgraph_edge0: Likewise.
7081 (make_speculative, clone): Update.
7082 (create_edge): Update.
7083 (init_lowered_empty_function): Update.
7084 * cgraphclones.c (cgraph_edge::clone): Update.
7085 (duplicate_thunk_for_node): Update.
7086 (cgraph_node::create_clone): Update.
7087 * cgraphunit.c (cgraph_node::analyze): Update.
7088 (cgraph_node::expand_thunk): Update.
7089 * final.c (dump_basic_block_info): Update.
7090 * gimple-streamer-in.c (input_bb): Update.
7091 * gimple-streamer-out.c (output_bb): Update.
7092 * graphite.c (print_global_statistics): Update.
7093 (print_graphite_scop_statistics): Update.
7094 * hsa-brig.c: Include basic-block.h.
7095 * hsa-dump.c: Include basic-block.h.
7096 * hsa-gen.c (T sum_slice): Update.
7097 (convert_switch_statements):Update.
7098 * hsa-regalloc.c: Include basic-block.h.
7099 * ipa-chkp.c (chkp_produce_thunks): Update.
7100 * ipa-cp.c (struct caller_statistics): Update.
7101 (init_caller_stats): Update.
7102 (gather_caller_stats): Update.
7103 (ipcp_cloning_candidate_p): Update.
7104 (good_cloning_opportunity_p): Update.
7105 (get_info_about_necessary_edges): Update.
7106 (dump_profile_updates): Update.
7107 (update_profiling_info): Update.
7108 (update_specialized_profile): Update.
7109 (perhaps_add_new_callers): Update.
7110 (decide_about_value): Update.
7111 (ipa_cp_c_finalize): Update.
7112 * ipa-devirt.c (struct odr_type_warn_count): Update.
7113 (struct decl_warn_count): Update.
7114 (struct final_warning_record): Update.
7115 (possible_polymorphic_call_targets): Update.
7116 (ipa_devirt): Update.
7117 * ipa-fnsummary.c (redirect_to_unreachable): Update.
7118 * ipa-icf.c (sem_function::merge): Update.
7119 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
7120 * ipa-inline.c (compute_uninlined_call_time): Update.
7121 (compute_inlined_call_time): Update.
7122 (want_inline_small_function_p): Update.
7123 (want_inline_self_recursive_call_p): Update.
7124 (edge_badness): Update.
7125 (lookup_recursive_calls): Update.
7126 (recursive_inlining): Update.
7127 (inline_small_functions): Update.
7128 (dump_overall_stats): Update.
7129 (dump_inline_stats): Update.
7130 * ipa-profile.c (ipa_profile_generate_summary): Update.
7131 (ipa_propagate_frequency): Update.
7132 (ipa_profile): Update.
7133 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7134 * ipa-utils.c (ipa_merge_profiles): Update.
7135 * loop-doloop.c (doloop_modify): Update.
7136 * loop-unroll.c (report_unroll): Update.
7137 (unroll_loop_runtime_iterations): Update.
7138 * lto-cgraph.c (lto_output_edge): Update.
7139 (lto_output_node): Update.
7140 (input_node): Update.
7141 (input_edge): Update.
7142 (merge_profile_summaries): Update.
7143 * lto-streamer-in.c (input_cfg): Update.
7144 * lto-streamer-out.c (output_cfg): Update.
7145 * mcf.c (create_fixup_graph): Update.
7146 (adjust_cfg_counts): Update.
7147 (sum_edge_counts): Update.
7148 * modulo-sched.c (sms_schedule): Update.
7149 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
7150 * predict.c (maybe_hot_count_p): Update.
7151 (probably_never_executed): Update.
7152 (dump_prediction): Update.
7153 (combine_predictions_for_bb): Update.
7154 (propagate_freq): Update.
7155 (handle_missing_profiles): Update.
7156 (counts_to_freqs): Update.
7157 (rebuild_frequencies): Update.
7158 (force_edge_cold): Update.
7159 * predict.h: Include profile-count.h
7160 (maybe_hot_count_p, counts_to_freqs): UPdate.
7161 * print-rtl-function.c: Do not include cfg.h
7162 * print-rtl.c: Include basic-block.h
7163 * profile-count.c: New file.
7164 * profile-count.h: New file.
7165 * profile.c (is_edge_inconsistent): Update.
7166 (correct_negative_edge_counts): Update.
7167 (is_inconsistent): Update.
7168 (set_bb_counts): Update.
7169 (read_profile_edge_counts): Update.
7170 (compute_frequency_overlap): Update.
7171 (compute_branch_probabilities): Update; Initialize and deinitialize
7173 (branch_prob): Update.
7174 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
7175 (edge_gcov_count): New.
7176 (bb_gcov_count): New.
7177 * shrink-wrap.c (try_shrink_wrapping): Update.
7178 * tracer.c (better_p): Update.
7179 * trans-mem.c (expand_transaction): Update.
7180 (ipa_tm_insert_irr_call): Update.
7181 (ipa_tm_insert_gettmclone_call): Update.
7182 * tree-call-cdce.c: Update.
7183 * tree-cfg.c (gimple_duplicate_sese_region): Update.
7184 (gimple_duplicate_sese_tail): Update.
7185 (gimple_account_profile_record): Update.
7186 (execute_fixup_cfg): Update.
7187 * tree-inline.c (copy_bb): Update.
7188 (copy_edges_for_bb): Update.
7189 (initialize_cfun): Update.
7190 (freqs_to_counts): Update.
7191 (copy_cfg_body): Update.
7192 (expand_call_inline): Update.
7193 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7194 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
7195 (try_unroll_loop_completely): Update.
7196 (try_peel_loop): Update.
7197 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7198 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
7199 * tree-ssa-loop-split.c (connect_loops): Update.
7200 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
7201 * tree-ssa-reassoc.c (branch_fixup): Update.
7202 * tree-ssa-tail-merge.c (replace_block_by): Update.
7203 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
7204 (compute_path_counts): Update.
7205 (update_profile): Update.
7206 (recompute_probabilities): Update.
7207 (update_joiner_offpath_counts): Update.
7208 (estimated_freqs_path): Update.
7209 (freqs_to_counts_path): Update.
7210 (clear_counts_path): Update.
7211 (ssa_fix_duplicate_block_edges): Update.
7212 (duplicate_thread_path): Update.
7213 * tree-switch-conversion.c (case_bit_test_cmp): Update.
7214 (struct switch_conv_info): Update.
7215 * tree-tailcall.c (decrease_profile): Update.
7216 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7217 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7218 * value-prof.c (check_counter): Update.
7219 (gimple_divmod_fixed_value): Update.
7220 (gimple_mod_pow2): Update.
7221 (gimple_mod_subtract): Update.
7222 (gimple_ic_transform): Update.
7223 (gimple_stringop_fixed_value): Update.
7224 * value-prof.h (gimple_ic): Update.
7226 2017-06-02 Carl Love <cel@us.ibm.com>
7228 * config/rs6000/rs6000-c: Add support for built-in functions
7229 vector double vec_doublee (vector signed int);
7230 vector double vec_doublee (vector unsigned int);
7231 vector double vec_doublee (vector float);
7232 vector double vec_doubleh (vector signed int);
7233 vector double vec_doubleh (vector unsigned int);
7234 vector double vec_doubleh (vector float);
7235 vector double vec_doublel (vector signed int);
7236 vector double vec_doublel (vector unsigned int);
7237 vector double vec_doublel (vector float);
7238 vector double vec_doubleo (vector signed int);
7239 vector double vec_doubleo (vector unsigned int);
7240 vector double vec_doubleo (vector float);.
7241 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
7242 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
7244 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
7245 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
7246 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
7248 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
7249 vec_doublel, vec_doubleh.
7250 * doc/extend.texi: Update the built-in documentation file for the
7251 new built-in functions.
7253 2017-06-02 David Malcolm <dmalcolm@redhat.com>
7256 * ipa-inline-analysis.c (free_growth_caches): Set
7257 edge_removal_hook_holder to NULL after removing it.
7259 2017-06-02 Sudakshina Das <sudi.das@arm.com>
7261 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
7262 comparision with zero.
7264 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
7265 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7266 for early expansion of vec_min and vec_max builtins.
7267 (builtin_function_type): Add min/max unsigned variants to those
7268 identified as having unsigned arguments.
7270 2017-06-02 Olivier Hainque <hainque@adacore.com>
7272 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
7274 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7276 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
7277 Use VALL_F16 iterator rather than VALL.
7279 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7281 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
7282 Emit CBNZ inside loop when doing a strong exchange and comparing
7283 against zero. Generate the CC flags after the loop.
7285 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
7287 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
7288 (dl_section_ref): New.
7289 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
7290 On AIX, append an expression to subtract the size of the
7291 section length to dl_section_ref.
7293 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
7295 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7296 for early expansion of vector absolute builtins.
7298 2017-06-02 Richard Biener <rguenther@suse.de>
7300 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
7301 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
7303 2017-06-02 Richard Biener <rguenther@suse.de>
7305 PR tree-optimization/80948
7306 * tree-tailcall.c (find_tail_calls): Track stmts to move in
7309 2017-06-02 Richard Biener <rguenther@suse.de>
7311 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
7313 * tree-vect-stmts.c (process_use): Do not mark backedge defs
7314 for inductions as relevant.
7316 2017-06-02 Richard Biener <rguenther@suse.de>
7318 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
7319 (vectorizable_induction): ... this. Remove dead code.
7321 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
7323 * builtins. (expand_builtin_alloca): Remove second parameter and
7324 infer its value from the first parameter instead.
7325 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
7327 2017-06-02 Jakub Jelinek <jakub@redhat.com>
7329 PR rtl-optimization/80903
7330 * loop-doloop.c (add_test): Unshare sequence.
7332 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
7334 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
7336 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
7338 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
7340 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
7341 xlogue_layout::get_instance, logue_layout::xlogue_layout,
7342 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
7343 (xlogue_layout::get_stub_rtx): Make static.
7344 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
7345 (xlogue_layout::compute_stub_managed_regs): Rename to...
7346 (xlogue_layout::count_stub_managed_regs): ...this.
7347 (xlogue_layout::is_stub_managed_reg): New function.
7348 (xlogue_layout::m_stub_names): Rename to...
7349 (xlogue_layout::s_stub_names): ...this, make static.
7350 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
7351 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
7352 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
7353 xlogue_layout::s_stub_names): Instantiate statics.
7354 (stub_managed_regs): Remove.
7355 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
7356 (disable_call_ms2sysv_xlogues): Rename to...
7357 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
7358 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
7360 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
7361 change after reload_completed.
7362 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
7364 (ix86_expand_prologue): Likewise.
7365 (ix86_expand_epilogue): Likewise.
7366 (ix86_expand_split_stack_prologue): Likewise.
7367 (ix86_compute_frame_layout): Remove frame parameter ...
7368 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
7369 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
7371 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
7372 (ix86_frame): Move from here ...
7373 * config/i386/i386.h (ix86_frame): ... to here.
7374 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
7375 complete ix86_frame data structure instead. Remove some_ld_name.
7377 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
7379 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
7380 symbols that hold a DECL_VALUE_EXPR.
7382 2017-06-01 Martin Jambor <mjambor@suse.cz>
7384 PR tree-optimization/80898
7385 * tree-sra.c (process_subtree_disqualification): Removed.
7386 (disqualify_candidate): Do not acll
7387 process_subtree_disqualification.
7388 (subtree_mark_written_and_enqueue): New function.
7389 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
7390 RHS has been disqualified and re-queue LHS if necessary. Apart
7391 from that, ignore disqualified RHS.
7393 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7395 * config/s390/s390.c (s390_emit_epilogue): Disable early return
7396 address fetch for z10 or later.
7398 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7400 * config/arc/arc.md (tst_movb): Add guard when splitting.
7402 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7404 * config/arc/arc.c (arc_can_eliminate): Test against
7405 arc_frame_pointer_needed.
7407 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7409 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
7410 to prevent store reordering.
7411 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
7412 (type): Add block type.
7413 (stack_tie): Define special instruction to be used in
7416 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7418 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
7419 constraint. It is not valid for the pattern.
7420 (noncommutative_binary_comparison): Likewise.
7422 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7424 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
7427 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7429 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
7430 be used by the reg-alloc.
7432 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7434 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
7435 reg-alloc when having mul64 or mul32x16 instructions.
7436 (mulsidi3): Likewise.
7437 (umulsidi3): Likewise.
7438 (mulsi32x16): New pattern.
7439 (mulsi64): Likewise.
7440 (mulsidi64): Likewise.
7441 (umulsidi64): Likewise.
7442 (MUL32x16_REG): Define.
7443 (mul64_600): Use MUL32x16_REG.
7444 (mac64_600): Likewise.
7445 (umul64_600): Likewise.
7446 (umac64_600): Likewise.
7448 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7450 * config/arc/arc.md (mulsi3_700): Make it commutative.
7452 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
7454 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
7456 (*sign_extendsidi2_insn): Likewise for movstosw.
7458 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
7460 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
7461 the type of the input discriminant value. Convert the
7462 discriminant value of signedness vary.
7464 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
7466 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
7467 Add to -Wall section.
7469 2017-06-01 Richard Biener <rguenther@suse.de>
7472 * fold-const.c (fold_plusminus_mult_expr): If the factored
7473 factor may be zero use a wrapping type for the inner operation.
7474 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
7475 and handle moved defs.
7476 (process_assignment): Properly guard the unary op case. Return a
7477 tri-state indicating that moving the stmt before the call may allow
7478 to continue. Pass through to_move.
7479 (find_tail_calls): Handle moving unrelated defs before
7482 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
7485 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
7486 splitter result in the canonical way.
7488 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
7490 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
7491 also for 32bit target. Update insn attributes.
7492 (zero-extendsidi2 splitter): Allow all registers for operand 1.
7494 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
7496 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
7497 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
7498 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
7499 (_mm_maskz_min_ss): New intrinsics.
7501 2017-05-31 Martin Liska <mliska@suse.cz>
7503 * tree-vect-loop.c (vect_create_epilog_for_reduction):
7504 Change comment style to one we normally use.
7505 (vectorizable_reduction): Likewise.
7506 (vectorizable_induction): Likewise.
7507 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7508 (vectorizable_call): Likewise.
7509 (vectorizable_simd_clone_call): Likewise.
7510 (vectorizable_conversion): Likewise.
7511 (vectorizable_assignment): Likewise.
7512 (vectorizable_shift): Likewise.
7513 (vectorizable_operation): Likewise.
7514 (vectorizable_store): Likewise.
7515 (vectorizable_load): Likewise.
7516 * tree-vectorizer.h: Likewise.
7518 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
7520 * passes.c (emergency_dump_function): New.
7521 * tree-pass.h (emergency_dump_function): Declare.
7522 * plugin.c (plugins_internal_error_function): Remove.
7523 * plugin.h (plugins_internal_error_function): Remove declaration.
7524 * toplev.c (internal_error_function): New static function. Use it...
7525 (general_init): ...here.
7527 2017-05-31 Graham Markall <graham.markall@embecosm.com>
7529 * config/arc/arc.c (arc_print_operand): Handle constant operands.
7530 (arc_rtx_costs): Add costs for new patterns.
7531 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
7532 * config/arc/predicates.md: Add _1_2_3_operand predicate.
7534 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
7536 * tree-ssa-strlen.c (get_next_strinfo): New function.
7537 (get_stridx_plus_constant): Use it.
7538 (zero_length_string): Likewise.
7539 (adjust_related_strinfos): Likewise.
7540 (adjust_last_stmt): Likewise.
7542 2017-05-31 Richard Biener <rguenther@suse.de>
7545 * config/i386/i386.c (ix86_expand_builtin): Remove assert
7546 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
7548 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
7550 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
7551 loop_vinfo argument and use of dependence distance vectors.
7552 Check instead whether the two references differ only in their
7553 initial value and assume that they have the same alignment if the
7554 difference is a multiple of the vector alignment.
7555 (vect_analyze_data_refs_alignment): Update call accordingly.
7557 2017-05-31 Martin Liska <mliska@suse.cz>
7560 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
7562 2017-05-31 Bin Cheng <bin.cheng@arm.com>
7564 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
7565 (create_intersect_range_checks): Move from ...
7566 * tree-data-ref.c (create_intersect_range_checks_index)
7567 (create_intersect_range_checks): ... to here.
7568 (create_runtime_alias_checks): New function factored from ...
7569 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
7570 here. Call above function.
7571 * tree-data-ref.h (create_runtime_alias_checks): New function.
7573 2017-05-31 Bin Cheng <bin.cheng@arm.com>
7575 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
7576 segment length for dr_b and compute it in wide_int.
7578 2017-05-31 Richard Biener <rguenther@suse.de>
7580 PR tree-optimization/80906
7581 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
7582 and pass through iv_map.
7583 (copy_bb_and_scalar_dependences): Adjust.
7584 (translate_pending_phi_nodes): Likewise.
7585 (copy_loop_close_phi_args): Handle code-generating IVs instead
7588 2017-05-30 David Malcolm <dmalcolm@redhat.com>
7590 * diagnostic-color.c (color_dict): Add "type-diff".
7591 (parse_gcc_colors): Update comment.
7592 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
7593 -fdiagnostics-show-template-tree and -fno-elide-type.
7594 (GCC_COLORS): Add type-diff to example.
7596 (-fdiagnostics-show-template-tree): New.
7597 (-fno-elide-type): New.
7598 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
7599 the pp_format_decoder callback. Call any m_format_postprocessor's
7601 (pretty_printer::pretty_printer): Initialize
7602 m_format_postprocessor.
7603 (pretty_printer::~pretty_printer): Delete any
7604 m_format_postprocessor.
7605 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
7606 (class format_postprocessor): New class.
7607 (struct pretty_printer::format_decoder): Document the new parameters.
7608 (struct pretty_printer::m_format_postprocessor): New field.
7609 * tree-diagnostic.c (default_tree_printer): Update for new
7610 bool and const char ** params.
7611 * tree-diagnostic.h (default_tree_printer): Likewise.
7613 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
7615 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
7616 (lwa_operand): Delete rs6000_gen_cell_microcode test.
7617 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
7618 rs6000_gen_cell_microcode code.
7619 (rs6000_final_prescan_insn): Delete.
7620 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
7621 "warn-cell-microcode" entries.
7622 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
7623 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
7624 throughout. Change cc_reg_not_micro_cr0_operand to
7625 cc_reg_not_cr0_operand throughout.
7626 (*extendhi<mode>2_noload): Delete.
7627 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
7628 (mwarn-cell-microcode): Delete.
7629 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
7630 -mgen-cell-microcode and -mwarn-cell-microcode.
7632 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
7635 * config/i386/constraints.md (Yd): New constraint.
7637 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
7638 and (?Yd, r) alternatives. Update insn attributes.
7639 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
7640 and (?*Yd, r) alternatives. Update insn attributes.
7641 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
7643 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
7645 * gimplify.c (gimplify_modify_expr): Don't create a
7646 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
7649 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
7651 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
7653 2017-05-30 Richard Biener <rguenther@suse.de>
7655 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
7656 and reduc_def fields.
7657 (STMT_VINFO_REDUC_TYPE): New define.
7658 (STMT_VINFO_REDUC_DEF): Likewise.
7659 (vect_force_simple_reduction): Adjust prototype.
7660 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
7661 (vect_is_simple_reduction): Remove check_reduction argument.
7662 (vect_force_simple_reduction): Adjust and set
7663 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
7664 (vectorizable_reduction): Do not re-do reduction analysis
7665 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
7666 * tree-parloops.c (gather_scalar_reductions): Adjust.
7668 2017-05-30 Richard Biener <rguenther@suse.de>
7671 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
7674 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
7676 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
7677 Introduce unknown_misalignment parameter and remove vf.
7678 (vect_peeling_hash_get_lowest_cost):
7679 Pass unknown_misalignment parameter.
7680 (vect_enhance_data_refs_alignment):
7681 Fix unsupportable data ref treatment.
7683 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7685 * tree-vect-data-refs.c (vect_get_data_access_cost):
7686 Workaround for SLP handling.
7687 (vect_enhance_data_refs_alignment):
7688 Compute costs for doing no peeling at all, compare to the best
7689 peeling costs so far and avoid peeling if cheaper.
7691 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7693 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
7694 Return peeling info and set costs to zero for unlimited cost
7696 (vect_enhance_data_refs_alignment): Also inspect all datarefs
7697 with unknown misalignment. Compute and costs for unknown
7698 misalignment, compare them to the costs for known misalignment
7699 and choose the cheapest for peeling.
7701 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7703 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
7704 (vect_get_peeling_costs_all_drs): Create function.
7705 (vect_peeling_hash_get_lowest_cost):
7706 Use vect_get_peeling_costs_all_drs.
7707 (vect_peeling_supportable): Create function.
7708 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
7710 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7712 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
7713 DR_HAS_NEGATIVE_STEP.
7714 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
7715 (vect_enhance_data_refs_alignment): Use.
7716 (vect_duplicate_ssa_name_ptr_info): Use.
7717 * tree-vectorizer.h (dr_misalignment): Use.
7718 (known_alignment_for_access_p): Use.
7720 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7723 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
7725 (has_section_name): New function.
7727 2017-05-30 Martin Liska <mliska@suse.cz>
7730 * auto-profile.c (get_function_decl_from_block): Fix
7733 2017-05-30 Richard Biener <rguenther@suse.de>
7736 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
7738 2017-05-30 Martin Liska <mliska@suse.cz>
7740 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
7741 * dumpfile.h (struct dump_file_info): Remove ctors.
7743 2017-05-30 Martin Liska <mliska@suse.cz>
7745 * predict.def: Fix GNU coding style.
7747 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7749 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
7750 Mark 'to' argument with ATTRIBUTE_UNUSED.
7752 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7754 * config/xtensa/xtensa.c (xtensa_emit_call): Use
7755 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
7756 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
7759 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
7761 * doc/install.texi (Options specification): Restore entry of
7762 --enable-sjlj-exceptions.
7764 2017-05-27 Michael Eager <eager@eagercon.com>
7767 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
7769 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
7771 * config/microblaze/microblaze.h
7772 (FIXED_REGISTERS): Update in macro.
7773 (CALL_USED_REGISTERS): Update in macro.
7775 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
7777 * doc/install.texi: Add links to macOS binary distributions.
7779 2017-05-27 Jakub Jelinek <jakub@redhat.com>
7783 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7785 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7787 2017-05-26 Martin Liska <mliska@suse.cz>
7789 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
7791 2017-05-26 Martin Liska <mliska@suse.cz>
7793 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
7794 always leading ';; '.
7795 (dump_bb_info): Likewise.
7796 (brief_dump_cfg): Likewise.
7797 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
7798 * dumpfile.c: Remove usage of TDF_VERBOSE.
7799 * dumpfile.h (enum dump_kind): Likewise.
7800 (dump_gimple_bb_header): Do not use TDF_COMMENT.
7801 * print-tree.c (debug_verbose): Remove.
7802 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
7803 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
7804 * tree-diagnostic.c (default_tree_printer): Replace
7805 TDF_DIAGNOSTIC with TDF_SLIM.
7807 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7809 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
7810 in parameter loop, rather than loop_vinfo.
7811 (create_intersect_range_checks): Ditto.
7812 (vect_create_cond_for_alias_checks): Update call to above functions.
7814 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7816 PR tree-optimization/80815
7817 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
7818 for merging runtime alias checks. Handle negative DR_STEPs.
7820 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7822 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
7824 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
7825 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
7826 out code pruning runtime alias checks.
7827 * tree-data-ref.c (prune_runtime_alias_test_list): New function
7828 factored out from above.
7829 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7831 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7833 (prune_runtime_alias_test_list): New decalaration.
7835 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7837 * tree-vect-data-refs.c (compare_tree): Rename and move ...
7838 * tree-data-ref.c (data_ref_compare_tree): ... to here.
7839 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
7840 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
7841 (operator==, comp_dr_with_seg_len_pair): Ditto.
7842 (vect_prune_runtime_alias_test_list): Ditto.
7844 2017-05-26 Martin Liska <mliska@suse.cz>
7847 * params.def: Bound partial-inlining-entry-probability param.
7849 2017-05-26 Marek Polacek <polacek@redhat.com>
7852 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
7855 2017-05-26 Richard Biener <rguenther@suse.de>
7857 PR tree-optimization/80842
7858 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
7861 2017-05-26 Richard Biener <rguenther@suse.de>
7863 PR tree-optimization/80844
7864 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
7866 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
7868 * doc/md.texi (Machine Constraints): Update x86 family
7869 machine constraints section to match 'config/i386/constraints.md'.
7871 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
7873 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
7875 2017-05-25 Nathan Sidwell <nathan@acm.org>
7877 * doc/invoke.texi (--enable-languages): Update documentation.
7879 2017-05-25 Martin Liska <mliska@suse.cz>
7881 * dumpfile.c: Add TDF_FOLDING.
7882 * dumpfile.h (enum dump_kind): Likewise.
7883 * genmatch.c (dt_simplify::gen_1): Use it.
7885 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7887 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
7889 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7891 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7892 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
7894 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7896 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
7897 * match.pd (X == C): Rewrite it here.
7898 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
7899 with_certain_nonzero_bits2): New predicates.
7900 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
7902 2017-05-24 Nathan Sidwell <nathan@acm.org>
7904 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
7907 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
7910 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7912 * config/powerpcspe: New port. Files are copied from the rs6000
7913 port, with "rs6000" in filenames replaced by "powerpcspe".
7915 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
7917 PR rtl-optimization/80754
7918 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
7920 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
7922 * config/sparc/sparc.md (length): Return the correct value for -mflat
7923 sibcalls to match output_sibcall.
7925 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7929 * config/rs6000/rs6000.c (struct machine_function): Add new field
7931 (rs6000_get_separate_components): Init that field, use it.
7932 (rs6000_components_for_bb): Use the field.
7934 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
7936 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
7938 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
7941 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
7943 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7946 * config/s390/s390.c (s390_check_qrst_address): Check incoming
7947 address against address_operand predicate.
7948 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
7950 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
7952 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
7953 parameters passed indirectly.
7955 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
7957 * config/i386/i386.md (*movdi_internal): Remove SSE4
7958 alternative 18 (?r, *v). Update insn attributes.
7959 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
7960 Update insn attributes.
7961 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
7962 Update insn attributes.
7963 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
7964 alternative 1 (r, v). Remove isa attribute.
7965 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
7966 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
7967 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
7969 2017-05-23 Tom de Vries <tom@codesourcery.com>
7971 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
7974 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7976 * cgraphunit.c (symbol_table::process_new_functions): Update.
7977 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
7978 (inline_generate_summary): Rename to ...
7979 (ipa_fn_summary_generate): ... this one.
7980 (inline_read_summary): Rename to ...
7981 (ipa_fn_summary_read): ... this one.
7982 (inline_write_summary): Rename to ...
7983 (ipa_fn_summary_write): ... this one.
7984 (inline_free_summary): Rename to ...
7985 (ipa_free_fn_summary): ... this one.
7986 (pass_data_local_fn_summary, pass_local_fn_summary,
7987 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
7988 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
7989 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
7990 make_pass_ipa_fn_summary): New.
7991 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
7992 inline_write_summary, inline_free_summary): Remove.
7993 (ipa_free_fn_summary) : New.
7994 * ipa-inline.c (ipa_inline): Update.
7995 (pass_ipa_inline): Do not generate summaries.
7996 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
7998 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
7999 and add pass_ipa_fn_summary.
8000 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
8002 (make_pass_inline_parameters): Remove.
8004 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
8006 * omp-low.c (struct omp_context): Remove "default_kind" member.
8009 * omp-offload.c (execute_oacc_device_lower): Remove the
8010 parallelism dimensions function attributes for unparallelized
8011 OpenACC kernels constructs.
8013 2017-05-23 Martin Liska <mliska@suse.cz>
8015 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
8017 (cgraph_edge::make_speculative): Likewise.
8018 (cgraph_edge::resolve_speculation): Likewise.
8019 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
8020 (cgraph_node::dump): Likewise.
8021 * cgraph.h: Likewise.
8022 * cgraphunit.c (analyze_functions): Likewise.
8023 (symbol_table::compile): Likewise.
8024 * ipa-cp.c (print_all_lattices): Likewise.
8025 (determine_versionability): Likewise.
8026 (initialize_node_lattices): Likewise.
8027 (ipcp_verify_propagated_values): Likewise.
8028 (estimate_local_effects): Likewise.
8029 (update_profiling_info): Likewise.
8030 (create_specialized_node): Likewise.
8031 (perhaps_add_new_callers): Likewise.
8032 (decide_about_value): Likewise.
8033 (decide_whether_version_node): Likewise.
8034 (identify_dead_nodes): Likewise.
8035 (ipcp_store_bits_results): Likewise.
8036 * ipa-devirt.c (dump_targets): Likewise.
8037 (ipa_devirt): Likewise.
8038 * ipa-icf.c (sem_item::dump): Likewise.
8039 (sem_function::equals): Likewise.
8040 (sem_variable::equals): Likewise.
8041 (sem_item_optimizer::read_section): Likewise.
8042 (sem_item_optimizer::execute): Likewise.
8043 (congruence_class::dump): Likewise.
8044 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
8045 (dump_inline_summary): Likewise.
8046 (estimate_node_size_and_time): Likewise.
8047 (inline_analyze_function): Likewise.
8048 * ipa-inline-transform.c (inline_call): Likewise.
8049 * ipa-inline.c (report_inline_failed_reason): Likewise.
8050 (want_early_inline_function_p): Likewise.
8051 (edge_badness): Likewise.
8052 (update_edge_key): Likewise.
8053 (inline_small_functions): Likewise.
8054 * ipa-profile.c (ipa_profile): Likewise.
8055 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
8056 (ipa_make_edge_direct_to_target): Likewise.
8057 (remove_described_reference): Likewise.
8058 (ipa_impossible_devirt_target): Likewise.
8059 (propagate_controlled_uses): Likewise.
8060 (ipa_print_node_params): Likewise.
8061 (ipcp_transform_function): Likewise.
8062 * ipa-pure-const.c (pure_const_read_summary): Likewise.
8063 (propagate_pure_const): Likewise.
8064 * ipa-reference.c (generate_summary): Likewise.
8065 (read_write_all_from_decl): Likewise.
8066 (propagate): Likewise.
8067 (ipa_reference_read_optimization_summary): Likewise.
8068 * ipa-utils.c (ipa_merge_profiles): Likewise.
8069 * ipa.c (walk_polymorphic_call_targets): Likewise.
8070 (symbol_table::remove_unreachable_nodes): Likewise.
8071 (ipa_single_use): Likewise.
8072 * passes.c (execute_todo): Likewise.
8073 * predict.c (drop_profile): Likewise.
8074 * symtab.c (symtab_node::get_dump_name): New function.
8075 (symtab_node::dump_name): Likewise.
8076 (symtab_node::dump_asm_name): Likewise.
8077 (symtab_node::dump_references): Likewise.
8078 (symtab_node::dump_referring): Likewise.
8079 (symtab_node::dump_base): Likewise.
8080 (symtab_node::debug_symtab): Likewise.
8081 * tree-sra.c (convert_callers_for_node): Likewise.
8082 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
8083 * value-prof.c (init_node_map): Likewise.
8085 2017-05-23 Martin Liska <mliska@suse.cz>
8087 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
8088 and symtab_node::debug_symtab to symbol_table::debug.
8089 * cgraphunit.c (analyze_functions): Use the renamed function.
8090 (symbol_table::compile): Likewise.
8091 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
8092 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
8093 * passes.c (execute_todo): Likewise.
8094 * symtab.c (symbol_table::dump): New function.
8095 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
8097 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
8099 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
8100 that nonconst implies exec.
8102 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
8104 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
8105 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
8106 (inline_edge_summary_vec): Turn into ...
8107 (ipa_call_summaries): ... this one.
8108 (redirect_to_unreachable, edge_set_predicate,
8109 evaluate_properties_for_edge, inline_summary_alloc,
8110 reset_ipa_call_summary, reset_inline_summary,
8111 inline_summary_t::duplicate): Update.
8112 (inline_edge_duplication_hook): Turn to ...
8113 (ipa_call_summary_t::duplicate): ... this one.
8114 (inline_edge_removal_hook): Turn to ...
8115 (ipa_call_summary_t::remove): ... this one.
8116 (dump_inline_edge_summary): Turn to ...
8117 (dump_ipa_call_summary): ... this one.
8118 (estimate_function_body_sizes): Update.
8119 (inline_update_callee_summaries): Update.
8120 (remap_edge_change_prob): Update.
8121 (remap_edge_summaries): Update.
8122 (inline_merge_summary): Update.
8123 (do_estimate_edge_time): Update.
8124 (inline_generate_summary): Update.
8125 (inline_read_section): Update.
8126 (inline_read_summary): Update.
8127 (inline_free_summary): Update.
8128 * ipa-inline.c (can_inline_edge_p): Update.
8129 (compute_inlined_call_time): Update.
8130 (want_inline_small_function_p): Update.
8131 (edge_badness): Update.
8132 (early_inliner): Update.
8133 * ipa-inline.h (inline_edge_summary): Turn to ...
8134 (ipa_call_summary): ... this one.
8135 (ipa_call_summary_t): New class.
8136 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
8137 (ipa_call_summaries): New.
8138 (inline_edge_summary): Remove.
8139 (estimate_edge_growth): Update.
8140 * ipa-profile.c (ipa_propagate_frequency_1): Update.
8141 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
8142 * ipa-split.c (execute_split_functions): Update.
8143 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
8145 2017-05-23 Tom de Vries <tom@codesourcery.com>
8147 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
8148 attributes): Document rdrand effective target.
8150 2017-05-23 Tom de Vries <tom@codesourcery.com>
8152 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
8153 attributes): Sort alphabetically.
8155 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
8157 * config/avr/genmultilib.awk: Use gsub instead of gensub.
8159 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
8162 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
8163 V2DF/V2DI splat into two separate patterns, one that handles
8164 registers, and the other that only handles memory. Drop support
8165 for splatting from a GPR on ISA 2.07 and then splitting the
8166 splat into direct move and splat.
8167 (vsx_splat_<mode>_reg): Likewise.
8168 (vsx_splat_<mode>_mem): Likewise.
8170 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
8172 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
8174 2017-05-22 Jakub Jelinek <jakub@redhat.com>
8177 * omp-low.c (finish_taskreg_remap): New function.
8178 (finish_taskreg_scan): If unit size of ctx->record_type
8179 is non-constant, unshare the size expression and replace
8180 decls in it with possible outer var refs.
8183 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
8184 GOVD_SHARED rather than GOVD_PRIVATE with it.
8185 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
8186 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
8189 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
8190 as last argument to build_outer_var_ref for pointer bases of array
8193 2017-05-19 Martin Sebor <msebor@redhat.com>
8195 * print-tree.c (print_node): Print DECL_READ_P flag.
8197 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
8199 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
8200 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
8201 * cgraph.c: Likewise.
8202 * cgraphunit.c: Likewise.
8203 * gengtype.c: Likewise.
8204 * ipa-cp.c: Likewise.
8205 * ipa-devirt.c: Likewise.
8206 * ipa-icf.c: Likewise.
8207 * ipa-predicate.c: Likewise.
8208 * ipa-profile.c: Likewise.
8209 * ipa-prop.c: Likewise.
8210 * ipa-split.c: Likewise.
8212 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
8213 edge_predicate_pool, dump_inline_hints,
8214 inline_summary::account_size_time, redirect_to_unreachable,
8215 edge_set_predicate, set_hint_predicate,
8216 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
8217 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
8218 inline_summary_t::remove, remap_hint_predicate_after_duplication,
8219 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
8220 ipa_call_summary_t::remove, initialize_growth_caches,
8221 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
8222 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
8223 mark_modified, unmodified_parm_1, unmodified_parm,
8224 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
8225 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
8226 compute_bb_predicates, will_be_nonconstant_expr_predicate,
8227 will_be_nonconstant_predicate, record_modified_bb_info,
8228 get_minimal_bb, record_modified, param_change_prob,
8229 phi_result_unknown_predicate, predicate_for_phi_result,
8230 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
8231 estimate_function_body_sizes, compute_inline_parameters,
8232 compute_inline_parameters_for_curren, pass_data_inline_parameters,
8233 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
8234 inline_update_callee_summaries, remap_edge_change_prob,
8235 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
8236 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
8237 inline_analyze_function, inline_summary_t::insert,
8238 inline_generate_summary, read_ipa_call_summary, inline_read_section,
8239 inline_read_summary, write_ipa_call_summary, inline_write_summary,
8240 inline_free_summary): Move to ipa-fnsummary.h
8241 (predicate_t): Remove.
8242 * ipa-fnsummary.c: New file.
8243 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
8244 (enum inline_hints_vals, inline_hints, agg_position_info,
8245 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
8246 inline_summaries, ipa_call_summary, ipa_call_summary_t,
8247 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
8248 dump_inline_summary, dump_inline_hints, inline_generate_summary,
8249 inline_read_summary, inline_write_summary, inline_free_summary,
8250 inline_analyze_function, initialize_inline_failed,
8251 inline_merge_summary, inline_update_overall_summary,
8252 compute_inline_parameters): Move to ipa-fnsummary.h
8253 * ipa-fnsummary.h: New file.
8254 * ipa-inline-transform.h: Include ipa-inline.h.
8255 * ipa-inline.c: LIkewise.
8257 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
8259 * ipa-inline.c (edge_badness): Use inlined_time instead of
8260 inline_summaries->get.
8262 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
8264 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
8266 2017-05-22 Nathan Sidwell <nathan@acm.org>
8268 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
8269 (fdump-lang): Document 'raw' option.
8270 * dumpfile.h (TDI_tu): Delete.
8271 * dumpfile.c (dump_files): Remove translation-unit.
8272 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
8274 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
8276 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
8277 command option from $(AWK) call.
8278 * config/avr/genmultilib.awk: Simplify and rewrite so that it
8279 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
8280 [FORMAT]: Remove handling of variable.
8281 * config/avr/t-multilib: Regenerate.
8283 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
8285 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
8287 (dump_inline_summary): Do not print self_time.
8288 (estimate_function_body_sizes): Do not set self_time.
8289 (compute_inline_parameters): Likewise.
8290 (inline_read_section, inline_write_summary): Do not stream self_time.
8291 * ipa-inline.h (inline_summary): Drop self_time.
8293 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
8295 * ipa-inline-analysis.c (account_size_time): Rename to ...
8296 (inline_summary::account_size_time): ... this one.
8297 (reset_ipa_call_summary): Turn to ...
8298 (ipa_call_summary::reset): ... this one.
8299 (reset_inline_summary): Turn to ...
8300 (inline_summary::reset): ... this one.
8301 (inline_summary_t::remove): Update.
8302 (inline_summary_t::duplicate): Update.
8303 (ipa_call_summary_t::remove): Update.
8304 (dump_inline_summary): Update.
8305 (estimate_function_body_sizes): Update.
8306 (compute_inline_parameters): Update.
8307 (estimate_node_size_and_time): Update.
8308 (inline_merge_summary): Update.
8309 (inline_update_overall_summary): Update.
8310 (inline_read_section): Update.
8311 (inline_write_summary): Update.
8312 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
8313 add account_size_time and reset member functions.
8314 (ipa_call_summary): Add reset function.
8315 * ipa-predicate.h (predicate::operator &): Constify.
8317 2017-05-22 Richard Biener <rguenther@suse.de>
8319 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
8321 2017-05-19 Jason Merrill <jason@redhat.com>
8323 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
8325 2017-05-19 Marek Polacek <polacek@redhat.com>
8328 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
8329 TYPE_OVERFLOW_WRAPS checks.
8331 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
8333 * tree-core.h (enum omp_clause_default_kind): Add
8334 "OMP_CLAUSE_DEFAULT_PRESENT".
8335 * tree-pretty-print.c (dump_omp_clause): Handle it.
8336 * gimplify.c (enum gimplify_omp_var_data): Add
8337 "GOVD_MAP_FORCE_PRESENT".
8338 (gimplify_adjust_omp_clauses_1): Map it to
8339 "GOMP_MAP_FORCE_PRESENT".
8340 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
8342 * gimplify.c (oacc_default_clause): Clarify.
8344 2017-05-19 Nathan Sidwell <nathan@acm.org>
8346 LANG_HOOK_REGISTER_DUMPS
8347 * toplev.c (general_init): Call register dump lang hook.
8348 * doc/invoke.texi: Document -fdump-lang option family.
8349 * dumpfile.c (dump_files): Remove class dump here.
8350 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
8351 * dumpfile.h (tree_dump_index): Remove TDI_class.
8352 * langhooks-def.h (lhd_register_dumps): Declare.
8353 (LANG_HOOKS_REGISTER_DUMPS): Define.
8354 (LANG_HOOKS_INITIALIZER): Add it.
8355 * langhooks.c (lhd_register_dumps): Define.
8356 * langhooks.h (struct lang_hooks): Add register_dumps.
8358 2017-05-19 Nathan Sidwell <nathan@acm.org>
8360 * context.h (context::set_passes): New.
8361 * context.c (context::context): Do not create pass manager.
8362 * toplev.c (general_init): Create pass manager here.
8364 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
8366 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
8367 use this splitter if two add or or instructions would also work for
8368 the constant we want to generate.
8370 2017-05-19 Richard Biener <rguenther@suse.de>
8373 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
8374 predicate evaluation.
8376 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
8378 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
8380 * ipa-inline.c (want_inline_small_function_p): Do not cast to
8383 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
8385 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
8386 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
8387 (inline_edge_summary_vec): Turn into ...
8388 (ipa_call_summaries): ... this one.
8389 (redirect_to_unreachable, edge_set_predicate,
8390 evaluate_properties_for_edge, inline_summary_alloc,
8391 reset_ipa_call_summary, reset_inline_summary,
8392 inline_summary_t::duplicate): Update.
8393 (inline_edge_duplication_hook): Turn to ...
8394 (ipa_call_summary_t::duplicate): ... this one.
8395 (inline_edge_removal_hook): Turn to ...
8396 (ipa_call_summary_t::remove): ... this one.
8397 (dump_inline_edge_summary): Turn to ...
8398 (dump_ipa_call_summary): ... this one.
8399 (estimate_function_body_sizes): Update.
8400 (inline_update_callee_summaries): Update.
8401 (remap_edge_change_prob): Update.
8402 (remap_edge_summaries): Update.
8403 (inline_merge_summary): Update.
8404 (do_estimate_edge_time): Update.
8405 (inline_generate_summary): Update.
8406 (inline_read_section): Update.
8407 (inline_read_summary): Update.
8408 (inline_free_summary): Update.
8409 * ipa-inline.c (can_inline_edge_p): Update.
8410 (compute_inlined_call_time): Update.
8411 (want_inline_small_function_p): Update.
8412 (edge_badness): Update.
8413 (early_inliner): Update.
8414 * ipa-inline.h (inline_edge_summary): Turn to ...
8415 (ipa_call_summary): ... this one.
8416 (ipa_call_summary_t): New class.
8417 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
8418 (ipa_call_summaries): New.
8419 (inline_edge_summary): Remove.
8420 (estimate_edge_growth): Update.
8421 * ipa-profile.c (ipa_propagate_frequency_1): Update.
8422 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
8423 * ipa-split.c (execute_split_functions): Update.
8424 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
8426 2017-05-19 Richard Biener <rguenther@suse.de>
8429 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
8431 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
8433 * config/rs6000/rs6000.c (struct machine_function): Add field
8434 fpr_is_wrapped_separately.
8435 (rs6000_get_separate_components): Use 64 components. Handle the
8437 (rs6000_components_for_bb): Handle the FPR components.
8438 (rs6000_emit_prologue_components): Handle the FPR components.
8439 (rs6000_emit_epilogue_components): Handle the FPR components.
8440 (rs6000_set_handled_components): Handle the FPR components.
8441 (rs6000_emit_prologue): Don't output prologue code for those FPRs
8442 that are already separately shrink-wrapped.
8443 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
8444 that are already separately shrink-wrapped.
8446 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
8449 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
8452 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
8453 (define_peephole2 for Altivec d-form load): Add peepholes to catch
8454 cases where the register allocator uses a move and an offsettable
8455 memory operation to/from a FPR register on ISA 2.06/2.07.
8456 (define_peephole2 for Altivec d-form store): Likewise.
8458 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
8461 * config/i386/mmx.md (*mov<mode>_internal): Enable
8462 alternatives 11, 12, 13 and 14 also for 32bit targets.
8463 Remove alternatives 15, 16, 17 and 18.
8464 * config/i386/sse.md (vec_concatv2di): Change
8465 alternative (!x, *y) to (x, ?!*Yn).
8467 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
8469 * dumpfile.h (enum dump_kind): Remove stray comma.
8471 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
8473 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
8474 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
8475 predicate::num_conditions
8476 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
8477 (CHANGED): turn into predicate::changed.
8478 (agg_position_info): Move to ipa-predicate.h
8479 (add_condition, predicate::add_clause, predicate::operator &=,
8480 predicate::or_with, predicate::evaluate, predicate::probability,
8481 dump_condition, dump_clause, predicate::dump,
8482 predicate::remap_after_duplication, predicate::remap_after_inlining,
8483 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
8484 (evaluate_conditions_for_known_args): Update.
8485 (set_cond_stmt_execution_predicate): Update.
8486 * ipa-inline.h: Include ipa-predicate.h
8487 (condition, inline_param_summary, conditions, agg_position_info,
8488 predicate): Move to ipa-predicate.h
8489 * ipa-predicate.c: New file.
8490 * ipa-predicate.h: New file.
8492 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
8494 * final.c (leaf_function_p): Check we are not in a sequence.
8496 2017-05-18 Martin Liska <mliska@suse.cz>
8498 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
8499 * dumpfile.c (dump_register): Use new enum dump_kind.
8500 (get_dump_file_name): Likewise.
8501 (dump_enable_all): Likewise.
8502 (dump_switch_p_1): Likewise.
8503 (enable_rtl_dump_file): Remove usage of TDF_RTL.
8504 * dumpfile.h (enum dump_kind): New enum type.
8505 (struct dump_file_info): Create constructor and
8506 format fields and comments.
8507 * passes.c (pass_manager::register_one_dump_file):
8509 * statistics.c (statistics_early_init): Likewise.
8510 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
8511 TDF_TREE with TDF_SLIM.
8512 (gather_memory_references_ref): Likewise.
8514 2017-05-18 Martin Liska <mliska@suse.cz>
8516 * vec.h (struct vnull): Use it.
8518 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
8520 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
8521 (true_predicate, false_predicate, true_predicate_p,
8522 false_predicate_p): Remove.
8523 (single_cond_predicate, not_inlined_predicate): Turn to member function
8525 (add_condition): Update.
8526 (add_clause): Turn to...
8527 (predicate::add_clause): ... this one; update; allow passing NULL
8529 (and_predicates): Turn to ...
8530 (predicate::operator &=): ... this one.
8531 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
8532 (or_predicates): Turn to ...
8533 (predicate::or_with): ... this one.
8534 (evaluate_predicate): Turn to ...
8535 (predicate::evaluate): ... this one.
8536 (predicate_probability): Turn to ...
8537 (predicate::probability): ... this one.
8538 (dump_condition): Update.
8539 (dump_predicate): Turn to ...
8540 (predicate::dump): ... this one.
8541 (account_size_time): Update.
8542 (edge_set_predicate): Update.
8543 (set_hint_predicate): UPdate.
8544 (evaluate_conditions_for_known_args): Update.
8545 (evaluate_properties_for_edge): Update.
8546 (remap_predicate_after_duplication): Turn to...
8547 (predicate::remap_after_duplication): ... this one.
8548 (remap_hint_predicate_after_duplication): Update.
8549 (inline_summary_t::duplicate): UPdate.
8550 (dump_inline_edge_summary): Update.
8551 (dump_inline_summary): Update.
8552 (set_cond_stmt_execution_predicate): Update.
8553 (set_switch_stmt_execution_predicate): Update.
8554 (compute_bb_predicates): Update.
8555 (will_be_nonconstant_expr_predicate): Update.
8556 (will_be_nonconstant_predicate): Update.
8557 (phi_result_unknown_predicate): Update.
8558 (predicate_for_phi_result): Update.
8559 (array_index_predicate): Update.
8560 (estimate_function_body_sizes): Update.
8561 (estimate_node_size_and_time): Update.
8562 (estimate_ipcp_clone_size_and_time): Update.
8563 (remap_predicate): Rename to ...
8564 (predicate::remap_after_inlining): ... this one.
8565 (remap_hint_predicate): Update.
8566 (inline_merge_summary): Update.
8567 (inline_update_overall_summary): Update.
8568 (estimate_size_after_inlining): Update.
8569 (read_predicate): Rename to ...
8570 (predicate::stream_in): ... this one.
8571 (read_inline_edge_summary): Update.
8572 (write_predicate): Rename to ...
8573 (predicate::stream_out): ... this one.
8574 (write_inline_edge_summary): Update.
8575 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
8576 (clause_t): Turn to uint32_t
8577 (predicate): Turn to class; implement constructor and operators
8579 (size_time_entry): Update.
8580 (inline_summary): Update.
8581 (inline_edge_summary): Update.
8583 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
8585 * fold-const.c (fold_binary_loc): Move transformation...
8586 * match.pd (C - X CMP X): ... here.
8588 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
8590 * config/sparc/sparc.c (sparc_option_override): Set function
8591 alignment for -mcpu=niagara7 to 64 to match the I$ line.
8592 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
8594 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
8596 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
8598 2017-05-18 Marek Polacek <polacek@redhat.com>
8601 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
8602 (pass_ubsan::execute): Call gimple_assign_single_p instead of
8603 gimple_assign_load_p.
8605 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
8608 * real.c (do_compare): Give decimal_do_compare preference over
8609 comparing just the signs.
8611 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
8613 * doc/md.texi (Canonicalization of Instructions): Describe the
8614 canonical form of instructions that inherently set a condition
8617 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
8620 * tree-cfg.c: Move deletion of unreachable case statements to after
8621 the merging of consecutive case labels.
8623 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
8625 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
8626 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
8627 restoring of callee-saved registers.
8629 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
8631 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
8632 * config/visium/visium.c (single_set_and_flags): Likewise.
8633 * config/visium/visium.md (Substitutions): Likewise.
8635 2017-05-17 Martin Liska <mliska@suse.cz>
8637 * cfg.c: Introduce dump_flags_t type and
8638 use it instead of int type.
8640 * cfghooks.c: Likewise.
8641 * cfghooks.h (struct cfg_hooks): Likewise.
8642 * cfgrtl.c: Likewise.
8643 * cfgrtl.h: Likewise.
8644 * cgraph.c (cgraph_node::get_body): Likewise.
8645 * coretypes.h: Likewise.
8646 * domwalk.c: Likewise.
8647 * domwalk.h: Likewise.
8648 * dumpfile.c (struct dump_option_value_info): Likewise.
8649 (dump_enable_all): Likewise.
8650 (dump_switch_p_1): Likewise.
8651 (opt_info_switch_p): Likewise.
8652 * dumpfile.h (enum tree_dump_index): Likewise.
8653 (struct dump_file_info): Likewise.
8654 * genemit.c: Likewise.
8655 * generic-match-head.c: Likewise.
8656 * gengtype.c (open_base_files): Likewise.
8657 * gimple-pretty-print.c: Likewise.
8658 * gimple-pretty-print.h: Likewise.
8659 * graph.c (print_graph_cfg): Likewise.
8660 * graphite-scop-detection.c (dot_all_sese): Likewise.
8661 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
8662 * loop-unroll.c (report_unroll): Likewise.
8663 * passes.c (pass_manager::register_one_dump_file): Likewise.
8664 * print-tree.c: Likewise.
8665 * statistics.c: Likewise.
8666 * tree-cfg.c: Likewise.
8667 * tree-cfg.h: Likewise.
8668 * tree-dfa.c: Likewise.
8669 * tree-dfa.h: Likewise.
8670 * tree-dump.c (dump_function): Likewise.
8671 * tree-dump.h (struct dump_info): Likewise.
8672 * tree-pretty-print.c: Likewise.
8673 * tree-pretty-print.h: Likewise.
8674 * tree-ssa-live.c: Likewise.
8675 * tree-ssa-live.h: Likewise.
8676 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
8677 * tree-vect-loop.c: Likewise.
8678 * tree-vect-slp.c: Likewise.
8680 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
8681 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8683 PR tree-optimization/80457
8684 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
8685 of all arguments to a statement as scalar_to_vec operations.
8686 (vectorizable_call): Adjust call to vect_model_simple_cost for
8688 (vectorizable_conversion): Likewise.
8689 (vectorizable_assignment): Likewise.
8690 (vectorizable_shift): Likewise.
8691 (vectorizable_operation): Likewise.
8692 (vectorizable_comparison): Likewise.
8693 (vect_is_simple_cond): Record the def types for operands.
8694 (vectorizable_condition): Likewise, call vect_model_simple_cost.
8695 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
8696 for statement argument count.
8698 2017-05-16 Carl Love <cel@us.ibm.com>
8700 * config/rs6000/rs6000-c: Add support for built-in functions
8701 vector unsigned long long vec_bperm (vector unsigned long long,
8702 vector unsigned char)
8703 vector signed long long vec_mule (vector signed int,
8705 vector unsigned long long vec_mule (vector unsigned int,
8706 vector unsigned int)
8707 vector signed long long vec_mulo (vector signed int,
8709 vector unsigned long long vec_mulo (vector unsigned int,
8710 vector unsigned int)
8711 vector signed char vec_sldw (vector signed char,
8714 vector unsigned char vec_sldw (vector unsigned char,
8715 vector unsigned char,
8717 vector signed short vec_sldw (vector signed short,
8718 vector signed short,
8720 vector unsigned short vec_sldw (vector unsigned short,
8721 vector unsigned short,
8723 vector signed int vec_sldw (vector signed int,
8726 vector unsigned int vec_sldw (vector unsigned int,
8727 vector unsigned int,
8729 vector signed long long vec_sldw (vector signed long long,
8730 vector signed long long,
8732 vector unsigned long long vec_sldw (vector unsigned long long,
8733 vector unsigned long long,
8735 * config/rs6000/rs6000-c: Add support for built-in functions
8736 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
8737 * config/rs6000/altivec.h: Add defintion for vec_sldw.
8738 * doc/extend.texi: Update the built-in documentation for the
8739 new built-in functions.
8741 2017-05-16 Marek Polacek <polacek@redhat.com>
8745 * tree.c (save_expr): Don't fold the expression.
8747 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
8749 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
8750 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
8751 and (?*y,m). Update insn attributes.
8753 2017-05-16 Martin Liska <mliska@suse.cz>
8755 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
8756 flags argument of print_gimple_stmt, print_gimple_expr,
8757 print_generic_stmt and print_generic_expr.
8758 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
8759 * coretypes.h: Likewise.
8760 * except.c (dump_eh_tree): Likewise.
8761 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8762 * gimple-pretty-print.h: Likewise.
8763 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
8764 (backprop::push_to_worklist): Likewise.
8765 (backprop::pop_from_worklist): Likewise.
8766 (backprop::process_use): Likewise.
8767 (backprop::intersect_uses): Likewise.
8768 (note_replacement): Likewise.
8769 * gimple-ssa-store-merging.c
8770 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
8771 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
8772 (pass_store_merging::execute): Likewise.
8773 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
8774 (ssa_base_cand_dump_callback): Likewise.
8775 (dump_incr_vec): Likewise.
8776 (replace_refs): Likewise.
8777 (replace_mult_candidate): Likewise.
8778 (create_add_on_incoming_edge): Likewise.
8779 (create_phi_basis): Likewise.
8780 (insert_initializers): Likewise.
8781 (all_phi_incrs_profitable): Likewise.
8782 (introduce_cast_before_cand): Likewise.
8783 (replace_one_candidate): Likewise.
8784 * gimplify.c (gimplify_expr): Likewise.
8785 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
8786 (set_rename): Likewise.
8787 (rename_uses): Likewise.
8788 (copy_loop_phi_nodes): Likewise.
8789 (add_close_phis_to_merge_points): Likewise.
8790 (copy_loop_close_phi_args): Likewise.
8791 (copy_cond_phi_args): Likewise.
8792 (graphite_copy_stmts_from_block): Likewise.
8793 (translate_pending_phi_nodes): Likewise.
8794 * graphite-poly.c (print_pdr): Likewise.
8795 (dump_gbb_cases): Likewise.
8796 (dump_gbb_conditions): Likewise.
8797 (print_scop_params): Likewise.
8798 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
8799 (build_cross_bb_scalars_use): Likewise.
8800 (gather_bbs::before_dom_children): Likewise.
8801 * hsa-dump.c (dump_hsa_immed): Likewise.
8802 * ipa-cp.c (print_ipcp_constant_value): Likewise.
8803 (get_replacement_map): Likewise.
8804 * ipa-inline-analysis.c (dump_condition): Likewise.
8805 (estimate_function_body_sizes): Likewise.
8806 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
8807 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
8808 * ipa-prop.c (ipa_dump_param): Likewise.
8809 (ipa_print_node_jump_functions_for_edge): Likewise.
8810 (ipa_modify_call_arguments): Likewise.
8811 (ipa_modify_expr): Likewise.
8812 (ipa_dump_param_adjustments): Likewise.
8813 (ipa_dump_agg_replacement_values): Likewise.
8814 (ipcp_modif_dom_walker::before_dom_children): Likewise.
8815 * ipa-pure-const.c (check_stmt): Likewise.
8816 (pass_nothrow::execute): Likewise.
8817 * ipa-split.c (execute_split_functions): Likewise.
8818 * omp-offload.c (dump_oacc_loop_part): Likewise.
8819 (dump_oacc_loop): Likewise.
8820 * trans-mem.c (tm_log_emit): Likewise.
8821 (tm_memopt_accumulate_memops): Likewise.
8822 (dump_tm_memopt_set): Likewise.
8823 (dump_tm_memopt_transform): Likewise.
8824 * tree-cfg.c (gimple_verify_flow_info): Likewise.
8825 (print_loop): Likewise.
8826 * tree-chkp-opt.c (chkp_print_addr): Likewise.
8827 (chkp_gather_checks_info): Likewise.
8828 (chkp_get_check_result): Likewise.
8829 (chkp_remove_check_if_pass): Likewise.
8830 (chkp_use_outer_bounds_if_possible): Likewise.
8831 (chkp_reduce_bounds_lifetime): Likewise.
8832 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
8833 (chkp_mark_completed_bounds): Likewise.
8834 (chkp_register_incomplete_bounds): Likewise.
8835 (chkp_mark_invalid_bounds): Likewise.
8836 (chkp_maybe_copy_and_register_bounds): Likewise.
8837 (chkp_build_returned_bound): Likewise.
8838 (chkp_get_bound_for_parm): Likewise.
8839 (chkp_build_bndldx): Likewise.
8840 (chkp_get_bounds_by_definition): Likewise.
8841 (chkp_generate_extern_var_bounds): Likewise.
8842 (chkp_get_bounds_for_decl_addr): Likewise.
8843 * tree-chrec.c (chrec_apply): Likewise.
8844 * tree-data-ref.c (dump_data_reference): Likewise.
8845 (dump_subscript): Likewise.
8846 (dump_data_dependence_relation): Likewise.
8847 (analyze_overlapping_iterations): Likewise.
8848 * tree-inline.c (expand_call_inline): Likewise.
8849 (tree_function_versioning): Likewise.
8850 * tree-into-ssa.c (dump_defs_stack): Likewise.
8851 (dump_currdefs): Likewise.
8852 (dump_names_replaced_by): Likewise.
8853 (dump_update_ssa): Likewise.
8854 (update_ssa): Likewise.
8855 * tree-object-size.c (pass_object_sizes::execute): Likewise.
8856 * tree-parloops.c (build_new_reduction): Likewise.
8857 (try_create_reduction_list): Likewise.
8858 (ref_conflicts_with_region): Likewise.
8859 (oacc_entry_exit_ok_1): Likewise.
8860 (oacc_entry_exit_single_gang): Likewise.
8861 * tree-pretty-print.h: Likewise.
8862 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
8863 (get_scalar_evolution): Likewise.
8864 (add_to_evolution): Likewise.
8865 (get_loop_exit_condition): Likewise.
8866 (analyze_evolution_in_loop): Likewise.
8867 (analyze_initial_condition): Likewise.
8868 (analyze_scalar_evolution): Likewise.
8869 (instantiate_scev): Likewise.
8870 (number_of_latch_executions): Likewise.
8871 (gather_chrec_stats): Likewise.
8872 (final_value_replacement_loop): Likewise.
8873 (scev_const_prop): Likewise.
8874 * tree-sra.c (dump_access): Likewise.
8875 (disqualify_candidate): Likewise.
8876 (create_access): Likewise.
8878 (maybe_add_sra_candidate): Likewise.
8879 (create_access_replacement): Likewise.
8880 (analyze_access_subtree): Likewise.
8881 (analyze_all_variable_accesses): Likewise.
8882 (sra_modify_assign): Likewise.
8883 (initialize_constant_pool_replacements): Likewise.
8884 (find_param_candidates): Likewise.
8885 (decide_one_param_reduction): Likewise.
8886 (replace_removed_params_ssa_names): Likewise.
8887 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
8888 * tree-ssa-copy.c (dump_copy_of): Likewise.
8889 (copy_prop_visit_cond_stmt): Likewise.
8890 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
8891 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
8892 (record_equivalences_from_stmt): Likewise.
8893 * tree-ssa-dse.c (compute_trims): Likewise.
8894 (delete_dead_call): Likewise.
8895 (delete_dead_assignment): Likewise.
8896 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
8897 (forward_propagate_into_cond): Likewise.
8898 (pass_forwprop::execute): Likewise.
8899 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8900 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
8902 (move_computations_worker): Likewise.
8903 (execute_sm): Likewise.
8904 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
8905 (remove_exits_and_undefined_stmts): Likewise.
8906 (remove_redundant_iv_tests): Likewise.
8907 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
8908 (adjust_iv_update_pos): Likewise.
8909 * tree-ssa-math-opts.c (bswap_replace): Likewise.
8910 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
8911 (value_replacement): Likewise.
8912 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
8913 * tree-ssa-pre.c (print_pre_expr): Likewise.
8914 (get_representative_for): Likewise.
8915 (create_expression_by_pieces): Likewise.
8916 (insert_into_preds_of_block): Likewise.
8917 (eliminate_insert): Likewise.
8918 (eliminate_dom_walker::before_dom_children): Likewise.
8919 (eliminate): Likewise.
8920 (remove_dead_inserted_code): Likewise.
8921 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
8922 * tree-ssa-reassoc.c (get_rank): Likewise.
8923 (eliminate_duplicate_pair): Likewise.
8924 (eliminate_plus_minus_pair): Likewise.
8925 (eliminate_not_pairs): Likewise.
8926 (undistribute_ops_list): Likewise.
8927 (eliminate_redundant_comparison): Likewise.
8928 (update_range_test): Likewise.
8929 (optimize_range_tests_var_bound): Likewise.
8930 (optimize_vec_cond_expr): Likewise.
8931 (rewrite_expr_tree): Likewise.
8932 (rewrite_expr_tree_parallel): Likewise.
8933 (linearize_expr): Likewise.
8934 (break_up_subtract): Likewise.
8935 (linearize_expr_tree): Likewise.
8936 (attempt_builtin_powi): Likewise.
8937 (attempt_builtin_copysign): Likewise.
8938 (transform_stmt_to_copy): Likewise.
8939 (transform_stmt_to_multiply): Likewise.
8940 (dump_ops_vector): Likewise.
8941 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
8942 (print_scc): Likewise.
8943 (set_ssa_val_to): Likewise.
8944 (visit_reference_op_store): Likewise.
8945 (visit_use): Likewise.
8946 (sccvn_dom_walker::before_dom_children): Likewise.
8947 (run_scc_vn): Likewise.
8948 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
8950 (expr_hash_elt::print): Likewise.
8951 (const_and_copies::pop_to_marker): Likewise.
8952 (const_and_copies::record_const_or_copy_raw): Likewise.
8953 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
8954 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
8955 (dump_predicates): Likewise.
8956 (find_uninit_use): Likewise.
8957 (warn_uninitialized_phi): Likewise.
8958 (pass_late_warn_uninitialized::execute): Likewise.
8959 * tree-ssa.c (verify_vssa): Likewise.
8960 (verify_ssa): Likewise.
8961 (maybe_optimize_var): Likewise.
8962 * tree-vrp.c (dump_value_range): Likewise.
8963 (dump_all_value_ranges): Likewise.
8964 (dump_asserts_for): Likewise.
8965 (register_edge_assert_for_2): Likewise.
8966 (vrp_visit_cond_stmt): Likewise.
8967 (vrp_visit_switch_stmt): Likewise.
8968 (vrp_visit_stmt): Likewise.
8969 (vrp_visit_phi_node): Likewise.
8970 (simplify_cond_using_ranges_1): Likewise.
8971 (fold_predicate_in): Likewise.
8972 (evrp_dom_walker::before_dom_children): Likewise.
8973 (evrp_dom_walker::push_value_range): Likewise.
8974 (evrp_dom_walker::pop_value_range): Likewise.
8975 (execute_early_vrp): Likewise.
8977 2017-05-16 Richard Biener <rguenther@suse.de>
8979 * dwarf2out.c (loc_list_from_tree_1): Do not create
8980 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
8982 2017-05-16 Richard Biener <rguenther@suse.de>
8984 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
8986 (note_variable_value_in_expr): If we resolved the decl ref
8987 do not push to the stack.
8989 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
8991 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
8992 operations in fast-math mode.
8993 (vaddq_f16): Likewise.
8994 (vmul_f16): Likewise.
8995 (vmulq_f16): Likewise.
8996 (vsub_f16): Likewise.
8997 (vsubq_f16): Likewise.
8998 * config/arm/neon.md (add<mode>3): New.
9000 (fma:<VH:mode>3): New. Also remove outdated comment.
9003 2017-05-16 Martin Liska <mliska@suse.cz>
9007 * ipa-devirt.c (warn_types_mismatch): Fix typo.
9008 (odr_types_equivalent_p): Likewise.
9010 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
9012 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
9014 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
9017 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
9018 non-interunit SSE move alternatives with '?'.
9019 (zero-extendsidi peephole2): New peephole to skip intermediate
9020 general register in SSE zero-extend sequence.
9022 2017-05-15 Jeff Law <law@redhat.com>
9024 * reorg.c (relax_delay_slots): Create a new variable to hold
9025 the temporary target rather than clobbering TARGET_LABEL.
9027 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
9028 missing argument to extract_bit_field call.
9029 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
9031 2017-05-15 Martin Liska <mliska@suse.cz>
9034 * gcc.c (process_command): Do not allow empty argument of -o option.
9036 2017-05-15 Renlin Li <renlin.li@arm.com>
9038 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
9039 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
9040 * config/aarch64/constraints.md (Usf): Add long call check.
9041 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
9042 (call_value): Likewise.
9043 (sibcall): Likewise.
9044 (sibcall_value): Likewise.
9046 (call_value_insn): New.
9047 (sibcall_insn): Update rtx pattern.
9048 (sibcall_value_insn): Likewise.
9049 (call_internal): Remove.
9050 (call_value_internal): Likewise.
9051 (sibcall_internal): Likewise.
9052 (sibcall_value_internal): Likewise.
9053 (call_reg): Likewise.
9054 (call_symbol): Likewise.
9055 (call_value_reg): Likewise.
9056 (call_value_symbol): Likewise.
9058 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
9061 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
9063 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
9065 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
9066 compatible with CCGOCmode and with CCZmode.
9068 2017-05-14 Martin Sebor <msebor@redhat.com>
9071 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
9072 (gimple_fold_builtin_snprintf): Same.
9073 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
9074 (gimple_fold_builtin_snprintf): Same.
9075 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
9077 (is_call_safe): New function.
9078 (try_substitute_return_value): Call it.
9079 (try_simplify_call): New function.
9080 (pass_sprintf_length::handle_gimple_call): Call it.
9082 2017-05-14 Martin Sebor <msebor@redhat.com>
9085 * builtins.c (expand_builtin_stpncpy): Simplify.
9087 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
9089 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
9090 * config/i386/i386.h
9091 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
9092 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
9093 (struct machine_function): Add new members call_ms2sysv,
9094 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
9095 (struct machine_frame_state): New fields sp_realigned and
9096 sp_realigned_offset.
9097 * config/i386/i386.c
9098 (enum xlogue_stub): New enum.
9099 (enum xlogue_stub_sets): New enum.
9100 (class xlogue_layout): New class.
9101 (struct ix86_frame): New fields stack_realign_allocate_offset,
9102 stack_realign_offset and outlined_save_offset. Modify comments to
9103 detail stack layout when using out-of-line stubs.
9104 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
9105 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
9106 -mcall-ms2sysv-xlogues.
9107 (stub_managed_regs): New static variable.
9108 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
9109 registers managed by out-of-line stub.
9110 (disable_call_ms2sysv_xlogues): New function.
9111 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
9112 m->call_ms2sysv when appropriate and compute frame layout for
9114 (sp_valid_at, fp_valid_at): New inline functions.
9115 (choose_basereg): New function.
9116 (choose_baseaddr): Add align parameter, use choose_basereg and modify
9118 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
9119 Use align parameter of choose_baseaddr to generated aligned SSE movs
9121 (pro_epilogue_adjust_stack): Modify to track
9122 machine_frame_state::sp_realigned.
9123 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
9124 (ix86_nsaved_sseregs): Likewise.
9125 (ix86_emit_save_regs): Likewise.
9126 (ix86_emit_save_regs_using_mov): Likewise.
9127 (ix86_emit_save_sse_regs_using_mov): Likewise.
9128 (get_scratch_register_on_entry): Likewise.
9129 (gen_frame_set): New function.
9130 (gen_frame_load): Likewise.
9131 (gen_frame_store): Likewise.
9132 (emit_outlined_ms2sysv_save): Likewise.
9133 (emit_outlined_ms2sysv_restore): Likewise.
9134 (ix86_expand_prologue): Modify stack re-alignment code and call
9135 emit_outlined_ms2sysv_save when appropriate.
9136 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
9137 parameter rtx_insn *insn, which allows the function to be used to only
9139 (ix86_expand_epilogue): Modify validity checks of frame and stack
9140 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
9141 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
9142 * config/i386/predicates.md
9143 (save_multiple): New predicate.
9144 (restore_multiple): Likewise.
9145 * config/i386/sse.md
9146 (save_multiple<mode>): New pattern.
9147 (save_multiple_realign<mode>): Likewise.
9148 (restore_multiple<mode>): Likewise.
9149 (restore_multiple_and_return<mode>): Likewise.
9150 (restore_multiple_leave_return<mode>): Likewise.
9151 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
9153 2017-05-14 Julia Koval <julia.koval@intel.com>
9155 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
9156 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
9157 (__builtin_ia32_xsetbv): New builtins.
9158 * config/i386/i386.c (ix86_expand_special_args_builtin):
9160 (ix86_expand_builtin): Special expand for new intrinsics.
9161 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
9162 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
9163 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
9165 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9167 * cfganal.c (inverted_post_order_compute): Change argument type
9169 * cfganal.h (inverted_post_order_compute): Adjust prototype.
9170 * df-core.c (rest_of_handle_df_initialize): Adjust.
9171 (rest_of_handle_df_finish): Likewise.
9172 (df_analyze_1): Likewise.
9173 (df_analyze): Likewise.
9174 (loop_inverted_post_order_compute): Change argument to be a vec *.
9175 (df_analyze_loop): Adjust.
9176 (df_get_n_blocks): Likewise.
9177 (df_get_postorder): Likewise.
9178 * df.h (struct df_d): Change field to be a vec.
9179 * lcm.c (compute_laterin): Adjust.
9180 (compute_available): Likewise.
9181 * lra-lives.c (lra_create_live_ranges_1): Likewise.
9182 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
9183 * tree-ssa-pre.c (compute_antic): Likewise.
9185 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9187 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
9188 (depth_first_search::depth_first_search): Change structure init
9189 function to this constructor.
9190 (depth_first_search::add_bb): Rename function to this member.
9191 (depth_first_search::execute): Likewise.
9192 (flow_dfs_compute_reverse_finish): Adjust.
9194 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9196 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
9197 (longest_simple_path): Likewise.
9198 * shrink-wrap.c (spread_components): Likewise.
9199 (disqualify_problematic_components): Likewise.
9200 (emit_common_heads_for_components): Likewise.
9201 (emit_common_tails_for_components): Likewise.
9202 (insert_prologue_epilogue_for_components): Likewise.
9204 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9206 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
9209 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9211 * df-core.c (df_set_blocks): Start using auto_bitmap.
9212 (df_compact_blocks): Likewise.
9213 * df-problems.c (df_rd_confluence_n): Likewise.
9214 * df-scan.c (df_insn_rescan_all): Likewise.
9215 (df_process_deferred_rescans): Likewise.
9216 (df_update_entry_block_defs): Likewise.
9217 (df_update_exit_block_uses): Likewise.
9218 (df_entry_block_bitmap_verify): Likewise.
9219 (df_exit_block_bitmap_verify): Likewise.
9220 (df_scan_verify): Likewise.
9221 * lra-constraints.c (lra_constraints): Likewise.
9222 (undo_optional_reloads): Likewise.
9223 (lra_undo_inheritance): Likewise.
9224 * lra-remat.c (calculate_gen_cands): Likewise.
9225 (do_remat): Likewise.
9226 * lra-spills.c (assign_spill_hard_regs): Likewise.
9227 (spill_pseudos): Likewise.
9228 * tree-ssa-pre.c (bitmap_set_and): Likewise.
9229 (bitmap_set_subtract_values): Likewise.
9231 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9233 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
9234 management with auto_bitmap.
9235 (fix_inter_tick): Likewise.
9236 (fix_recovery_deps): Likewise.
9237 * ira.c (add_store_equivs): Likewise.
9238 (find_moveable_pseudos): Likewise.
9239 (split_live_ranges_for_shrink_wrap): Likewise.
9240 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
9241 (rtx_reuse_manager::seen_def_p): Likewise.
9242 (rtx_reuse_manager::set_seen_def): Likewise.
9243 * print-rtl.h (class rtx_reuse_manager): Likewise.
9245 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9247 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
9249 (migrate_btr_def): Likewise.
9250 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
9251 * df-core.c (loop_post_order_compute): Likewise.
9252 (loop_inverted_post_order_compute): Likewise.
9253 * hsa-common.h: Likewise.
9254 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
9255 * init-regs.c (initialize_uninitialized_regs): Likewise.
9256 * ipa-inline.c (resolve_noninline_speculation): Likewise.
9257 (inline_small_functions): Likewise.
9258 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
9259 * ira.c (combine_and_move_insns): Likewise.
9260 (build_insn_chain): Likewise.
9261 * loop-invariant.c (find_invariants): Likewise.
9262 * lower-subreg.c (propagate_pseudo_copies): Likewise.
9263 * predict.c (tree_predict_by_opcode): Likewise.
9264 (predict_paths_leading_to): Likewise.
9265 (predict_paths_leading_to_edge): Likewise.
9266 (estimate_loops_at_level): Likewise.
9267 (estimate_loops): Likewise.
9268 * shrink-wrap.c (try_shrink_wrapping): Likewise.
9269 (spread_components): Likewise.
9270 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
9271 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
9272 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
9273 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
9274 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
9275 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
9276 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
9277 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
9278 (mark_threaded_blocks): Likewise.
9279 (thread_through_all_blocks): Likewise.
9280 * tree-ssa.c (verify_ssa): Likewise.
9281 (execute_update_addresses_taken): Likewise.
9282 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
9284 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9286 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
9288 (post_order_compute): Likewise.
9289 (inverted_post_order_compute): Likewise.
9290 (pre_and_rev_post_order_compute_fn): Likewise.
9292 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9294 * genrecog.c (int_set::int_set): Explicitly construct our
9295 auto_vec base class.
9296 * vec.h (auto_vec::auto_vec): New constructor.
9298 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9300 * bitmap.h (class auto_bitmap): New constructor taking
9301 bitmap_obstack * argument.
9303 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9305 * bitmap.h (class auto_bitmap): Change type of m_bits to
9306 bitmap_head, and adjust ctor / dtor and member operators.
9308 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
9310 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
9311 when returned register mode doesn't match original mode.
9313 2017-05-12 Jeff Law <law@redhat.com>
9314 Jakub Jelinek <jakub@redhat.com>
9316 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
9317 we look for cc setter after the compare-elim changes.
9318 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
9319 within the vector to match what compare-elim now expects.
9320 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
9321 (xorsi3_flags, one_cmplsi2_flags): Likewise.
9323 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
9324 after the compare-elim changes.
9325 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
9326 the vector to match what compare-elim now expects.
9327 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
9328 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
9329 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
9330 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
9331 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
9333 * config/visium/visium.c (single_set_and_flags): Fix where
9334 we look for cc setter after the compare-elim changes.
9335 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
9336 with the vector to match what compare-elim now expects.
9337 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
9338 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
9339 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
9340 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
9341 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
9342 (neg<mode>2_insn_set_overflow): Likewise.
9344 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
9347 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
9348 maybe_expand_insn call, set ops[0].target. If still set after call,
9349 set alt_rtl. Add extra arg to recursive calls.
9350 (extract_bit_field): Add alt_rtl argument. Pass to
9352 * expmed.h (extract_bit_field): Fix prototype.
9353 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
9354 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
9355 to extract_bit_field_calls.
9356 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
9357 Pass alt_rtl to extract_bit_field calls.
9358 * calls.c (store_unaligned_arguments_into_psuedos)
9359 load_register_parameters): Pass extra NULL to extract_bit_field calls.
9360 * optabs.c (maybe_legitimize_operand): Clear op->target when call
9362 * optabs.h (struct expand_operand): Add target bitfield.
9364 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
9366 * compare-elim.c (try_eliminate_compare): Canonicalize
9367 operation with embedded compare to
9368 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
9369 (set (reg) (operation)].
9371 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
9373 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
9376 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
9377 cost of adding a carry flag for ADC instruction.
9378 [case MINUS]: Ignore the cost of subtracting a carry flag
9379 for SBB instruction.
9381 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
9383 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
9385 * config/rs6000/bmiintrin.h: New file.
9386 * config/rs6000/bmi2intrin.h: New file.
9387 * config/rs6000/x86intrin.h: New file.
9389 2017-05-12 Jeff Law <law@redhat.com>
9391 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
9394 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
9397 * tree-cfg.c: Remove cfg edges of unreachable case statements.
9399 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9401 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
9402 early expansion of vector divide builtins.
9403 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
9404 builtins identified as having unsigned arguments.
9406 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9408 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
9409 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
9410 expansion of vector logical operations (and, andc, or, xor,
9413 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9415 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
9416 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
9418 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9420 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
9421 early GIMPLE expansion of vector multiplies.
9423 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9425 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
9426 TARGET_HAVE_MOVT conditional.
9427 (movt splitter): Likewise.
9429 2017-05-12 Richard Biener <rguenther@suse.de>
9431 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
9432 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9433 Fold all stmts not inplace.
9435 2017-05-12 Richard Biener <rguenther@suse.de>
9437 PR tree-optimization/80713
9438 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
9439 inserted_exprs bit for not removed stmts.
9441 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
9444 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
9445 parallelized" attribute for parallelized OpenACC kernels.
9446 * omp-offload.c (execute_oacc_device_lower): Use it.
9448 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
9449 Set "oacc kernels" attribute.
9450 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
9451 parameter. Adjust all users.
9452 (oacc_fn_attrib_kernels_p): Remove function.
9453 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
9454 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
9455 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
9456 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
9457 assert "oacc kernels" attribute is set.
9459 2017-05-11 Carl Love <cel@us.ibm.com>
9461 * config/rs6000/rs6000-c: Add support for built-in functions
9462 vector unsigned char vec_popcnt (vector signed char)
9463 vector unsigned char vec_popcnt (vector unsigned char)
9464 vector unsigned short vec_popcnt (vector signed short)
9465 vector unsigned short vec_popcnt (vector unsigned short)
9466 vector unsigned int vec_popcnt (vector signed int)
9467 vector unsigned int vec_popcnt (vector unsigned int)
9468 vector unsigned long long vec_popcnt (vector signed long long)
9469 vector unsigned long long vec_popcnt (vector unsigned long long)
9470 vector signed long long vec_slo (vector signed long long,
9472 vector signed long long vec_slo (vector signed long long,
9473 vector unsigned char)
9474 vector unsigned long long vec_slo (vector unsigned long long,
9476 vector unsigned long long vec_slo (vector unsigned long long,
9477 vector unsigned char)
9478 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
9479 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
9480 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
9481 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
9482 * doc/extend.texi: Update the built-in documentation file for the
9483 new built-in functions.
9485 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
9487 * attribs.h (sorted_attr_string): Move machine independent
9488 functions for target clone support from the i386 port to common
9489 code. Rename ix86_function_versions to common_function_versions.
9490 Rename make_name to make_unique_name.
9491 (common_function_versions): Likewise.
9492 (make_unique_name): Likewise.
9493 (make_dispatcher_decl): Likewise.
9494 (is_function_default_version): Likewise.
9495 * attribs.c (attr_strcmp): Likewise.
9496 (sorted_attr_string): Likewise.
9497 (common_function_versions): Likewise.
9498 (make_unique_name): Likewise.
9499 (make_dispatcher_decl): Likewise.
9500 (is_function_default_version): Likewise.
9501 * config/i386/i386.c (attr_strcmp): Likewise.
9502 (sorted_attr_string): Likewise.
9503 (ix86_function_versions): Likewise.
9504 (make_name): Likewise.
9505 (make_dispatcher_decl): Likewise.
9506 (is_function_default_version): Likewise.
9507 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
9509 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9512 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
9513 Account for direct move costs for vec_construct of integer
9516 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
9519 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
9520 (UNSPEC_STX_ATOMIC): Ditto.
9521 (loaddi_via_sse): New insn.
9522 (storedi_via_sse): Ditto.
9523 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
9524 Update corresponding peephole2 patterns.
9525 (atomic_storedi_fpu): Ditto.
9527 2017-05-11 Julia Koval <julia.koval@intel.com>
9529 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
9530 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
9532 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
9533 (__builtin_ia32_rsqrt14ss_mask): New builtins.
9534 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
9536 2017-05-11 Nathan Sidwell <nathan@acm.org>
9538 * graphite-poly.c: Include dumpfile.h.
9540 * dumpfle.h (dump_function): Declare here ...
9541 * tree-dump.h (dump_function): ... not here.
9542 * dumpfile.c: #include tree-cfg.h.
9543 (dump_function): Move here from ...
9544 * tree-dump.c (dump_function): ... here.
9545 * gimplify.c: #include splay-tree.h, not tree-dump.h.
9546 * graphite-poly.c: Don't include tree-dump.h.
9547 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
9548 * print-tree.c: Likewise.
9549 * stor-layout.c: Likewise.
9550 * tree-nested.c: Likewise.
9552 * dumpfile.c (dump_start): Use TDF_FLAGS.
9553 (dump_enable_all): Fix TDF_KIND check thinko.
9555 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
9557 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9558 array entries to represent two legal parameterizations of the
9559 overloaded __builtin_cmpb function, as represented by the
9560 P6_OV_BUILTIN_CMPB constant.
9561 (altivec_resolve_overloaded_builtin): Add special case handling
9562 for the __builtin_cmpb function, as represented by the
9563 P6_OV_BUILTIN_CMPB constant.
9564 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
9565 (BU_P6_64BIT_2): New macro.
9566 (BU_P6_OVERLOAD_2): New macro
9567 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
9568 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
9569 (CMPB): Add overload support to represent both 32-bit and 64-bit
9570 compare-bytes function.
9571 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9572 support for TARGET_CMPB.
9573 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
9574 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9575 documentation of the __builtin_cmpb overloaded built-in function.
9577 2017-05-11 Richard Biener <rguenther@suse.de>
9579 PR tree-optimization/80705
9580 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
9581 bases are not vectorizable.
9583 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9585 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
9586 when counting register pressure.
9588 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9590 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
9591 (iv_ca_more_deps): Renamed to ...
9592 (iv_ca_compare_deps): ... this.
9593 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
9595 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9597 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
9599 (determine_group_iv_costs): ... here.
9600 (find_inv_vars_cb): Record inv var if it's not recorded before.
9602 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9604 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
9605 (get_shiftadd_cost): Ditto.
9607 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9609 * tree-ssa-address.c: Include header file.
9610 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
9612 (add_to_parts): Refactor.
9613 (addr_to_parts): New parameter. Update use of move_hint_to_base.
9614 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
9617 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9619 PR tree-optimization/53090
9620 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
9622 (extract_cond_operands): Detect condition with IV on both sides
9623 and return COMP_IV_EXPR_2.
9624 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
9625 (rewrite_use_compare): Simplify by removing call to function
9626 extract_cond_operands.
9628 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9630 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
9631 (extract_cond_operands): Detect condition comparing against non-
9632 invariant bound and return appropriate enum value.
9633 (find_interesting_uses_cond): Update use of extract_cond_operands.
9634 Handle its return value accordingly.
9635 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
9637 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9639 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
9640 nonlinear iv_use computation in loop invariant sensitive way.
9642 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9644 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
9645 (find_iv_candidates): Call relate_compare_use_with_all_cands.
9647 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9649 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
9650 (dump_cand): Support iv_cand.inv_exprs.
9651 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
9653 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
9656 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9658 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
9660 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
9661 as local function. Include necessary header files.
9662 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
9664 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9666 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
9668 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9670 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
9671 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
9672 RSHIFT_EXPR and BIT_NOT_EXPR.
9674 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9676 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
9677 (adjust_setup_cost): New parameter supporting round up adjustment.
9678 (struct address_cost_data): Delete.
9679 (force_expr_to_var_cost): Don't bound cost with spill_cost.
9680 (split_address_cost, ptr_difference_cost): Delete.
9681 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
9682 (struct ainc_cost_data): New struct.
9683 (get_address_cost_ainc): New function.
9684 (get_address_cost, get_computation_cost): Reimplement.
9685 (determine_group_iv_cost_address): Record inv_expr for all uses of
9687 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
9688 (iv_ca_has_deps): Reimplemented to ...
9689 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
9691 (iv_ca_extend): Call iv_ca_more_deps.
9693 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9695 * tree-ssa-address.c (struct mem_address): Move to header file.
9696 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
9697 * tree-ssa-address.h (struct mem_address): Move from C file.
9698 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
9700 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9702 * tree-affine.h (aff_combination_type): New interface.
9703 (aff_combination_zero_p): Remove static.
9704 (aff_combination_const_p): New interface.
9705 (aff_combination_singleton_var_p): New interfaces.
9707 2017-05-11 Richard Biener <rguenther@suse.de>
9709 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9710 Skip unreachable blocks and destinations.
9711 (eliminate): Move stmt removal and fixup ...
9712 (fini_eliminate): ... here. Skip inserted exprs.
9713 (pass_pre::execute): Move fini_pre after fini_eliminate.
9714 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
9715 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
9716 PRE to get rid of dead code that has invalid SSA form and
9717 split critical edges again.
9719 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9721 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9723 2017-05-11 Richard Biener <rguenther@suse.de>
9725 * passes.c (execute_function_todo): Verify loops if they are
9726 said to be up-to-date.
9727 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
9728 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
9730 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
9733 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
9734 handle calling assemble_external ourself.
9737 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
9738 modes with zero size. Enhance comment.
9740 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9742 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
9743 built-ins for vec_xl and vec_xst with short and char pointer
9746 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9748 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
9749 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
9750 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
9751 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
9752 (_mm_maskz_min_round_ss): New intrinsics.
9753 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
9754 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
9755 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
9756 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
9757 (__builtin_ia32_minss_mask_round): New builtins.
9758 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9759 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9760 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
9762 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
9763 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
9765 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
9768 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9770 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
9771 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
9772 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
9773 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
9774 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
9775 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
9776 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
9777 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9778 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
9779 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
9780 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
9781 (__builtin_ia32_mulss_mask_round): New builtins.
9782 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9783 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9784 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
9786 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
9787 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
9789 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
9792 2017-05-10 Julia Koval <julia.koval@intel.com>
9794 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
9795 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
9796 (_mm256_setr_m128i): New intrinsics.
9798 2017-05-10 Julia Koval <julia.koval@intel.com>
9800 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
9801 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
9802 (_mm_maskz_rcp14_ss): New intrinsics.
9803 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
9804 (__builtin_ia32_rcp14ss_mask): New builtins.
9805 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
9807 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
9809 PR tree-optimization/51513
9810 * tree-cfg.c (gimple_seq_unreachable_p): New function.
9811 (assert_unreachable_fallthru_edge_p): Use it.
9812 (group_case_labels_stmt): Likewise.
9813 * tree-cfg.h: Prototype it.
9814 * stmt.c: Include cfghooks.h and tree-cfg.h.
9815 (emit_case_dispatch_table) <gap_label>: New local variable.
9816 Use it to fill dispatch table gaps.
9817 Test for default_label before updating probabilities.
9818 (expand_case) <default_label>: Remove unneeded initialization.
9819 Test for unreachable default case statement and remove its edge.
9820 Set default_label accordingly.
9821 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
9823 2017-05-10 Carl Love <cel@us.ibm.com>
9825 * config/rs6000/rs6000-c: Add support for built-in functions
9826 vector signed char vec_neg (vector signed char)
9827 vector signed short int vec_neg (vector short int)
9828 vector signed int vec_neg (vector signed int)
9829 vector signed long long vec_neg (vector signed long long)
9830 vector float vec_neg (vector float)
9831 vector double vec_neg (vector double)
9832 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
9834 * config/rs6000/altivec.h: Add define for vec_neg
9835 * doc/extend.texi: Update the built-in documentation for the
9836 new built-in functions.
9838 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9840 PR tree-optimization/77644
9841 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
9843 2017-05-10 Nathan Sidwell <nathan@acm.org>
9845 * dumpfile.h (TDI_lang_all): New.
9846 (TDF_KIND): New. Renumber others
9847 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
9849 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
9851 (get_dump_file_name): Adjust suffix generation.
9852 (dump_enable_all): Use TDF_KIND.
9853 * doc/invoke.texi (-fdump-lang-all): Document.
9855 * dumpfile.h: Tabify.
9857 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
9860 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
9861 Move member access before delete.
9863 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
9865 * tree-inline.c (expand_call_inline): Split block at stmt
9868 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9871 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
9872 are now unused after splitting mov{sf,sd}_hardfloat.
9873 (f32_lr2): Likewise.
9875 (f32_lm2): Likewise.
9877 (f32_li2): Likewise.
9880 (f32_sr2): Likewise.
9882 (f32_sm2): Likewise.
9884 (f32_si2): Likewise.
9887 (f32_vsx): Likewise.
9889 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
9890 For movsf, order stores so the VSX stores occur before the GPR
9891 store which encourages the register allocator to use a traditional
9892 FPR instead of a GPR. For movsd, order the stores so that the GPR
9893 store comes before the VSX stores to allow the power6 to work.
9894 This is due to the power6 not having a 32-bit integer store
9895 instruction from a FPR.
9896 (movsf_hardfloat): Likewise.
9897 (movsd_hardfloat): Likewise.
9899 2017-05-09 Martin Sebor <msebor@redhat.com>
9901 PR translation/80280
9902 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
9905 PR translation/80280
9906 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
9907 data member added in r247778.
9908 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
9910 2017-05-09 Nathan Sidwell <nathan@acm.org>
9912 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
9914 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
9917 2017-05-09 Marek Polacek <polacek@redhat.com>
9919 * doc/invoke.texi: Fix typo.
9921 2017-05-09 Richard Biener <rguenther@suse.de>
9923 * tree-vrp.c (vrp_val_is_max): Adjust comment.
9924 (vrp_val_is_min): Likewise.
9925 (set_value_range_to_value): Likewise.
9926 (set_value_range_to_nonnegative): Likewise.
9927 (gimple_assign_nonzero_p): Likewise.
9928 (gimple_stmt_nonzero_p): Likewise.
9929 (vrp_int_const_binop): Likewise. Remove unreachable case.
9930 (adjust_range_with_scev): Adjust comments.
9931 (compare_range_with_value): Likewise.
9932 (extract_range_from_phi_node): Likewise.
9933 (test_for_singularity): Likewise.
9935 2017-05-09 Richard Biener <rguenther@suse.de>
9937 * tree-vrp.c (get_single_symbol): Add assert that we don't
9938 get overflowed constants as invariant part.
9939 (compare_values_warnv): Add comment before the TREE_NO_WARNING
9940 checks. Use wi::cmp instead of recursing for integer constants.
9941 (compare_values): Just ignore whether we assumed undefined
9942 overflow instead of failing the compare.
9943 (extract_range_for_var_from_comparison_expr): Add comment before the
9944 TREE_NO_WARNING sets.
9945 (test_for_singularity): Likewise.
9946 (extract_range_from_comparison): Do not disable optimization
9947 when we assumed undefined overflow.
9948 (extract_range_basic): Remove init of unused var.
9950 2017-05-09 Richard Biener <rguenther@suse.de>
9952 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
9953 (extract_range_from_multiplicative_op_1): Adjust.
9954 (extract_range_from_binary_expr_1): Use int_const_binop.
9956 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
9959 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
9960 rs6000_store_data_bypass_p in seven define_bypass directives and
9961 in several comments.
9962 * config/rs6000/rs6000-protos.h: Add prototype for
9963 rs6000_store_data_bypass_p function.
9964 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
9965 function implements slightly different (rs6000-specific) semantics
9966 than store_data_bypass_p, returning false rather than aborting
9967 with assertion error when arguments do not satisfy the
9968 requirements of store data bypass.
9969 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
9970 rs6000_store_data_bypass_p.
9972 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
9974 * config/xtensa/xtensa-protos.h
9975 (xtensa_initial_elimination_offset): New declaration.
9976 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
9977 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
9978 macro definition, add case for FRAME_POINTER_REGNUM when
9979 FRAME_GROWS_DOWNWARD.
9980 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
9981 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
9982 xtensa_initial_elimination_offset.
9984 2017-05-08 Nathan Sidwell <nathan@acm.org>
9986 * doc/invoke.texi: Alphabetize -fdump options.
9988 2017-05-08 Martin Sebor <msebor@redhat.com>
9990 PR translation/80280
9991 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
9993 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
9995 * target.def (compute_frame_layout): New optional target hook.
9996 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
9997 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
9998 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
10000 * reload1.c (verify_initial_elim_offsets): Likewise.
10001 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
10002 (use_simple_return_p): Call arm_compute_frame_layout if needed.
10003 (arm_get_frame_offsets): Split up into this ...
10004 (arm_compute_frame_layout): ... and this function.
10006 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
10008 * config/aarch64/constraints.md (Usa): New constraint.
10009 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
10011 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
10013 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
10014 with_multilib_list after it has been checked.
10016 2017-05-08 Richard Biener <rguenther@suse.de>
10018 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
10019 (bitmap_set_subtract_values): Likewise.
10021 2017-05-08 Richard Biener <rguenther@suse.de>
10023 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
10024 (gimple_assign_nonzero): ... this and remove strict_overflow_p
10026 (gimple_stmt_nonzero_warnv_p): Rename to ...
10027 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
10029 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
10030 (extract_range_basic): Adjust, do not disable propagation on
10031 strict overflow sensitive simplification.
10032 (vrp_visit_cond_stmt): Likewise.
10034 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
10036 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
10037 body size unconditionally.
10039 2017-05-07 Jeff Law <law@redhat.com>
10042 2017-05-06 Jeff Law <law@redhat.com>
10043 PR tree-optimization/78496
10044 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
10047 PR tree-optimization/78496
10048 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
10049 (simplify_stmt_using_ranges): Call it.
10050 (vrp_dom_walker::before_dom_children): Extract equivalences
10051 from an ASSERT_EXPR with an equality comparison against a
10054 2017-05-06 Jeff Law <law@redhat.com>
10056 PR tree-optimization/78496
10057 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
10060 PR tree-optimization/78496
10061 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
10062 (simplify_stmt_using_ranges): Call it.
10063 (vrp_dom_walker::before_dom_children): Extract equivalences
10064 from an ASSERT_EXPR with an equality comparison against a
10067 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
10069 * lra-constraints.c (lra_copy_reg_equiv): New function.
10070 (split_reg): Use it to copy equivalence information from the
10071 original register to the spill register.
10073 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
10075 PR rtl-optimization/75964
10076 * simplify-rtx.c (simplify_const_relational_operation): Remove
10077 invalid handling of comparisons of integer ABS.
10079 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
10081 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
10082 initialize to zero.
10083 (init_regs): Remove declaration.
10084 (function_arg_advance_32): Initialize error_p as boolean variable.
10086 2017-05-05 Nathan Sidwell <nathan@acm.org>
10088 * store-motion.c (remove_reachable_equiv_notes): Reformat long
10089 lines. Use for (;;).
10091 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10093 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
10094 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
10095 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
10096 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
10097 VF=2 that require versioning.
10099 2017-05-05 David Malcolm <dmalcolm@redhat.com>
10101 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
10104 2017-05-05 David Malcolm <dmalcolm@redhat.com>
10106 * diagnostic.h (diagnostic_override_option_index): Convert from
10107 macro to inline function.
10109 2017-05-05 David Malcolm <dmalcolm@redhat.com>
10111 * diagnostic.c (last_module_changed_p): New function.
10112 (set_last_module): New function.
10113 (diagnostic_report_current_module): Convert macro usage to
10114 the above functions.
10115 * diagnostic.h (diagnostic_context::last_module): Strengthen
10116 from const line_map * to const line_map_ordinary *.
10117 (diagnostic_last_module_changed): Delete macro.
10118 (diagnostic_set_last_module): Delete macro.
10120 2017-05-05 David Malcolm <dmalcolm@redhat.com>
10122 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
10123 with diagnostic_report_diagnostic.
10124 (diagnostic_n_impl_richloc): Likewise.
10125 * diagnostic.h (report_diagnostic): Delete macro.
10126 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
10127 with diagnostic_report_diagnostic.
10128 * substring-locations.c (format_warning_va): Likewise.
10130 2017-05-05 David Malcolm <dmalcolm@redhat.com>
10132 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
10133 save/restor of format_spec. Move option-printing code to...
10134 (print_option_information): ...this new function, and
10135 reimplement by simply printing to the pretty_printer,
10136 rather than appending to the format string.
10138 2017-05-05 David Malcolm <dmalcolm@redhat.com>
10140 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
10141 handling logic into...
10142 (update_effective_level_from_pragmas): ...this new function.
10144 2017-05-04 Andrew Waterman <andrew@sifive.com>
10146 * config/riscv/riscv.opt (mstrict-align): New option.
10147 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
10148 (SLOW_UNALIGNED_ACCESS): Define.
10149 (riscv_slow_unaligned_access): Declare.
10150 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
10152 (riscv_slow_unaligned_access): New variable.
10153 (rocket_tune_info): Set slow_unaligned_access to true.
10154 (optimize_size_tune_info): Set slow_unaligned_access to false.
10155 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
10156 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
10157 (riscv_option_override): Set riscv_slow_unaligned_access.
10158 * doc/invoke.texi: Add -mstrict-align to RISC-V.
10160 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
10162 * config/riscv/riscv.md: Unify indentation.
10164 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
10169 * config/rs6000/rs6000.md (u code attribute): Add FIX and
10171 (extendsi<mode>2): Add support for doing sign extension via
10172 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
10173 don't have ISA 3.0 instructions.
10174 (extendsi<mode>2 splitter): Likewise.
10175 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
10176 generate the normal insns since SImode can now go in vector
10177 registers. Disallow the special UNSPECs needed for previous
10178 machines to hide SImode being used. Add new insns
10179 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
10180 (fix_trunc<mode>si2_stfiwx): Likewise.
10181 (fix_trunc<mode>si2_internal): Likewise.
10182 (fixuns_trunc<mode>si2): Likewise.
10183 (fixuns_trunc<mode>si2_stfiwx): Likewise.
10184 (fctiw<u>z_<mode>_smallint): Likewise.
10185 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
10186 of floating point to 32-bit integer from doing a direct move to
10187 the GPR registers to do a store.
10188 (fctiwz_<mode>): Break long line.
10190 2017-05-05 Bin Cheng <bin.cheng@arm.com>
10192 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
10193 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
10194 (addr_list, addr_offset_valid_p): New.
10195 (split_address_groups): Check offset validity with above function.
10196 (gt-tree-ssa-loop-ivopts.h): Include header file.
10198 2017-05-05 Nathan Sidwell <nathan@acm.org>
10200 * config.gcc (arm*-*-*): Add missing 'fi'.
10202 2017-05-05 Steve Ellcey <sellcey@cavium.com>
10204 * doc/invoke.texi (-fopt-info): Explicitly say order of options
10205 included in -fopt-info does not matter.
10206 * doc/optinfo.texi (-fopt-info): Fix description of default
10207 behavour. Explicitly say order of options included in -fopt-info
10210 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
10212 * config.gcc: Allow combinations of aprofile and rmprofile values for
10213 --with-multilib-list.
10214 * config/arm/t-multilib: New file.
10215 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
10216 variables. Remove setting of ISA and floating-point ABI in
10217 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
10218 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
10219 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
10220 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
10222 * config/arm/t-rmprofile: Likewise except for the matches changes.
10223 * doc/install.texi (--with-multilib-list): Document the combination of
10224 aprofile and rmprofile values and warn about pitfalls in doing that.
10226 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
10228 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
10229 (movdi_aarch64): Likewise.
10231 2017-05-05 Jakub Jelinek <jakub@redhat.com>
10233 PR tree-optimization/80632
10234 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
10236 (build_arrays): Initialize it for virtual phis.
10237 (fix_phi_nodes): Use it for virtual phis.
10239 PR tree-optimization/80558
10240 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
10241 [x, y] op z into [x op, y op z] for op & or | if conditions
10244 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
10245 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10248 * config/arm/arm.md (use_literal_pool): Remove.
10249 (64-bit immediate split): No longer takes cost into consideration
10250 if arm_disable_literal_pool is enabled.
10251 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
10252 used when arm_disable_literal_pool is enabled.
10253 (arm_max_const_double_inline_cost): Remove use of
10254 arm_disable_literal_pool.
10255 (push_minipool_fix): Add assert.
10256 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
10257 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
10258 (no_literal_pool_sf_immediate): New.
10260 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10262 PR tree-optimization/80613
10263 * tree-ssa-dce.c (propagate_necessity): Remove cases for
10264 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
10266 2017-05-05 Richard Biener <rguenther@suse.de>
10268 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
10270 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
10272 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
10273 of this flag from insn conditions due to removal from r247495.
10275 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
10277 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
10279 (arm_early_store_addr_dep_ptr): Likewise.
10280 * config/arm/aarch-common-protos.h
10281 (arm_early_load_addr_dep_ptr): Add prototype.
10282 (arm_early_store_addr_dep_ptr): Likewise.
10283 * config/arm/cortex-a53.md: Add new bypasses.
10285 2017-05-05 Jakub Jelinek <jakub@redhat.com>
10287 * tree.c (next_type_uid): Change type to unsigned.
10288 (type_hash_canon): Decrement back next_type_uid if
10289 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
10290 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
10293 2017-05-04 Martin Sebor <msebor@redhat.com>
10295 * builtins.c: Fix a trivial typo in a comment.
10297 PR middle-end/79234
10298 * builtins.c (check_sizes): Adjust to handle reading past the end.
10299 Avoid printing excessive upper bound of ranges. Use %E to print
10300 tree nodes instead of converting them to %wu.
10301 (expand_builtin_memchr): New function.
10302 (compute_dest_size): Rename...
10303 (compute_objsize): ...to this.
10304 (expand_builtin_memcpy): Adjust.
10305 (expand_builtin_mempcpy): Adjust.
10306 (expand_builtin_strcat): Adjust.
10307 (expand_builtin_strcpy): Adjust.
10308 (check_strncat_sizes): Adjust.
10309 (expand_builtin_strncat): Adjust.
10310 (expand_builtin_strncpy): Adjust and simplify.
10311 (expand_builtin_memset): Adjust.
10312 (expand_builtin_bzero): Adjust.
10313 (expand_builtin_memcmp): Adjust.
10314 (expand_builtin): Handle memcmp.
10315 (maybe_emit_chk_warning): Check strncat just once.
10317 2017-05-04 Martin Sebor <msebor@redhat.com>
10319 PR preprocessor/79214
10320 PR middle-end/79222
10321 PR middle-end/79223
10322 * builtins.c (check_sizes): Add inlining context and issue
10323 warnings even when -Wno-system-headers is set.
10324 (check_strncat_sizes): Same.
10325 (expand_builtin_strncat): Same.
10326 (expand_builtin_memmove): New function.
10327 (expand_builtin_stpncpy): Same.
10328 (expand_builtin): Handle memmove and stpncpy.
10330 2017-05-04 Bin Cheng <bin.cheng@arm.com>
10332 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
10333 which is not used any more.
10335 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
10337 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
10339 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
10341 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
10342 (cortexa53_tunings): Likewise.
10343 (cortexa57_tunings): Likewise.
10344 (cortexa72_tunings): Likewise.
10345 (cortexa73_tunings): Likewise.
10347 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
10349 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
10350 Set loop alignment to 8.
10352 2017-05-04 Martin Sebor <msebor@redhat.com>
10354 PR translation/80280
10355 * builtins.c (expand_builtin_object_size): Add missing quoting to
10356 %D and like directives.
10357 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
10358 (hsa_type_for_tree_type): Same.
10359 (verify_function_arguments): Same.
10360 * symtab.c (symbol_table::change_decl_assembler_name): Same.
10361 * varasm.c (get_section): Same.
10364 2017-05-04 Martin Sebor <msebor@redhat.com>
10366 PR translation/80280
10367 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
10369 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
10371 * config/aarch64/aarch64.c (generic_addrcost_table):
10372 Change HI/TI mode setting.
10374 2017-05-04 Martin Jambor <mjambor@suse.cz>
10376 PR tree-optimization/80622
10377 * tree-sra.c (comes_initialized_p): New function.
10378 (build_accesses_from_assign): Only set write lazily when
10379 comes_initialized_p is false.
10380 (analyze_access_subtree): Use comes_initialized_p.
10381 (propagate_subaccesses_across_link): Assert !comes_initialized_p
10382 instead of testing for PARM_DECL.
10384 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10386 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
10387 constraint on operand 0 to allow more general addressing modes.
10388 Adjust output template.
10389 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
10391 * config/aarch64/aarch64-protos.h
10392 (aarch64_address_valid_for_prefetch_p): Declare prototype.
10393 * config/aarch64/constraints.md (Dp): New address constraint.
10394 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
10397 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
10399 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
10400 update use of estimate_ipcp_clone_size_and_time.
10401 (estimate_local_effects): Update use of
10402 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
10403 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
10404 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
10405 Return nonspecialized time.
10407 2017-05-04 Richard Biener <rguenther@suse.de>
10409 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
10410 for the last VUSE which def dominates the PHI. Directly call
10412 (get_continuation_for_phi_1): Remove.
10414 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
10416 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
10417 to explain the use of truncating division. Cap the number of
10418 iterations to the maximum given by nb_iterations_upper_bound,
10421 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10423 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
10424 * configure: Regenerate.
10425 * config.in: Regenerate.
10426 * config/i386/driver-mingw32.c: new file.
10427 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
10428 * config.host: Link driver-mingw32.o on MinGW host.
10429 * doc/install.texi: Document new --enable-mingw-wildcard configure
10432 2017-05-04 Marek Polacek <polacek@redhat.com>
10434 PR tree-optimization/80612
10435 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
10437 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10438 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
10440 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
10441 (movt splitter): Likewise.
10442 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
10443 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
10444 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
10445 block for Thumb-1 with MOVT.
10446 (thumb2_legitimate_address_p): Move code block ...
10447 (can_avoid_literal_pool_for_label_p): ... into this new function.
10448 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
10450 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
10451 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
10452 "M-profile targets with the MOVT instruction".
10454 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10456 * config/arm/arm-builtins.c (arm_init_builtins): Rename
10457 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
10458 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
10460 2017-05-04 Martin Liska <mliska@suse.cz>
10462 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
10463 variable cond_code.
10465 2017-05-04 Richard Biener <rguenther@suse.de>
10467 * tree.c (array_at_struct_end_p): Handle arrays at struct
10468 end with flexarrays more conservatively. Refactor and treat
10469 arrays of arrays or aggregates more strict. Fix
10470 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
10471 * tree.c (array_at_struct_end_p): Adjust prototype.
10472 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
10473 * gimple-fold.c (get_range_strlen): Likewise.
10474 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
10476 2017-05-04 Richard Biener <rguenther@suse.de>
10478 PR tree-optimization/31130
10479 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
10481 (supports_overflow_infinity): Likewise.
10482 (is_negative_overflow_infinity): Likewise.
10483 (is_positive_overflow_infinity): Likewise.
10484 (is_overflow_infinity): Likewise.
10485 (stmt_overflow_infinity): Likewise.
10486 (overflow_infinity_range_p): Likewise.
10487 (usable_range_p): Remove as always returning true.
10488 (make_overflow_infinity): Remove.
10489 (negative_overflow_infinity): Likewise.
10490 (positive_overflow_infinity): Likewise.
10491 (avoid_overflow_infinity): Likewise.
10492 (set_value_range): Adjust accordingly.
10493 (set_value_range_to_nonnegative): Likewise, remove now unused
10494 overflow_infinity arg.
10495 (vrp_operand_equal_p): Adjust.
10496 (update_value_range): Likewise.
10497 (range_int_cst_singleton_p): Likewise.
10498 (operand_less_p): Likewise.
10499 (compare_values_warnv): Likewise.
10500 (extract_range_for_var_from_comparison_expr): Likewise.
10501 (vrp_int_const_binop): Likewise.
10502 (zero_nonzero_bits_from_vr): Likewise.
10503 (extract_range_from_multiplicative_op_1): Likewise.
10504 (extract_range_from_binary_expr_1): Likewise.
10505 (extract_range_from_unary_expr): Likewise.
10506 (extract_range_from_comparison): Likewise.
10507 (extract_range_basic): Likewise.
10508 (adjust_range_with_scev): Likewise.
10509 (compare_ranges): Likewise.
10510 (compare_range_with_value): Likewise.
10511 (dump_value_range): Likewise.
10512 (test_for_singularity): Likewise, remove strict_overflow_p parameter
10514 (simplify_cond_using_ranges): Adjust.
10516 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
10518 * brig-builtins.def: Added a builtin for class_f64.
10519 * builtin-types.def: Added a builtin type needed by class_f64.
10521 2017-05-03 Jason Merrill <jason@redhat.com>
10523 * timevar.def: Add TV_CONSTEXPR.
10525 2017-05-03 David Malcolm <dmalcolm@redhat.com>
10527 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
10529 2017-05-03 Martin Jambor <mjambor@suse.cz>
10531 * ipa-prop.c (ipa_update_after_lto_read): Removed.
10532 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
10533 * ipa-cp.c (ipcp_propagate_stage): Do not call
10534 ipa_update_after_lto_read.
10535 * ipa-inline.c (ipa_inline): Likewise.
10537 2017-05-03 Martin Jambor <mjambor@suse.cz>
10539 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
10540 tag. Added a default constructor and a destructor.
10541 (ipa_edge_args_sum_t): New class;
10542 (ipa_edge_args_sum): Declare.
10543 (ipa_edge_args_vector): Remove declaration.
10544 (IPA_EDGE_REF): Use ipa_edge_args_sum.
10545 (ipa_free_edge_args_substructures): Remove declaration.
10546 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
10547 (ipa_edge_args_info_available_for_edge_p): Likewise.
10548 * ipa-prop.c (ipa_edge_args_vector): Removed.
10549 (edge_removal_hook_holder): Likewise.
10550 (edge_duplication_hook_holder): Likewise.
10551 (ipa_edge_args_sum): New variable.
10552 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
10553 ipa_edge_args_vector.
10554 (ipa_free_edge_args_substructures): Likewise.
10555 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
10556 ipa_edge_args_vector.
10557 (ipa_edge_removal_hook): Turned into method
10558 ipa_edge_args_sum_t::remove.
10559 (ipa_edge_duplication_hook): Turned into method
10560 ipa_edge_args_sum_t::duplicate.
10561 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
10562 registering edge hooks.
10563 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
10564 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
10565 ipa_edge_args_sum instead of ipa_edge_args_vector.
10566 * ipa-profile.c (ipa_profile): Likewise.
10568 2017-05-03 Martin Jambor <mjambor@suse.cz>
10570 * symbol-summary.h (function_summary): New method exists.
10571 (function_summary::symtab_removal): Deallocate through release.
10572 (call_summary): New class.
10573 (gt_ggc_mx): New overload.
10574 (gt_pch_nx): Likewise.
10575 (gt_pch_nx): Likewise.
10577 2017-05-03 Jeff Law <law@redhat.com>
10579 PR tree-optimization/78496
10580 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
10581 from simplify_cond_using_ranges. Split off code to walk
10582 backwards through casts into ...
10583 (simplify_cond_using_ranges_2): New function.
10584 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
10585 (execute_vrp): After identifying jump threads, call
10586 simplify_cond_using_ranges_2.
10588 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
10591 * ipa-inline.h (inline_summary): Add ctor.
10592 (create_ggc): Do not use ggc_cleared_alloc.
10594 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
10595 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10597 * gcc.c (handle_braces): Support escaping in switch matching
10599 * doc/invoke.texi (Spec Files): Document it.
10600 Remove superfluous @code markup in items.
10602 2017-05-03 David Malcolm <dmalcolm@redhat.com>
10604 * diagnostic-show-locus.c (struct column_range): New struct.
10605 (get_affected_columns): New function.
10606 (get_printed_columns): New function.
10607 (struct correction): New struct.
10608 (correction::ensure_capacity): New function.
10609 (correction::ensure_terminated): New function.
10610 (struct line_corrections): New struct.
10611 (line_corrections::~line_corrections): New dtor.
10612 (line_corrections::add_hint): New function.
10613 (layout::print_trailing_fixits): Reimplement in terms of the new
10615 (selftest::test_overlapped_fixit_printing): New function.
10616 (selftest::diagnostic_show_locus_c_tests): Call it.
10618 2017-05-03 Nathan Sidwell <nathan@acm.org>
10620 Canonicalize canonical type hashing
10621 * tree.h (type_hash_canon_hash): Declare.
10622 * tree.c (type_hash_list, attribute_hash_list): Move into
10623 type_hash_canon_hash.
10624 (build_type_attribute_qual_variant): Break out hash code calc into
10625 type_hash_canon_hash.
10626 (type_hash_canon_hash): New. Generic type hash computation.
10627 (build_range_type_1, build_array_type_1, build_function_type,
10628 build_method_type_directly, build_offset_type, build_complex_type,
10629 make_vector_type): Call it.
10631 2017-05-03 Richard Biener <rguenther@suse.de>
10633 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
10634 When all DRs have unknown misaligned do not always peel
10635 when there is a store but apply the same costing model as if
10636 there were only loads.
10638 2017-05-03 Richard Biener <rguenther@suse.de>
10641 PR tree-optimization/80492
10642 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
10643 compare_base_decls returning dont-know properly.
10645 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
10647 * config/arm/iterators.md (CCSI): New mode iterator.
10648 (arch): New mode attribute.
10649 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
10650 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
10651 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
10652 code iterator for success result mode.
10653 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
10654 the corresponding new insn generators.
10656 2017-05-03 Bin Cheng <bin.cheng@arm.com>
10659 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10660 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10662 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
10664 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
10665 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
10666 (DDR_A): Wrap DDR argument in brackets.
10667 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
10668 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
10669 (DDR_REVERSED_P): Likewise.
10671 2017-05-03 Jakub Jelinek <jakub@redhat.com>
10673 PR tree-optimization/79472
10674 * tree-switch-conversion.c (struct switch_conv_info): Add
10675 contiguous_range and default_case_nonstandard fields.
10676 (collect_switch_conv_info): Compute contiguous_range and
10677 default_case_nonstandard fields, don't clear final_bb if
10678 contiguous_range and only the default case doesn't have the required
10680 (check_all_empty_except_final): Set default_case_nonstandard instead
10681 of failing if contiguous_range and the default case doesn't have empty
10683 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
10684 and only the default case doesn't have the required constants. Skip
10686 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
10687 if default_case_nonstandard.
10688 (build_constructors): Build constant 1 just once. Assert that default
10689 values aren't inserted in between cases if contiguous_range. Skip
10691 (build_arrays): Skip virtual phis.
10692 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
10693 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
10694 Handle virtual phis.
10695 (gen_inbound_check): Handle default_case_nonstandard case.
10696 (process_switch): Adjust check_final_bb caller. Call
10697 gather_default_values with the first non-default case instead of
10698 default case if default_case_nonstandard.
10700 2017-05-02 Nathan Sidwell <nathan@acm.org>
10702 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
10703 check. Fix formatting.
10705 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
10707 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
10708 errors when comparing specialized and unspecialized times.
10710 2017-05-02 David Malcolm <dmalcolm@redhat.com>
10712 * diagnostic-show-locus.c
10713 (layout::should_print_annotation_line_p): Make private.
10714 (layout::print_annotation_line): Make private.
10715 (layout::annotation_line_showed_range_p): Make private.
10716 (layout::show_ruler): Make private.
10717 (layout::print_source_line): Make private. Pass in line and
10718 line_width, rather than calling location_get_source_line. Drop
10720 (layout::print_leading_fixits): New method.
10721 (layout::print_any_fixits): Rename to...
10722 (layout::print_trailing_fixits): ...this, and make private.
10723 Don't print newline fixits.
10724 (diagnostic_show_locus): Move logic for printing one row into...
10725 (layout::print_line): ...this new function. Move the
10726 location_get_source_line call and error-handling from
10727 print_source_line to here. Call print_leading_fixits, and rename
10728 print_any_fixits to print_trailing_fixits.
10729 (selftest::test_fixit_insert_containing_newline): Update now that
10730 newlines are partially supported.
10731 (selftest::test_fixit_insert_containing_newline_2): New test.
10732 (selftest::test_fixit_replace_containing_newline): Update comments.
10733 (selftest::diagnostic_show_locus_c_tests): Call the new test.
10734 * edit-context.c (class added_line): New class.
10735 (class edited_line): Describe newline handling in comment.
10736 (edited_line::actually_edited_p): New method.
10737 (edited_line::print_content): Delete redundant decl.
10738 (edited_line::m_predecessors): New field.
10739 (edited_file::print_content): Call edited_line::print_content.
10740 (edited_file::print_diff): Update to support newlines.
10741 (edited_file::print_diff_hunk): Likewise.
10742 (edited_file::print_run_of_changed_lines): New function.
10743 (edited_file::print_diff_line): Convert to...
10744 (print_diff_line): ...this.
10745 (edited_file::get_effective_line_count): New function.
10746 (edited_line::edited_line): Initialize new field m_predecessors.
10747 (edited_line::~edited_line): Clean up m_predecessors.
10748 (edited_line::apply_fixit): Handle newlines.
10749 (edited_line::get_effective_line_count): New function.
10750 (edited_line::print_content): New function.
10751 (edited_line::print_diff_lines): New function.
10752 (selftest::test_applying_fixits_insert_containing_newline): New
10754 (selftest::test_applying_fixits_replace_containing_newline): New
10756 (selftest::insert_line): New function.
10757 (selftest::test_applying_fixits_multiple_lines): Add example of
10759 (selftest::edit_context_c_tests): Call the new tests.
10761 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10763 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
10764 parameter cand. Update dump information.
10765 (get_computation_cost): Update uses.
10767 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10769 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
10770 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
10771 (get_computation_at, rewrite_use_address): Update use of
10772 get_computation_aff.
10774 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10776 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
10777 (get_computation): Delete.
10778 (get_computation_cost): Implement like get_computation_cost_at.
10779 Use get_computation_at.
10780 (get_computation_cost_at): Delete.
10781 (rewrite_use_nonlinear_expr): Use get_computation_at.
10782 (rewrite_use_compare, remove_unused_ivs): Ditto.
10784 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10786 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
10788 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10790 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
10791 (ivopts_global_cost_for_size): Rename parameter and update uses.
10792 (iv_ca_recount_cost): Update uses.
10793 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
10794 candidates seperately in n_invs and n_cands.
10795 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
10797 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10799 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
10800 (find_inv_vars_cb): New.
10801 (find_depends): Renamed to ...
10802 (find_inv_vars): ... this.
10803 (add_candidate_1, force_var_cost): Call find_inv_vars.
10804 (split_address_cost, determine_group_iv_cost_cond): Ditto.
10806 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10808 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
10809 inv_vars. Add inv_exprs.
10810 (struct iv_cand): Rename depends_on to inv_vars.
10811 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
10812 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
10813 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
10814 (dump_cand): Dump inv_vars.
10815 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
10816 (record_invariant, find_depends, add_candidate_1): Ditto.
10817 (set_group_iv_cost, force_var_cost): Ditto.
10818 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
10819 (get_computation_cost_at, get_computation_cost): Ditto.
10820 (determine_group_iv_cost_generic): Ditto.
10821 (determine_group_iv_cost_address): Ditto.
10822 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
10823 (determine_group_iv_costs): Ditto.
10824 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
10825 (iv_ca_set_remove_invariants): Renamed to ...
10826 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
10827 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
10828 (iv_ca_set_add_invariants): Renamed to ...
10829 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
10830 (iv_ca_set_cp): Use iv_ca_set_add_invs.
10831 (iv_ca_has_deps): Support inv_vars and inv_exprs.
10832 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
10833 (create_new_ivs): Remove useless dump.
10835 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10837 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
10839 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
10840 (iv_ca_set_no_cp, create_new_iv): Ditto.
10842 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10844 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10846 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10848 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
10849 function tree_check2.
10851 2017-05-02 Martin Liska <mliska@suse.cz>
10853 * doc/gcov.texi: Add missing preposition.
10854 * gcov.c (function_info::function_info): Properly fill up
10855 all member variables.
10857 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10859 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
10861 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10863 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
10865 2017-05-02 Martin Liska <mliska@suse.cz>
10868 * lto-streamer-in.c (lto_read_tree_1): Remove
10869 LTO_STREAMER_DEBUG.
10870 * lto-streamer.c (struct tree_hash_entry): Likewise.
10871 (struct tree_entry_hasher): Likewise.
10872 (tree_entry_hasher::hash): Likewise.
10873 (tree_entry_hasher::equal): Likewise.
10874 (lto_streamer_init): Likewise.
10875 (lto_orig_address_map): Likewise.
10876 (lto_orig_address_get): Likewise.
10877 (lto_orig_address_remove): Likewise.
10878 * lto-streamer.h: Likewise.
10879 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
10880 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
10882 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
10884 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
10885 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
10886 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
10887 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
10888 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
10889 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
10890 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
10891 (mm_maskz_sub_ss): New intrinsics.
10892 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10893 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
10894 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
10895 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
10896 (__builtin_ia32_subss_mask_round): New builtins.
10897 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10898 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
10899 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
10901 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
10902 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
10904 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
10907 2017-05-02 Martin Jambor <mjambor@suse.cz>
10909 PR tree-optimization/78687
10910 * tree-sra.c (access): New field parent.
10911 (process_subtree_disqualification): New function.
10912 (disqualify_candidate): Call it.
10913 (build_accesses_from_assign): Reset write flag if creating an
10915 (build_access_subtree): Fill in parent field and also prpagate
10916 down grp_write flag.
10917 (create_artificial_child_access): New parameter set_grp_write, set
10918 grp_write to its value.
10919 (propagate_subaccesses_across_link): Also propagate grp_write flag
10921 (propagate_all_subaccesses): Push the closest parent back to work
10922 queue if add_access_to_work_queue returned true.
10924 2017-05-02 Richard Biener <rguenther@suse.de>
10926 * common.opt (fstrict-overflow): Alias negative to fwrapv.
10927 * doc/invoke.texi (fstrict-overflow): Remove all traces of
10928 -fstrict-overflow documentation.
10929 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
10930 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
10931 flag_strict_overflow.
10932 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
10933 * lto-opts.c (lto_write_options): Do not stream it.
10934 * lto-wrapper.c (merge_and_complain): Do not handle it.
10935 * opts.c (default_options_table): Do not set -fstrict-overflow.
10936 (finish_options): Likewise do not clear it when sanitizing.
10937 * simplify-rtx.c (simplify_const_relational_operation): Do not
10938 test flag_strict_overflow.
10940 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10942 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
10943 using enabled attribute.
10944 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
10945 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
10946 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
10947 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
10948 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
10949 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
10950 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
10951 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
10952 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
10953 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
10955 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10957 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
10959 2017-05-02 Richard Biener <rguenther@suse.de>
10961 PR tree-optimization/80591
10963 2017-04-10 Richard Biener <rguenther@suse.de>
10965 * tree-ssa-structalias.c (find_func_aliases): Properly handle
10968 2017-05-02 Richard Biener <rguenther@suse.de>
10970 PR tree-optimization/80549
10971 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
10972 (cleanup_tree_cfg_noloop): Create forwarders to known loop
10973 headers if they do not have a preheader.
10975 2017-05-02 Martin Liska <mliska@suse.cz>
10978 * common.opt: Fix typo.
10979 * doc/invoke.texi: Likewise.
10981 2017-05-01 Jan Beulich <jbeulich@suse.com>
10983 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
10984 swapping, add (x,x,m,x,n) alternative.
10986 2017-05-01 Nathan Sidwell <nathan@acm.org>
10988 * calls.c (combine_pending_stack_adjustment_and_call): Remove
10989 unnecessary unadjusted_alignment check.
10991 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10994 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
10996 * gimplify.c (gimplify_cilk_detach): New function.
10997 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
10998 * tree-core.h: Document EXPR_CILK_SPAWN.
10999 * tree.h (EXPR_CILK_SPAWN): Define.
11001 2017-05-01 David Malcolm <dmalcolm@redhat.com>
11003 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
11004 to use new fixit_hint representation, using the "replace" logic.
11005 (get_line_span_for_fixit_hint): Likewise.
11006 (layout::print_any_fixits): Likewise.
11007 (selftest::test_one_liner_many_fixits): Rename to...
11008 (selftest::test_one_liner_many_fixits_1): ...this, and update
11009 comment and expected output to reflect that the multiple fix-it
11010 hints are now consolidated into one insertion.
11011 (selftest::test_one_liner_many_fixits_2): New test.
11012 (selftest::test_diagnostic_show_locus_one_liner): Update for
11014 (selftest::test_fixit_consolidation): Update for fix-it API
11016 * diagnostic.c (print_parseable_fixits): Likewise.
11017 * edit-context.c (edited_line::m_line_events): Convert from
11018 auto_vec <line_event *> to auto_vec <line_event>.
11019 (class line_event): Convert from abstract base class to a concrete
11020 class, taking over the role of replace_event.
11021 (class insert_event): Delete.
11022 (class replace_event): Rename to class line_event. Convert to
11024 (edit_context::add_fixits): Reimplement.
11025 (edit_context::apply_insert): Delete.
11026 (edit_context::apply_replace): Rename to...
11027 (edit_context::apply_fixit): ...this. Convert to half-open range.
11028 (edited_file::apply_insert): Delete.
11029 (edited_file::apply_replace): Rename to...
11030 (edited_file::apply_fixit): ...this.
11031 (edited_line::~edited_line): Drop deletion of events.
11032 (edited_line::apply_insert): Delete.
11033 (edited_line::apply_replace): Rename to...
11034 (edited_line::apply_fixit): ...this. Convert to half-open range.
11035 Update for change to type of m_line_events.
11036 * edit-context.h (edit_context::apply_insert): Delete.
11037 (edit_context::apply_replace): Rename to...
11038 (edit_context::apply_fixit): ...this.
11040 2017-05-01 Martin Sebor <msebor@redhat.com>
11042 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
11045 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
11048 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
11049 __get_cpuid_max returns 0.
11050 (__get_cpuid_count): Ditto.
11052 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
11054 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
11055 replacement expression is another instance of one of its arguments.
11057 2017-05-01 Jakub Jelinek <jakub@redhat.com>
11060 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
11061 check for stack push/pop autoinc.
11062 * config/i386/i386.c (ix86_agi_dependent): Return false
11063 if the only reason why modified_in_p returned true is that
11064 addr is SP based and set_insn is a push or pop.
11066 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
11068 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
11071 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
11074 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
11075 (account_size_time): Use two predicates - exec_pred and
11077 (evaluate_conditions_for_known_args): Compute both clause and
11079 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
11080 (inline_summary_t::duplicate): Update.
11081 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
11083 (compute_inline_parameters): Likewise.
11084 (estimate_edge_size_and_time): Update caluclation of time.
11085 (estimate_node_size_and_time): Compute both time and nonspecialized
11087 (estimate_ipcp_clone_size_and_time): Update.
11088 (inline_merge_summary): Update.
11089 (do_estimate_edge_time): Update.
11090 (do_estimate_edge_size): Update.
11091 (do_estimate_edge_hints): Update.
11092 (inline_read_section, inline_write_summary): Stream both new predicates.
11093 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
11095 (compute_inlined_call_time): Cleanup.
11096 (big_speedup_p): Update.
11097 (edge_badness): Update.
11098 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
11099 (size_time_entry): Replace predicate by exec_predicate and
11100 nonconst_predicate.
11101 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
11102 (estimate_edge_time): Return also nonspec_time.
11103 (reset_edge_growth_cache): Update.
11105 2017-04-29 Jakub Jelinek <jakub@redhat.com>
11107 PR rtl-optimization/80491
11108 * ifcvt.c (noce_process_if_block): When looking for x setter
11109 with missing else_bb, don't check only the insn right before
11110 cond_earliest, but look for the last insn that x is modified in
11111 within the same bb.
11113 PR rtl-optimization/80491
11114 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
11116 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
11118 PR tree-optimization/80487
11119 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
11121 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11123 PR tree-optimization/79697
11124 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
11125 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
11126 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
11128 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
11129 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
11131 2017-04-28 Martin Sebor <msebor@redhat.com>
11133 PR tree-optimization/80523
11134 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
11135 (init_target_to_host_charmap, target_to_host, target_strtol10): New
11137 (maybe_warn, format_directive, parse_directive): Use new functions.
11138 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
11140 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
11142 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
11144 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
11146 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
11147 target_header_dir): Set correctly.
11148 * configure: Regenerated.
11149 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
11150 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
11151 instead of SYSTEM_HEADER_DIR.
11153 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
11155 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
11156 (estimate_local_effects): Likewise.
11157 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
11158 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
11159 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
11160 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
11161 do_estimate_edge_time, estimate_edge_time): Likewise.
11162 * ipa-inline-analysis.c (estimate_node_size_and_time,
11163 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
11164 (estimate_time_after_inlining): Remove.
11166 2017-04-28 Martin Liska <mliska@suse.cz>
11168 * doc/gcov.texi: Enhance documentation of gcov.
11170 2017-04-28 Martin Liska <mliska@suse.cz>
11172 * doc/gcov.texi: Sort options in alphabetic order.
11173 * doc/gcov-dump.texi: Likewise.
11174 * doc/gcov-tool.texi: Likewise.
11175 * gcov.c (print_usage): Likewise.
11176 * gcov-dump.c (print_usage): Likewise.
11177 * gcov-tool.c (print_merge_usage_message): Likewise.
11178 (print_rewrite_usage_message): Likewise.
11179 (print_overlap_usage_message): Likewise.
11181 2017-04-28 Martin Liska <mliska@suse.cz>
11183 PR gcov-profile/53915
11184 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
11186 2017-04-28 Martin Liska <mliska@suse.cz>
11188 PR gcov-profile/79891
11189 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
11190 is marked by compiler as living on a line.
11191 (get_cycles_count): Remove usage of the union.
11192 (output_intermediate_file): Likewise.
11193 (find_source): Fix GNU coding style.
11194 (accumulate_line_counts): Remove old non-all block mode.
11195 (output_lines): Remove usage of the union.
11196 * profile.c (output_location): Include all BBs, even if
11197 belonging to a same line (and file) as a previous BB.
11199 2017-04-28 Martin Liska <mliska@suse.cz>
11201 * gcov.c (process_args): Handle new argument 'w'.
11202 (read_graph_file): Assign ID to BBs.
11203 (output_branch_count): Display BB # if verbose flag is set.
11204 (output_lines): Likewise for arcs.
11205 (print_usage): Add '--verbose' option help.
11206 * doc/gcov.texi: Document --verbose (-w) option.
11208 2017-04-28 Martin Liska <mliska@suse.cz>
11210 * gcov.c (struct block_location_info): New struct.
11211 (process_file): Fill up the new structure.
11212 (read_graph_file): Replace usage of encoding by the newly added
11214 (add_line_counts): Likewise.
11215 (accumulate_line_counts): Remove usage of the union.
11216 (function_info::function_info): New function.
11217 (function_info::~function_info): Likewise.
11218 (process_file): Call delete instead of release_function.
11219 (release_function): Release the function.
11220 (release_structures): Call delete instead of release_function.
11221 (solve_flow_graph): Replace usage of num_blocks.
11222 (find_exception_blocks): Likewise.
11223 (output_lines): Fix GNU coding style.
11225 2017-04-28 Martin Liska <mliska@suse.cz>
11228 * coverage.c (coverage_remove_note_file): New function.
11229 * coverage.h: Declare the function.
11230 * toplev.c (finalize): Clean if an error has been seen.
11232 2017-04-28 Martin Liska <mliska@suse.cz>
11234 PR gcov-profile/80031
11235 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
11236 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
11237 * gcov.c (read_graph_file): Read just number of blocks.
11238 * profile.c (branch_prob): Do not stream 0 flags per a basic
11241 2017-04-28 Martin Liska <mliska@suse.cz>
11243 * gcov-dump.c (tag_*): Add new argument to declarations.
11244 (dump_gcov_file): Likewise.
11245 (tag_blocks): Add and use new argument depth.
11246 (tag_arcs): Likewise.
11247 (tag_lines): Likewise.
11248 (tag_counters): Likewise.
11249 (tag_summary): Likewise.
11250 (dump_working_sets): Use depth to do a proper indentation.
11252 2017-04-28 Jakub Jelinek <jakub@redhat.com>
11255 * cgraph.h (symtab_node::debug_symtab): No longer inline.
11256 * symtab.c (symtab_node::debug_symtab): Move definition here.
11258 2017-04-28 Richard Biener <rguenther@suse.de>
11260 * lto-streamer.h (LTO_major_version): Bump to 7.
11262 2017-04-28 Richard Biener <rguenther@suse.de>
11264 * tree-vrp.c (assert_info): New struct.
11265 (add_assert_info): New helper.
11266 (register_edge_assert_for_2): Refactor to add asserts to a vector
11268 (register_edge_assert_for_1): Likewise.
11269 (register_edge_assert_for): Likewise.
11270 (finish_register_edge_assert_for): New helper actually registering
11271 asserts where live on edge.
11272 (find_conditional_asserts): Adjust.
11273 (find_switch_asserts): Likewise.
11274 (evrp_dom_walker::try_find_new_range): Generalize.
11275 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
11277 2017-04-27 Marek Polacek <polacek@redhat.com>
11280 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
11281 arg10 and arg11 to itype.
11283 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
11285 * doc/extend.texi (Object Size Checking): Improve grammar.
11287 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
11290 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
11291 that the logic for permitting reciprocal estimates matches that
11294 2017-04-27 Jakub Jelinek <jakub@redhat.com>
11297 * tree.c (type_cache_hasher::equal): Only compare
11298 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
11299 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
11300 non-aggregate element types.
11301 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
11302 about the flag on ARRAY_TYPEs in the comment, formatting fix.
11304 2017-04-27 Richard Biener <rguenther@suse.de>
11306 PR middle-end/80533
11307 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
11308 stripping ARRAY_REFs from MEM_EXPR make sure we're not
11309 keeping a reference to a trailing array.
11311 2017-04-27 Richard Biener <rguenther@suse.de>
11313 PR middle-end/80539
11314 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
11315 being in loop-closed SSA form conservatively.
11316 (chrec_fold_multiply_poly_poly): Likewise.
11318 2017-04-27 Tamar Christina <tamar.christina@arm.com>
11320 PR middle-end/79665
11321 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
11322 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
11324 2017-04-27 Jakub Jelinek <jakub@redhat.com>
11327 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
11328 (aarch64_function_arg_alignment): Return unsigned int again, but still
11329 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
11330 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
11331 Don't emit -Wpsabi note.
11332 (aarch64_function_arg_boundary): Likewise.
11333 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
11336 2017-04-26 Nathan Sidwell <nathan@acm.org>
11338 * tree.h (crc32_unsigned_n): Declare.
11339 (crc32_unsigned, crc32_unsigned): Make inline.
11340 * tree.c (crc32_unsigned_bits): Replace with ...
11341 (crc32_unsigned_n): ... this.
11342 (crc32_unsigned, crc32_byte): Remove.
11343 (crc32_string): Remove unnecessary braces.
11345 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
11347 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
11348 * ipa-inline-analysis.c (MAX_TIME): Remove.
11349 (account_size_time): Use sreal for time.
11350 (dump_inline_summary): Update.
11351 (estimate_function_body_sizes): Update.
11352 (estimate_edge_size_and_time): Update.
11353 (estimate_calls_size_and_time): Update.
11354 (estimate_node_size_and_time): Update.
11355 (inline_merge_summary): Update.
11356 (inline_update_overall_summary): Update.
11357 (estimate_time_after_inlining): Update.
11358 (inline_read_section): Update.
11359 (inline_write_summary): Update.
11360 * ipa-inline.c (compute_uninlined_call_time): Update.
11361 (compute_inlined_call_time): Update.
11362 (recursive_inlining): Update.
11363 (inline_small_functions): Update.
11364 (dump_overall_stats): Update.
11365 * ipa-inline.h: Include sreal.h.
11366 (size_time_entry): Turn time to sreal.
11367 (inline_summary): Turn self_time nad time to sreal.
11369 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
11371 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
11373 (sreal::stream_out, sreal::stream_in): New.
11374 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
11376 2017-04-25 Jakub Jelinek <jakub@redhat.com>
11378 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
11381 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
11384 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
11386 (dimode_scalar_chain::compute_convert_gain): Ditto.
11387 (dimode_scalar_chain::make_vector_copies): Ditto.
11388 (dimode_scalar_chain::convert_reg): Ditto.
11389 (dimode_scalar_chain::convert_insn): Ditto.
11390 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
11391 (VI248_AVX512BW_1): New mode iterator.
11392 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
11393 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
11396 2017-04-25 Martin Sebor <msebor@redhat.com>
11398 PR tree-optimization/80497
11399 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
11400 constants are representable in HOST_WIDE_INT.
11401 (parse_directive): Ditto.
11403 2017-04-25 Martin Sebor <msebor@redhat.com>
11406 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
11407 (new_zero_array): Adjust signature.
11408 (dom_info::dom_init): Used unsigned rather that size_t.
11409 (dom_info::dom_info): Same.
11411 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11412 Jakub Jelinek <jakub@redhat.com>
11415 * config/arm/arm.c: Include gimple.h.
11416 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
11417 returns negative, increment ncrn only if it returned positive.
11418 (arm_needs_doubleword_align): Return int instead of bool,
11419 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
11420 members, but if there is any such non-FIELD_DECL
11421 > PARM_BOUNDARY aligned decl, return -1 instead of false.
11422 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
11423 returns negative, increment nregs only if it returned positive.
11424 (arm_setup_incoming_varargs): Likewise.
11425 (arm_function_arg_boundary): Emit -Wpsabi note if
11426 arm_needs_doubleword_align returns negative, return
11427 DOUBLEWORD_ALIGNMENT only if it returned positive.
11429 2017-04-25 Marek Polacek <polacek@redhat.com>
11432 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11433 first argument to type.
11435 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
11438 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
11439 type checks to test for compatibility instead of equality.
11441 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11442 Jakub Jelinek <jakub@redhat.com>
11445 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
11447 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
11448 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
11449 the alignment computation, but return their maximum in warn_alignment.
11450 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
11451 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
11453 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
11454 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
11457 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11459 * config/arc/simdext.md (dmpyh): Fix typo.
11461 2017-04-25 Richard Biener <rguenther@suse.de>
11463 PR tree-optimization/80492
11464 * alias.c (compare_base_decls): Handle registers with asm
11465 specification conservatively.
11466 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
11467 compare_base_decls returning dont-know properly.
11469 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11471 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
11472 (legitimate_offset_address_p): New function.
11473 (arc_legitimate_address_p): Use above function.
11475 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11477 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
11479 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11481 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
11482 ACCH registers whenever they are available.
11484 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11486 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
11487 double regs fix when not used.
11489 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11491 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
11493 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
11494 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
11496 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11498 * config/arc/arc.c (arc_output_addsi): Check for h-register class
11499 when emitting short ADD instructions.
11501 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11503 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
11505 (cmpsi_cc_c_insn): Likewise.
11506 (cbranchsi4_scratch): Compute proper instruction length using
11507 compact_hreg_operand.
11508 * config/arc/predicates.md (compact_hreg_operand): New predicate.
11510 2017-04-25 Richard Biener <rguenther@suse.de>
11512 PR middle-end/80509
11513 * passes.c (pass_manager::pass_manager): Initialize
11514 m_name_to_pass_map.
11516 2017-04-25 Richard Biener <rguenther@suse.de>
11518 PR tree-optimization/79201
11519 * tree-ssa-sink.c (statement_sink_location): Handle calls.
11521 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11524 * config/s390/vector.md: Split MEM->GPR vector moves for
11525 non-s_operand addresses.
11527 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11530 * config/s390/predicates.md (reload_const_wide_int_operand): New
11532 * config/s390/s390.md ("movti"): Remove d/P alternative.
11533 ("movti_bigconst"): New pattern definition.
11535 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11538 * s390-protos.h (s390_expand_cs_hqi): Removed.
11539 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
11540 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
11541 modes as well as CCZ1mode and CCZmode.
11542 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
11543 signature of s390_emit_compare_and_swap.
11544 (s390_expand_cs_hqi): Likewise, make static.
11545 (s390_expand_cs_tdsi): Generate an explicit compare before trying
11546 compare-and-swap, in some cases.
11547 (s390_expand_cs): Wrapper function.
11548 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
11550 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
11551 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
11552 patterns for small and large integers. Forbid symref memory operands.
11553 Move expander to s390.c. Require cc register.
11554 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
11555 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
11556 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
11557 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
11558 symref memory operands. Remove CC mode and call s390_match_ccmode
11560 ("atomic_exchange<mode>"): Allow and implement all integer modes.
11562 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11564 * config/s390/s390.md (define_peephole2): New peephole to help
11565 combining the load-and-test pattern with volatile memory.
11567 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11569 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
11570 with CCZmode for TARGET_Z196.
11572 2017-04-25 Jakub Jelinek <jakub@redhat.com>
11574 PR rtl-optimization/80501
11575 * combine.c (make_compound_operation_int): Set subreg_code to SET
11576 even for AND with mask of the sign bit of mode.
11578 PR rtl-optimization/80500
11579 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
11580 sum's initial value.
11582 2017-04-25 Julian Brown <julian@codesourcery.com>
11583 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11585 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
11587 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
11589 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
11591 2017-04-25 Julian Brown <julian@codesourcery.com>
11592 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11594 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
11595 (thunderx2t99_sha): New Reservation.
11597 2017-04-25 Julian Brown <julian@codesourcery.com>
11598 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11600 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
11601 type for 1-element load.
11603 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
11605 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
11607 2017-04-24 Martin Jambor <mjambor@suse.cz>
11609 PR tree-optimization/80293
11610 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
11611 char arrays not totally scalarizable if it is false.
11612 (analyze_all_variable_accesses): Pass correct value in the new
11613 parameter. Add a statistics counter.
11615 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
11617 PR middle-end/79931
11618 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
11620 2017-04-24 Richard Biener <rguenther@suse.de>
11622 PR tree-optimization/80494
11623 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
11624 out for complex types.
11626 2017-04-24 Richard Biener <rguenther@suse.de>
11628 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
11629 * tree-ssa-sccvn.c (print_scc): Print SCC size.
11630 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
11631 (DFS): Adjust and never fail.
11632 (sccvn_dom_walker::fail): Remove.
11633 (sccvn_dom_walker::before_dom_children): Adjust.
11634 (run_scc_vn): Likewise and never fail.
11635 * tree-ssa-pre.c (pass_pre::execute): Adjust.
11636 (pass_fre::execute): Likewise.
11638 2017-04-24 Richard Biener <rguenther@suse.de>
11640 PR tree-optimization/79725
11641 * tree-ssa-sink.c (statement_sink_location): Return whether
11642 failure reason was zero uses. Move that check later.
11643 (sink_code_in_bb): Deal with zero uses by removing the stmt
11646 2017-04-24 Richard Biener <rguenther@suse.de>
11649 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
11650 pointer-based references.
11652 2017-04-24 Richard Biener <rguenther@suse.de>
11655 * pass_manager.h (pass_manager::operator new): Remove.
11656 (pass_manager::operator delete): Likewise.
11657 * passes.c (pass_manager::operator new): Remove.
11658 (pass_manager::operator delete): Likewise.
11659 (pass_manager::pass_manager): Zero individual pass members.
11661 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
11664 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
11665 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
11666 Check "XEXP (src, 1)" operand here.
11667 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
11668 Check "XEXP (src, 1)" operand here.
11669 (dimode_scalar_chain::make_vector_copies): Detect count register
11670 of a shift instruction. Zero extend count register from QImode
11671 to DImode to satisfy vector shift pattern count operand predicate.
11672 Substitute vector shift count operand with a DImode copy.
11673 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
11676 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
11678 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
11679 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
11680 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
11681 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
11682 (UNSPEC_NOREX_MEM): Remove definition.
11684 2017-04-21 Richard Biener <rguenther@suse.de>
11686 PR tree-optimization/79547
11687 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
11688 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
11689 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
11690 without any constraints.
11692 2017-04-21 Richard Biener <rguenther@suse.de>
11694 PR tree-optimization/78847
11695 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
11697 2017-04-21 Richard Biener <rguenther@suse.de>
11699 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
11700 (build_distinct_type_copy): Likewise.
11701 (build_variant_type_copy): Likewise.
11702 * tree.c (build_qualified_type): Pass down mem-stat info.
11703 (build_distinct_type_copy): Likewise.
11704 (build_variant_type_copy): Likewise.
11706 2017-04-21 Richard Biener <rguenther@suse.de>
11708 PR tree-optimization/80237
11709 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
11711 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
11712 for a simplified result.
11714 2016-04-21 Richard Biener <rguenther@suse.de>
11716 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
11717 sth as strict as a simple_iv but a chrec without symbols and an
11718 operand defined in the loop we are peeling (and not some subloop).
11719 (propagate_constants_for_unrolling): Propagate all constants.
11721 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11724 * config/i386/i386.c (print_reg): Remove assert for disalowed
11725 regno values, call output_operand_lossage instead.
11727 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11730 * config/i386/constraints.md (Yc): New register constraint.
11731 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
11732 Use Yc constraint for alternative 2 of operand 0. Remove
11733 preferred_for_speed attribute.
11735 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11737 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
11738 lastprivate clauses in SIMT case.
11740 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
11742 * doc/invoke.texi (-Wextra-semi): Document new warning option.
11744 2017-04-20 Richard Biener <rguenther@suse.de>
11746 PR tree-optimization/57796
11747 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
11748 as N scalar stores.
11749 (vect_model_load_cost): Cost gathers as N scalar loads.
11751 2017-04-20 Richard Biener <rguenther@suse.de>
11753 * ggc-page.c (ggc_allocated_p): Rename to ...
11754 (safe_lookup_page_table_entry): ... this and return the lookup
11756 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
11758 2017-04-20 Richard Biener <rguenther@suse.de>
11760 PR tree-optimization/80453
11761 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
11762 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
11763 from the conditions.
11764 (vn_phi_eq): Pass them down.
11765 (vn_phi_lookup): Record them.
11766 (vn_phi_insert): Likewise.
11768 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
11770 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
11771 uninitialized variable warning to avoid buffer overrun.
11773 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11776 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
11777 is suppressed for '{ 0 }' in C.
11779 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11781 * BASE-VER: Set to 8.0.0.
11783 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
11785 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
11786 priority .init_array and .fini_array section with SECTION_NOTYPE
11789 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11791 PR middle-end/80423
11792 * tree.h (build_array_type): Add typeless_storage default argument.
11793 * tree.c (type_cache_hasher::equal): Also compare
11794 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
11795 (build_array_type): Add typeless_storage argument, set
11796 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
11798 (build_nonshared_array_type): Adjust build_array_type_1 caller.
11799 (build_array_type): Likewise. Add typeless_storage argument.
11801 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11802 Jakub Jelinek <jakub@redhat.com>
11804 PR tree-optimization/80426
11805 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
11806 operation on symbolic operands, also compute the overflow for the
11807 invariant part when the operation degenerates into a negation.
11809 2017-04-19 Jakub Jelinek <jakub@redhat.com>
11812 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
11813 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
11816 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
11818 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
11821 * config/avr/avr.c (tree.h): Include it.
11822 (cgraph.h): Include it.
11823 (avr_encode_section_info): Don't warn for uninitialized progmem
11824 variable if it's just an alias.
11826 2017-04-19 Richard Biener <rguenther@suse.de>
11829 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
11830 when needed by AutoPGO.
11832 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
11835 * doc/lto.texi: Remove an extra 'that'.
11837 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
11839 PR rtl-optimization/80429
11840 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
11841 are only used in debug insns.
11843 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11844 Vladimir Makarov <vmakarov@redhat.com>
11846 * config/sparc/predicates.md (input_operand): Add comment. Return
11847 true for any memory operand when LRA is in progress.
11848 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
11850 2017-04-18 Jeff Law <law@redhat.com>
11853 * mips.md ({return,simple_return}_internal): Do not overwrite
11856 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11858 PR tree-optimization/80443
11859 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
11860 instead of adding 1, subtract -1 and similarly instead of subtracting
11863 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
11865 PR rtl-optimization/80357
11866 * haifa-sched.c (tmp_bitmap): New variable.
11867 (model_recompute): Handle duplicate use records.
11868 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
11869 (free_global_sched_pressure_data): Free it.
11871 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
11874 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
11875 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
11876 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
11877 instead of SYSTEM_HEADER_DIR.
11879 2017-04-18 Jeff Law <law@redhat.com>
11881 PR middle-end/80422
11882 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
11883 predecessors after walking up the insn chain.
11885 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11888 * dwarf2out.c (modified_type_die): Try harder not to emit internal
11889 sizetype type into debug info.
11891 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
11894 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
11895 unneeded test for TARGET_UPPER_REGS_SF.
11896 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
11898 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11901 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
11902 instead of gsi_after_labels.
11904 2017-04-18 Jeff Law <law@redhat.com>
11906 * regcprop.c (maybe_mode_change): Avoid creating copies of the
11910 2017-04-13 Jeff Law <law@redhat.com>
11911 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11912 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11914 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
11917 * config/avr/avr.c (intl.h): Include it.
11918 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
11920 2017-04-18 Martin Liska <mliska@suse.cz>
11922 PR gcov-profile/78783
11923 * gcov-tool.c (gcov_output_files): Validate that destination
11924 file is either removed by the tool or by a user.
11926 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11927 Guy Benyei <guybe@mellanox.com>
11929 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
11930 block, and do not negate it, the stored id is already negative.
11932 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11934 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
11936 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11939 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
11940 masks of options that should be turned off if the VSX vector
11941 options are turned off.
11942 (OTHER_P8_VECTOR_MASKS): Likewise.
11943 (OTHER_VSX_VECTOR_MASKS): Likewise.
11944 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
11945 rs6000_disable_incompatible_switches to validate no type switches
11947 (rs6000_incompatible_switch): New function to disallow turning on
11948 other vector options if -mno-vsx, -mno-power8-vector, or
11949 -mno-power9-vector are specified.
11951 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11953 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
11955 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11957 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
11958 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
11959 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
11960 (ARG_POINTER_CFA_OFFSET): Likewise.
11962 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11964 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
11965 conditions to take advantage of various optimizations.
11967 2017-04-13 Jeff Law <law@redhat.com>
11969 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11970 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11971 (zero_extendsidi2_dext): Likewise.
11973 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11976 * fold-const.c (fold_ternary_loc): Revert
11977 use op0 instead of fold_convert_loc (loc, type, arg0) part of
11980 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
11982 PR rtl-optimization/80343
11983 * lra-remat.c (update_scratch_ops): Assign original hard reg to
11984 new scratch pseudo.
11986 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
11989 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
11990 to ubsan_encode_value.
11992 2017-04-13 Jeff Law <law@redhat.com>
11994 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
11995 appearing in DEBUG_INSNs.
11997 2017-04-13 Martin Liska <mliska@suse.cz>
11999 PR gcov-profile/80413
12000 * gcov-io.c (gcov_write_string): Copy to buffer just when
12001 allocated size is greater than zero.
12003 2017-04-13 Jakub Jelinek <jakub@redhat.com>
12006 * dwarf2out.c (decls_for_scope): Ignore declarations of
12007 current_function_decl in BLOCK_NONLOCALIZED_VARS.
12009 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
12012 * ipa-visibility.c (non_local_p): Fix typos.
12013 (localize_node): When localizing symbol in same comdat group,
12014 dissolve the group only when we know external symbols are going
12016 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
12018 2017-04-12 Jakub Jelinek <jakub@redhat.com>
12020 PR tree-optimization/79390
12021 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
12022 order does not result in usable sequence, retry with reversed operand
12028 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
12029 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
12030 op0 instead of fold_convert_loc (loc, type, arg0).
12032 2017-04-12 Jeff Law <law@redhat.com>
12034 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
12035 has a delay slot in the generated code.
12037 * config/cris/cris.md (cris_preferred_reload_class): Return
12038 GENNONACR_REGS rather than GENERAL_REGS.
12040 2017-04-12 Jakub Jelinek <jakub@redhat.com>
12043 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
12044 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
12045 signedness of the result type.
12047 2017-04-12 Richard Biener <rguenther@suse.de>
12048 Jeff Law <law@redhat.com>
12050 PR tree-optimization/80359
12051 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
12052 trim stores to TARGET_MEM_REFs.
12054 2017-04-12 Richard Biener <rguenther@suse.de>
12056 PR tree-optimization/79390
12057 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
12058 threading case even more.
12060 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
12063 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
12064 for quad_address_p for TImode, instead of just not indexed_address.
12066 2017-04-12 Richard Biener <rguenther@suse.de>
12067 Bernd Edlinger <bernd.edlinger@hotmail.de>
12069 PR middle-end/79671
12070 * alias.c (component_uses_parent_alias_set_from): Handle
12071 TYPE_TYPELESS_STORAGE.
12072 (get_alias_set): Likewise.
12073 * tree-core.h (tree_type_common): Add typeless_storage flag.
12074 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
12075 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
12076 for types containing members with TYPE_TYPELESS_STORAGE.
12077 (place_field): Likewise.
12078 (layout_type): Likewise for ARRAY_TYPE.
12079 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
12080 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
12081 TYPE_TYPELESS_STORAGE.
12082 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
12084 2017-04-12 Jakub Jelinek <jakub@redhat.com>
12087 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
12088 first argument to type.
12090 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12094 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
12095 CONST0_RTX (mode) rather than const0_rtx where appropriate.
12096 (rs6000_expand_binop_builtin): Likewise.
12097 (rs6000_expand_ternop_builtin): Likewise; also add missing
12098 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
12099 vshasigma built-ins.
12100 * doc/extend.texi: Document that vec_xxpermdi's third argument
12101 must be a constant.
12103 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
12105 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
12106 Use shift_const cost parameter when calculating gain of STV shifts.
12108 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
12110 PR rtl-optimization/70478
12111 * lra-constraints.c (process_alt_operands): Check memory for
12112 disfavoring memory insn operand.
12114 2017-04-11 Jakub Jelinek <jakub@redhat.com>
12116 PR middle-end/80100
12117 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
12118 left shift in unsigned HOST_WIDE_INT type.
12120 PR rtl-optimization/80385
12121 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
12122 (not (neg X)) into (plus X -1) for complex or non-integral modes.
12125 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
12126 if they have any depend clauses.
12128 2017-04-11 Martin Liska <mliska@suse.cz>
12131 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
12132 * ipa-split.c (split_function): Create a local comdat symbol
12133 if caller is in a comdat group.
12135 2017-04-11 Martin Liska <mliska@suse.cz>
12138 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
12141 2017-04-11 Martin Sebor <msebor@redhat.com>
12143 PR middle-end/80364
12144 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
12145 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
12147 (directive::set_width, directive::set_precision, format_character):
12149 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
12152 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
12155 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
12156 conflict, set target->arch_name instead of target->cpu_name.
12158 2017-04-11 Richard Biener <rguenther@suse.de>
12160 PR tree-optimization/80374
12161 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
12162 build_zero_cst, remove fold_convertible_p check again.
12164 2017-04-11 Martin Liska <mliska@suse.cz>
12167 * ubsan.c (instrument_object_size): Do not instrument register
12170 2017-04-11 Jakub Jelinek <jakub@redhat.com>
12173 * config/i386/i386-builtin-types.def
12174 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
12175 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
12176 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
12177 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
12178 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
12179 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
12180 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
12181 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
12182 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
12183 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
12184 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
12185 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
12186 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
12187 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
12188 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
12189 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
12190 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
12191 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
12192 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
12193 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
12194 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
12195 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
12196 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
12197 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
12198 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
12199 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
12200 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
12201 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
12202 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
12203 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
12204 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
12205 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
12206 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
12207 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
12208 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
12209 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
12210 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
12211 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
12212 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
12213 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
12214 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
12215 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
12216 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
12217 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
12218 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
12219 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
12221 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
12222 flag to second_arg_count, handle 4 argument function type _COUNT
12223 aliases, handle second_arg_count on second argument rather than last.
12225 2017-04-10 Jeff Law <law@redhat.com>
12227 PR tree-optimization/80374
12228 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
12229 record anything if we can not convert integer_zero_node to the
12232 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
12235 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12236 Enhance special handling given to the TARGET_P9_MINMAX option in
12237 relation to certain other options.
12239 2017-04-10 Bin Cheng <bin.cheng@arm.com>
12241 PR tree-optimization/80153
12242 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
12243 remove POINTER_PLUS_EXPR's base part directly, rather than through
12246 2017-04-10 Richard Biener <rguenther@suse.de>
12247 Bin Cheng <bin.cheng@arm.com>
12249 PR tree-optimization/80153
12250 * tree-affine.c (aff_combination_to_tree): Get base pointer from
12251 the first element of pointer type aff_tree. Build result expr in
12253 (add_elt_to_tree): Convert to type unconditionally. Remove other
12254 fold_convert calls.
12255 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
12256 (rewrite_use_nonlinear_expr): Check invariant using iv information.
12258 2017-04-10 Richard Biener <rguenther@suse.de>
12260 * tree-ssa-structalias.c (find_func_aliases): Properly handle
12263 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
12265 PR rtl-optimization/70478
12266 * lra-constraints.c (curr_small_class_check): New.
12267 (update_and_check_small_class_inputs): New.
12268 (process_alt_operands): Update curr_small_class_check. Disfavor
12269 alternative insn memory operands. Check available regs for small
12272 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
12275 * config/mips/mips.opt (-mvirt): Update description.
12276 * doc/invoke.texi (-mvirt): Likewise.
12278 2017-04-10 Richard Biener <rguenther@suse.de>
12280 PR middle-end/80362
12281 * fold-const.c (fold_binary_loc): Look at unstripped ops when
12282 looking for NEGATE_EXPR in -A / -B to A / B folding.
12284 2017-04-10 Martin Liska <mliska@suse.cz>
12286 PR gcov-profile/80224
12287 * gcov.c (print_usage): Fix usage string.
12288 (get_gcov_intermediate_filename): Remove.
12289 (output_gcov_file): Use both for normal and intermediate format.
12290 (generate_results): Do not initialize special file for
12291 intermediate format.
12293 2017-04-10 Richard Biener <rguenther@suse.de>
12295 PR tree-optimization/80304
12296 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
12299 2017-04-10 Nathan Sidwell <nathan@acm.org>
12302 * config/rs6000/rs6000.c (rs6000_vector_type): New.
12303 (rs6000_init_builtins): Use it.
12305 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12307 * config/arm/arm.md (<mrc>): Add mode to SET source.
12308 (<mrrc>): Likewise.
12310 2017-04-10 Richard Biener <rguenther@suse.de>
12312 PR middle-end/80344
12313 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
12315 2017-04-10 Jakub Jelinek <jakub@redhat.com>
12318 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
12319 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
12320 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
12321 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
12322 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
12323 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
12324 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
12325 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
12326 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
12327 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
12328 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
12329 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
12330 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
12331 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
12332 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
12333 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
12334 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
12335 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
12336 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
12337 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
12338 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
12339 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
12340 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
12342 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
12344 PR rtl-optimization/70478
12345 * lra-constraints.c: Reverse the last patch.
12347 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
12349 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
12350 Add comment for WCHAR_T.
12352 2017-04-08 Martin Liska <mliska@suse.cz>
12355 2017-04-07 Martin Liska <mliska@suse.cz>
12358 * ipa-split.c (split_function): Add function part to a same comdat
12361 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12364 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
12366 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
12368 * rs6000/rs6000.c (vec_load_pendulum): Rename...
12369 (vec_pairing): ...to this.
12370 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
12371 (rs6000_sched_init): Adjust for name change.
12372 (struct rs6000_sched_context): Likewise.
12373 (rs6000_init_sched_context): Likewise.
12374 (rs6000_set_sched_context): Likewise.
12376 2017-04-07 Jakub Jelinek <jakub@redhat.com>
12382 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
12384 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
12385 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
12386 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
12388 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
12390 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
12392 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
12394 PR rtl-optimization/70703
12395 * ira-color.c (update_conflict_hard_regno_costs): Use
12396 int64_t instead of HOST_WIDE_INT.
12398 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
12400 PR rtl-optimization/70478
12401 * lra-constraints.c (process_alt_operands): Disfavor alternative
12402 insn memory operands.
12404 2017-04-07 Jeff Law <law@redhat.com>
12406 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
12407 CALL and NOTE_INSN_CALL_ARG_LOCATION.
12409 2017-04-07 Martin Liska <mliska@suse.cz>
12412 * config/aarch64/aarch64.c (aarch64_process_target_attr):
12413 Show error message instead of an ICE.
12415 2017-04-07 Martin Liska <mliska@suse.cz>
12418 * ipa-split.c (split_function): Add function part to a same comdat
12421 2017-04-07 Richard Biener <rguenther@suse.de>
12423 PR middle-end/80341
12424 * tree.c (get_unwidened): Also handle ! for_type case for
12426 * convert.c (do_narrow): Split out from ...
12427 (convert_to_integer_1): ... here. Do not pass final truncation
12428 type to get_unwidened for TRUNC_DIV_EXPR.
12430 2017-04-07 Richard Biener <rguenther@suse.de>
12432 * tree-affine.c (wide_int_ext_for_comb): Take type rather
12434 (aff_combination_const): Adjust.
12435 (aff_combination_scale): Likewise.
12436 (aff_combination_add_elt): Likewise.
12437 (aff_combination_add_cst): Likewise.
12438 (aff_combination_convert): Likewise.
12439 (add_elt_to_tree): Likewise. Remove unused argument.
12440 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
12442 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
12444 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
12446 * config/arm/arm.c (arm_default_short_enums): Use
12447 ARM_DEFAULT_SHORT_ENUMS.
12448 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
12450 2017-04-06 Jakub Jelinek <jakub@redhat.com>
12453 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
12454 members with redundant out-of-class redeclaration.
12456 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
12459 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
12460 * config/i386/i386.md (*zero_extendsidi2):
12461 Add (?*x,*x) and (?*v,*v) alternatives.
12463 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
12466 * config/i386/i386.c (ix86_expand_builtin)
12467 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
12468 mode from insn data. Convert operands to insn operand mode.
12469 Copy operands that don't satisfy insn predicate to a register.
12471 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
12473 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
12476 2017-04-06 Richard Biener <rguenther@suse.de>
12478 PR tree-optimization/80334
12479 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
12480 preserve alignment of accesses.
12482 2017-04-06 Richard Biener <rguenther@suse.de>
12484 PR tree-optimization/80262
12485 * tree-sra.c (build_ref_for_offset): Preserve address-space
12487 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
12488 Drop useless address-space information on MEM_REF offsets.
12490 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
12492 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
12494 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
12496 PR rtl-optimization/70703
12497 * ira-color.c (update_conflict_hard_regno_costs): Use
12498 HOST_WIDE_INT instead of long.
12500 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
12503 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
12504 not defined for x86_64 target. Add -mmmx target option when __SSE2__
12506 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
12507 for x86_64 target. Handle -m3dnowa option.
12509 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
12511 PR rtl-optimization/70703
12512 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
12513 (update_conflict_hard_regno_costs): Use long instead of unsigned
12514 arithmetic for cost calculation.
12516 2017-04-05 Jakub Jelinek <jakub@redhat.com>
12517 Bernd Edlinger <bernd.edlinger@hotmail.de>
12520 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
12523 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
12526 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
12527 ptr_mode with Pmode throughout.
12528 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
12529 into probe_stack_range and use DImode.
12531 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
12534 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
12535 call_eh_return is true.
12537 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12539 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
12540 Initialize last_match_fntype_index.
12542 2017-04-05 Jakub Jelinek <jakub@redhat.com>
12545 * tree-nvr.c: Include internal-fn.h.
12546 (pass_return_slot::execute): Ignore internal calls without
12549 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12550 Richard Biener <rguenther@suse.de>
12553 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
12554 captures used multiple times, except for the last use.
12555 * generic-match-head.c: Include gimplify.h.
12557 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12559 PR tree-optimization/79390
12560 * target.h (struct noce_if_info): Declare.
12561 * targhooks.h (default_noce_conversion_profitable_p): Declare.
12562 * target.def (noce_conversion_profitable_p): New target hook.
12563 * ifcvt.h (struct noce_if_info): New type, moved from ...
12564 * ifcvt.c (struct noce_if_info): ... here.
12565 (noce_conversion_profitable_p): Renamed to ...
12566 (default_noce_conversion_profitable_p): ... this. No longer
12568 (noce_try_store_flag_constants, noce_try_addcc,
12569 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
12570 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
12571 instead of noce_conversion_profitable_p.
12572 * config/i386/i386.c: Include ifcvt.h.
12573 (ix86_option_override_internal): Don't override
12574 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
12575 (ix86_noce_conversion_profitable_p): New function.
12576 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
12577 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
12578 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
12579 * doc/tm.texi: Regenerated.
12581 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12583 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
12586 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12589 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
12590 instructions for small multiply cores.
12592 2017-04-04 Jeff Law <law@redhat.com>
12594 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
12596 (mips_expand_vec_perm_const): Initialize elements in orig_perm
12597 that are not set by the loop over the elements.
12599 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12602 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
12603 int mode, convert_modes it to mode as unsigned, otherwise use
12604 lowpart_subreg to mode rather than SImode.
12605 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
12606 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
12607 Use DImode instead of SImode for the shift count operand.
12608 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
12611 2017-04-04 Richard Biener <rguenther@suse.de>
12613 PR middle-end/80281
12614 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
12615 arithmetic done for the negate or the plus. Simplify.
12616 (A - (-B) -> A + B): Likewise.
12617 * fold-const.c (split_tree): Make sure to not negate pointers.
12619 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
12621 PR rtl-optimization/60818
12622 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
12623 a compare of comparisons with the thing compared if this results
12624 in a different machine mode.
12626 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
12628 * alias.c (base_alias_check): Fix typo in comment.
12629 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
12630 * cgraphunit.c (symbol_table::compile): Likewise.
12631 * collect2.c (maybe_run_lto_and_relink): Likewise.
12632 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
12633 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
12634 * config/avr/avr.c (avr_map_op_t): Likewise.
12635 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
12636 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
12637 * config/epiphany/epiphany.md (movcc): Likewise.
12638 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
12639 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
12641 * config/mips/mips.c (mips_save_restore_reg): Likewise.
12642 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
12643 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
12644 * config/sh/sh.c (sh_rtx_costs): Likewise.
12645 * fold-const.c (fold_truth_andor): Likewise.
12646 * genautomata.c (collapse_flag): Likewise.
12647 * gengtype.h (struct type::u::s): Likewise.
12648 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
12649 * input.c (FORMAT_AMOUNT): Likewise.
12650 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
12651 (known_aggs_to_agg_replacement_list): Likewise.
12652 * ipa-inline-analysis.c: Likewise.
12653 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
12654 * ipa-polymorphic-call.c
12655 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
12656 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
12657 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
12659 * modulo-sched.c (apply_reg_moves): Likewise.
12660 * omp-expand.c (build_omp_regions_1): Likewise.
12661 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
12662 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
12663 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
12664 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
12665 * value-prof.c: Likewise.
12666 * var-tracking.c (val_reset): Likewise.
12668 2017-04-03 Richard Biener <rguenther@suse.de>
12670 PR tree-optimization/80275
12671 * fold-const.c (split_address_to_core_and_offset): Handle
12674 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
12676 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
12677 descriptors is at least equal to that of functions.
12679 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
12681 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
12683 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
12686 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
12687 (mov<IMOD4:mode>): New expander.
12688 (*mov<IMOD4:mode>_internal): New insn and split pattern.
12690 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
12692 PR rtl-optimization/79405
12693 * fwprop.c (propagations_left): New variable.
12694 (forward_propagate_into): Decrement it.
12695 (fwprop_init): Initialize it.
12696 (fw_prop): If the variable has reached zero, stop propagating.
12697 (fwprop_addr): Ditto.
12699 2017-03-31 Jakub Jelinek <jakub@redhat.com>
12702 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
12703 a FUNCTION_DECL, pass it as decl instead of origin to
12706 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
12708 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
12711 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
12714 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
12715 TARGET_VSX_SMALL_INTEGER.
12717 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12719 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
12720 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
12722 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
12724 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
12725 extraction from odd-numbered MSA register.
12727 2017-03-31 Jakub Jelinek <jakub@redhat.com>
12729 PR middle-end/80173
12730 * expmed.c (store_bit_field_1): Don't attempt to create
12731 a word subreg out of hard registers wider than word if they
12732 have HARD_REGNO_NREGS of 1 for their mode.
12734 PR middle-end/80163
12735 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
12736 conversions to integer types wider than word and pointer.
12739 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
12740 (rtx_equal_for_cselib_p): Pass 0 to it.
12741 * cselib.c (cselib_hasher::equal): Likewise.
12742 (rtx_equal_for_cselib_1): Add depth argument. If depth
12743 is 128, don't look up VALUE locs and punt. Increment
12744 depth in recursive calls when walking VALUE locs.
12746 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
12748 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
12749 (make_gcov_file_name): Use the canonical path name for generating
12751 (read_line): Fix handling of files with ascii null bytes.
12753 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
12755 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
12756 to initialise a vector register instead
12757 of using a const_int.
12759 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12761 PR translation/80189
12762 * gimplify.c (omp_default_clause): Use %qs instead of %s in
12763 diagnostic messages.
12765 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
12768 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
12769 (dfp_diex_<mode>): Update mode of operand 1.
12770 * doc/extend.texi (dxex, dxexq): Document change to return type.
12771 (diex, diexq): Document change to argument type.
12773 2017-03-30 Martin Jambor <mjambor@suse.cz>
12776 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
12777 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
12778 it reflects the signature changes performed at the callee side.
12779 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
12780 to cgraph_build_function_type_skip_args.
12781 (build_function_decl_skip_args): Adjust call to the above function.
12783 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12786 * config/i386/sse.md
12787 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
12788 register as dest whenever it is a MEM not rtx_equal_p to the
12789 corresponding dup operand, and when forcing into reg move the
12790 reg into the memory afterwards.
12791 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
12792 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
12793 for the force_reg mode.
12794 (avx512vl_vextractf128<mode>): Use register as dest either
12795 always when a MEM, or when it is a MEM not rtx_equal_p to the
12796 corresponding dup operand, or even not when it is a CONST_VECTOR
12797 depending on the mode and lo vs. hi.
12798 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
12800 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
12801 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
12802 Likewise. Require that operands[2] is even.
12803 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
12804 Remove extraneous parens. Require that operands[2] is a multiple
12806 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
12807 operands[0] is a MEM if <mask_applied>, the predicates/constraints
12808 disallow memory then.
12810 2017-03-30 Richard Biener <rguenther@suse.de>
12812 PR tree-optimization/77498
12813 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
12814 to non-constants over backedges.
12816 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
12818 PR rtl-optimization/80233
12819 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
12820 as last_combined_insn. Do not test for BARRIER_P separately.
12822 2017-03-29 Andreas Schwab <schwab@suse.de>
12825 * calls.c (prepare_call_address): Convert funexp to Pmode before
12826 copying to temp reg.
12828 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12830 PR tree-optimization/80158
12831 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
12832 Handle possible future case of more than one alternate
12834 (replace_rhs_if_not_dup): Likewise.
12835 (replace_one_candidate): Likewise.
12837 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
12839 PR rtl-optimization/80193
12840 * ira.c (ira): Do not check allocation for LRA.
12842 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
12844 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
12845 (nvptx_output_simt_exit): Declare.
12846 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
12847 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
12848 (init_softstack_frame): Move initialization of crtl->is_leaf to...
12849 (nvptx_declare_function_name): ...here. Emit declaration of local
12850 memory space buffer for omp_simt_enter insn.
12851 (nvptx_output_unisimt_switch): New.
12852 (nvptx_output_softstack_switch): New.
12853 (nvptx_output_simt_enter): New.
12854 (nvptx_output_simt_exit): New.
12855 * config/nvptx/nvptx.h (struct machine_function): New fields
12856 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
12857 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
12858 (UNSPECV_SIMT_EXIT): Ditto.
12859 (omp_simt_enter_insn): New insn.
12860 (omp_simt_enter): New expansion.
12861 (omp_simt_exit): New insn.
12862 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
12864 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
12865 (expand_GOMP_SIMT_ENTER_ALLOC): New.
12866 (expand_GOMP_SIMT_EXIT): New.
12867 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
12868 (GOMP_SIMT_ENTER_ALLOC): Ditto.
12869 (GOMP_SIMT_EXIT): Ditto.
12870 * target-insns.def (omp_simt_enter): New insn.
12871 (omp_simt_exit): Ditto.
12872 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
12874 (lower_rec_simd_input_clauses): Implement SIMT privatization.
12875 (lower_rec_input_clauses): Likewise.
12876 (lower_lastprivate_clauses): Handle SIMT privatization.
12878 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
12879 (ompdevlow_adjust_simt_enter): New.
12880 (find_simtpriv_var_op): New.
12881 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
12882 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
12884 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
12885 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
12886 (copy_decl_for_dup_finish): Ditto.
12888 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
12890 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
12893 * config/i386/i386.c (ix86_option_override_internal): Always
12894 allow -mpreferred-stack-boundary=3 for 64-bit targets.
12896 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12898 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
12900 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12902 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
12903 mark new edge's irreducible flag accordign to it.
12904 (vect_do_peeling): Check loop preheader edge's irreducible flag
12905 and pass it to function slpeel_add_loop_guard.
12907 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
12909 PR tree-optimization/80218
12910 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
12911 Update block frequencies and counts.
12913 2017-03-28 Richard Biener <rguenther@suse.de>
12915 PR tree-optimization/78644
12916 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
12917 of a simplification result we may not use it at all.
12919 2017-03-28 Richard Biener <rguenther@suse.de>
12922 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
12923 without arguments, generate default definition of a SSA name.
12925 2017-03-28 Richard Biener <rguenther@suse.de>
12927 PR middle-end/80222
12928 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
12929 TYPE_REF_CAN_ALIAS_ALL references.
12930 * fold-const.c (fold_indirect_ref_1): Likewise.
12932 2017-03-28 Martin Liska <mliska@suse.cz>
12935 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
12936 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
12938 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12939 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
12941 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
12942 (EXTRA_SPECS): Define.
12943 (SUBTARGET_EXTRA_SPECS): Likewise.
12944 (SUBTARGET_CPP_SPEC): Likewise.
12945 * config/arc/elf.h (EXTRA_SPECS): Renamed to
12946 SUBTARGET_EXTRA_SPECS.
12947 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
12949 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12951 * config/arc/simdext.md (vst64_insn): Update pattern.
12952 (vld32wh_insn): Likewise.
12953 (vld32wl_insn): Likewise.
12954 (vld64_insn): Likewise.
12955 (vld32_insn): Likewise.
12957 2017-03-28 Marek Polacek <polacek@redhat.com>
12960 * fold-const.c (fold_comparison): Use protected_set_expr_location
12961 instead of SET_EXPR_LOCATION.
12963 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
12965 * tree.c (add_expr): Avoid name lookup warning.
12967 2017-03-27 Jeff Law <law@redhat.com>
12969 PR tree-optimization/80216
12970 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
12971 function name. Limit recursion depth.
12972 (record_temporary_equivalences): Corresponding changes.
12974 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
12976 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
12979 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12982 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
12984 * cfgcleanup.c (reg_note_cfa_p): New array.
12985 (insns_have_identical_cfa_notes): New function.
12986 (old_insns_match_p): Don't cross-jump in between /f
12987 and non-/f instructions. If both i1 and i2 are frame related,
12988 verify all CFA notes, their order and content.
12990 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
12993 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
12994 HImode and SImode with zero extend to DImode to one insn.
12995 (bswap<mode>2_extenddi): Likewise.
12996 (bswapsi2_extenddi): Likewise.
12997 (bswaphi2_extendsi): Likewise.
12998 (bswaphi2): Combine bswap HImode and SImode into one insn.
12999 Separate memory insns from swapping register.
13000 (bswapsi2): Likewise.
13001 (bswap<mode>2): Likewise.
13002 (bswaphi2_internal): Delete, no longer used.
13003 (bswapsi2_internal): Likewise.
13004 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
13005 store, and gpr<-gpr swap insns.
13006 (bswap<mode>2_store): Likewise.
13007 (bswaphi2_reg): Register only splitter, combine with the splitter.
13008 (bswaphi2 splitter): Likewise.
13009 (bswapsi2_reg): Likewise.
13010 (bswapsi2 splitter): Likewise.
13011 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
13012 the insns into load, store, and register/register insns.
13013 (bswapdi2_ldbrx): Likewise.
13014 (bswapdi2_load): Likewise.
13015 (bswapdi2_store): Likewise.
13016 (bswapdi2_reg): Likewise.
13018 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
13020 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
13021 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
13023 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13026 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
13028 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
13029 special handling for target option conflicts between dform
13030 options (-mpower9-dform, -mpower9-dform-vector,
13031 -mpower9-dform-scalar) and -mno-direct-move.
13033 2017-03-27 Richard Biener <rguenther@suse.de>
13035 PR tree-optimization/80181
13036 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
13038 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
13040 * config/arc/predicates.md (move_double_src_operand): Replace the
13041 call to move_double_src_operand with a call to address_operand.
13043 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
13045 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
13046 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
13047 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
13049 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
13051 * config/arc/predicates.md (long_immediate_loadstore_operand):
13052 Consider scaled addresses cases.
13054 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
13056 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
13057 restored when in interrupt.
13058 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
13059 doesn't have delay slot.
13061 2017-03-27 Richard Biener <rguenther@suse.de>
13064 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
13065 inlined thunk clones.
13067 2017-03-27 Jakub Jelinek <jakub@redhat.com>
13070 * asan.c (instrument_derefs): Copy over last operand from
13071 original COMPONENT_REF to the new COMPONENT_REF with
13072 DECL_BIT_FIELD_REPRESENTATIVE.
13073 * ubsan.c (instrument_object_size): Likewise.
13075 2017-03-27 Richard Biener <rguenther@suse.de>
13077 PR tree-optimization/80170
13078 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
13079 sure DR/SCEV didnt fold in constants we do not see when looking
13080 at the reference base alignment.
13082 2017-03-27 Richard Biener <rguenther@suse.de>
13084 PR middle-end/80171
13085 * gimple-fold.c (fold_ctor_reference): Properly guard against
13086 NULL return value from canonicalize_constructor_val.
13088 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
13091 * config/i386/i386.c (ix86_expand_builtin)
13092 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
13093 flags reg setting and flags reg using instructions.
13094 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
13095 clobbering instructions to zero extend op2.
13097 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
13099 * doc/install.texi (Configuration) <--with-aix-soname>:
13100 Update link to AIX ld.
13102 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
13104 PR rtl-optimization/80160
13105 PR rtl-optimization/80159
13106 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
13107 reg_alternate_class into account.
13109 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
13112 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
13113 to consider in curr_insn_transform.
13115 2017-03-24 Jakub Jelinek <jakub@redhat.com>
13117 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
13118 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
13119 and emit_mode_inner.
13121 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13123 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
13124 argument to the overloaded builtin variants. Use the new flag to
13125 deprecate certain builtin variants.
13126 * config/s390/s390-builtin-types.def: Add new builtin types.
13127 * config/s390/s390-builtins.h: Support new flags field for
13128 overloaded builtins.
13129 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
13130 (s390_macro_to_expand): Enable vector float data type.
13131 (s390_cpu_cpp_builtins_internal): Indicate support of the new
13132 builtins by incrementing the __VEC__ version number.
13133 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
13135 (s390_resolve_overloaded_builtin): Emit error messages depending
13136 on the builtin flags.
13137 * config/s390/s390.c (s390_expand_builtin): Support additional
13138 flags argument. Change error message to match the messages
13139 emitted in s390-c.c.
13140 * config/s390/s390.md: New UNSPEC_* constants.
13141 (op_type): Add new instruction types.
13142 * config/s390/vecintrin.h: Add new builtins and test data class
13144 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
13145 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
13146 (VEC_INEXACT, VEC_NOINEXACT): New constants.
13147 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
13148 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
13149 ("vec_mergel<mode>"): V_HW -> VEC_HW.
13151 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
13152 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
13153 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
13154 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
13156 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
13157 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
13158 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
13159 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
13161 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
13162 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
13163 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
13164 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
13165 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
13166 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
13167 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
13169 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
13170 ("vec_scatter_element<V_HW_4:mode>_DI")
13171 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
13172 ("vec_fpint<mode>", "vflls")
13173 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
13174 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
13175 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
13176 ("*vec_cmphe<mode>_cc"): ... these.
13178 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
13179 mode constant instead of magic value.
13181 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13183 * config/s390/s390.c (s390_expand_vec_compare): Support other
13184 vector floating point modes than just V2DF.
13185 (s390_expand_vcond): Likewise.
13186 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
13187 (s390_cannot_change_mode_class): Prevent mode changes between TF
13188 and V1TF in vector registers.
13189 * config/s390/s390.md (DF, SF): New mode attributes.
13190 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
13191 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
13192 SFmode support for VRs.
13193 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
13195 (VFT, VF_HW): New mode iterators.
13196 (vw, sdx): New mode attributes.
13197 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
13198 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
13199 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
13200 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
13201 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
13202 also the new vector floating point modes. Renaming to ...
13204 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
13205 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
13206 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
13207 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
13208 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
13209 ("vec_unordered<mode>"): ... these.
13211 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
13212 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
13213 ("*vec_extendv2df"): New insn definitions.
13215 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13217 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
13218 ("mulditi3_2", "*muldi3_sign"): New patterns.
13219 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
13220 rename the pattern definition.
13222 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13224 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
13226 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
13228 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13230 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
13231 instruction if possible.
13232 * config/s390/vector.md (vec_halfnumelts): New mode
13234 ("*vec_vllezlf<mode>"): New pattern.
13236 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13238 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
13239 ("popcountv4si2", "popcountv2di2"): Rename to ...
13240 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
13241 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
13243 ("popcount<mode>2_vxe"): New pattern.
13245 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13247 * common/config/s390/s390-common.c (processor_flags_table): Add
13249 * config.gcc: Add arch12.
13250 * config/s390/driver-native.c (s390_host_detect_local_cpu):
13251 Default to arch12 for unknown CPU model numbers.
13252 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
13253 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
13254 PROCESSOR_max sanity check.
13255 * config/s390/s390-opts.h (enum processor_type): Add
13257 * config/s390/s390.c (processor_table): Add arch12.
13258 (s390_expand_builtin): Add check for B_VXE flag.
13259 (s390_issue_rate): Add PROCESSOR_ARCH12.
13260 (s390_get_sched_attrmask): Likewise.
13261 (s390_get_unit_mask): Likewise.
13262 (s390_sched_score): Enable z13 scheduling for arch12.
13263 (s390_sched_reorder): Likewise.
13264 (s390_sched_variable_issue): Likewise.
13265 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
13267 (s390_tune_attr): Use z13 scheduling also for arch12.
13268 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
13269 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
13270 (TARGET_VXE_P): New macros.
13271 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
13272 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
13273 * config/s390/s390.opt: Add arch12 as processor_type.
13275 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13277 * config/s390/s390.md
13278 ("fixuns_truncdddi2", "fixuns_trunctddi2")
13279 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
13280 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
13282 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
13283 Rename expanders to ...
13285 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
13286 ("fixuns_truncdddi2_emu"): ... these.
13288 ("fixuns_trunc<mode>si2_emu"): New expander.
13290 ("*fixuns_truncdfdi2_z13"): Rename to ...
13291 ("*fixuns_truncdfdi2_vx"): ... this.
13293 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13295 * config/s390/2964.md: Remove the single element vector compare
13296 instructions which are no longer used.
13297 * config/s390/s390.c (s390_select_ccmode): Remove handling of
13299 (s390_canonicalize_comparison): Remove handling of DFmode
13301 (s390_expand_vec_compare_scalar): Remove function.
13302 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
13303 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
13305 ("*cmp<mode>_ccs"): Add wfcdb instruction.
13307 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13309 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
13311 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
13312 will anyway by matched by mov<mode>_64dfp.
13314 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13316 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
13317 vlef/vstef. Add missing operand to vleif.
13319 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13321 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
13322 pair for all vector types with 64 bit elements.
13323 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
13324 * config/s390/vector.md (V_HW_64): ... here.
13325 (V_128_NOSINGLE): New mode iterator.
13326 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
13327 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
13328 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
13329 ("*vec_load_pairv2di"): Change to ...
13330 ("*vec_load_pair<mode>"): ... this one.
13332 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13334 * config/s390/constraints.md: Add comments.
13335 (jKK): Reject element sizes > 8 bytes.
13336 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
13338 * config/s390/s390.md: Add the s_operand checks formerly in
13339 s390_split_ok_p to various splitters where they are still
13341 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
13342 for 128 bit vectors. Plus two splitters.
13344 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13346 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
13349 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13352 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
13353 error if the boundary argument is not constant.
13355 2017-03-24 Jakub Jelinek <jakub@redhat.com>
13357 PR rtl-optimization/80112
13358 * loop-doloop.c (doloop_condition_get): Don't check condition
13359 if cmp isn't SET with IF_THEN_ELSE src.
13361 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13363 PR tree-optimization/80158
13364 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
13365 replacing a candidate statement, also replace it for the
13366 candidate's alternate interpretation.
13367 (replace_rhs_if_not_dup): Likewise.
13368 (replace_one_candidate): Likewise.
13370 2017-03-24 Richard Biener <rguenther@suse.de>
13372 PR tree-optimization/80167
13373 * graphite-isl-ast-to-gimple.c
13374 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
13376 (translate_isl_ast_to_gimple::get_rename): Likewise.
13378 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
13380 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
13381 handling of certain combinations of target options, including the
13382 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
13383 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
13385 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13388 * config/arm/arm.md (*load_multiple): Add reload_completed to
13389 matching condition.
13391 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13392 Richard Biener <rguenth@suse.de>
13394 PR tree-optimization/79908
13395 PR tree-optimization/80136
13396 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
13397 been cast away, gimplify_and_add suffices.
13399 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
13401 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
13403 2017-03-23 Richard Biener <rguenther@suse.de>
13405 PR tree-optimization/80032
13406 * gimplify.c (gimple_push_cleanup): Forced unconditional
13407 cleanups still have to go to the conditional_cleanups
13410 2017-03-22 Jakub Jelinek <jakub@redhat.com>
13412 PR tree-optimization/80072
13413 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
13415 (next_operand_entry_id): Change type to unsigned int.
13416 (sort_by_operand_rank): Make sure to return the right return value
13417 even if unsigned fields are bigger than INT_MAX.
13418 (struct oecount): Change cnt and id type to unsigned int.
13419 (oecount_hasher::equal): Formatting fix.
13420 (oecount_cmp): Make sure to return the right return value
13421 even if unsigned fields are bigger than INT_MAX.
13422 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
13425 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
13426 TREE_READONLY on result if writing it more than once.
13429 * doc/invoke.texi (-fsanitize=thread): Document that with
13430 -fnon-call-exceptions atomics are not able to throw
13434 * tsan.c: Include tree-eh.h.
13435 (instrument_builtin_call): Call maybe_clean_eh_stmt or
13436 maybe_clean_or_replace_eh_stmt where needed.
13437 (instrument_memory_accesses): Add cfg_changed argument.
13438 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
13439 if it returned true.
13440 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
13442 PR rtl-optimization/63191
13443 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
13444 wrapper function, moved the whole old content into ...
13445 (ix86_delegitimize_address_1): ... this. New inline function.
13446 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
13447 true as last argument instead of ix86_delegitimize_address.
13449 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
13451 * config/aarch64/aarch64.c (generic_branch_cost): Copy
13452 cortexa57_branch_cost.
13454 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
13456 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
13458 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13461 * doc/md.texi (Constraints): Document wA constraint.
13462 * config/rs6000/constraints.md (wA): New.
13463 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
13464 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
13465 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
13466 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
13468 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
13471 * gimplify.c (is_oacc_declared): New function.
13472 (oacc_default_clause): Use it to set default flags for acc declared
13473 variables inside parallel regions.
13474 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
13475 declared variables.
13476 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
13477 declare attribute to any decl as necessary.
13479 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
13482 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
13483 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
13484 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
13485 (arm_arch_lpae): This.
13486 * config/arm/arm.c (arm_arch7ve): Rename into ...
13487 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
13488 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
13491 2017-03-22 Martin Liska <mliska@suse.cz>
13494 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
13495 error message instead of an ICE.
13497 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13499 * doc/extend.texi (6.11 Additional Floating Types): Revise.
13501 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
13503 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
13505 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
13508 2017-03-21 Martin Sebor <msebor@redhat.com>
13510 * doc/extend.texi: Use "cannot" instead of "can't."
13511 * doc/hostconfig.texi: Same.
13512 * doc/install.texi: Same.
13513 * doc/invoke.texi: Same.
13514 * doc/loop.texi: Same.
13515 * doc/md.texi: Same.
13516 * doc/objc.texi: Same.
13517 * doc/rtl.texi: Same.
13518 * doc/tm.texi: Same.
13519 * doc/tm.texi.in: Same.
13520 * doc/trouble.texi: Same.
13522 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
13525 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
13526 (collect_checksum_attributes): Set it.
13527 (die_checksum_ordered): Use it.
13529 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13531 PR tree-optimization/79908
13532 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
13533 change: For a VA_ARG whose LHS has been cast away, use
13534 force_gimple_operand to construct the side effects.
13536 2017-03-21 David Malcolm <dmalcolm@redhat.com>
13538 PR translation/80001
13539 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
13540 more amenable to translation.
13541 (oacc_loop_auto_partitions): Likewise.
13543 2017-03-21 Marek Polacek <polacek@redhat.com>
13544 Martin Sebor <msebor@redhat.com>
13546 PR tree-optimization/80109
13547 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
13548 on INTEGRAL_TYPE_P.
13550 2017-03-21 Jakub Jelinek <jakub@redhat.com>
13551 Segher Boessenkool <segher@kernel.crashing.org>
13554 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
13555 check reg_used_between_p between insn and one of succ or succ2
13556 depending on if succ is artificial insn not inserted into insn
13559 2017-03-21 Martin Liska <mliska@suse.cz>
13561 PR gcov-profile/80081
13562 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
13563 * doc/gcc.texi: Include gcov-dump stuff.
13564 * doc/gcov-dump.texi: New file.
13566 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
13568 PR rtl-optimization/79150
13569 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
13570 conditional jump, if the jump is the last insn of the loop.
13572 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13573 Richard Biener <rguenth@suse.de>
13575 PR tree-optimization/79908
13576 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
13577 been cast away, use force_gimple_operand to construct the side
13580 2017-03-21 Martin Liska <mliska@suse.cz>
13582 PR libfortran/79956
13583 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
13586 2017-03-21 Brad Spengler <spender@grsecurity.net>
13589 * plugin.c (htab_hash_plugin): New function.
13590 (add_new_plugin): Use it and adjust.
13591 (parse_plugin_arg_opt): Adjust.
13592 (init_one_plugin): Likewise.
13594 2017-03-21 Richard Biener <rguenther@suse.de>
13596 PR tree-optimization/80032
13597 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
13598 if set force the cleanup to happen unconditionally.
13599 (gimplify_target_expr): Push inserted clobbers with force_uncond
13600 to avoid them being removed by control-dependent DCE.
13602 2017-03-21 Richard Biener <rguenther@suse.de>
13604 PR tree-optimization/80122
13605 * tree-inline.c (copy_bb): Do not expans va-arg packs or
13606 va_arg_pack_len when the inlined call stmt requires pack
13608 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
13610 2017-03-21 Jakub Jelinek <jakub@redhat.com>
13613 * tsan.c (instrument_builtin_call): If the memory model argument
13614 is not a constant, assume it is valid.
13617 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
13620 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
13622 PR rtl-optimization/79910
13623 * combine.c (can_combine_p): Do not allow combining an I0 or I1
13624 if its dest is used by an insn before I2 (other than the combined
13625 insns themselves, which are properly handled already).
13627 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
13630 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13632 * combine.c (record_used_regs): New static function.
13633 (try_combine): Handle situations where there is an additional
13634 instruction between I2 and I3 which needs to have a LOG_LINK
13638 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
13640 * combine.c (try_combine): Delete redundant i1 test. Call
13641 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
13643 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13646 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
13647 alternatives 13/14.
13649 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13651 PR tree-optimization/80054
13652 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
13653 the optimization if a PHI or any of its arguments is not dominated
13654 by the candidate's basis. Use gphi* rather than gimple* as
13656 (replace_profitable_candidates): Clean up a gimple* variable that
13657 should be a gphi* variable.
13659 2017-03-20 Martin Sebor <msebor@redhat.com>
13662 * doc/extend.texi (attribute constructor): Document present limitation.
13664 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
13667 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
13668 __POWER9_VECTOR__ #ifdef control, change template definition to
13669 use Power9-specific built-in function.
13670 (vec_any_eq): Likewise.
13671 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
13672 to control outcomes from this test.
13673 (vector_ae_<mode>p): For VEC_F modes, likewise.
13675 2017-03-20 Ian Lance Taylor <iant@google.com>
13677 * config/i386/i386.c (ix86_function_regparm): Save an extra
13678 register for -fsplit-stack with DECL_STATIC_CHAIN.
13680 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
13683 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
13684 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
13686 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
13688 * config/riscv/riscv.c (riscv_print_operand): Use "fence
13690 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
13693 2017-03-20 Marek Polacek <polacek@redhat.com>
13696 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
13698 2017-03-20 Richard Biener <rguenther@suse.de>
13700 PR tree-optimization/80113
13701 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
13702 allocate extra SSA name for PHI def.
13703 (add_close_phis_to_outer_loops): Likewise.
13704 (add_close_phis_to_merge_points): Likewise.
13705 (copy_loop_close_phi_args): Likewise.
13706 (copy_cond_phi_nodes): Likewise.
13708 2017-03-20 Martin Liska <mliska@suse.cz>
13710 PR middle-end/79753
13711 * tree-chkp.c (chkp_build_returned_bound): Do not build
13712 returned bounds for a LHS that's not a BOUNDED_P type.
13714 2017-03-20 Martin Liska <mliska@suse.cz>
13718 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
13719 COMPLEX_CST and VECTOR_CST.
13721 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13724 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
13725 target operand. A new splitter adds the clobber statement in case
13726 the target operand is dead anyway.
13728 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
13730 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
13731 to age-old versions of binutils and glibc.
13733 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
13735 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
13737 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13739 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
13741 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13743 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
13744 requirement for binutils 2.13.
13746 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
13748 * combine.c (try_combine): Delete redundant i1 test. Call
13749 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
13751 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
13753 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
13754 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
13756 <riscv64-*-elf>: Re-arrange section
13757 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
13758 <riscv32-*-linux>: Likewise.
13759 <riscv64-*-elf>: Likewise
13760 <riscv64-*-linux>: Likewise.
13762 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
13765 * aarch64.opt(verbose-cost-dump): Fix typo.
13767 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
13770 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
13771 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
13773 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13775 * reload.c (find_reloads): When reloading a nonoffsettable address,
13776 use RELOAD_OTHER for it and its address reloads.
13778 PR rtl-optimization/79910
13779 * combine.c (record_used_regs): New static function.
13780 (try_combine): Handle situations where there is an additional
13781 instruction between I2 and I3 which needs to have a LOG_LINK
13784 2017-03-17 Jeff Law <law@redhat.com>
13786 PR tree-optimization/71437
13787 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
13788 conditional in the hash table first.
13789 (vrp_dom_walker::before_dom_children): Extract condition from
13790 ASSERT_EXPR. Record condition, its inverion and any implied
13791 conditions as well.
13793 2017-03-17 Marek Polacek <polacek@redhat.com>
13794 Markus Trippelsdorf <markus@trippelsdorf.de>
13796 PR tree-optimization/80079
13797 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
13800 2017-03-17 Richard Biener <rguenther@suse.de>
13802 PR middle-end/80075
13803 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
13804 Properly verify the LHS before the RHS possibly claims to be
13806 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
13809 2017-03-17 Martin Jambor <mjambor@suse.cz>
13811 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
13812 (List of -O2 options): Likewise.
13813 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
13816 2017-03-17 Tom de Vries <tom@codesourcery.com>
13818 * gcov-dump.c (print_usage): Print bug_report_url.
13820 2017-03-17 Richard Biener <rguenther@suse.de>
13822 PR middle-end/80050
13823 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
13824 (parser::peek): Likewise.
13826 2017-03-17 Richard Biener <rguenther@suse.de>
13828 PR tree-optimization/80048
13829 * sese.c (free_sese_info): Properly release rename_map and
13830 copied_bb_map elements.
13832 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
13834 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
13835 Add linked-list forward and backlinks. Insert on
13836 construction, remove on destruction.
13837 (class pass_store_merging): Add m_stores_head field.
13838 (pass_store_merging::terminate_and_process_all_chains):
13839 Iterate over m_stores_head list.
13840 (pass_store_merging::terminate_all_aliasing_chains):
13842 (pass_store_merging::execute): Check for debug stmts first.
13843 Push new chains onto the m_stores_head stack.
13845 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
13848 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
13849 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
13850 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
13852 2017-03-16 Jeff Law <law@redhat.com>
13854 PR tree-optimization/71437
13855 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
13856 member function. Implementation moved into after_dom_children
13857 member function and into the threader's thread_outgoing_edges
13859 (dom_opt_dom_walker::after_dom_children): Simplify by moving
13860 some code into new thread_outgoing_edges.
13861 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
13862 definition. Simplify marker handling (do it here). Assume we always
13863 have the available expression and the const/copies tables.
13864 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
13866 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
13867 * tree-vrp.c (equiv_stack): No longer file scoped.
13868 (vrp_dom_walker): New class.
13869 (vrp_dom_walker::before_dom_children): New member function.
13870 (vrp_dom_walker::after_dom_children): Likewise.
13871 (identify_jump_threads): Setup domwalker. Use it rather than
13872 walking edges in a random order by hand. Simplify setup/finalization.
13873 (finalize_jump_threads): Remove.
13874 (vrp_finalize): Do not call identify_jump_threads here.
13875 (execute_vrp): Do it here instead and call thread_through_all_blocks
13878 PR tree-optimization/71437
13879 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
13881 (simplify_stmt_for_jump_threading): Add basic_block argument. All
13883 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
13884 (dom_opt_dom_walker::thread_across_edge): Remove
13885 handle_dominating_asserts argument. All callers changed.
13886 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
13887 changes. Remove calls to lhs_of_dominating_assert. Other
13888 uses of handle_dominating_asserts turn into unconditional code
13889 (simplify_control_stmt_condition_1): Likewise.
13890 (simplify_control_stmt_condition): Likewise.
13891 (thread_through_normal_block, thread_across_edge): Likewise.
13892 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
13893 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
13894 object if it is not an SSA_NAME.
13895 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
13896 before calling into the VRP specific simplifiers.
13897 (identify_jump_threads): Remove handle_dominating_asserts
13900 2017-03-16 Jakub Jelinek <jakub@redhat.com>
13903 * tree-diagnostic.c (default_tree_printer): No longer static.
13904 * tree-diagnostic.h (default_tree_printer): New prototype.
13906 2017-03-16 Tamar Christina <tamar.christina@arm.com>
13908 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
13909 Change ins into fmov.
13911 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13913 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
13914 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
13915 Use h_con constraint for operand 1.
13916 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
13917 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
13919 2017-03-15 Jeff Law <law@redhat.com>
13921 PR tree-optimization/71437
13922 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
13923 (record_temporary_equivalences): Use it.
13925 PR tree-optimization/71437
13926 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
13927 tree-ssa-scopedtables.
13928 (lookup_avail_expr, build_and_record_new_cond): Likewise.
13929 (record_conditions, record_cond, vuse_eq): Likewise.
13930 (record_edge_info): Adjust to API tweak of record_conditions.
13931 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
13932 (record_temporary_equivalences, optimize_stmt): Likewise.
13933 (eliminate_redundant_computations): Likewise.
13934 (record_equivalences_from_stmt): Likewise.
13935 * tree-ssa-scopedtables.c: Include options.h and params.h.
13936 (vuse_eq): New function, moved from tree-ssa-dom.c
13937 (build_and_record_new_cond): Likewise.
13938 (record_conditions): Likewise. Accept vector of conditions rather
13939 than edge_equivalence structure for first argument.
13940 for the first argument.
13941 (avail_exprs_stack::lookup_avail_expr): New member function, moved
13942 from tree-ssa-dom.c.
13943 (avail_exprs_stack::record_cond): Likewise.
13944 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
13945 from tree-ssa-dom.c.
13946 (avail_exprs_stack): Add new member functions lookup_avail_expr
13948 (record_conditions): Declare.
13950 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
13953 * lra-constraints.c (process_alt_operands): Increase reject for
13954 reloading an input/output operand.
13956 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
13959 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
13960 insns to convert from signed/unsigned char/short to IEEE 128-bit
13962 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
13964 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
13967 * config/i386/i386.c (ix86_vector_duplicate_value): Create
13968 subreg of inner mode for values already in registers.
13970 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
13972 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
13973 iteration reg is used after the loop.
13975 2017-03-14 Martin Sebor <msebor@redhat.com>
13977 PR tree-optimization/79800
13978 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
13979 precision in negative-positive range.
13980 (format_floating): Call non-const overload with adjusted precision.
13982 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13985 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
13988 2017-03-14 Martin Sebor <msebor@redhat.com>
13990 PR middle-end/80020
13991 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
13992 * builtins.def (aligned_alloc): Use it.
13995 * Makefile.in (GTFILES): Add calls.c.
13996 * calls.c: Include "gt-calls.h".
13998 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
14000 PR rtl-optimization/79728
14001 * regs.h (struct target_regs): New field
14002 x_contains_allocatable_regs_of_mode.
14003 (contains_allocatable_regs_of_mode): New macro.
14004 * reginfo.c (init_reg_sets_1): Initialize it, and change
14005 contains_reg_of_mode so it includes global regs as well.
14006 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
14007 rather than contains_regs_of_mode.
14009 2017-03-14 Martin Liska <mliska@suse.cz>
14011 * doc/invoke.texi: Document options that can't be combined with
14012 -fcheck-pointer-bounds.
14014 2017-03-14 Martin Liska <mliska@suse.cz>
14016 PR middle-end/79831
14017 * doc/invoke.texi (-Wchkp): Document the option.
14019 2017-03-14 Martin Liska <mliska@suse.cz>
14021 * Makefile.in: Install gcov-dump.
14023 2017-03-14 Martin Liska <mliska@suse.cz>
14025 * multiple_target.c (expand_target_clones): Bail out for
14026 an invalid attribute.
14028 2017-03-14 Richard Biener <rguenther@suse.de>
14030 * alias.c (struct alias_set_entry): Pack properly.
14031 * cfgloop.h (struct loop): Likewise.
14032 * cse.c (struct set): Likewise.
14033 * ipa-utils.c (struct searchc_env): Likewise.
14034 * loop-invariant.c (struct invariant): Likewise.
14035 * lra-remat.c (struct cand): Likewise.
14036 * recog.c (struct change_t): Likewise.
14037 * rtl.h (struct address_info): Likewise.
14038 * symbol-summary.h (function_summary): Likewise.
14039 * tree-loop-distribution.c (struct partition): Likewise.
14040 * tree-object-size.c (struct object_size_info): Likewise.
14041 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
14042 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
14043 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
14044 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
14045 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
14046 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
14047 (struct _stmt_vec_info): Likewise.
14049 2017-03-14 Martin Liska <mliska@suse.cz>
14052 * multiple_target.c (create_dispatcher_calls): Check that
14053 a target can create a function dispatcher.
14055 2017-03-14 Martin Liska <mliska@suse.cz>
14058 * multiple_target.c (expand_target_clones): Drop local.local
14059 flag for default implementation.
14061 2017-03-14 Richard Biener <rguenther@suse.de>
14063 PR tree-optimization/80030
14064 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
14066 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
14068 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
14069 gcc_fallthrough() instead of __attribute__((fallthrough));
14071 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
14073 * doc/gcc.texi: Remove "up" link to (DIR).
14074 * doc/gccint.texi: Ditto.
14076 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
14078 * doc/install.texi (Specific) <avr>: Remove reference to
14081 2017-03-13 Jeff Law <law@redhat.com>
14083 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
14084 attribute rather than comments.
14086 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
14087 match_scratch operand is highest.
14089 2017-03-13 Martin Liska <mliska@suse.cz>
14091 PR middle-end/78339
14092 * ipa-pure-const.c (warn_function_noreturn): If the declarations
14093 is a CHKP clone, use original declaration.
14095 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
14097 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
14098 (arc_conditional_register_usage): Use a different allocation order
14099 when optimizing for size.
14100 * common/config/arc/arc-common.c (arc_option_optimization_table):
14101 Section anchors default on when optimizing for size.
14103 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
14105 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
14107 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
14109 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
14110 * config/arc/arc.md (cpu_facility): Add cd variant.
14111 (*movqi_insn): Add code density variant.
14112 (*movhi_insn): Likewise.
14113 (*movqi_insn): Likewise.
14114 (*addsi3_mixed): Likewise.
14115 (subsi3_insn): Likewise.
14117 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
14119 * config/arc/arc.md (movsi_cond_exec): Update constraint.
14121 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
14123 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
14124 expressions with MINUS and UNARY ops.
14126 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14129 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
14131 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
14132 between vec_select and vector argument.
14133 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
14134 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
14135 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
14136 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
14137 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
14138 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
14140 2017-03-13 Richard Biener <rguenther@suse.de>
14143 * params.def (vect-max-peeling-for-alignment): Fix typo.
14145 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
14147 * doc/install.texi (Specific) <mips-*-*>: Remove description of
14148 issue that only occurred with binutils below 2.18.
14150 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
14152 * doc/install.texi (Specific) <cris-axis-elf>: No longer
14153 refer to binutils 2.11/2.12 minimum.
14155 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
14157 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
14158 ftp.kernel.org and simplify binutils requirement.
14160 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
14162 * doc/invoke.texi (Warning Options): Fix spelling of link-time
14164 (Optimize Options): Ditto. Also remove redundancy.
14166 2017-03-10 David Malcolm <dmalcolm@redhat.com>
14168 PR translation/79848
14169 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
14171 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
14172 to G_ to avoid double translation.
14174 2017-03-10 David Malcolm <dmalcolm@redhat.com>
14176 PR translation/79923
14177 * auto-profile.c (get_combined_location): Convert leading
14178 character of diagnostics to lower case and remove trailing period.
14179 (read_profile): Likewise for various diagnostics.
14180 * config/arm/arm.c (arm_option_override): Remove trailing period
14181 from various diagnostics.
14182 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
14183 (msp430_expand_delay_cycles): Likewise.
14185 2017-03-10 David Malcolm <dmalcolm@redhat.com>
14188 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
14189 full command-line argument, rather than just "str".
14190 (aarch64_validate_march): Likewise.
14191 (aarch64_validate_mtune): Likewise.
14193 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
14195 PR rtl-optimization/78911
14196 * lra-assigns.c (must_not_spill_p): New function.
14197 (spill_for): Use it.
14199 2017-03-10 Jakub Jelinek <jakub@redhat.com>
14201 PR tree-optimization/79981
14202 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
14203 ATOMIC_COMPARE_EXCHANGE ifn result.
14204 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
14205 IFN_ATOMIC_COMPARE_EXCHANGE.
14207 2017-03-10 David Malcolm <dmalcolm@redhat.com>
14210 * opts.c (parse_sanitizer_options): Add missing question mark to
14211 "did you mean" message.
14213 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14215 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
14217 (VMULEUH_UNS): Likewise.
14218 (VMULOUB_UNS): Likewise.
14219 (VMULOUH_UNS): Likewise.
14220 * config/rs6000/rs6000.c (builtin_function_type): Remove
14221 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
14223 2017-03-10 David Malcolm <dmalcolm@redhat.com>
14226 * read-rtl-function.c (function_reader::read_rtx_operand): Update
14227 x with result of extra_parsing_for_operand_code_0.
14228 (function_reader::extra_parsing_for_operand_code_0): Convert
14229 return type from void to rtx, returning x. When reading
14230 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
14231 larger size containing struct block_symbol.
14233 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
14235 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
14236 -mfloat128-hardware without -m64.
14238 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
14241 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
14242 entries to the case statement that marks unsigned arguments to
14243 overloaded functions.
14245 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
14247 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
14248 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
14250 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
14253 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
14254 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
14256 2017-03-10 Martin Liska <mliska@suse.cz>
14260 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
14261 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
14264 2017-03-10 Olivier Hainque <hainque@adacore.com>
14266 * tree-switch-conversion (array_value_type): Start by resetting
14267 candidate type to it's main variant.
14269 2017-03-10 Jakub Jelinek <jakub@redhat.com>
14271 PR rtl-optimization/79909
14272 * combine.c (try_combine): Use simplify_replace_rtx on individual
14273 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
14274 of the whole CALL_INSN_FUNCTION_USAGE.
14276 PR tree-optimization/79972
14277 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
14278 get_range_info on SSA_NAMEs. Formatting fixes.
14280 2017-03-10 Richard Biener <rguenther@suse.de>
14281 Jakub Jelinek <jakub@redhat.com>
14283 PR tree-optimization/77975
14284 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
14285 edge to be constant.
14286 (get_val_for): For constant x return it. Formatting fix.
14287 (loop_niter_by_eval): Avoid pointless looping if the next iteration
14288 would use the same bases as the current one.
14290 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14292 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
14293 instead of vec_select for V1TImode.
14294 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
14296 (VSX_LE_128): Add V1TI to this mode iterator.
14297 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
14298 (*vsx_le_perm_store_<mode>): Likewise.
14299 (pre-reload splitter for VSX stores): Likewise.
14300 (post-reload splitter for VSX stores): Likewise.
14301 (*vsx_xxpermdi2_le_<mode>): Likewise.
14302 (*vsx_lxvd2x2_le_<mode>): Likewise.
14303 (*vsx_stxvd2x2_le_<mode>): Likewise.
14305 2017-03-09 Michael Eager <eager@eagercon.com>
14307 Correct failures with --enable-checking=yes,rtl.
14309 * config/microblaze/microblaze.c (microblaze_expand_shift):
14310 Replace GET_CODE test with CONST_INT_P and INTVAL test with
14311 test for const0_rtx.
14312 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
14313 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
14315 2017-03-09 Richard Biener <rguenther@suse.de>
14317 PR tree-optimization/79977
14318 * graphite-scop-detection.c (scop_detection::merge_sese):
14319 Handle the case of extra exits to blocks dominating the entry.
14321 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
14323 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
14326 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
14328 PR rtl-optimization/79949
14329 * lra-constraints.c (process_alt_operands): Check memory when
14330 trying to predict a cycle. Print about the overall increase.
14332 2017-03-09 Richard Biener <rguenther@suse.de>
14334 PR middle-end/79971
14335 * gimple-expr.c (useless_type_conversion_p): Preserve
14336 TYPE_SATURATING for fixed-point types.
14338 2017-03-09 Richard Biener <rguenther@suse.de>
14341 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
14342 alignment of BLKmode params.
14344 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14347 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
14348 (VALL_NO_V2Q): Likewise.
14350 * config/aarch64/aarch64-simd.md
14351 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
14353 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
14354 VALL_NO_V2Q mode iterator.
14355 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
14357 2017-03-09 Martin Liska <mliska@suse.cz>
14359 PR tree-optimization/79631
14360 * tree-chkp-opt.c (chkp_is_constant_addr): Call
14361 tree_int_cst_sign_bit just for INTEGER constants.
14363 2017-03-09 Martin Liska <mliska@suse.cz>
14367 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
14370 2017-03-09 Marek Polacek <polacek@redhat.com>
14373 * tree.c (inchash::add_expr): Handle TREE_VEC.
14375 2017-03-09 Martin Liska <mliska@suse.cz>
14378 (chkp_narrow_size_and_offset): New function.
14379 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
14380 (void chkp_parse_bit_field_ref): New function.
14381 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
14382 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
14384 2017-03-09 Martin Liska <mliska@suse.cz>
14387 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
14388 (chkp_find_bounds_1): Remove gcc_unreachable.
14390 2017-03-09 Jakub Jelinek <jakub@redhat.com>
14393 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
14394 BUILT_IN_SYNC*, determine the access type from the size suffix and
14395 always build a MEM_REF with that type. Handle forgotten
14396 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
14399 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
14400 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
14401 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
14402 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
14403 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
14404 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
14405 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
14406 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
14407 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
14408 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
14409 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
14410 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
14411 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
14412 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
14413 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
14414 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
14415 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
14416 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
14417 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
14418 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
14419 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
14420 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
14421 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
14422 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
14423 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
14424 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
14425 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
14426 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
14427 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
14428 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
14429 definitions outside of __OPTIMIZE__ guarded section.
14432 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
14433 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
14434 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
14435 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
14438 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14440 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
14441 ("vfenez<mode>"): Add missing constraints.
14443 2017-03-08 Martin Sebor <msebor@redhat.com>
14446 * config/nds32/nds32.c (nds32_option_override):
14447 Fix misspelled diagnostic.
14449 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14452 * gimplify.c (gimplify_omp_for): Replace index var in outer
14453 taskloop statement with an artificial variable and add
14454 OMP_CLAUSE_PRIVATE clause for it.
14456 2017-03-08 Richard Biener <rguenther@suse.de>
14458 PR tree-optimization/79955
14459 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
14460 for accesses that are completely outside of the variable.
14462 2017-03-08 Andrew Haley <aph@redhat.com>
14464 PR tree-optimization/79943
14465 * tree-ssa-loop-split.c (compute_new_first_bound): When
14466 calculating the new upper bound, (END-BEG) should be added, not
14469 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14471 * config/avr/avr.md (setmemhi): Make sure match_dup
14472 operand number comes before match_scratch.
14474 2017-03-08 Richard Biener <rguenther@suse.de>
14476 PR tree-optimization/79920
14477 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
14478 with ncopies == 1 to ...
14479 (vect_transform_slp_perm_load): ... here. Properly compute
14480 all element loads by iterating VF times over the group. Do
14481 not handle ncopies (computed in a broken way) in
14482 vect_create_mask_and_perm.
14484 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14487 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
14488 is a uniform vector, use uniform_vector_p return value instead of
14489 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
14491 2017-03-07 Marek Polacek <polacek@redhat.com>
14493 PR middle-end/79809
14494 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
14495 (alloca_call_type): Likewise.
14497 2017-03-07 Martin Liska <mliska@suse.cz>
14499 * gcov.c (process_args): Put comment to correct location.
14501 2017-03-07 Martin Liska <mliska@suse.cz>
14503 PR middle-end/68270
14504 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
14505 Use array_at_struct_end_p instead of DECL_CHAIN (field).
14506 (chkp_narrow_bounds_for_field): Likewise.
14507 (chkp_parse_array_and_component_ref): Pass one more argument to
14510 2017-03-07 Richard Biener <rguenther@suse.de>
14512 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
14515 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
14517 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
14518 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
14520 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14523 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
14524 to end of description.
14525 (PARAM_MAX_STORES_TO_MERGE): Likewise.
14527 2017-03-07 Jakub Jelinek <jakub@redhat.com>
14529 PR rtl-optimization/79901
14530 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
14532 (*avx512f_<code><mode>3<mask_name>): ... this.
14533 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
14534 iterator instead of VI8_AVX2_AVX512BW.
14536 PR rtl-optimization/79901
14537 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
14538 min/max expander, expand it using expand_vec_cond_expr.
14541 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
14544 2017-03-06 Jakub Jelinek <jakub@redhat.com>
14547 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
14548 to void * for PCH reasons.
14549 * dwarf2out.c (output_loc_operands, output_die): Cast
14550 v.val_vec.array to unsigned char *.
14552 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
14555 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
14558 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
14560 PR rtl-optimization/79571
14561 * lra-constraints.c (process_alt_operands): Calculate static
14562 reject and subtract it from overall when only addresses will be
14565 2017-03-06 Julia Koval <julia.koval@intel.com>
14568 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
14569 incoming stack boundary to 128 for 64-bit targets.
14571 2017-03-06 Richard Biener <rguenther@suse.de>
14573 PR tree-optimization/79894
14574 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
14575 to NULL after folding it.
14577 2017-03-06 Richard Biener <rguenther@suse.de>
14579 PR tree-optimization/79824
14580 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
14581 check disabling peeling for gaps.
14583 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
14585 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
14586 attributes): Document gettimeofday.
14588 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
14590 * config/s390/s390.c (s390_option_override_internal): Set
14591 PARAM_MIN_VECT_LOOP_BOUND
14593 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
14595 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
14596 * config/s390/s390.md: Likewise.
14598 2017-03-06 Jakub Jelinek <jakub@redhat.com>
14601 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
14602 (<avx2_avx512>_perm<mode>): Rename to ...
14603 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
14605 (<avx512>_perm<mode>_mask): Rename to ...
14606 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
14608 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
14609 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
14610 instead of VI8F_256_512.
14611 (avx512f_perm<mode>): New define_expand.
14612 (avx512f_perm<mode>_mask): Likewise.
14613 (avx512f_perm<mode>_1<mask_name>): New define_insn.
14614 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
14616 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14618 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
14619 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
14621 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
14623 2017-03-06 Martin Liska <mliska@suse.cz>
14626 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
14627 when having a SSA NAME w/o VAR_DECL assigned to it.
14629 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14631 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
14632 msa_dpsub_<su>_d): Fix MODE for vec_select.
14634 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14636 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
14638 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
14640 2017-03-06 Richard Biener <rguenther@suse.de>
14642 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
14643 * plugin.c (register_plugin_info): Likewise.
14644 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
14646 2017-03-05 Jakub Jelinek <jakub@redhat.com>
14648 * config/i386/sse.md (sse_storehps, sse_storelps,
14649 avx_<castmode><avxsizesuffix>_<castmode>,
14650 avx512f_<castmode><avxsizesuffix>_<castmode>,
14651 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
14652 in condition that at least one operand is not a MEM.
14654 2017-03-03 Jakub Jelinek <jakub@redhat.com>
14656 PR middle-end/79805
14657 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
14658 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
14660 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
14661 gimple_call_nothrow_p flag based on whether original builtin can throw.
14662 If it can, emit following stmts on the fallthrough edge.
14663 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
14664 don't create new bb if inserting just debug stmts on the edge, try to
14665 insert them on the fallthru bb or just reset debug stmts.
14667 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
14670 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
14671 restore recog_data (including the operand rtxes inside it) around
14672 the call to get_insn_template.
14674 2017-03-03 Martin Sebor <msebor@redhat.com>
14676 PR tree-optimization/79699
14677 * context.c (context::~context): Free MPFR caches to avoid
14678 a memory leak on program exit.
14680 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14682 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
14683 Use wide_int::ulow () instead of .elt (0).
14685 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
14687 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
14688 (*pushxf): Limit oF constraint to 32bit targets and add oC
14689 constraint for 64bit targets.
14690 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
14691 (*pushdf): Change rmF constraint to rmC.
14693 2017-03-03 Martin Liska <mliska@suse.cz>
14695 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
14696 Remove unused variable.
14698 2017-03-03 Jakub Jelinek <jakub@redhat.com>
14701 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
14702 is a memory operand, increase num_memory.
14703 (ix86_expand_args_builtin): Likewise.
14705 2017-03-03 Jan Hubicka <jh@suse.cz>
14708 * ipa-devirt.c (maybe_record_node): Properly handle
14709 __cxa_pure_virtual visibility.
14711 2017-03-03 Martin Liska <mliska@suse.cz>
14713 PR tree-optimization/79803
14714 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
14716 (pass_loop_prefetch::execute): Disabled optimization if an
14717 assumption about L1 cache size is not met.
14719 2017-03-03 Martin Liska <mliska@suse.cz>
14721 PR rtl-optimization/79574
14722 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
14723 (hash_scan_set): Likewise.
14724 (dump_hash_table): Likewise.
14725 (hoist_code): Likewise.
14727 2017-03-03 Richard Biener <rguenther@suse.de>
14729 * fixed-value.c (fixed_from_string): Restore use of elt (1)
14730 in place of uhigh ().
14731 (fixed_convert_from_real): Likewise.
14733 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
14736 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
14738 2017-03-03 Richard Biener <rguenther@suse.de>
14740 PR middle-end/79818
14741 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
14742 TYPE_OVERFLOW_UNDEFINED check.
14744 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14746 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
14748 (vector_ae_<mode>_p): Likewise.
14749 (vector_nez_<mode>_p): Likewise.
14750 (vector_ne_v2di_p): Likewise.
14751 (vector_ae_v2di_p): Likewise.
14752 (vector_ne_<mode>_p): Likewise.
14753 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
14755 (vsx_tsqrt<mode>2_fe): Likewise.
14757 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
14760 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
14762 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14764 PR rtl-optimization/79780
14765 * cprop.c (one_cprop_pass): When second and further conditional trap
14766 in a single basic block is turned into an unconditional trap, turn it
14767 into a deleted note to avoid RTL verification failures.
14769 2017-03-02 Richard Biener <rguenther@suse.de>
14771 * fold-const.c (const_binop): Use ulow () instead of elt (0).
14773 2017-03-02 Richard Biener <rguenther@suse.de>
14775 PR tree-optimization/79345
14777 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
14778 param and abort the walk, returning -1 if it is hit.
14779 (walk_aliased_vdefs): Take a limit param and pass it on.
14780 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
14781 defaulting to 0 and return a signed int.
14782 * tree-ssa-uninit.c (struct check_defs_data): New struct.
14783 (check_defs): New helper.
14784 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
14785 about uninitialized memory.
14786 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
14787 bogus uninitialized warning.
14788 (fixed_convert_from_real): Likewise.
14790 2017-03-02 Bin Cheng <bin.cheng@arm.com>
14792 PR tree-optimization/66768
14793 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
14794 iv_use if base object can't be determined.
14796 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14798 PR tree-optimization/79345
14799 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
14800 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
14801 (get_pattern_stats): Initialize it.
14802 * genemit.c (gen_expand): Verify match_scratch numbers come after
14803 match_operand/match_dup numbers.
14804 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
14805 match_scratch numbers.
14806 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
14808 * config/s390/s390.md (trunctdsd2): Likewise.
14810 2017-03-02 Richard Biener <rguenther@suse.de>
14812 * wide-int.h (wide_int_storage::operator=): Implement in terms
14815 2017-03-02 Richard Biener <rguenther@suse.de>
14817 PR tree-optimization/79777
14818 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
14819 the to insert expression to sth existing.
14821 2017-03-01 Martin Sebor <msebor@redhat.com>
14823 PR middle-end/79692
14824 * gimple-ssa-sprintf.c
14825 (directive::known_width_and_precision): New function.
14826 (format_integer): Use it.
14827 (get_mpfr_format_length): Consider the full range of precision
14828 when computing %g output with the # flag. Set the likely byte
14829 count to 3 rather than 1 when precision is indeterminate.
14830 (format_floating): Correct the lower bound of precision.
14832 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14834 * doc/invoke.texi: Document default code model for 64-bit Linux.
14836 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14839 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
14840 udiv rather than div since input pattern is unsigned.
14842 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
14844 * config/i386/i386.c (print_reg): Warn for values of
14845 unsupported size in integer register.
14847 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
14850 * config/rs6000/predicates.md (current_file_function_operand): Do
14851 not allow self calls to be local if the function is replaceable.
14853 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
14856 * config/rs6000/altivec.h (vec_ctz and others): Change the
14857 preprocessor macro that controls conditional compilation from
14858 _ARCH_PWR9 to __POWER9_VECTOR__.
14859 (vec_all_ne): Change parameterization of __altivec_scalar_pred
14860 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
14861 control (instead of _ARCH_PWR9 control) so that template
14862 definition uses power9-specific function.
14863 (vec_any_eq): Likewise.
14864 (vec_all_ne): Change macro definition to use a power9-specific
14865 expansion under #ifdef __POWER9_VECTOR__ control (instead of
14866 _ARCH_PWR9 control).
14867 (vec_any_eq) Likewise.
14868 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
14869 expansion for CMPNEF to remove support for xvcmpnesp instruction.
14870 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
14871 support for xvcmpnedp instruction.
14872 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
14873 macro expansion so that Power9 implementation of vec_all_ne does
14874 not use the AltiVec predicate framework.
14875 (VCMPNEH_P): Likewise.
14876 (VCMPNEW_P): Likewise.
14877 (VCMPNED_P): Likewise.
14878 (VCMPNEFP_P): Likewise.
14879 (VCMPNEDP_P): Likewise.
14880 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
14881 implementation of vec_any_eq to not use AltiVec predicate
14883 (VCMPAEH_P): Likewise.
14884 (VCMPAEW_P): Likewise.
14885 (VCMPAED_P): Likewise.
14886 (VCMPAEFP_P): Likewise.
14887 (VCMPAEDP_P): Likewise.
14888 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
14889 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
14890 not use the AltiVec predicate framework.
14891 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
14892 of vec_any_eq to not use AltiVec predicate framework.
14893 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
14894 support for predefined __POWER9_VECTOR__ macro to indicate that
14895 Power9 instruction selection is enabled.
14896 (altivec_overloaded_builtins): Remove extraneous
14897 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
14898 function argument types RS6000_BTI_bool_V16QI and
14899 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
14900 entry for overloaded function argument types RS6000_BTI_bool_V4SI
14901 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
14902 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
14903 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
14904 Power9 for implementations of vec_cmpne. Change the signature for
14905 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
14906 (representing vec_all_ne) to remove the previously described first
14907 argument of type RS6000_BTI_INTSI, as this was an artifact of
14908 reliance on the AltiVec predicate framework, which is no longer
14909 used in the implementation of these functions. Add
14910 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
14911 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
14912 since, unlike the AltiVec predicate framework implementation, we
14913 do not share function descriptors between vec_alle and vec_anyeq.
14914 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
14915 set of modes that receive special treatment even when
14916 TARGET_P9_VECTOR is true. The special treatment emits code that
14917 does not depend on Power9 instructions.
14918 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
14919 define_expand to not rely on AltiVec predicate framework.
14920 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14922 (vector_ne_v2di_p): Change this define_expand to not rely on
14923 AltiVec predicate framework.
14924 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
14926 (vector_ne_<mode>_p): Change this define_expand to not rely on
14927 AltiVec predicate framework.
14928 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14930 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
14931 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
14932 define_insn pattern.
14933 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
14934 define_insn pattern because the xvcmpne<VSs>. instruction is not
14936 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
14937 instruction is not supported.
14939 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14941 * config/nvptx/nvptx.c: Include intl.h.
14943 2017-03-01 Martin Jambor <mjambor@suse.cz>
14946 * ipa-prop.h (ipa_bits): Removed field known.
14947 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
14948 to pointers. Adjusted their comments to warn about their sharing.
14949 (ipcp_transformation_summary): Change bits to a vector of pointers.
14950 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
14951 (ipa_get_ipa_bits_for_value): Declare.
14952 * tree-vrp.h (value_range): Mark as GTY((for_user)).
14953 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
14954 (ipa_bits_hash_table): Likewise.
14955 (ipa_vr_ggc_hash_traits): Likewise.
14956 (ipa_vr_hash_table): Likewise.
14957 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
14958 being pointers and vr_known being removed.
14959 (ipa_set_jf_unknown): Likewise.
14960 (ipa_get_ipa_bits_for_value): New function.
14961 (ipa_set_jfunc_bits): Likewise.
14962 (ipa_get_value_range): New overloaded functions.
14963 (ipa_set_jfunc_vr): Likewise.
14964 (ipa_compute_jump_functions_for_edge): Use the above functions to
14965 construct bits and vr parts of jump functions.
14966 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
14967 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14969 (ipcp_grow_transformations_if_necessary): Also allocate
14970 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14972 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
14973 them. Fix too long lines.
14974 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
14975 vr_known being removed.
14976 (ipa_read_jump_function): Use new setter functions to construct bits
14977 and vr parts of jump functions or set them to NULL.
14978 (write_ipcp_transformation_info): Adjust for bits being pointers.
14979 (read_ipcp_transformation_info): Likewise.
14980 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
14982 Include gt-ipa-prop.h.
14983 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
14985 (ipcp_store_bits_results): Likewise.
14986 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
14987 Do not write to existing jump functions but use a temporary instead.
14989 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14992 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
14993 attempt to use its first operand as BIT_FIELD_REF base.
14995 2017-03-01 Richard Biener <rguenther@suse.de>
14997 PR middle-end/79721
14998 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
14999 interpolating formula in wrapping arithmetic.
15000 (chrec_apply): Convert chrec_evaluate return value to wanted type.
15002 2017-03-01 Jakub Jelinek <jakub@redhat.com>
15004 PR tree-optimization/79734
15005 * tree-vect-generic.c (expand_vector_condition): Optimize
15006 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
15007 Handle VEC_COND_EXPR where comparison has different inner width from
15008 type's inner width.
15010 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
15012 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
15013 markup, and similar issues. Remove @opindex entries for things
15014 that aren't options. Add missing -mmpy-option entries.
15016 2017-02-28 Jakub Jelinek <jakub@redhat.com>
15018 PR tree-optimization/79737
15019 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
15020 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
15021 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
15022 instead of byte_size. Formatting fix.
15023 (shift_bytes_in_array_right): Formatting fix.
15025 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
15028 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
15029 condition on optimize for the leaf function test.
15031 2017-02-28 Martin Liska <mliska@suse.cz>
15034 * read-rtl-function.c (function_reader::handle_unknown_directive):
15035 Bail out when one uses -flto.
15037 2017-02-28 Martin Liska <mliska@suse.cz>
15039 * common.opt: Replace space with tabular for options of <number>
15041 * config/i386/i386.opt: Show <number> value for
15042 -mlarge-data-threshold.
15043 * opts.c (print_filtered_help): Do not display number in hexadecimal
15046 2017-02-28 Martin Liska <mliska@suse.cz>
15048 * common.opt: Fix --help=option -Q for options which are of
15051 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
15053 * config/i386/i386.c (print_reg): Error out for values
15054 of 8-bit size in invalid integer register.
15056 2017-02-28 Martin Sebor <msebor@redhat.com>
15058 PR tree-optimization/79691
15059 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
15061 2017-02-28 Jakub Jelinek <jakub@redhat.com>
15064 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
15065 gcc_unreachable with output_operand_lossage.
15067 2017-02-28 Richard Biener <rguenther@suse.de>
15069 PR tree-optimization/79740
15070 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
15072 (visit_nary_op): Insert the nary into the hashtable if we
15073 pattern-matched sth.
15074 * tree-ssa-pre.c (eliminate_insert): Robustify.
15076 2017-02-28 Richard Biener <rguenther@suse.de>
15078 PR middle-end/79731
15079 * fold-const.c (decode_field_reference): Reject out-of-bound
15082 2017-02-28 Jakub Jelinek <jakub@redhat.com>
15084 * config/i386/i386.c: Include intl.h.
15085 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
15086 instead of just cond ? "..." : "...".
15087 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
15088 * coverage.c (read_counts_file): Likewise.
15089 * omp-offload.c: Include intl.h.
15090 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
15091 of just cond ? "..." : "...".
15092 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
15093 of just cond ? "..." : "...".
15095 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
15098 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
15100 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
15101 'tune for' CPU name.
15102 * config/arm/arm-cpu-data.h: Regenerated.
15104 2017-02-28 Richard Biener <rguenther@suse.de>
15106 PR tree-optimization/79732
15107 * tree-inline.c (expand_call_inline): Do not shadow var.
15109 2017-02-28 Richard Biener <rguenther@suse.de>
15111 PR tree-optimization/79723
15112 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
15113 address-space properly.
15115 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
15117 * doc/optinfo.texi (Optimization groups): Fix option used for
15119 * doc/invoke.texi (-fopt-info): Document "omp".
15120 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
15121 (OPTGROUP_ALL): Add OPTGROUP_OMP.
15122 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
15123 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
15124 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
15126 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
15128 * dumpfile.c (optgroup_options): Instead of "openmp", associate
15129 OPTGROUP_OMP with "omp".
15131 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
15134 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
15135 for arithmetic shift of unsigned V2DI.
15137 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
15139 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
15140 arc/linux.h headers.
15141 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
15142 (LINK_SPEC): Likewise.
15143 (ARC_TLS_EXTRA_START_SPEC): Likewise.
15144 (EXTRA_SPECS): Likewise.
15145 (STARTFILE_SPEC): Likewise.
15146 (ENDFILE_SPEC): Likewise.
15147 (LIB_SPEC): Likewise.
15148 (TARGET_SDATA_DEFAULT): Likewise.
15149 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
15150 (MULTILIB_DEFAULTS): Likewise.
15151 (DWARF2_UNWIND_INFO): Likewise.
15152 * config/arc/big.h: New file.
15153 * config/arc/elf.h: Likewise.
15154 * config/arc/linux.h: Likewise.
15155 * config/arc/t-uClibc: Remove.
15157 2017-02-27 Bin Cheng <bin.cheng@arm.com>
15159 PR tree-optimization/77536
15160 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
15161 (tree_transform_and_unroll_loop): Use above function to compute the
15162 estimated niter of unrolled loop and use it when scaling profile.
15163 Also use count info rather than frequency if it's non-zero.
15164 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
15165 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
15166 (vect_transform_loop): Call above function.
15168 2017-02-27 Richard Biener <rguenther@suse.de>
15170 PR tree-optimization/45397
15171 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
15172 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
15173 (visit_nary_op): Add pattern matching for CSEing sign-changed
15174 or truncated operations with wider ones.
15176 2017-02-27 Richard Biener <rguenther@suse.de>
15178 PR tree-optimization/79690
15179 * tree-vect-stmts.c (vectorizable_store): Use vector type
15180 built from the DR with address-space.
15182 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
15184 * doc/invoke.texi (Optimize Options): Refine the description
15185 of asan-use-after-return.
15187 2017-02-25 Alan Modra <amodra@gmail.com>
15189 PR rtl-optimization/79584
15190 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
15191 base, not ad->base_term, the reg within base. Remove assertion
15192 that ad->base == ad->base_term. Replace gen_int_mode using
15193 bogus mode with const0_rtx.
15195 2017-02-25 Jakub Jelinek <jakub@redhat.com>
15197 PR middle-end/79396
15198 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
15199 FMA_EXPR like tcc_binary or tcc_unary.
15201 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
15204 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
15206 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
15207 (output_loc_operands): Handle DW_OP_call_ref and
15208 DW_OP_GNU_variable_value.
15209 (struct variable_value_struct): New type.
15210 (struct variable_value_hasher): Likewise.
15211 (variable_value_hash): New variable.
15212 (string_types): Remove.
15213 (copy_loc_descr): New function.
15214 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
15215 (prepend_loc_descr_to_each): New function.
15216 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
15217 instead of add_loc_descr_to_each if the first argument is single
15218 location list and the second has multiple.
15219 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
15220 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
15221 when looking for variable value which doesn't have other location info.
15222 (loc_list_from_tree): Formatting fix.
15223 (gen_array_type_die): Simplify DW_AT_string_length handling.
15224 (adjust_string_types): Remove.
15225 (gen_subprogram_die): Don't call adjust_string_types nor test/set
15226 string_types. Call resolve_variable_values.
15227 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
15228 (resolve_addr_in_expr): Likewise. Add A argument.
15229 (copy_deref_exprloc): Remove deref argument. Adjust for the
15230 original expression being DW_OP_GNU_variable_value with optionally
15231 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
15232 optionally after it.
15233 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
15234 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
15235 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
15236 (variable_value_hasher::hash, variable_value_hasher::equal): New
15238 (resolve_variable_value_in_expr, resolve_variable_value,
15239 resolve_variable_values, note_variable_value_in_expr,
15240 note_variable_value): New functions.
15241 (dwarf2out_early_finish): Call note_variable_value on all toplevel
15244 2017-02-24 Jakub Jelinek <jakub@redhat.com>
15247 * opts.h (handle_generated_option): Add GENERATED_P argument.
15248 * opts-common.c (handle_option): Adjust function comment.
15249 (handle_generated_option): Add GENERATED_P argument, pass it to
15251 (control_warning_option): Pass false to handle_generated_option
15253 * opts.c (maybe_default_option): Pass true to handle_generated_option
15255 * optc-gen.awk: Likewise.
15257 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
15259 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
15260 a REG, look at the REG it is a SUBREG of.
15261 (splitter for cmpeqsi_t): Ditto.
15263 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
15265 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
15266 the special USEs with the pattern of the insn, not the insn itself.
15268 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
15271 * doc/invoke.texi: Document -mload-store-pairs.
15273 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
15274 Sandra Loosemore <sandra@codesourcery.com>
15276 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
15277 argument isn't a CONST_INT.
15278 (nios2_alternate_compare_const): Assert op is a CONST_INT.
15279 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
15280 (nios2_validate_compare): Bypass alternate compare logic if *op2
15281 is not a CONST_INT.
15282 (ldstwm_operation_p): Return false if first_base is not a REG or
15283 if first_offset is not a CONST_INT.
15285 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
15287 * config/cris/cris.md: Use correct operand in a define_peephole2.
15289 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
15291 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
15293 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
15295 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
15296 this_insn if it is an INSN or JUMP_INSN.
15297 (force_offsettable): Look at base, not at addr.
15298 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
15299 on things that aren't necessarily CONST_INTs.
15301 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
15303 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
15304 -mfpmath=sse is the default also for x86-32 targets with SSE2
15305 instruction set when @option{-ffast-math} is enabled
15307 2017-02-24 Jeff Law <law@redhat.com>
15309 PR rtl-optimizatoin/79286
15310 * ira.c (update_equiv_regs): Drop may_trap_p exception to
15313 2017-02-24 Richard Biener <rguenther@suse.de>
15315 PR tree-optimization/79389
15316 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
15319 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
15321 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
15322 function comment to reflect reality.
15323 (loop_exits_before_overflow): Fix typo in function description.
15325 2017-02-24 Richard Biener <rguenther@suse.de>
15327 PR tree-optimization/79389
15328 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
15329 properly that a threading opportunity exists. Detect conditional
15330 copy/constant propagation opportunities.
15332 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
15334 * config/visium/visium.md (type): Add trap.
15335 (b): New mode attribute.
15336 (*btst): Rename into...
15337 (*btst<mode>): ...this and adjust.
15338 (*cbranchsi4_btst_insn): Rename into...
15339 (*cbranch<mode>4_btst_insn): ...this and adjust.
15340 (trap): New define_insn.
15342 2017-02-23 Jakub Jelinek <jakub@redhat.com>
15344 PR tree-optimization/79389
15345 * ifcvt.c (struct noce_if_info): Add rev_cond field.
15346 (noce_reversed_cond_code): New function.
15347 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
15348 reversed_comparison_code. Formatting fix.
15349 (noce_try_store_flag): Test rev_cond != NULL in addition to
15350 reversed_comparison_code.
15351 (noce_try_store_flag_constants): Likewise.
15352 (noce_try_store_flag_mask): Likewise.
15353 (noce_try_addcc): Use rev_cond if non-NULL instead of
15354 reversed_comparison_code.
15355 (noce_try_cmove_arith): Likewise. Formatting fixes.
15356 (noce_try_minmax, noce_try_abs): Clear rev_cond.
15357 (noce_find_if_block): Initialize rev_cond.
15358 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
15359 instead of false as last argument never attempt to reverse it
15362 2017-02-23 Bin Cheng <bin.cheng@arm.com>
15364 PR tree-optimization/79663
15365 * tree-predcom.c (combine_chains): Process refs in reverse order
15366 only for ZERO length chains, and add explaining comment.
15368 2017-02-23 Jeff Law <law@redhat.com>
15370 PR tree-optimization/79578
15371 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
15372 in call to operand_equal_p.
15374 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
15377 * config/i386/cpuid.h: Fix another undefined behavior.
15379 2017-02-23 Richard Biener <rguenther@suse.de>
15381 PR tree-optimization/79683
15382 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
15383 vector types for data-refs.
15385 2017-02-23 Martin Liska <mliska@suse.cz>
15387 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
15389 2017-02-23 Jakub Jelinek <jakub@redhat.com>
15391 PR middle-end/79665
15392 * internal-fn.c (get_range_pos_neg): Moved to ...
15393 * tree.c (get_range_pos_neg): ... here. No longer static.
15394 * tree.h (get_range_pos_neg): New prototype.
15395 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
15396 are known to be in between 0 and signed maximum inclusive, try to
15397 expand both unsigned and signed divmod and use the cheaper one from
15400 2017-02-22 Jeff Law <law@redhat.com>
15402 PR tree-optimization/79578
15403 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
15404 to compare base operands.
15406 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
15409 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
15410 gpc_reg_operand instead of fpr_reg_operand.
15412 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
15414 * config/mips/mips.c (mips_return_in_memory): Force FP
15415 vector types to be returned in memory for o32 ABI.
15417 2017-02-22 Jakub Jelinek <jakub@redhat.com>
15419 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
15420 instead of DW_TAG_member for static data member declarations and don't
15421 set no_linkage_name for static inline data members.
15422 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
15425 2017-02-22 Martin Liska <mliska@suse.cz>
15427 * doc/invoke.texi: Replace inequality signs with square brackets
15430 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15432 PR tree-optimization/68644
15433 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
15435 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
15438 * lra-constraints.c (simplify_operand_subreg): Handle
15439 WORD_REGISTER_OPERATIONS targets.
15441 2017-02-22 Jakub Jelinek <jakub@redhat.com>
15444 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
15445 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
15446 elimination by swapping fld*.
15448 2017-02-22 Richard Biener <rguenther@suse.de>
15450 PR tree-optimization/79673
15451 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
15452 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
15453 irrelevant address-space qualifiers and avoiding a
15454 ADDR_SPACE_CONVERT_EXPR from fold_convert.
15456 2017-02-22 Richard Biener <rguenther@suse.de>
15458 PR tree-optimization/79666
15459 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
15460 to not symbolically negate if that may introduce undefined
15463 2017-02-22 Martin Liska <mliska@suse.cz>
15466 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
15467 * data-streamer-out.c (streamer_write_gcov_count_stream):
15469 * value-prof.c (stream_out_histogram_value): Make assert more
15470 precise based on type of counter.
15472 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
15475 * config/i386/i386.md (standard_x87sse_constant_load splitter):
15476 Use nonimmediate_operand instead of memory_operand for operand 1.
15477 (float-extend standard_x87sse_constant_load splitter): Ditto.
15479 2017-02-21 Jeff Law <law@redhat.com>
15481 PR tree-optimization/79621
15482 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
15483 blocks with edges to themselves.
15485 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15488 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
15489 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
15490 Use gimple_call_builtin_p.
15493 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
15494 on temporarily removed DEBUG_INSNs.
15496 PR tree-optimization/79649
15497 * tree-loop-distribution.c (classify_partition): Give up on
15498 non-generic address space loads/stores.
15500 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
15502 * doc/loop.texi (Loop manipulation): Remove nonexistent
15503 tree_ssa_loop_version from the documentation.
15504 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
15506 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15509 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
15510 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
15511 * config/rs6000/rs6000.c: Include except.h.
15512 (rs6000_expand_split_stack_prologue): Call
15513 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
15515 2017-02-21 Martin Jambor <mjambor@suse.cz>
15518 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
15519 have been analyzed.
15521 2017-02-21 Martin Jambor <mjambor@suse.cz>
15523 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
15524 for backward compatibility only.
15525 * doc/invoke.texi (Option Summary): Remove all references to
15526 -fipa-cp-alignment.
15528 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
15532 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15534 * lra-constraints.c (curr_insn_transform): Handle
15535 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
15537 2017-02-21 Martin Liska <mliska@suse.cz>
15539 * config/i386/i386.opt: Replace -masm-dialect with -masm.
15541 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
15543 PR translation/79638
15544 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
15546 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
15549 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
15550 (arm_function_ok_for_sibcall): Return false for an indirect call by
15551 descriptor if all the argument registers are used.
15552 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
15553 alignment of the function.
15555 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15557 PR tree-optimization/61441
15558 * simplify-rtx.c (simplify_const_unary_operation): For
15559 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
15560 the sNaN unmodified.
15562 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
15564 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15565 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15566 instead of SYSTEM_HEADER_DIR.
15568 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
15569 Martin Liška <mliska@suse.cz>
15571 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
15572 Fix typos and grammar, use active voice, and clarify.
15574 2017-02-20 Marek Polacek <polacek@redhat.com>
15576 PR middle-end/79537
15577 * gimplify.c (gimplify_expr): Handle unused *&&L;.
15580 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
15582 2017-02-20 Jakub Jelinek <jakub@redhat.com>
15585 * config/i386/i386.c (ix86_expand_builtin): Handle
15586 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
15587 ix86_builtins_isa[fcode].isa as a requirement of those
15588 flags and any other flag in the bitmask.
15589 (ix86_init_mmx_sse_builtins): Use 0 instead of
15590 ~OPTION_MASK_ISA_64BIT as mask.
15591 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
15592 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
15593 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
15594 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
15596 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15599 * lra-constraints.c (split_reg): Check requested split mode
15600 is supported by the register.
15602 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15604 * lra-constraints.c (simplify_operand_subreg): Remove early
15607 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15610 * lra-constraints.c (curr_insn_transform): Tighten condition
15611 for converting SUBREG reloads from OP_OUT to OP_INOUT.
15613 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15616 * lra-constraints.c (curr_insn_transform): Handle
15617 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
15619 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
15622 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
15624 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
15626 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
15629 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
15632 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15634 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
15637 2017-02-18 Jakub Jelinek <jakub@redhat.com>
15639 * final.c (last_columnnum, override_columnnum): New variables.
15640 (final_start_function): Set last_columnnum, pass it to begin_prologue
15641 hook and pass 0 to dwarf2out_begin_prologue.
15642 (final_scan_insn): Update override_columnnum. Pass last_columnnum
15643 to source_line debug hook.
15644 (notice_source_line): Compute last_columnnum and for debug_column_info
15645 return true on column changes.
15646 * debug.h (struct gcc_debug_hooks): Add column argument to
15647 source_line and begin_prologue hooks.
15648 (debug_nothing_int_charstar_int_bool): Remove prototype.
15649 (debug_nothing_int_int_charstar,
15650 debug_nothing_int_int_charstar_int_bool): New prototypes.
15651 (dwarf2out_begin_prologue): Add column argument.
15652 * debug.c (do_nothing_debug_hooks): Adjust source_line and
15653 begin_prologue hooks.
15654 (debug_nothing_int_charstar_int_bool): Remove.
15655 (debug_nothing_int_int_charstar,
15656 debug_nothing_int_int_charstar_int_bool): New functions.
15657 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
15658 through to dwarf2out_source_line.
15659 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
15660 (dwarf2out_source_line): Add column argument, emit it if requested.
15661 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
15663 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
15664 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
15665 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
15666 through to dwarf2out_begin_prologue.
15667 (vmsdbgout_source_line): Add column argument, pass it through to
15668 dwarf2out_source_line.
15669 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
15670 dbxout_source_line caller.
15671 (dbxout_source_line): Add column argument.
15673 * common.opt (gno-column-info, gcolumn-info): New options.
15674 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
15675 (check_die): Also test for multiple DW_AT_decl_column attributes.
15676 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
15677 DW_AT_decl_column if requested.
15678 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
15680 (gen_variable_die): Likewise.
15681 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
15682 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
15685 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
15686 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
15687 (ix86_handle_option): Handle OPT_m3dnowa.
15688 * doc/invoke.texi (-m3dnowa): Document.
15689 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
15690 -m3dnowa instead of -m3dnow -march=athlon.
15693 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
15694 instead of gcc_assert for K, r and R code checks. Formatting fixes.
15696 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15699 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
15700 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
15701 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
15702 generator for vsx_xxpermdi_<mode>_be.
15703 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
15704 force big-endian semantics.
15705 (vsx_xxpermdi_<mode>_be): New define_expand with same
15706 implementation as previous version of vsx_xxpermdi_<mode>.
15708 2017-02-17 Jakub Jelinek <jakub@redhat.com>
15710 PR tree-optimization/79327
15711 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
15712 variable, its initialization and use.
15714 2017-02-17 Julia Koval <julia.koval@intel.com>
15716 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
15717 (OPTION_MASK_ISA_PKU_UNSET): New.
15718 (ix86_handle_option): Handle -mrdpid.
15719 * config/i386/cpuid.h (bit_RDPID): New.
15720 * config/i386/driver-i386.c (host_detect_local_cpu):
15721 Detect RDPID feature.
15722 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
15723 * config/i386/i386-c.c (ix86_target_macros_internal):
15725 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
15726 (ix86_valid_target_attribute_inner_p): Add "rdpid".
15727 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
15728 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
15729 * config/i386/i386.md (define_insn "rdpid"): New.
15730 * config/i386/i386.opt Add -mrdpid.
15731 * config/i386/immintrin.h (_rdpid_u32): New.
15733 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
15735 PR rtl-optimization/79541
15736 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
15737 instead of transforming it into USE.
15739 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
15741 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
15742 If HONOR_SNANS (SFmode) force the input to a register.
15743 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
15744 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
15745 an frsp or similar insn.
15747 2017-02-17 Martin Liska <mliska@suse.cz>
15749 PR rtl-optimization/79577
15750 * params.def (selsched-max-sched-times): Increase minimum to 1.
15752 2017-02-17 Martin Liska <mliska@suse.cz>
15754 PR rtl-optimization/79574
15755 * gcse.c (want_to_gcse_p): Prevent integer overflow.
15757 2017-02-17 Martin Liska <mliska@suse.cz>
15759 PR tree-optimization/79529
15760 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
15761 ssa_defined_default_def_p to handle cases which are implicitly
15763 * tree-ssa.c (ssa_defined_default_def_p): New function.
15764 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
15765 which are implicitly defined.
15766 * tree-ssa.h (ssa_defined_default_def_p): Declare.
15768 2017-02-17 Richard Biener <rguenther@suse.de>
15770 PR middle-end/79576
15771 * params.def (max-ssa-name-query-depth): Limit to 10.
15773 2017-02-17 Richard Biener <rguenther@suse.de>
15775 PR tree-optimization/79552
15776 * tree-ssa-structalias.c (visit_loadstore): Properly verify
15779 2017-02-17 Richard Biener <rguenther@suse.de>
15782 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
15784 2017-02-17 Marek Polacek <polacek@redhat.com>
15786 PR middle-end/79536
15787 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
15788 (fold_negate_expr): New wrapper.
15790 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
15792 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
15793 Correct terminology and de-emphasize pre-standard behavior.
15795 2017-02-16 Alan Modra <amodra@gmail.com>
15797 PR rtl-optimization/79286
15798 * ira.c (def_dominates_uses): New function.
15799 (update_equiv_regs): Don't create an equivalence for insns that
15800 may trap where the register def does not dominate the use.
15802 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
15804 PR rtl-optimization/78127
15805 * lra.c (lra): Call lra_eliminate before finish the loop after
15808 2017-02-16 Richard Biener <rguenther@suse.de>
15810 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
15812 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
15813 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
15814 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
15815 (isl_val_int_from_wi): New function.
15816 (extract_affine_gmp): Rename to ...
15817 (extract_affine_wi): ... this, take a widest_int.
15818 (extract_affine_int): Just wrap extract_affine_wi.
15819 (add_param_constraints): Use isl_val_int_from_wi.
15820 (add_loop_constraints): Likewise, and extract_affine_wi.
15822 2017-02-15 Jeff Law <law@redhat.com>
15824 PR middle-end/79521
15825 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
15826 ira_init_register_move_cost_if_necessary.
15828 2017-02-15 Martin Sebor <msebor@redhat.com>
15830 PR middle-end/32003
15831 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
15832 removed in a prior commit.
15834 2017-02-15 Bin Cheng <bin.cheng@arm.com>
15836 PR tree-optimization/79347
15837 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
15838 counters during peeling.
15840 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
15842 * Makefile.in (site.exp): Remove "set ISLVER".
15844 2017-02-15 Jakub Jelinek <jakub@redhat.com>
15847 * real.c (real_from_integer): Call real_convert even for decimal.
15849 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
15852 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
15854 2017-02-14 Andrew Pinski <apinski@cavium.com>
15856 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
15857 cores and change the partno/implementer to be correct.
15858 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
15859 the 'B" as the implementer.
15860 * config/aarch64/aarch64-tune.md: Regenerate.
15862 2017-02-14 Carl Love <cel@us.ibm.com>
15864 * config/rs6000/rs6000.c: Add case statement entry to make the
15865 xvcvuxdsp built-in argument unsigned.
15866 * config/rs6000/vsx.md: Fix the source and return operand types so they
15867 match the instruction definitions from the ISA document. Fix typo
15868 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
15871 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
15874 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
15875 member early_clobber_alts.
15876 * lra-lives.c (reg_early_clobber_p): New.
15877 (process_bb_lives): Use it.
15878 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
15879 (debug_operand_data): Initialize early_clobber_alts.
15880 (setup_operand_alternative): Set up early_clobber_alts.
15881 (collect_non_operand_hard_regs): Ditto. Pass early clobber
15882 alternatives to new_insn_reg.
15883 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
15885 (lra_update_insn_regno_info): Pass the new arg.
15887 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15889 PR middle-end/79505
15890 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
15891 (new_oacc_loop_raw): Don't clear already cleared fields.
15894 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
15895 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
15896 _mm512_prefetch_i64gather_ps): New inline functions and macros.
15898 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
15901 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
15903 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
15906 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
15907 the extra instruction to the right place to store 128-bit constant
15910 2017-02-14 Martin Sebor <msebor@redhat.com>
15912 PR middle-end/79448
15913 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
15914 warning for strings of unknown length.
15916 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
15918 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
15920 2017-02-14 Jeff Law <law@redhat.com>
15923 * ira-costs.c (scan_one_insn): Initialize register move costs
15924 for pseudos seen in USE/CLOBBER insns.
15926 PR tree-optimization/79095
15927 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
15928 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
15929 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
15930 if the operands are known to be not equal, then the resulting range
15932 (intersect_ranges): If the new range is ~[0,0] and the old range is
15933 wide, then prefer ~[0,0].
15934 * tree-vrp.c (overflow_comparison_p_1): New function.
15935 (overflow_comparison_p): New function.
15936 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
15937 if NAME is used in an overflow test.
15938 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
15939 overflow check that can be expressed as an equality test, then adjust
15940 ops to be that equality test.
15942 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15944 * config/s390/s390-builtin-types.def: Remove flags argument.
15945 * config/s390/s390.c (s390_init_builtins): Likewise.
15947 2017-02-14 Martin Liska <mliska@suse.cz>
15949 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
15950 vector. Fix trailing white spaces.
15952 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
15954 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
15957 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15959 PR rtl-optimization/68664
15960 * config/arm/arm.c (arm_sched_can_speculate_insn):
15961 New function. Declare prototype.
15962 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15964 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15966 PR rtl-optimization/68664
15967 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
15969 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15971 2017-02-14 Amit Pawar <amit.pawar@amd.com>
15973 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
15974 max skip bytes for function, loop and jump.
15976 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15978 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
15979 ABS_EXPR for gimple dump.
15981 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15984 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
15986 PR tree-optimization/79408
15987 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
15988 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
15989 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
15990 also if rhs1 is INTEGER_CST.
15992 2017-02-14 Richard Biener <rguenther@suse.de>
15994 PR middle-end/79432
15995 * tree-into-ssa.c (insert_phi_nodes): When the function can
15996 have abnormal edges rewrite SSA names with broken use-def
15997 dominance out of SSA and register them for PHI insertion.
15999 2017-02-13 Martin Sebor <msebor@redhat.com>
16001 PR middle-end/79496
16002 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
16003 clearing info.nowrite flag when snprintf size argument is a range.
16005 2017-02-13 Jakub Jelinek <jakub@redhat.com>
16007 * cprop.c (cprop_jump): Add missing space in string literal.
16008 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
16009 (get_constraint_for_component_ref): Likewise.
16010 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
16011 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
16012 * lra-constraints.c (process_alt_operands): Likewise.
16013 * ipa-inline.c (inline_small_functions): Likewise.
16014 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
16015 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
16016 * trans-mem.c (diagnose_tm_1_op): Likewise.
16017 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
16018 (grid_parallel_clauses_gridifiable): Likewise.
16020 * config/nvptx/mkoffload.c (process): Add space in between
16023 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
16024 "MOD4_SSE_REGS" and "ALL_REGS".
16026 * spellcheck.c (test_data): Add , in between "foo" and "food".
16028 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16031 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
16032 boundary crossing check and subsequent code generation agree.
16034 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16036 * config/aarch64/aarch64.c (has_memory_op): Delete.
16037 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
16040 2017-02-13 Jakub Jelinek <jakub@redhat.com>
16042 PR rtl-optimization/79388
16043 PR rtl-optimization/79450
16044 * combine.c (distribute_notes): When removing TEM_INSN for which
16045 corresponding dest has last value recorded, invalidate that last
16048 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16050 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
16051 of explicit '@'. Add missing assembly comment marker on branch costs
16054 2017-02-13 Nathan Sidwell <nathan@acm.org>
16056 * gengtype-lex.l (<in_struct>): Add '/'.
16058 2017-02-13 Martin Liska <mliska@suse.cz>
16061 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
16063 2017-02-13 Richard Biener <rguenther@suse.de>
16065 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
16067 * configure: Re-generate.
16068 * config.in: Likewise.
16069 * graphite-dependences.c: Simplify as if
16070 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
16071 * graphite-isl-ast-to-gimple.c: Likewise.
16072 * graphite-optimize-isl.c: Likewise.
16073 * graphite-poly.c: Likewise.
16074 * graphite-sese-to-poly.c: Likewise.
16075 * graphite.h: Likewise.
16076 * toplev.c: Include isl/version.h and use isl_version () for
16077 printing the ISL version.
16078 * doc/install.texi: Update ISL requirement.
16080 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
16082 * doc/standards.texi (Standards): Update reference to
16085 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
16087 * doc/extend.texi (Named Address Spaces): sourceware.org now
16089 * doc/install.texi (Binaries): Ditto.
16092 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
16094 * doc/cpp.texi: Replace "stringify"/"stringification" with C
16095 standard terminology "stringize"/"stringizing" throughout.
16096 * doc/cppinternals.texi: Likewise.
16098 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
16100 * doc/extend.texi: Fix some spelling mistakes and typos.
16101 * doc/invoke.texi: Likewise.
16103 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
16106 * params.def (inline-min-speedup) Change from 10 to 8.
16108 2017-02-11 Jakub Jelinek <jakub@redhat.com>
16110 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
16113 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
16116 * ipa-inline-analysis.c (get_minimal_bb): New function.
16117 (record_modified): Use it.
16118 (remap_edge_change_prob): Handle also ancestor functions.
16120 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
16122 * doc/contrib.texi (Contributors): Remove broken link into
16123 the Mauve CVS repository.
16125 2017-02-11 Jakub Jelinek <jakub@redhat.com>
16127 PR middle-end/79454
16128 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
16129 result computation whenever lhs doesn't have vector mode, not
16130 just when it has BLKmode.
16132 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
16134 * doc/makefile.texi (profiledbootstrap): Refer to the
16135 installation instructions only in textual form.
16137 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16140 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
16142 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
16144 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
16145 (Specific): Update mingw-w64 reference.
16147 (Specific): Remove broken link to Renesas RX processor.
16149 2017-02-10 Richard Biener <rguenther@suse.de>
16151 * toplev.c (process_options): Do not mention obsolete graphite
16152 options when printing sorry message about missing graphite support.
16153 Mention -floop-nest-optimize.
16155 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
16157 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
16158 (vtst_p16): Likewise.
16159 (vtstq_p8): Likewise.
16160 (vtstq_p16): Likewise.
16162 (vtstq_p64): Likewise.
16163 * config/arm/arm_neon.h (vgetq_lane_p64): New.
16164 (vset_lane_p64): New.
16165 (vsetq_lane_p64): New.
16167 2017-02-10 Jakub Jelinek <jakub@redhat.com>
16169 PR tree-optimization/79411
16170 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
16171 stmt operands are SSA_NAMEs used in abnormal phis.
16172 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
16175 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
16178 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
16181 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
16183 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
16185 2017-02-09 Jakub Jelinek <jakub@redhat.com>
16187 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
16191 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
16192 not arbitrary TREE_CONSTANT.
16195 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
16196 "omp declare target link" attribute unless is_global_var.
16197 * omp-offload.c (find_link_var_op): Likewise.
16199 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
16200 Chung-Lin Tang <cltang@codesourcery.com>
16202 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
16204 (gimplify_adjust_omp_clauses): Don't delete TILE.
16205 (gimplify_omp_for): Deal with TILE.
16206 * internal-fn.c (expand_GOACC_TILE): New function.
16207 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
16209 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
16210 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
16212 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
16213 avoid DIV for outermost collapse var.
16214 (expand_oacc_for): Insert tile element loop as needed. Adjust.
16215 Remove out of date comments, fix whitespace.
16216 * omp-general.c (omp_extract_for_data): Deal with tiling.
16217 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
16218 adjust OLF_DIM_BASE value.
16219 (struct omp_for_data): Add tiling field.
16220 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
16221 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
16222 for auto loops. Remove default auto determining, moved to
16223 oacc_loop_fixed_partitions.
16224 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
16225 stmts, add e_mask field.
16226 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
16227 (oacc_thread_numbers): Use oacc_dim_call.
16228 (oacc_xform_tile): New.
16229 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
16230 (finish_oacc_loop): Adjust for ifns vector.
16231 (oacc_loop_discover_walk): Append loop abstraction sites to list,
16232 add case for GOACC_TILE fns.
16233 (oacc_loop_xform_loop): Delete.
16234 (oacc_loop_process): Iterate over call list directly, and add
16235 handling for GOACC_TILE fns.
16236 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
16238 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
16239 vector partitioning to outer loops. Assign 2 partitions to loops
16240 when available. Add TILE handling.
16241 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
16242 (execite_oacc_device_lower): Process GOACC_TILE fns,
16243 ignore unknown specs.
16244 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
16245 * tree.c (omp_clause_num_ops): Adjust TILE ops.
16246 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
16248 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
16250 * configure.ac (ACX_BUGURL): Update.
16251 * configure: Regenerate.
16253 2017-02-09 Richard Biener <rguenther@suse.de>
16255 PR tree-optimization/69823
16256 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
16257 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
16259 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
16261 * config/arc/arc-c.def: Add __NPS400__ definition.
16262 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
16263 (TARGET_NPS400): Define.
16265 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
16267 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
16269 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
16270 pointer, arch_info.
16271 (arc_cpu_types): Fill the arch_info field with a pointer into the
16272 arc_arch_types table.
16273 (arc_selected_cpu): Declare.
16274 * config/arc/arc.c (arc_selected_cpu): Make global.
16275 (arc_selected_arch): Delete.
16276 (arc_base_cpu): Delete.
16277 (arc_override_options): Remove references to deleted variables,
16278 update access to arch information.
16279 (ARC_OPT): Update access to arch information.
16280 (ARC_OPTX): Likewise.
16281 * config/arc/arc.h (arc_base_cpu): Remove declaration.
16282 (TARGET_ARC600): Update access to arch information.
16283 (TARGET_ARC601): Likewise.
16284 (TARGET_ARC700): Likewise.
16285 (TARGET_EM): Likewise.
16286 (TARGET_HS): Likewise.
16287 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
16290 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
16293 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
16294 condition/operands for integer GE/LE/GEU/LEU operations.
16296 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
16298 PR translation/79397
16299 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
16302 2017-02-08 Martin Jambor <mjambor@suse.cz>
16305 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
16306 whether allocation happened.
16307 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
16308 nothing was allocated.
16310 2017-02-08 Jakub Jelinek <jakub@redhat.com>
16312 PR tree-optimization/79408
16313 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
16314 constant, but SSA_NAME with a known integer range, use the minimum
16315 of that range instead of op1 to determine if modulo can be replaced
16316 with its first operand.
16318 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16320 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
16322 2017-02-08 Richard Biener <rguenther@suse.de>
16324 PR tree-optimization/71824
16325 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
16326 Check all loops contained in the merged region.
16328 2017-02-07 Andrew Pinski <apinski@cavium.com>
16330 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
16332 2017-02-07 Andrew Pinski <apinski@cavium.com>
16334 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
16335 (thunderxt88): Likewise.
16336 (thunderxt81): Disable LSE and change v8.1 to v8.
16337 (thunderxt83): Likewise.
16339 2017-02-07 Jakub Jelinek <jakub@redhat.com>
16340 Richard Biener <rguenther@suse.de>
16342 PR middle-end/79399
16343 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
16344 type from int to size_t.
16345 * ira-costs.c (struct_costs_size): Change type from int to size_t.
16347 2017-02-07 Jakub Jelinek <jakub@redhat.com>
16349 PR rtl-optimization/79386
16350 * cprop.c (bypass_conditional_jumps): Initialize
16351 bypass_last_basic_block already before splitting bbs after
16352 unconditional traps...
16353 (bypass_conditional_jumps): ... rather than here.
16356 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
16357 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
16358 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
16359 fix -masm=intel patterns.
16361 2017-02-07 Richard Biener <rguenther@suse.de>
16363 PR tree-optimization/79256
16364 PR middle-end/79278
16365 * builtins.c (get_object_alignment_2): Use min_align_of_type
16366 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
16367 and ADJUST_FIELD_ALIGN.
16369 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
16371 * doc/tm.texi: Regenerate.
16372 * stor-layout.c (layout_decl): Adjust.
16373 (update_alignment_for_field): Likewise.
16374 (place_field): Likewise.
16375 (min_align_of_type): Likewise.
16376 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
16377 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
16378 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
16379 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
16380 * config/frv/frv.c (frv_adjust_field_align): Likewise.
16381 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
16382 * config/i386/i386.c (x86_field_alignment): Likewise.
16383 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
16384 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
16385 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
16386 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
16387 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
16388 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
16392 2017-01-30 Richard Biener <rguenther@suse.de>
16394 PR tree-optimization/79256
16395 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16396 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16399 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
16401 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
16402 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
16403 builtins to SImode and emit a zero-extend, if necessary.
16405 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16407 * docs/invoke.texi (RISC-V Options): Alphabetize.
16409 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16411 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
16414 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16416 * config/riscv/riscv.c: New file.
16417 * common/config/riscv/riscv-common.c: Likewise.
16418 * config.gcc: Likewise.
16419 * config/riscv/constraints.md: Likewise.
16420 * config/riscv/elf.h: Likewise.
16421 * config/riscv/generic.md: Likewise.
16422 * config/riscv/linux.h: Likewise.
16423 * config/riscv/multilib-generator: Likewise.
16424 * config/riscv/peephole.md: Likewise.
16425 * config/riscv/pic.md: Likewise.
16426 * config/riscv/predicates.md: Likewise.
16427 * config/riscv/riscv-builtins.c: Likewise.
16428 * config/riscv/riscv-c.c: Likewise.
16429 * config/riscv/riscv-ftypes.def: Likewise.
16430 * config/riscv/riscv-modes.def: Likewise.
16431 * config/riscv/riscv-opts.h: Likewise.
16432 * config/riscv/riscv-protos.h: Likewise.
16433 * config/riscv/riscv.h: Likewise.
16434 * config/riscv/riscv.md: Likewise.
16435 * config/riscv/riscv.opt: Likewise.
16436 * config/riscv/sync.md: Likewise.
16437 * config/riscv/t-elf-multilib: Likewise.
16438 * config/riscv/t-linux: Likewise.
16439 * config/riscv/t-linux-multilib: Likewise.
16440 * config/riscv/t-riscv: Likewise.
16441 * configure.ac: Likewise.
16442 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
16443 Waterman as RISC-V maintainers.
16444 * doc/install.texi: Add RISC-V entries.
16445 * doc/invoke.texi: Add RISC-V options section.
16446 * doc/md.texi: Add RISC-V constraints section.
16447 * configure: Regenerated.
16449 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
16452 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
16453 false values to be constant vectors with all 0 or all 1 bits set.
16454 (vcondu<mode><mode>): Likewise.
16455 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
16457 (fpmask_comparison_operator): Update comment.
16458 (vecint_comparison_operator): New predicate.
16459 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
16460 vector conditionals when the true and false values are constant
16461 vectors with all 0 bits or all 1 bits set.
16463 2017-02-06 Martin Sebor <msebor@redhat.com>
16465 PR tree-optimization/79376
16466 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
16468 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
16470 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
16471 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
16472 to simplify split condition.
16474 2017-02-06 Jakub Jelinek <jakub@redhat.com>
16476 * omp-expand.c (oxpand_omp_atomic_fetch_op,
16477 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
16480 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
16482 PR rtl-optimization/68664
16483 * target.def (can_speculate_insn): New hook.
16484 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
16485 * doc/tm.texi: Regenerate.
16486 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
16487 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
16488 (rs6000_sched_can_speculate_insn): New function.
16490 2017-02-06 Jakub Jelinek <jakub@redhat.com>
16492 PR tree-optimization/79284
16493 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
16494 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
16495 vectorizable_mask_load_store, vectorizable_operation,
16496 vect_is_simple_cond, get_same_sized_vectype): Use it instead
16497 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
16498 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
16499 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
16500 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
16501 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
16502 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
16503 is_gimple_assign (stmt). Replace another such test with
16504 is_gimple_assign (stmt).
16506 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
16509 * config/avr/avr.c (rtl-iter.h): Include it.
16510 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
16511 (avr_legitimate_combined_insn): ...and implementation.
16513 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
16515 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
16516 * config/s390/s390.c (s390_const_operand_ok)
16517 (s390_canonicalize_comparison, s390_extract_part)
16518 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
16519 (s390_contiguous_bitmask_p, s390_rtx_costs)
16520 (legitimize_pic_address): Likewise.
16521 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
16522 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
16523 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
16524 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
16525 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
16527 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
16529 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
16530 REGNO($0) == REGNO($1).
16532 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16534 * config/s390/linux.h(SIZE_TYPE): Add comment.
16536 2017-02-06 Julian Brown <julian@codesourcery.com>
16537 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16538 Virendra Pathak <virendra.pathak@broadcom.com>
16540 * config/aarch64/aarch64-cores.def: Change the scheduler
16542 * config/aarch64/aarch64.md: Include thunderx2t99.md.
16543 * config/aarch64/thunderx2t99.md: New file.
16545 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16547 * doc/standards.texi (Go Language): Update link to language
16550 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
16552 * tree-eh.c (lower_resx): Sanitize profile.
16553 (cleanup_empty_eh_move_lp): Likewise.
16555 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
16558 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
16560 * cfgloopmanip.h (loop_version): Update prototype.
16561 * modulo-sched.c (sms_schedule): Update call of loop_version.
16562 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
16563 * tree-parloops.c (gen_parallel_loop): Likewise.
16564 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
16565 * tree-ssa-loop-split.c (split_loop): Likewise.
16566 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
16567 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
16569 2017-02-05 Martin Liska <mliska@suse.cz>
16572 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
16574 (print_operand_address): Initialize a struct to zero.
16576 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16578 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
16579 garbage collector only in textual form.
16581 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16583 * doc/extend.texi (x86 specific memory model extensions for
16584 transactional memory): Simplify a phrase.
16586 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
16589 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
16590 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
16591 (atomic_storedi_1): Likewise.
16593 2017-02-04 Jakub Jelinek <jakub@redhat.com>
16595 PR tree-optimization/79338
16596 * tree-parloops.c (gather_scalar_reductions): Don't call
16597 vect_analyze_loop_form for loop->inner before destroying loop's
16600 2017-02-03 Martin Sebor <msebor@redhat.com>
16602 PR tree-optimization/79327
16603 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
16604 when precision has resulted in leading zeros.
16605 (format_integer): Adjust the likely counter to assume an unknown
16606 argument that may be zero is non-zero.
16608 2017-02-03 Jason Merrill <jason@redhat.com>
16611 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
16612 avoid copying non-taken branch.
16614 2017-02-03 Jakub Jelinek <jakub@redhat.com>
16616 PR tree-optimization/79340
16617 * tree-vect-loop.c (vectorizable_reduction): Release
16618 vec_defs elements after safe_splicing them into other vectors.
16621 PR tree-optimization/79327
16622 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
16623 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
16625 (format_integer): Use wide_int_to_tree instead of build_int_cst
16626 + to_?hwi. If argmin is NULL, just set argmin and argmax to
16627 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
16628 of shortest and longest sequence.
16630 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
16632 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
16633 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
16635 2017-02-03 Walter Lee <walt@tilera.com>
16638 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
16639 after initial stackframe link reg save.
16640 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
16642 2017-02-03 Jakub Jelinek <jakub@redhat.com>
16645 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
16646 wu for stxssp alternative.
16648 2017-02-03 Martin Sebor <msebor@redhat.com>
16650 PR tree-optimization/79352
16651 * gimple-fold.c (get_range_strlen): Add argument.
16652 (get_range_strlen): Change return type to bool.
16653 (get_maxval_strlen): Pass in a dummy argument.
16654 * gimple-fold.h (get_range_strlen): Change return type to bool.
16655 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
16656 * tree.h (array_at_struct_end_p): Add argument.
16657 * tree.c (array_at_struct_end_p): Handle it.
16659 2017-02-03 Martin Liska <mliska@suse.cz>
16662 * multiple_target.c (create_dispatcher_calls): Redirect edge
16663 from a caller of a dispatcher.
16664 (expand_target_clones): Make the clones local.
16665 (ipa_target_clone): Do both target clones and resolvers.
16666 (ipa_dispatcher_calls): Remove the pass.
16667 (pass_dispatcher_calls::gate): Likewise.
16668 (make_pass_dispatcher_calls): Likewise.
16669 * passes.def (pass_target_clone): Put as very first IPA early
16672 2017-02-03 Martin Liska <mliska@suse.cz>
16674 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
16675 in case of a function with ifunc attribute.
16677 2017-02-03 Martin Liska <mliska@suse.cz>
16679 * cgraph.c (cgraph_node::dump): Dump function version info.
16680 * symtab.c (symtab_node::dump_base): Add missing new line.
16682 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
16684 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
16685 (ifcombine_ifandif): Use it.
16687 2017-02-03 Martin Liska <mliska@suse.cz>
16689 * doc/invoke.texi: Document default value for
16690 use-after-scope-direct-emission-threshold.
16692 2017-02-03 Martin Liska <mliska@suse.cz>
16694 PR tree-optimization/79339
16695 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
16696 (format_floating): Likewise.
16698 2017-02-03 Martin Liska <mliska@suse.cz>
16701 * ipa-prop.c (ipa_node_params_t::insert): Remove current
16703 (ipa_node_params_t::remove): Likewise.
16704 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
16705 initialization from removed ipa_node_params_t::insert.
16706 (ipa_node_params::~ipa_node_params): Move from removed
16707 ipa_node_params_t::release.
16708 * symbol-summary.h (symbol_summary::m_released): New member.
16709 Do not release a summary twice. Do not allow to call finalizer
16710 for types of a summary that live in GGC memory.
16712 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16714 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
16717 2017-02-02 Martin Sebor <msebor@redhat.com>
16719 PR middle-end/79275
16720 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
16721 (format_string): Tighten up the range of output for non-constant
16722 strings and correct the expected range for wide non-constant strings.
16724 2017-02-02 Martin Sebor <msebor@redhat.com>
16726 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
16728 PR middle-end/32003
16729 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
16731 (-fdump-tree-@var): Add to index and document how to come up
16732 with pass-specific option and dump file names.
16733 (-fdump-passes): Clarify where to look for output.
16735 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
16737 PR middle-end/77445
16738 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
16739 statistics of the analyzed path; allow threading for speed when
16740 any of BBs along the path are optimized for speed.
16742 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
16744 PR middle-end/78468
16745 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
16746 settings of the virtual registers.
16749 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
16751 * explow.c (get_dynamic_stack_size): Take known alignment of stack
16752 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
16755 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16757 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
16758 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
16760 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16762 * config/s390/s390.md: Add missing comments with the expanded
16764 * config/s390/vector.md: Likewise.
16765 * config/s390/vx-builtins.md: Likewise.
16767 2017-02-02 Jakub Jelinek <jakub@redhat.com>
16770 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
16771 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
16772 conditions on a single line.
16774 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16776 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16777 __S390_VX__ to __VX__.
16779 2017-02-01 Andrew Pinski <apinski@cavium.com>
16781 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
16782 stmt_info to record_stmt_cost.
16783 (vect_get_known_peeling_cost): Pass stmt_info if known to
16785 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
16786 cpu_vector_cost field into
16787 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
16788 field into vec_int_stmt_cost and vec_fp_stmt_cost.
16789 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
16790 splitting of scalar_stmt_cost and vec_stmt_cost.
16791 (thunderx_vector_cost): Likewise.
16792 (cortexa57_vector_cost): LIkewise.
16793 (exynosm1_vector_cost): Likewise.
16794 (xgene1_vector_cost): Likewise.
16795 (thunderx2t99_vector_cost): Improve after the splitting of the two
16797 (aarch64_builtin_vectorization_cost): Update for the splitting of
16798 scalar_stmt_cost and vec_stmt_cost.
16800 2017-02-01 Torvald Riegel <triegel@redhat.com>
16801 Richard Henderson <rth@redhat.com>
16803 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
16804 conditional on existance of a fast atomic load.
16805 * optabs-query.c (can_atomic_load_p): New function.
16806 * optabs-query.h (can_atomic_load_p): Declare it.
16807 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
16808 no fast atomic load is available for the particular size of access.
16809 (expand_atomic_compare_and_swap): Likewise.
16810 (expand_atomic_load): Likewise.
16811 (expand_atomic_store): Likewise.
16812 (expand_atomic_fetch_op): Likewise.
16813 * testsuite/lib/target-supports.exp
16814 (check_effective_target_sync_int_128): Remove x86 because it provides
16815 no fast atomic load.
16816 (check_effective_target_sync_int_128_runtime): Likewise.
16818 2017-02-01 Richard Biener <rguenther@suse.de>
16820 * graphite.c: Include tree-vectorizer.h for find_loop_location.
16821 (graphite_transform_loops): Provide opt-info for optimized nests.
16822 * tree-parloop.c (parallelize_loops): Provide opt-info for
16823 parallelized loops.
16825 2017-02-01 Richard Biener <rguenther@suse.de>
16827 PR middle-end/79315
16828 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
16829 was not set before.
16831 2017-02-01 Richard Biener <rguenther@suse.de>
16833 PR tree-optimization/71824
16834 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
16835 Verify the loops are valid in the merged SESE region.
16836 (scop_detection::can_represent_loop_1): Check analyzing the
16837 evolution of the number of iterations in the region succeeds.
16839 2017-01-31 Ian Lance Taylor <iant@golang.org>
16841 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
16842 REG_ARGS_SIZE note to 32-bit push insns and call insn.
16844 2017-01-31 David Malcolm <dmalcolm@redhat.com>
16846 PR preprocessor/79210
16847 * input.c (get_substring_ranges_for_loc): Replace line_width
16848 assertion with error-handling.
16850 2017-01-31 Richard Biener <rguenther@suse.de>
16852 PR tree-optimization/77318
16853 * graphite-sese-to-poly.c (extract_affine): Fix assert.
16854 (create_pw_aff_from_tree): Take loop parameter.
16855 (add_condition_to_pbb): Pass loop of the condition to
16856 create_pw_aff_from_tree.
16858 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16860 * config/s390/s390.c (s390_asan_shadow_offset): New function.
16861 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
16863 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
16867 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
16869 (convert_int_to_float128): Likewise.
16870 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
16871 (convert_int_to_float128): Likewise.
16872 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
16873 (UNSPEC_IEEE128_CONVERT): Likewise.
16874 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
16875 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
16876 Use local variables for IBM extended format.
16877 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
16878 (fix_trunc<mode>si2_fprs): Likewise.
16879 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
16880 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
16881 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
16882 to know that we can now have integers of all sizes in vector
16884 (fix<uns>_<mode>di2_hw): Likewise.
16885 (float<uns>_<mode>si2_hw): Likewise.
16886 (fix_<mode>si2_hw): Likewise.
16887 (fixuns_<mode>si2_hw): Likewise.
16888 (float<uns>_<mode>di2_hw): Likewise.
16889 (float_<mode>di2_hw): Likewise.
16890 (float_<mode>si2_hw): Likewise.
16891 (floatuns_<mode>di2_hw): Likewise.
16892 (floatuns_<mode>si2_hw): Likewise.
16893 (xscvqp<su>wz_<mode>): Delete, no longer used.
16894 (xscvqp<su>dz_<mode>): Likewise.
16895 (xscv<su>dqp_<mode>): Likewise.
16896 (ieee128_mfvsrd_64bit): Likewise.
16897 (ieee128_mfvsrd_32bit): Likewise.
16898 (ieee128_mfvsrwz): Likewise.
16899 (ieee128_mtvsrw): Likewise.
16900 (ieee128_mtvsrd_64bit): Likewise.
16901 (ieee128_mtvsrd_32bit): Likewise.
16903 2017-01-31 Martin Liska <mliska@suse.cz>
16906 * ipa-prop.c (ipa_free_all_node_params): Call release method
16907 instead of ~sumbol_summary to not to trigger double times
16910 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
16912 PR tree-optimization/71691
16913 * bitmap.h (class auto_bitmap): New.
16914 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
16915 is_maybe_undefined instead of ssa_undefined_value_p.
16917 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16919 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16920 __S390_ARCH_LEVEL__ to __ARCH__.
16922 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16924 PR tree-optimization/79267
16925 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
16926 if should_remove_lhs_p is true.
16928 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
16931 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
16932 (add_alignment_attribute): New.
16933 (base_type_die): Add alignment attribute.
16934 (subrange_type_die): Likewise.
16935 (modified_type_die): Likewise.
16936 (gen_array_type_die): Likewise.
16937 (gen_descr_array_type_die: Likewise.
16938 (gen_enumeration_type_die): Likewise.
16939 (gen_subprogram_die): Likewise.
16940 (gen_variable_die): Likewise.
16941 (gen_field_die): Likewise.
16942 (gen_ptr_to_mbr_type_die): Likewise.
16943 (gen_struct_or_union_type_die): Likewise.
16944 (gen_subroutine_type_die): Likewise.
16945 (gen_typedef_die): Likewise.
16946 (base_type_cmp): Compare alignment attribute.
16948 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16951 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
16952 (setb_unsigned) New pattern for setb with CCUNS.
16953 * config/rs6000/rs6000.c (expand_block_compare): Use a different
16954 subfc./subfe sequence to avoid overflow problems. Generate a
16955 shorter sequence with cmpld/setb for power9.
16956 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
16957 for generating subfc. instruction.
16958 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
16959 now uses this instruction.
16961 2017-01-30 Ian Lance Taylor <iant@google.com>
16964 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
16965 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
16967 2017-01-30 Martin Sebor <msebor@redhat.com>
16969 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
16970 Move constant to the right of a relational operator.
16971 (get_mpfr_format_length, format_character, format_string): Ditto.
16972 (should_warn_p, maybe_warn): Same.
16974 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
16976 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
16979 * asan.c (get_translation_unit_decl): Remove function.
16980 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
16982 2017-01-30 Martin Liska <mliska@suse.cz>
16984 PR gcov-profile/79259
16985 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
16986 -fprofile-generate.
16988 2017-01-30 Martin Liska <mliska@suse.cz>
16991 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
16992 Initialize variables with NULL value.
16994 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
16997 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
16999 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
17001 2017-01-30 Richard Biener <rguenther@suse.de>
17003 PR tree-optimization/79276
17004 * tree-vrp.c (process_assert_insertions): Properly adjust common
17005 when removing a duplicate.
17007 * gcc.dg/torture/pr79276.c: New testcase.
17009 2017-01-30 Richard Biener <rguenther@suse.de>
17011 PR tree-optimization/79256
17012 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
17013 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
17015 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
17017 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
17020 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
17021 ("*r<noxa>sbg_<mode>_sll_bitmask")
17022 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
17023 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
17024 Use contiguous_bitmask_nowrap_operand.
17026 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17029 * config/rs6000/altivec.h (vec_xl): Revise #define.
17030 (vec_xst): Likewise.
17032 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
17034 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
17036 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
17038 PR rtl-optimization/79194
17039 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
17040 traps before call to bypass_conditional_jumps.
17042 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
17044 PR tree-optimization/71374
17045 * lra-constraints.c (check_conflict_input_operands): New.
17046 (match_reload): Use it.
17048 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
17051 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
17052 account to calculate conflict_set.
17054 2017-01-27 Bin Cheng <bin.cheng@arm.com>
17056 PR rtl-optimization/78559
17057 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
17058 other_insn in combine.
17060 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17062 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
17063 uint16_type_node for BT_UINT16.
17065 2017-01-27 David Malcolm <dmalcolm@redhat.com>
17067 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
17068 "RTL Tests" to menu.
17069 (GIMPLE Tests): New node.
17070 (RTL Tests): New node.
17072 2017-01-27 Richard Biener <rguenther@suse.de>
17074 PR tree-optimization/79245
17075 * tree-loop-distribution.c (distribute_loop): Apply cost
17076 modeling also to detected patterns.
17078 2017-01-27 Richard Biener <rguenther@suse.de>
17080 PR tree-optimization/71433
17081 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
17082 (compare_assert_loc): New function.
17083 (process_assert_insertions): Sort and optimize assert locations
17084 to remove duplicates and push down identical assertions on
17085 edges to their destination block.
17087 2017-01-27 Richard Biener <rguenther@suse.de>
17089 PR tree-optimization/79244
17090 * tree-vrp.c (remove_range_assertions): Forcefully propagate
17091 out SSA names even if abnormal.
17093 2017-01-27 Jakub Jelinek <jakub@redhat.com>
17095 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
17096 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
17097 instead of MPFR_RNDN.
17099 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
17102 * arm.c (arm_option_override): Don't call build_target_option_node
17103 until after doing all option overrides.
17104 (arm_valid_target_attribute_tree): Likewise.
17106 2017-01-27 Martin Liska <mliska@suse.cz>
17108 * doc/invoke.texi (-fprofile-arcs): Document profiling support
17109 for {cd}tors and C++ {cd}tors.
17111 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
17113 * config/s390/s390.md ("*setmem_long_and")
17114 ("*setmem_long_and_31z"): Use zero_extend instead of and.
17116 2017-01-26 Martin Sebor <msebor@redhat.com>
17118 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
17121 2017-01-26 Martin Sebor <msebor@redhat.com>
17123 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
17124 HAVE_DFmode before using XFmode or DFmode.
17125 (parse_directive): Avoid using the z length modifier to avoid
17126 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
17128 PR middle-end/78703
17129 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
17130 to accept adjustment as an array.
17131 (get_int_range): New function.
17132 (struct directive): Make width and prec arrays.
17133 (directive::set_width, directive::set_precision): Call get_int_range.
17134 (format_integer, format_floating): Handle width and precision ranges.
17135 (format_string, parse_directive): Same.
17137 2017-01-26 Jakub Jelinek <jakub@redhat.com>
17140 * dwarf2out.c (generate_skeleton_bottom_up): For children with
17141 comdat_type_p set, just clone them, but keep the children in the
17145 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
17146 which have direct callers with -fvar-tracking-assignments enabled
17148 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
17149 inside of type units.
17151 2017-01-26 Martin Sebor <msebor@redhat.com>
17153 PR middle-end/78703
17154 * gimple-ssa-sprintf.c (struct result_range): Add likely and
17156 (struct format_result): Replace number_chars, number_chars_min,
17157 and number_chars_max with a single member of struct result_range.
17159 (format_result::operator+=): Adjust.
17160 (struct fmtresult): Remove bounded. Handle likely and unlikely
17162 (fmtresult::adjust_for_width_or_precision): New function.
17163 (fmtresult:type_max_digits): New function.
17164 (bytes_remaining): Handle likely and unlikely counters.
17165 (min_bytes_remaining): Remove.
17166 (format_percent): Simplify.
17167 (format_integer, format_floating): Set likely and unlikely counters.
17168 (get_string_length, format_character, format_string): Same.
17169 (format_plain, should_warn_p): New function.
17170 (maybe_warn): Call should_warn_p. Update diagnostic messages
17171 and handle those for all directives, including plain strings.
17172 (format_directive): Handle likely and unlikely counters.
17173 Remove unnecessary quoting from diagnostics. Add an informational
17175 (add_bytes): Remove.
17176 (pass_sprintf_length::compute_format_length): Simplify.
17177 (try_substitute_return_value): Handle likely and unlikely counters.
17179 2017-01-26 Carl Love <cel@us.ibm.com>
17181 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
17182 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
17184 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
17187 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
17188 endianess for subregs into account.
17189 * lra-constraints.c (lra_constraints): Do risky transformations
17190 always on the first iteration.
17191 * lra-lives.c (check_pseudos_live_through_calls): Add arg
17192 last_call_used_reg_set.
17193 (process_bb_lives): Define and use last_call_used_reg_set.
17194 * lra.c (lra): Always continue after lra_constraints on the first
17197 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
17199 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
17201 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
17203 2017-01-26 Jakub Jelinek <jakub@redhat.com>
17205 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
17206 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
17207 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
17208 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
17209 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
17210 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
17211 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
17212 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
17213 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
17215 2017-01-26 Marek Polacek <polacek@redhat.com>
17218 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
17219 for the third operand.
17221 2017-01-26 Jakub Jelinek <jakub@redhat.com>
17223 PR middle-end/79236
17224 * omp-low.c (struct omp_context): Add simt_stmt field.
17225 (scan_omp_for): Return omp_context *.
17226 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
17227 context to the _simt_ SIMD stmt.
17228 (lower_omp_for): For combined SIMD with sibling _simt_
17229 SIMD, make sure to use the same decls in _looptemp_
17230 clauses as in the sibling.
17232 2017-01-26 David Sherwood <david.sherwood@arm.com>
17234 PR middle-end/79212
17235 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
17238 2017-01-26 Jakub Jelinek <jakub@redhat.com>
17241 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
17242 emit fld b; fld a; if possible.
17244 * brig-builtins.def: Update copyright years.
17245 * config/arm/arm_acle_builtins.def: Update copyright years.
17247 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
17250 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
17251 constraint instead of o for the stxsd instruction.
17253 2017-01-25 Carl Love <cel@us.ibm.com>
17255 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
17256 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
17258 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
17260 * doc/invoke.texi (C++ Dialect Options): Fix typo.
17262 2017-01-25 Richard Biener <rguenther@suse.de>
17264 PR tree-optimization/69264
17265 * target.def (vector_alignment_reachable): Improve documentation.
17266 * doc/tm.texi: Regenerate.
17267 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
17269 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
17270 earlier changes with respect to TYPE_USER_ALIGN.
17271 (vector_alignment_reachable_p): Likewise. Improve dumping.
17273 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17276 * config/arm/arm.md (xordi3): Force constant operand into a register
17279 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17281 * doc/invoke.texi (-fstore-merging): Correct default optimization
17282 levels at which it is enabled.
17283 (-O): Move -fstore-merging from list to...
17286 2017-01-25 Richard Biener <rguenther@suse.de>
17289 * omp-expand.c: Include debug.h.
17290 (expand_omp_taskreg): Make sure to generate early debug before
17291 outlining anything from a function.
17292 (expand_omp_target): Likewise.
17293 (grid_expand_target_grid_body): Likewise.
17295 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
17298 * asan.c (get_translation_unit_decl): New function.
17299 (asan_add_global): Extract modules file name from globals
17300 TRANSLATION_UNIT_DECL name.
17302 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
17305 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
17306 for long calls with APCS frame and VFP.
17308 2017-01-24 David Malcolm <dmalcolm@redhat.com>
17310 * cfg.c (original_copy_tables_initialized_p): New function.
17311 * cfg.h (original_copy_tables_initialized_p): New decl.
17312 * cfgrtl.c (relink_block_chain): Guard the call to
17313 free_original_copy_tables with a call to
17314 original_copy_tables_initialized_p.
17315 * cgraph.h (symtab_node::native_rtl_p): New decl.
17316 * cgraphunit.c (symtab_node::native_rtl_p): New function.
17317 (symtab_node::needed_p): Don't assert for early assembly output
17318 for __RTL functions.
17319 (cgraph_node::finalize_function): Set "force_output" for __RTL
17321 (cgraph_node::analyze): Bail out early for __RTL functions.
17322 (analyze_functions): Update assertion to support __RTL functions.
17323 (cgraph_node::expand): Bail out early for __RTL functions.
17324 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
17326 * function.h (struct function): Update comment for field
17328 * gimple-expr.c: Include "tree-pass.h".
17329 (gimple_has_body_p): Return false for __RTL functions.
17330 * Makefile.in (OBJS): Add run-rtl-passes.o.
17331 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
17333 (gcc::pass_manager::get_clean_slate): New accessor.
17334 * passes.c: Include "insn-addr.h".
17335 (should_skip_pass_p): Add logging. Update logic for running
17336 "expand" to be compatible with both __GIMPLE and __RTL. Guard
17337 property-provider override so it is only done for gimple passes.
17339 (skip_pass): New function.
17340 (execute_one_pass): Call skip_pass when skipping passes.
17341 * read-md.c (md_reader::read_char): Support filtering
17342 the input to a subset of line numbers.
17343 (md_reader::md_reader): Initialize fields
17344 m_first_line and m_last_line.
17345 (md_reader::read_file_fragment): New function.
17346 * read-md.h (md_reader::read_file_fragment): New decl.
17347 (md_reader::m_first_line): New field.
17348 (md_reader::m_last_line): New field.
17349 * read-rtl-function.c (function_reader::create_function): Only
17350 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
17351 curr_properties. Set DECL_INITIAL to a dummy block.
17352 (read_rtl_function_body_from_file_range): New function.
17353 * read-rtl-function.h (read_rtl_function_body_from_file_range):
17355 * run-rtl-passes.c: New file.
17356 * run-rtl-passes.h: New file.
17358 2017-01-24 Jeff Law <law@redhat.com>
17360 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
17363 2017-01-24 Bin Cheng <bin.cheng@arm.com>
17365 PR tree-optimization/79159
17366 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
17367 (record_nonwrapping_iv): Improve boundary using above function if no
17368 value range information.
17370 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
17371 Martin Jambor <mjambor@suse.cz>
17373 * brig-builtins.def: New file.
17374 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
17375 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
17376 (DEF_HSAIL_SAT_BUILTIN): Likewise.
17377 (DEF_HSAIL_INTR_BUILTIN): Likewise.
17378 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
17379 * builtin-types.def (BT_INT8): New.
17380 (BT_INT16): Likewise.
17381 (BT_UINT8): Likewise.
17382 (BT_UINT16): Likewise.
17383 (BT_FN_ULONG): Likewise.
17384 (BT_FN_UINT_INT): Likewise.
17385 (BT_FN_UINT_ULONG): Likewise.
17386 (BT_FN_UINT_LONG): Likewise.
17387 (BT_FN_UINT_PTR): Likewise.
17388 (BT_FN_ULONG_PTR): Likewise.
17389 (BT_FN_INT8_FLOAT): Likewise.
17390 (BT_FN_INT16_FLOAT): Likewise.
17391 (BT_FN_UINT32_FLOAT): Likewise.
17392 (BT_FN_UINT16_FLOAT): Likewise.
17393 (BT_FN_UINT8_FLOAT): Likewise.
17394 (BT_FN_UINT64_FLOAT): Likewise.
17395 (BT_FN_UINT16_UINT32): Likewise.
17396 (BT_FN_UINT32_UINT16): Likewise.
17397 (BT_FN_UINT16_UINT16_UINT16): Likewise.
17398 (BT_FN_INT_PTR_INT): Likewise.
17399 (BT_FN_UINT_PTR_UINT): Likewise.
17400 (BT_FN_LONG_PTR_LONG): Likewise.
17401 (BT_FN_ULONG_PTR_ULONG): Likewise.
17402 (BT_FN_VOID_UINT64_UINT64): Likewise.
17403 (BT_FN_UINT8_UINT8_UINT8): Likewise.
17404 (BT_FN_INT8_INT8_INT8): Likewise.
17405 (BT_FN_INT16_INT16_INT16): Likewise.
17406 (BT_FN_INT_INT_INT): Likewise.
17407 (BT_FN_UINT_FLOAT_UINT): Likewise.
17408 (BT_FN_FLOAT_UINT_UINT): Likewise.
17409 (BT_FN_ULONG_UINT_UINT): Likewise.
17410 (BT_FN_ULONG_UINT_PTR): Likewise.
17411 (BT_FN_ULONG_ULONG_ULONG): Likewise.
17412 (BT_FN_UINT_UINT_UINT): Likewise.
17413 (BT_FN_VOID_UINT_PTR): Likewise.
17414 (BT_FN_UINT_UINT_PTR: Likewise.
17415 (BT_FN_UINT32_UINT64_PTR): Likewise.
17416 (BT_FN_INT_INT_UINT_UINT): Likewise.
17417 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
17418 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
17419 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
17420 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
17421 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
17422 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
17423 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
17424 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
17425 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
17426 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
17427 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
17428 * doc/frontends.texi: List BRIG FE.
17429 * doc/install.texi (Testing): Add BRIG tesring requirements.
17430 * doc/invoke.texi (Overall Options): Mention BRIG.
17431 * doc/standards.texi (Standards): Doucment BRIG HSA version.
17433 2017-01-24 Richard Biener <rguenther@suse.de>
17435 PR translation/79208
17436 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
17438 2017-01-24 Martin Jambor <mjambor@suse.cz>
17441 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
17442 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
17443 and known_contexts.
17445 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
17447 PR middle-end/79123
17448 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
17449 casts from signed to unsigned really don't have a range.
17451 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
17453 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
17454 GMP_RNDx for compatiblity.
17456 2017-01-24 Martin Liska <mliska@suse.cz>
17459 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
17460 that would prevent us to call alloca with -1 as argument.
17462 2017-01-24 Jakub Jelinek <jakub@redhat.com>
17464 * dwarf2out.c (output_compilation_unit_header, output_file_names):
17465 Avoid -Wformat-security warning.
17467 2017-01-23 Andrew Pinski <apinski@cavium.com>
17469 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
17472 2017-01-23 Martin Sebor <msebor@redhat.com>
17474 PR middle-end/78703
17475 * gimple-ssa-sprintf.c (warn_level): New global.
17476 (format_integer): Use it here and throughout the rest of the file.
17477 Use the same switch to compute sign as base.
17478 (maybe_warn): New function.
17479 (format_directive): Factor out warnings into maybe_warn.
17480 Add debugging output. Use warn_level.
17481 (add_bytes): Use warn_level.
17482 (pass_sprintf_length::compute_format_length): Add debugging output.
17483 (try_substitute_return_value): Same.
17484 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
17486 PR middle-end/78703
17487 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
17488 (struct fmtresult, format_integer, format_floating): Adjust.
17489 (fmtresult::fmtresult): Set max correctly in two argument ctor.
17490 (get_string_length, format_string,format_directive): Same.
17491 (pass_sprintf_length::compute_format_length): Same.
17492 (try_substitute_return_value): Simplify slightly.
17494 PR middle-end/78703
17495 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
17496 (fmtresult::operator+=): Outlined.
17497 (struct fmtresult): Add ctors.
17498 (struct conversion_spec): Rename...
17499 (struct directive): ...to this. Add and remove data members.
17500 (directive::set_width, directive::set_precision): New functions.
17501 (format_percent): Use fmtresult ctor.
17502 (get_width_and_precision): Remove.
17503 (format_integer): Make naming changes. Avoid computing width and
17505 (format_floating): Same. Adjust indentation.
17506 (format_character, format_none): New functions.
17507 (format_string): Moved character handling to format_character.
17508 (format_directive): Remove arguments, change return type.
17509 (parse_directive): New function.
17510 (pass_sprintf_length::compute_format_length): Move directive
17511 parsing to parse_directive.
17513 2017-01-23 Jakub Jelinek <jakub@redhat.com>
17515 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
17516 (assign_assembler_name_if_needed): ... this.
17517 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
17518 (assign_assembler_name_if_needed): ... this.
17519 (free_lang_data_in_cgraph): Adjust callers.
17520 * cgraphunit.c (cgraph_node::analyze): Likewise.
17521 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
17524 2017-01-23 Richard Biener <rguenther@suse.de>
17526 PR tree-optimization/79088
17527 PR tree-optimization/79188
17528 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
17529 resetting loop bounds after last path deletion. Reset loop
17530 bounds of the target loop, make code match the comments.
17531 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
17532 Make sure loops need no fixups.
17534 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
17536 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
17537 exponent support with double type for first argument.
17538 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
17539 type returned by __builtin_vec_extract_sig,
17540 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
17541 functions from "vector int" to "vector unsigned int" or from
17542 "vector long long int" to "vector unsigned long long int".
17543 Changed type returned by __builtin_vec_extract_exp,
17544 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
17545 functions from "vector int" to "vector unsigned int" or from
17546 "vector long long int" to "vector unsigned long long int".
17547 Changed return type of __builtin_vec_test_data_class,
17548 __builtin_vec_test_data_class_sp, and
17549 __builtin_vec_test_data_class_dp from "vector int" to
17550 "vector bool int" or from "vector long long int" to "vector bool
17551 long long int" and changed second argument type from "unsigned
17552 int" to "int". Added new overloaded function forms "vector float
17553 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
17554 "vector float __builtin_vec_insert_exp_sp (vector float, vector
17555 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
17556 double, vector unsigned long long int)" and "vector double
17557 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
17558 long int)". Changed return type of
17559 __builtin_scalar_test_data_class and
17560 __builtin_scalar_test_data_class_sp and
17561 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
17562 int" and changed second argument from "unsigned int" to "int".
17563 Changed type returned by __builtin_scalar_test_neg,
17564 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
17565 from "int" to "bool int". Added new overloaded function form
17566 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
17567 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
17568 exponent double-precision with floating point first argument.
17569 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
17570 documentation of scalar_test_data_class, scalar_test_neg,
17571 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
17572 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
17573 vec_test_data_class built-in functions to reflect refinements in
17574 their type signatures.
17576 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
17578 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
17580 (aarch64_elf_asm_destructor): Likewise.
17582 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
17584 PR rtl-optimization/78634
17585 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
17586 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
17587 * ifcvt.c (noce_try_cmove): Add missing cost check.
17589 PR rtl-optimization/71724
17590 * combine.c (if_then_else_cond): Look for situations where it is
17591 beneficial to undo the work of one of the recursive calls.
17593 2017-01-23 Bin Cheng <bin.cheng@arm.com>
17595 PR tree-optimization/70754
17596 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
17597 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
17598 combined stmt before it if not NULL.
17599 (combine_chains): Process refs reversely and compute dominance point
17602 2017-01-23 Martin Liska <mliska@suse.cz>
17604 PR tree-optimization/79196
17605 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
17606 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
17608 (strlen_optimize_stmt): Call the renamed function.
17610 2017-01-23 Michael Matz <matz@suse.de>
17612 PR tree-optimization/78384
17613 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
17615 2017-01-23 Richard Biener <rguenther@suse.de>
17617 PR tree-optimization/79186
17618 * tree-vrp.c (register_new_assert_for): Make sure we've seen
17619 both incoming edges before moving an assert.
17621 2017-01-23 Martin Jambor <mjambor@suse.cz>
17623 * ipa-prop.c (load_from_param_1): Removed.
17624 (load_from_unmodified_param): Bits from load_from_param_1 put back
17626 (load_from_param): Removed.
17627 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
17628 with stmt. Reverted back to use of load_from_unmodified_param.
17630 2017-01-23 Martin Jambor <mjambor@suse.cz>
17633 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
17634 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
17635 field a pointer to garbage collected vector, mark lattices and
17636 ipcp_orig_node with GTY((skip)).
17637 (ipa_get_param_count): Adjust to descriptors being a pointer.
17638 (ipa_get_param): Likewise.
17639 (ipa_get_type): Likewise.
17640 (ipa_get_param_move_cost): Likewise.
17641 (ipa_set_param_used): Likewise.
17642 (ipa_get_controlled_uses): Likewise.
17643 (ipa_set_controlled_uses): Likewise.
17644 (ipa_is_param_used): Likewise.
17645 (ipa_node_params_t): Move into garbage collector. New methods insert
17647 (ipa_node_params_sum): Annotate wth GTY(()).
17648 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
17650 (ipa_load_from_parm_agg): Adjust declaration.
17651 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
17652 * ipa-profile.c (ipa_profile): Likewise.
17653 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
17654 (ipa_populate_param_decls): Make descriptors parameter garbage
17656 (ipa_dump_param): Adjust to descriptors being a pointer.
17657 (ipa_alloc_node_params): Likewise.
17658 (ipa_initialize_node_params): Likewise.
17659 (load_from_param_1): Make descriptors parameter garbage collected.
17660 (load_from_unmodified_param): Likewise.
17661 (load_from_param): Likewise.
17662 (ipa_load_from_parm_agg): Likewise.
17663 (ipa_node_params::~ipa_node_params): Removed.
17664 (ipa_free_all_node_params): Remove call to delete operator.
17665 (ipa_node_params_t::insert): New.
17666 (ipa_node_params_t::remove): Likewise.
17667 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
17668 copy known_csts and known_contexts vectors.
17669 (ipa_read_node_info): Adjust to descriptors being a pointer.
17670 (ipcp_modif_dom_walker): Make m_descriptors field garbage
17672 (ipcp_transform_function): Make descriptors variable garbage
17675 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
17677 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
17678 * config/i386/avx512dqintrin.h: Ditto.
17679 * config/i386/avx512fintrin.h: Ditto.
17680 * config/i386/i386.c: Handle new builtins.
17681 * config/i386/i386-builtin.def: Add new builtins.
17682 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
17683 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
17685 2017-01-23 Jakub Jelinek <jakub@redhat.com>
17686 Martin Liska <mliska@suse.cz>
17688 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
17689 * asan.c (asan_expand_poison_ifn): Support stores and use
17690 appropriate ASAN report function.
17691 * internal-fn.c (expand_ASAN_POISON_USE): New function.
17692 * internal-fn.def (ASAN_POISON_USE): Declare.
17693 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
17694 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
17695 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
17696 ASAN_POISON calls w/o LHS.
17697 * tree-ssa.c (execute_update_addresses_taken): Create clobber
17698 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
17699 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
17700 * gimplify.c (asan_poison_variables): Add attribute
17701 use_after_scope_memory to variables that really needs to live
17703 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
17704 having the attribute.
17706 2017-01-23 Martin Liska <mliska@suse.cz>
17708 * asan.c (create_asan_shadow_var): New function.
17709 (asan_expand_poison_ifn): Likewise.
17710 * asan.h (asan_expand_poison_ifn): New declaration.
17711 * internal-fn.c (expand_ASAN_POISON): Likewise.
17712 * internal-fn.def (ASAN_POISON): New builtin.
17713 * sanopt.c (pass_sanopt::execute): Expand
17714 asan_expand_poison_ifn.
17715 * tree-inline.c (copy_decl_for_dup_finish): Make function
17717 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
17718 * tree-ssa.c (is_asan_mark_p): New function.
17719 (execute_update_addresses_taken): Rewrite local variables
17720 (identified just by use-after-scope as addressable) into SSA.
17722 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
17724 * doc/install.texi (Specific): opensource.apple.com uses https
17725 now. Remove trailing slash.
17727 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
17729 * README.Portability: Remove note on an Irix compatibility issue.
17731 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
17733 * gcov.c (INCLUDE_ALGORITHM): Define.
17734 (INCLUDE_VECTOR): Define.
17735 No longer include <vector> and <algorithm> directly.
17737 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
17739 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
17741 * doc/invoke.texi (Code Gen Options): Ditto.
17743 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
17746 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
17748 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
17750 rtl-optimization/79125
17751 * cprop.c (local_cprop_pass): Handle cases where we make an
17752 unconditional trap.
17754 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
17758 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
17759 read from, for big endian.
17761 2017-01-20 Jiong Wang <jiong.wang@arm.com>
17763 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
17764 register pauth builtins for LP64 only.
17766 2017-01-20 Marek Polacek <polacek@redhat.com>
17769 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
17772 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17774 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
17776 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
17777 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
17778 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
17780 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17783 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
17784 in big-endian mode when they are not a single duplicated value.
17786 2017-01-20 Richard Biener <rguenther@suse.de>
17788 * BASE-VER: Bump to 7.0.1.
17790 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17792 * omp-low.c (omplow_simd_context): New struct. Use it...
17793 (lower_rec_simd_input_clauses): ...here and...
17794 (lower_rec_input_clauses): ...here to hold common data. Adjust all
17795 references to idx, lane, max_vf, is_simt.
17797 2017-01-20 Graham Markall <graham.markall@embecosm.com>
17799 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
17802 2017-01-20 Martin Jambor <mjambor@suse.cz>
17804 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
17805 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
17807 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
17808 (GTFILES): Rename hsa.c to hsa-common.c.
17809 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
17810 * hsa-dump.c: Likewise.
17811 * hsa-gen.c: Likewise.
17812 * hsa-regalloc.c: Likewise.
17813 * ipa-hsa.c: Likewise.
17814 * omp-expand.c: Likewise.
17815 * omp-low.c: Likewise.
17816 * toplev.c: Likewise.
17818 2017-01-20 Marek Polacek <polacek@redhat.com>
17821 * doc/invoke.texi: Document -Wduplicated-branches.
17822 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
17823 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
17824 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
17825 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
17826 return 0 only when not OEP_LEXICOGRAPHIC.
17827 (fold_build_cleanup_point_expr): Use the expression
17828 location when building CLEANUP_POINT_EXPR.
17829 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
17830 * tree.c (add_expr): Handle error_mark_node.
17832 2017-01-20 Martin Liska <mliska@suse.cz>
17835 * tree-profile.c (init_ic_make_global_vars): Do not call
17837 (gimple_init_gcov_profiler): Likewise.
17839 2017-01-20 Martin Liska <mliska@suse.cz>
17842 * cgraph.h (maybe_create_reference): Remove argument and
17844 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
17846 * ipa-cp.c (create_specialized_node): Likewise.
17847 * symtab.c (symtab_node::maybe_create_reference): Handle
17848 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
17850 2017-01-20 Martin Liska <mliska@suse.cz>
17852 * read-rtl-function.c (function_reader::create_function): Use
17853 build_decl instread of build_decl_stat.
17855 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
17857 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
17858 * config/i386/avx512dqintrin.h: Ditto.
17859 * config/i386/avx512fintrin.h: Ditto.
17860 * config/i386/i386-builtin-types.def: Add new types.
17861 * config/i386/i386.c: Handle new types.
17862 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
17863 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
17864 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
17865 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
17866 (__builtin_ia32_kshiftridi): New.
17867 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
17869 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
17873 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
17874 define to rs6000_init_stack_protect_guard.
17875 (rs6000_init_stack_protect_guard): New function.
17877 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17878 Yunqiang Su <yunqiang.su@imgtec.com>
17880 * config.gcc (supported_defaults): Add madd4.
17881 (with_madd4): Add validation.
17882 (all_defaults): Add madd4.
17883 * config/mips/mips.opt (mmadd4): New option.
17884 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17886 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17888 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
17889 (ISA_HAS_FUSED_MADD4): Likewise.
17890 * doc/invoke.texi (-mmadd4): Document the new option.
17891 * doc/install.texi (--with-madd4): Document the new option.
17893 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17895 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
17896 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
17897 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
17898 (aarch64_init_pauth_hint_builtins): New.
17899 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
17900 (aarch64_expand_builtin): Expand new builtins.
17902 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17904 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
17905 * combine-stack-adj.c (no_unhandled_cfa): Handle
17906 REG_CFA_TOGGLE_RA_MANGLE.
17907 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
17908 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
17909 info for return address signing.
17910 (aarch64_expand_epilogue): Likewise.
17912 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17914 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
17915 * config/aarch64/aarch64-protos.h
17916 (aarch64_return_address_signing_enabled): New declaration.
17917 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
17919 (aarch64_expand_prologue): Sign return address before it's pushed onto
17921 (aarch64_expand_epilogue): Authenticate return address fetched from
17923 (aarch64_override_options): Sanity check for ILP32 and ISA level.
17924 (aarch64_attributes): New function attributes for "sign-return-address".
17925 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
17926 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
17927 ("*do_return"): Generate combined instructions according to key index.
17928 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
17929 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
17931 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
17932 * config/aarch64/aarch64.opt (msign-return-address=): New.
17933 * doc/extend.texi (AArch64 Function Attributes): Documents
17934 "sign-return-address=".
17935 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
17937 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17939 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
17940 overall option summary.
17942 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17944 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
17945 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
17946 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
17947 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
17949 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
17951 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
17952 -mpower9-minmax by default for -mcpu=power9.
17953 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
17954 128-bit floating point.
17956 2017-01-20 Alan Modra <amodra@gmail.com>
17958 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
17959 optimizing for size.
17961 2017-01-20 Alan Modra <amodra@gmail.com>
17964 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
17965 for strcmp and strncmp from corresponding builtin decl.
17967 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17969 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
17970 instead of i386/rtems-64.h.
17971 * config/i386/rtems-64.h: Remove.
17973 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17977 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
17979 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
17981 2017-01-19 Tamar Christina <tamar.christina@arm.com>
17983 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
17984 Change int to HOST_WIDE_INT.
17985 * config/aarch64/aarch64-protos.h
17986 (aarch64_simd_gen_const_vector_dup): Likewise.
17987 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
17989 2017-01-19 David Malcolm <dmalcolm@redhat.com>
17991 * langhooks-def.h (lhd_type_for_size): New decl.
17992 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
17993 * langhooks.c (lhd_type_for_size): New function, taken from
17996 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
17998 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
17999 define_bypass for CR latency.
18000 (power9-cracked-alu): Update bypass latency and remove power9-branch.
18001 (power9-alu2): Add define_bypass for CR latency.
18003 (power9-mul): Update insn latency.
18004 (power9-mul-compare): Update insn latency, bypass latency and remove
18007 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18009 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
18011 * config/aarch64/aarch64.md
18012 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
18013 aarch64_nopcrelative_literal_loads.
18014 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
18016 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
18018 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
18019 TARGET_LOONGSON_3A.
18020 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
18022 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
18025 * config.gcc (supported_defaults): Add lxc1-sxc1.
18026 (with_lxc1_sxc1): Add validation.
18027 (all_defaults): Add lxc1-sxc1.
18028 * config/mips/mips.opt (mlxc1-sxc1): New option.
18029 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
18031 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
18032 __mips_no_lxc1_sxc1.
18033 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
18034 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
18035 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
18037 2017-01-19 Richard Biener <rguenther@suse.de>
18039 PR tree-optimization/72488
18040 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
18041 sure to restore SSA info.
18042 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
18044 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
18046 PR rtl-optimization/79121
18047 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
18048 of the inner type when shifting an extended value.
18050 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
18053 * symtab.c (symtab_node::equal_address_to): Fix comparing of
18054 interposable aliases.
18056 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
18059 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
18060 Use the evmergelohi instruction.
18061 (mov_si<mode>_e500_subreg4_2_le): Likewise.
18062 (mov_sitf_e500_subreg8_2_be): Likewise.
18063 (mov_sitf_e500_subreg12_2_le): Likewise.
18064 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
18065 (mov_si<mode>_e500_subreg4_2_be): Likewise.
18066 (mov_sitf_e500_subreg8_2_le): Likewise.
18067 (mov_sitf_e500_subreg12_2_be): Likewise.
18069 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18071 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
18072 attribute from vecsimple to vecperm.
18073 (altivec_vbpermq2): Likewise.
18075 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18078 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
18080 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18081 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
18082 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
18083 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
18084 case where N arg is SIZE_MAX.
18085 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
18086 (cmpstrsi): Add pattern.
18088 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18090 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18091 __builtin_vec_revb builtins.
18092 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
18093 built-in functions to support generation of the ISA 3.0 XXBR<x>
18094 vector byte reverse instructions.
18095 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
18096 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
18097 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
18098 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
18099 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
18100 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
18101 (P9V_BUILTIN_VEC_REVB): Likewise.
18102 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
18103 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
18104 (p9_xxbrq_v16qi): Likewise.
18105 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
18106 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
18107 (p9_xxbrh_v8hi): Likewise.
18108 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
18109 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
18110 vec_revb built-in functions.
18112 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
18114 PR rtl-optimization/78952
18115 * config/i386/i386.md (any_extract): New code iterator.
18116 (*insvqi_2): Use any_extract for source operand.
18117 (*insvqi_3): Use any_shiftrt for source operand.
18119 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
18121 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
18123 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
18125 2017-01-18 Matthias Klose <doko@ubuntu.com>
18127 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
18129 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18131 * config/rs6000/altivec.h (vec_bperm): Change #define.
18132 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
18133 (altivec_vbpermq2): New define_insn.
18134 (altivec_vbpermd): Likewise.
18135 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
18136 function interface.
18137 (VBPERMD): Likewise.
18138 (VBPERM): New polymorphic function interface.
18139 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
18140 Add entries for P9V_BUILTIN_VEC_VBPERM.
18141 * doc/extend.texi: Add interfaces for vec_bperm.
18143 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18145 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
18146 first letter of error messages.
18147 (s390_resolve_overloaded_builtin): Likewise.
18148 * config/s390/s390.c (s390_expand_builtin): Likewise.
18149 (s390_invalid_arg_for_unprototyped_fn): Likewise.
18150 (s390_valid_target_attribute_inner_p): Likewise.
18151 * config/s390/s390.md ("tabort"): Likewise.
18153 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
18155 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
18156 (ISA_AVOID_DIV_HILO): New macro.
18157 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
18158 (ISA_HAS_DDIV): Likewise.
18160 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
18162 * doc/invoke.texi (fabi-version): Correct number of occurrences.
18164 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
18166 * doc/invoke.texi (fabi-version): Spelling fix.
18168 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
18171 * doc/invoke.texi (fabi-version): Mention mangling fix for
18174 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
18177 * doc/invoke.texi (fabi-version): Document discriminator mangling.
18179 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
18182 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
18183 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
18185 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
18187 (stack_protect_test): Ditto.
18188 * config/rs6000/rs6000.opt (mstack-protector-guard=,
18189 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
18191 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
18192 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
18193 -mstack-protector-guard-offset=.
18194 (RS/6000 and PowerPC Options): Ditto.
18196 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
18198 * config/i386/i386.h (MASK_CLASS_P): New define.
18199 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
18200 there are no registers from different register sets also when
18201 mask registers are used. Update function comment.
18202 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
18203 to (*k/*r) and (*k/*km) alternatives.
18205 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
18207 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
18208 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
18209 (EH_RETURN_HANDLER_RTX): New define.
18210 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
18211 Force frame pointer in EH return functions.
18212 (aarch64_expand_epilogue): Add barrier for eh_return.
18213 (aarch64_final_eh_return_addr): Remove.
18214 (aarch64_eh_return_handler_rtx): New function.
18215 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
18217 (aarch64_eh_return_handler_rtx): New prototype.
18219 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18221 * config/rs6000/altivec.h (vec_rlmi): New #define.
18222 (vec_vrlnm): Likewise.
18223 (vec_rlnm): Likewise.
18224 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
18225 (UNSPEC_VRLNM): Likewise.
18226 (VIlong): New mode iterator.
18227 (altivec_vrl<VI_char>mi): New define_insn.
18228 (altivec_vrl<VI_char>nm): Likewise.
18229 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
18231 (VRLDNM): Likewise.
18232 (RLNM): New polymorphic function entry.
18233 (VRLWMI): New monomorphic function entry.
18234 (VRLDMI): Likewise.
18235 (RLMI): New polymorphic function entry.
18236 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
18237 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
18238 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
18241 2017-01-17 Jakub Jelinek <jakub@redhat.com>
18244 * dwarf2out.c (field_byte_offset): Restore the
18245 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
18246 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
18247 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
18250 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
18253 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
18255 2017-01-17 Jakub Jelinek <jakub@redhat.com>
18258 * dwarf2out.c (add_data_member_location_attribute): For constant
18259 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
18260 instead of DW_AT_data_member_location, DW_AT_bit_offset and
18261 DW_AT_byte_size attributes.
18263 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
18265 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
18266 after forcing to constant memory when the code model is medium.
18268 2017-01-17 Julia Koval <julia.koval@intel.com>
18271 * config/i386/avx512fintrin.h
18272 (_mm512_i32gather_ps): Change __addr type to void const*.
18273 (_mm512_mask_i32gather_ps): Ditto.
18274 (_mm512_i32gather_pd): Ditto.
18275 (_mm512_mask_i32gather_pd): Ditto.
18276 (_mm512_i64gather_ps): Ditto.
18277 (_mm512_mask_i64gather_ps): Ditto.
18278 (_mm512_i64gather_pd): Ditto.
18279 (_mm512_mask_i64gather_pd): Ditto.
18280 (_mm512_i32gather_epi32): Ditto.
18281 (_mm512_mask_i32gather_epi32): Ditto.
18282 (_mm512_i32gather_epi64): Ditto.
18283 (_mm512_mask_i32gather_epi64): Ditto.
18284 (_mm512_i64gather_epi32): Ditto.
18285 (_mm512_mask_i64gather_epi32): Ditto.
18286 (_mm512_i64gather_epi64): Ditto.
18287 (_mm512_mask_i64gather_epi64): Ditto.
18288 (_mm512_i32scatter_ps): Change __addr type to void*.
18289 (_mm512_mask_i32scatter_ps): Ditto.
18290 (_mm512_i32scatter_pd): Ditto.
18291 (_mm512_mask_i32scatter_pd): Ditto.
18292 (_mm512_i64scatter_ps): Ditto.
18293 (_mm512_mask_i64scatter_ps): Ditto.
18294 (_mm512_i64scatter_pd): Ditto.
18295 (_mm512_mask_i64scatter_pd): Ditto.
18296 (_mm512_i32scatter_epi32): Ditto.
18297 (_mm512_mask_i32scatter_epi32): Ditto.
18298 (_mm512_i32scatter_epi64): Ditto.
18299 (_mm512_mask_i32scatter_epi64): Ditto.
18300 (_mm512_i64scatter_epi32): Ditto.
18301 (_mm512_mask_i64scatter_epi32): Ditto.
18302 (_mm512_i64scatter_epi64): Ditto.
18303 (_mm512_mask_i64scatter_epi64): Ditto.
18304 * config/i386/avx512pfintrin.h
18305 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
18306 (_mm512_mask_prefetch_i32gather_ps): Ditto.
18307 (_mm512_mask_prefetch_i64gather_pd): Ditto.
18308 (_mm512_mask_prefetch_i64gather_ps): Ditto.
18309 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
18310 (_mm512_prefetch_i32scatter_ps): Ditto.
18311 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
18312 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
18313 (_mm512_prefetch_i64scatter_pd): Ditto.
18314 (_mm512_prefetch_i64scatter_ps): Ditto.
18315 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
18316 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
18317 * config/i386/avx512vlintrin.h
18318 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
18319 (_mm_mmask_i32gather_ps): Ditto.
18320 (_mm256_mmask_i32gather_pd): Ditto.
18321 (_mm_mmask_i32gather_pd): Ditto.
18322 (_mm256_mmask_i64gather_ps): Ditto.
18323 (_mm_mmask_i64gather_ps): Ditto.
18324 (_mm256_mmask_i64gather_pd): Ditto.
18325 (_mm_mmask_i64gather_pd): Ditto.
18326 (_mm256_mmask_i32gather_epi32): Ditto.
18327 (_mm_mmask_i32gather_epi32): Ditto.
18328 (_mm256_mmask_i32gather_epi64): Ditto.
18329 (_mm_mmask_i32gather_epi64): Ditto.
18330 (_mm256_mmask_i64gather_epi32): Ditto.
18331 (_mm_mmask_i64gather_epi32): Ditto.
18332 (_mm256_mmask_i64gather_epi64): Ditto.
18333 (_mm_mmask_i64gather_epi64): Ditto.
18334 (_mm256_i32scatter_ps): Change __addr type to void*.
18335 (_mm256_mask_i32scatter_ps): Ditto.
18336 (_mm_i32scatter_ps): Ditto.
18337 (_mm_mask_i32scatter_ps): Ditto.
18338 (_mm256_i32scatter_pd): Ditto.
18339 (_mm256_mask_i32scatter_pd): Ditto.
18340 (_mm_i32scatter_pd): Ditto.
18341 (_mm_mask_i32scatter_pd): Ditto.
18342 (_mm256_i64scatter_ps): Ditto.
18343 (_mm256_mask_i64scatter_ps): Ditto.
18344 (_mm_i64scatter_ps): Ditto.
18345 (_mm_mask_i64scatter_ps): Ditto.
18346 (_mm256_i64scatter_pd): Ditto.
18347 (_mm256_mask_i64scatter_pd): Ditto.
18348 (_mm_i64scatter_pd): Ditto.
18349 (_mm_mask_i64scatter_pd): Ditto.
18350 (_mm256_i32scatter_epi32): Ditto.
18351 (_mm256_mask_i32scatter_epi32): Ditto.
18352 (_mm_i32scatter_epi32): Ditto.
18353 (_mm_mask_i32scatter_epi32): Ditto.
18354 (_mm256_i32scatter_epi64): Ditto.
18355 (_mm256_mask_i32scatter_epi64): Ditto.
18356 (_mm_i32scatter_epi64): Ditto.
18357 (_mm_mask_i32scatter_epi64): Ditto.
18358 (_mm256_i64scatter_epi32): Ditto.
18359 (_mm256_mask_i64scatter_epi32): Ditto.
18360 (_mm_i64scatter_epi32): Ditto.
18361 (_mm_mask_i64scatter_epi32): Ditto.
18362 (_mm256_i64scatter_epi64): Ditto.
18363 (_mm256_mask_i64scatter_epi64): Ditto.
18364 (_mm_i64scatter_epi64): Ditto.
18365 (_mm_mask_i64scatter_epi64): Ditto.
18366 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
18367 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
18368 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
18369 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
18370 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
18371 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
18372 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
18373 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
18374 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
18375 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
18376 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
18377 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
18378 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
18379 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
18380 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
18381 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
18382 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
18383 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
18384 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
18385 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
18386 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
18387 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
18388 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
18389 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
18390 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
18391 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
18392 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
18393 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
18394 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
18395 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
18396 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
18397 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
18398 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
18399 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
18400 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
18401 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
18402 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
18403 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
18404 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
18405 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
18406 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
18407 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
18408 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
18409 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
18410 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
18411 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
18412 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
18413 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
18414 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
18415 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
18416 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
18417 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
18418 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
18419 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
18420 definitions accordingly.
18422 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
18423 Kuan-Lin Chen <kuanlinchentw@gmail.com>
18426 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
18429 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
18432 * ira-conflicts.c (ira_build_conflicts): Update total conflict
18433 hard regs for inner regno.
18435 2017-01-17 Martin Liska <mliska@suse.cz>
18438 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
18439 assumption and add comment.
18441 2017-01-17 Nathan Sidwell <nathan@acm.org>
18443 * ipa-visibility.c (localize_node): New function, broken out of ...
18444 (function_and_variable_visibility): ... here. Call it.
18446 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
18448 PR middle-end/77445
18449 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
18450 correctly set frequency of oudgoing edge.
18451 (duplicate_thread_path): Fix profile updating.
18453 2017-01-17 Jakub Jelinek <jakub@redhat.com>
18456 * configure.ac: Add GCC_BASE_VER.
18457 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
18458 version from BASE-VER file.
18459 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
18460 (gcc.o): Depend on $(BASEVER).
18461 * common.opt (dumpfullversion): New option.
18462 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
18463 * doc/invoke.texi: Document -dumpfullversion.
18464 * doc/install.texi: Document --with-gcc-major-version-only.
18465 * configure: Regenerated.
18467 2017-01-17 Richard Biener <rguenther@suse.de>
18469 PR tree-optimization/71433
18470 * tree-vrp.c (register_new_assert_for): Merge same asserts
18471 on all incoming edges.
18472 (process_assert_insertions_for): Handle insertions at the
18475 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
18477 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
18478 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
18480 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
18483 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
18486 2017-01-17 Alan Modra <amodra@gmail.com>
18489 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
18490 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
18491 symbolic stack limit when pic.
18493 2017-01-16 Martin Sebor <msebor@redhat.com>
18495 PR tree-optimization/78608
18496 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
18498 2017-01-16 Jeff Law <law@redhat.com>
18501 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
18502 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
18503 for several include directories that may be relative to sysroot.
18504 * config/i386/x-mingw32 (gplus_includedir): Define.
18505 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
18506 (native_system_includedir): Likewise.
18507 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
18508 override if TARGET_SYSTEM_ROOT is defined.
18509 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
18511 PR tree-optimization/79090
18512 PR tree-optimization/33562
18513 PR tree-optimization/61912
18514 PR tree-optimization/77485
18515 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
18516 and computed trims into the dump file.
18518 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
18520 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
18522 2017-01-16 Jakub Jelinek <jakub@redhat.com>
18525 * gimplify.c (gimplify_init_constructor): If want_value and
18526 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
18530 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
18531 sequence. Formatting fixes.
18532 (doloop_optimize): Formatting fixes.
18535 * gcc.c (debug_level_greater_than_spec_func): New function.
18536 (static_spec_functions): Add debug-level-gt spec function.
18537 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
18539 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
18540 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
18541 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
18542 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
18543 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
18544 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
18546 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
18548 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
18549 QImode fixups to general and mask registers only.
18551 2017-01-16 Carl Love <cel@us.ibm.com>
18553 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
18554 for built-in functions
18555 vector signed char vec_nabs (vector signed char)
18556 vector signed short vec_nabs (vector signed short)
18557 vector signed int vec_nabs (vector signed int)
18558 vector signed long long vec_nabs (vector signed long long)
18559 vector float vec_nabs (vector float)
18560 vector double vec_nabs (vector double)
18561 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
18563 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
18564 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
18565 * doc/extend.texi: Update the documentation file for the new built-in
18568 2017-01-16 Martin Sebor <msebor@redhat.com>
18570 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
18573 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18575 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
18576 UNSPEC_VSX__XXSPLTD to require special splat handling.
18578 2017-01-16 David Malcolm <dmalcolm@redhat.com>
18581 * system.h: Poison strndup.
18583 2017-01-16 Alan Modra <amodra@gmail.com>
18586 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
18589 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
18591 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
18593 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
18595 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
18596 call recog here. Assert that INSN_CODE (insn) is non-negative.
18598 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
18601 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
18603 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
18604 in the currently scheduled RTL fragment.
18606 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
18608 PR rtl-optimization/78751
18609 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
18612 2017-01-14 Jeff Law <law@redhat.com>
18614 PR tree-optimization/79090
18615 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
18616 variable length stores.
18617 (compute_trims): Delete dead assignment to *trim_tail.
18618 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
18621 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
18623 PR rtl-optimization/78626
18624 PR rtl-optimization/78727
18625 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
18626 of a block, and split such blocks after everything else is finished.
18628 2017-01-14 Alan Modra <amodra@gmail.com>
18631 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
18632 target legitimate_combined_insn.
18633 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
18634 (rs6000_legitimate_combined_insn): New function.
18635 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
18637 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
18638 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
18639 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
18641 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
18643 * doc/frontends.texi (G++ and GCC): Remove references to Java.
18645 2017-01-13 Jeff Law <law@redhat.com>
18647 PR tree-optimization/33562
18648 PR tree-optimization/61912
18649 PR tree-optimization/77485
18650 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
18652 (delete_dead_assignment): Likewise.
18653 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
18654 statement to delete_dead_call and delete_dead_assignment.
18656 2017-01-13 David Malcolm <dmalcolm@redhat.com>
18659 * substring-locations.c (format_warning_va): Strengthen case 1 so
18660 that both endpoints of the substring must be within the format
18661 range for just the substring to be printed.
18663 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
18665 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
18666 * config/i386/i386.c (ix86_target_string): Add missing options
18667 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
18668 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
18669 flags_other and ix86_target_other to flags2_other. Display unknown
18671 (ix86_valid_target_attribute_inner_p): Add missing options and
18672 reorder options by implied ISAs, as in ix86_target_string.
18674 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
18676 * hash-table.h (hash_table::too_empty_p): New function.
18677 (hash_table::expand): Use it.
18678 (hash_table::traverse): Likewise.
18679 (hash_table::empty_slot): Use sizeof (value_type) instead of
18680 sizeof (PTR) to convert bytes to elements. Shrink the table
18681 if the current size is excessive for the current number of
18684 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
18686 * ira-costs.c (record_reg_classes): Break from the inner loop
18687 early once alt_fail is known to be true. Update outer loop
18688 handling accordingly.
18690 2017-01-13 Jeff Law <law@redhat.com>
18692 * tree-ssa-dse.c (decrement_count): New function.
18693 (increment_start_addr, maybe_trim_memstar_call): Likewise.
18694 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
18695 when we know the partially dead statement is a mem* function.
18697 PR tree-optimization/61912
18698 PR tree-optimization/77485
18699 * tree-ssa-dse.c: Include expr.h.
18700 (maybe_trim_constructor_store): New function.
18701 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
18703 PR tree-optimization/33562
18704 PR tree-optimization/61912
18705 PR tree-optimization/77485
18706 * doc/invoke.texi: Document new dse-max-object-size param.
18707 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
18708 * tree-ssa-dse.c: Include params.h.
18709 (dse_store_status): New enum.
18710 (initialize_ao_ref_for_dse): New, partially extracted from
18712 (valid_ao_ref_for_dse, normalize_ref): New.
18713 (setup_live_bytes_from_ref, compute_trims): Likewise.
18714 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
18715 (maybe_trim_partially_dead_store): Likewise.
18716 (maybe_trim_complex_store): Likewise.
18717 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
18718 Track what bytes live from the original store. Return tri-state
18719 for dead, partially dead or live.
18720 (dse_dom_walker): Add constructor, destructor and new private members.
18721 (delete_dead_call, delete_dead_assignment): New extracted from
18723 (dse_optimize_stmt): Make a member of dse_dom_walker.
18724 Use initialize_ao_ref_for_dse.
18726 PR tree-optimization/33562
18727 PR tree-optimization/61912
18728 PR tree-optimization/77485
18729 * sbitmap.h (bitmap_count_bits): Prototype.
18730 (bitmap_clear_range, bitmap_set_range): Likewise.
18731 * sbitmap.c (bitmap_clear_range): New function.
18732 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
18734 2017-01-13 Martin Liska <mliska@suse.cz>
18737 * function.c (set_cfun): Add new argument force.
18738 * function.h (set_cfun): Likewise.
18739 * ipa-inline-transform.c (inline_call): Use the function when
18740 strict alising from is dropped for function we inline to.
18742 2017-01-13 Richard Biener <rguenther@suse.de>
18744 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
18745 for dumping GIMPLE INTEGER_CSTs.
18747 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18749 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
18750 to 201112L since C++17.
18752 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18755 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
18756 if -fsanitize=kernel-address is present.
18758 2017-01-13 Richard Biener <rguenther@suse.de>
18760 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
18761 as _Literal ( type ) number in case usual suffixes do not
18762 preserve all information.
18764 2017-01-13 Richard Biener <rguenther@suse.de>
18766 PR tree-optimization/77283
18767 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
18768 and ssa-iterators.h.
18769 (is_feasible_trace): Implement a cost model based on joiner
18772 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18775 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
18776 char or short to __float128/_Float128 directly.
18778 2017-01-12 Martin Sebor <msebor@redhat.com>
18780 to -Wformat-overflow.
18781 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
18782 (min_bytes_remaining): Same.
18783 (get_string_length): Same.
18784 (format_string): Same.
18785 (format_directive): Same.
18787 (pass_sprintf_length::handle_gimple_call): Same.
18789 2017-01-12 Jakub Jelinek <jakub@redhat.com>
18791 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
18792 info.nowrite calls with no lhs that can't throw. Return bool
18793 whether gsi_remove has been called or not.
18794 (pass_sprintf_length::handle_gimple_call): Return bool whether
18795 try_substitute_return_value called gsi_remove. Formatting fix.
18796 (pass_sprintf_length::execute): Don't use gsi_remove if
18797 handle_gimple_call returned true.
18800 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
18801 be removed due to side-effects, don't remove following barrier nor
18802 turn the successor edge into fallthru edge.
18804 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18807 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
18808 element-reversing loads and stores as not swappable.
18810 2017-01-12 Nathan Sidwell <nathan@acm.org>
18811 Nicolai Stange <nicstange@gmail.com>
18813 * combine.c (try_combine): Don't ignore result of overlap checking
18814 loop. Combine overlap & asm check into single loop.
18816 2017-01-12 Richard Biener <rguenther@suse.de>
18818 * tree-pretty-print.c (dump_generic_node): Provide -gimple
18819 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
18821 2017-01-12 Richard Biener <rguenther@suse.de>
18823 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
18824 and TS_TARGET_OPTION directly derive from TS_BASE.
18825 * tree-core.h (tree_optimization_option): Derive from tree_base.
18826 (tree_target_option): Likewise.
18828 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
18830 * config/i386/i386.c (memory_address_length): Increase len
18831 only when rip_relative_addr_p returns false.
18833 2017-01-11 Julia Koval <julia.koval@intel.com>
18835 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
18836 (OPTION_MASK_ISA_SGX_SET): New.
18837 (ix86_handle_option): Handle OPT_msgx.
18838 * config.gcc: Added sgxintrin.h.
18839 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
18840 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
18841 * config/i386/i386.c (ix86_target_string): Add -msgx.
18843 (ix86_option_override_internal): Handle new options.
18844 (ix86_valid_target_attribute_inner_p): Add sgx.
18845 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
18846 * config/i386/i386.opt: Add msgx.
18847 * config/i386/sgxintrin.h: New file.
18848 * config/i386/x86intrin.h: Add sgxintrin.h.
18850 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18853 * fold-const.c (maybe_nonzero_address): Return 1 for function
18855 (tree_single_nonzero_warnv_p): Don't handle function local objects
18859 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
18862 2017-01-11 David Malcolm <dmalcolm@redhat.com>
18865 * opts.c: Include "spellcheck.h"
18866 (struct string_fragment): New struct.
18867 (struct edit_distance_traits<const string_fragment &>): New
18869 (get_closest_sanitizer_option): New function.
18870 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
18872 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18874 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
18876 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
18877 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
18878 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
18879 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
18880 for initial die_offset if dwarf_split_debug_info.
18881 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
18882 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
18884 (output_skeleton_debug_sections): Formatting fix. Use
18885 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
18886 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
18888 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
18890 * config/arm/cortex-a53.md: Add bypasses for
18891 cortex_a53_r2f_cvt.
18892 (cortex_a53_r2f): Only use for transfers.
18893 (cortex_a53_f2r): Likewise.
18894 (cortex_a53_r2f_cvt): Add reservation for conversions.
18895 (cortex_a53_f2r_cvt): Likewise.
18897 2017-01-11 Tamar Christina <tamar.christina@arm.com>
18899 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
18900 to all inlined functions, change static to extern.
18902 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
18905 * config/arm/arm.c (legitimize_pic_address): Handle reference to
18907 (arm_assemble_integer): Likewise.
18909 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
18911 * config.gcc: Use new awk script to check CPU, FPU and architecture
18912 parameters for --with-... options.
18913 * config/arm/parsecpu.awk: New file
18914 * config/arm/arm-cpus.in: New file.
18915 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
18917 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
18919 * config/arm/t-arm: Update dependency rules.
18920 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
18921 of processing .def files.
18922 * config/arm/genopt.sh: Deleted.
18923 * config/arm/gentune.sh: Deleted.
18924 * config/arm/arm-cores.def: Deleted.
18925 * config/arm/arm-arches.def: Deleted.
18926 * config/arm/arm-fpus.def: Deleted.
18927 * config/arm/arm-tune.md: Regenerated.
18928 * config/arm/arm-tables.opt: Regenerated.
18929 * config/arm/arm-cpu.h: New generated file.
18930 * config/arm/arm-cpu-data.h: New generated file.
18931 * config/arm/arm-cpu-cdata.h: New generated file.
18933 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
18936 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
18938 (input_varpool_node): Unpack dynamically_initialized bit.
18940 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
18942 PR rtl-optimization/79032
18943 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
18944 the alignment of the adjusted memory reference against that of MODE,
18945 instead of the alignment of the original memory reference.
18947 2017-01-11 Martin Jambor <mjambor@suse.cz>
18949 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
18951 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
18952 decorated functions.
18954 2017-01-11 Richard Biener <rguenther@suse.de>
18956 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
18957 set range/nonnull info for PHI results. Do not set it on
18958 stmts marked for removal.
18960 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
18962 * expr.c (store_field): In the bitfield case, fetch the return value
18963 from the registers before applying a single big-endian adjustment.
18964 Always do a final load for a BLKmode value not larger than a word.
18966 2017-01-10 David Malcolm <dmalcolm@redhat.com>
18969 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18970 that we correctly handle column numbers greater than
18971 LINE_MAP_MAX_COLUMN_NUMBER.
18973 2017-01-10 Martin Sebor <msebor@redhat.com>
18975 PR middle-end/78245
18976 * gimple-ssa-sprintf.c (get_destination_size): Call
18977 {init,fini}object_sizes.
18978 * tree-object-size.c (addr_object_size): Adjust.
18979 (pass_through_call): Adjust.
18980 (pass_object_sizes::execute): Adjust.
18981 * tree-object-size.h (fini_object_sizes): Declare.
18983 2017-01-10 Martin Sebor <msebor@redhat.com>
18985 PR tree-optimization/78775
18986 * builtins.c (get_size_range): Move...
18987 * calls.c: ...to here.
18988 (alloc_max_size): Accept zero argument.
18989 (operand_signed_p): Remove.
18990 (maybe_warn_alloc_args_overflow): Call get_size_range.
18991 * calls.h (get_size_range): Declare.
18993 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
18995 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
18996 from TI's devices.csv file as of September 2016.
18997 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
18999 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
19001 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
19002 * doc/invoke.texi: Likewise.
19003 * doc/md.texi: Likewise.
19004 * doc/objc.texi: Likewise.
19006 2017-01-10 Joshua Conner <joshconner@google.com>
19008 * config/arm/fuchsia-elf.h: New file.
19009 * config/fuchsia.h: New file.
19010 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
19011 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
19013 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
19015 2016-01-10 Richard Biener <rguenther@suse.de>
19017 PR tree-optimization/79034
19018 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
19019 Propagate out degenerate PHIs in the joiner.
19021 2017-01-10 Martin Liska <mliska@suse.cz>
19023 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
19024 (sort_congruence_classes_by_decl_uid): Likewise.
19025 (sort_congruence_class_groups_by_decl_uid): Likewise.
19026 (sem_item_optimizer::merge_classes): Sort class, groups in these
19027 classes and members in the groups by DECL_UID of declarations.
19028 This would make merge operations stable.
19030 2017-01-10 Martin Liska <mliska@suse.cz>
19032 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
19033 usage of m_classes_vec.
19034 (sem_item_optimizer::~sem_item_optimizer): Likewise.
19035 (sem_item_optimizer::get_group_by_hash): Likewise.
19036 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
19037 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
19038 (sem_item_optimizer::verify_classes): Likewise.
19039 (sem_item_optimizer::process_cong_reduction): Likewise.
19040 (sem_item_optimizer::dump_cong_classes): Likewise.
19041 (sem_item_optimizer::merge_classes): Likewise.
19042 * ipa-icf.h (congruence_class_hash): Rename from
19043 congruence_class_group_hash. Remove declaration of m_classes_vec.
19045 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
19047 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
19048 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
19049 * config.gcc: Add avx512vpopcntdqintrin.h.
19050 * config/i386/avx512vpopcntdqintrin.h: New.
19051 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
19052 * config/i386/i386-builtin-types.def: Add new types.
19053 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
19054 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
19055 __builtin_ia32_vpopcountq_v8di_mask): New.
19056 * config/i386/i386-c.c (ix86_target_macros_internal): Define
19057 __AVX512VPOPCNTDQ__.
19058 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
19059 (PTA_AVX512VPOPCNTDQ): Define.
19060 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
19061 TARGET_AVX512VPOPCNTDQ_P): Define.
19062 * config/i386/i386.opt: Add mavx512vpopcntdq.
19063 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
19064 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
19066 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19068 PR middle-end/77484
19069 * predict.def (PRED_CALL): Set to 67.
19071 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
19073 * expr.c (store_field): In the bitfield case, if the value comes from
19074 a function call and is of an aggregate type returned in registers, do
19075 not modify the field mode; extract the value in all cases if the mode
19076 is BLKmode and the size is not larger than a word.
19078 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
19081 * config/i386/cpuid.h: Fix undefined behavior.
19083 2017-01-04 Jeff Law <law@redhat.com>
19085 PR tree-optimization/79007
19086 PR tree-optimization/67955
19087 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
19088 conservative for pt.null when flag_non_call_exceptions is on.
19090 2017-01-09 Jakub Jelinek <jakub@redhat.com>
19092 PR translation/79019
19093 PR translation/79020
19094 * params.def (PARAM_INLINE_MIN_SPEEDUP,
19095 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
19096 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
19098 * config/avr/avr.opt (maccumulate-args): Likewise.
19099 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
19100 * common.opt (freport-bug): Likewise.
19101 * cif-code.def (CIF_FINAL_ERROR): Likewise.
19102 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
19103 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
19104 translatable string.
19105 * config/i386/i386.c (function_value_32): Likewise.
19106 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
19107 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
19109 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
19110 * common/config/msp430/msp430-common.c (msp430_handle_option):
19112 * symtab.c (symtab_node::verify_base): Likewise.
19113 * opts.c (set_debug_level): Likewise.
19114 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
19115 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
19116 missing whitespace to translatable strings.
19117 * config/avr/avr.md (bswapsi2): Fix typo in comment.
19118 * config/sh/superh.h: Likewise.
19119 * config/i386/xopintrin.h: Likewise.
19120 * config/i386/znver1.md: Likewise.
19121 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
19122 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
19123 * double-int.h (struct double_int): Likewise.
19124 * double-int.c (div_and_round_double): Likewise.
19125 * wide-int.cc: Likewise.
19126 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
19127 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
19128 * cfgcleanup.c (crossjumps_occured): Renamed to ...
19129 (crossjumps_occurred): ... this.
19130 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
19133 PR tree-optimization/78899
19134 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
19135 returning bool return struct loop *, NULL for failure and the new
19137 (versionable_outer_loop_p): Don't version outer loop if it has
19138 dont_vectorized bit set.
19139 (tree_if_conversion): When versioning outer loop, ensure
19140 tree_if_conversion is performed also on the inner loop of the
19141 non-vectorizable outer loop copy.
19142 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
19143 LOOP_VECTORIZED in inner loop of the scalar outer loop and
19144 prevent vectorization of it.
19145 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
19146 the outer loop vectorization of the non-scalar version is attempted
19147 before vectorization of the inner loop in scalar version. If
19148 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
19149 vectorization of its inner loop.
19150 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
19151 has 2 inner loops, rename also on edges from bb whose single pred
19152 is outer_loop->header. Fix typo in function comment.
19154 2017-01-09 Martin Sebor <msebor@redhat.com>
19157 * asan.c (asan_emit_stack_protection): Increase local buffer size
19158 to avoid snprintf truncation warning.
19160 2017-01-09 Andrew Pinski <apinski@cavium.com>
19162 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
19163 to reference thunderx2t99 for the tuning structure
19164 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
19166 (thunderx2t99_extra_costs): This.
19167 * config/aarch64/aarch64-tune.md: Regenerate.
19168 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
19169 (vulcan_addrcost_table): This.
19170 (vulcan_regmove_cost): Rename to ...
19171 (thunderx2t99_regmove_cost): This.
19172 (vulcan_vector_cost): Rename to ...
19173 (thunderx2t99_vector_cost): this.
19174 (vulcan_branch_cost): Rename to ...
19175 (thunderx2t99_branch_cost): This.
19176 (vulcan_tunings): Rename to ...
19177 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
19178 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
19180 2017-01-09 Martin Jambor <mjambor@suse.cz>
19184 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
19185 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
19186 (propagate_vr_accross_jump_function): Use the above function for all
19187 value range computations for pass-through jump functions and type
19188 converasion from explicit value range values.
19189 (ipcp_propagate_stage): Do not attempt to deduce types of formal
19190 parameters from TYPE_ARG_TYPES.
19191 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
19192 (ipa_write_node_info): Stream type of the actual argument.
19193 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
19195 2017-01-09 Martin Liska <mliska@suse.cz>
19198 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
19199 (lookup_compiler): Do not show error message with have_E.
19201 2017-01-09 Jakub Jelinek <jakub@redhat.com>
19203 PR tree-optimization/78938
19204 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
19205 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
19206 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
19207 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
19210 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19212 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
19215 2017-01-09 Richard Biener <rguenther@suse.de>
19217 PR tree-optimization/78997
19218 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
19219 name condition properly.
19221 2017-01-09 Richard Biener <rguenther@suse.de>
19224 * dwarf2out.c (is_cxx): New overload with context.
19225 (is_naming_typedef_decl): Use it.
19227 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
19229 * invoke.texi (Option Summary): Correct spacing in option lists
19230 and add line breaks to fix over-long lines.
19232 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
19234 PR middle-end/17660
19236 * extend.texi (Common Variable Attributes): Add xref to GCC
19237 Internals manual to explain mode attribute keywords.
19239 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
19242 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
19243 and Preprocessor Options.
19244 (Options for Linking): Document -pthread here....
19245 (RS/6000 and PowerPC Options): ...not here.
19246 (Solaris 2 Options): ...or here.
19247 * doc/cppopts.texi: Document -pthread.
19249 2017-01-08 Martin Sebor <msebor@redhat.com>
19251 PR middle-end/77708
19252 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
19253 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
19254 New member functions.
19255 (format_directive): Used them.
19257 (pass_sprintf_length::handle_gimple_call): Same.
19258 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
19259 to avoid truncation for any argument.
19260 (extract_affine_mul): Same.
19261 * tree.c (get_file_function_name): Same.
19263 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19265 PR middle-end/77484
19266 * predict.def (PRED_INDIR_CALL): Set to 86.
19268 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
19270 PR preprocessor/54124
19271 * doc/cppopts.texi: Reformat -d subtable to list the full name
19272 of the options. Add cross-reference to the docs for the general
19273 compiler -d options.
19274 * doc/invoke.texi (Developer Options): Add cross-reference to the
19275 preprocessor-specific -d option documentation.
19277 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
19279 PR preprocessor/13498
19280 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
19281 redudant material, and reflect new command-line options.
19282 (System Headers): Likewise.
19284 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
19286 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
19287 -isystem, and -idirafter. Copy-edit.
19288 * doc/cppopts.texi: Copy-edit. Remove contradiction about
19289 default for -ftrack-macro-expansion. Delete obsolete and
19290 badly-formatted implementation details about -fdebug-cpp output.
19291 * doc/cppwarnopts.texi: Copy-edit.
19293 2017-01-07 David Malcolm <dmalcolm@redhat.com>
19296 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
19297 that the transition from a max line width >= 1<<10 to narrower
19298 lines works correctly.
19300 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
19302 * doc/options.texi (PerFunction): New.
19303 * opt-functions.awk (switch_flags): Map both Optimization and
19304 PerFunction to CL_OPTIMIZATION.
19305 * opth-gen.awk: Test for PerFunction flag along with
19307 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
19308 it only when the latter is present. Skip those that don't in
19309 the hash function generator.
19310 * common.opt (fvar-tracking): Mark as PerFunction instead of
19312 (fvar-tracking-assignments): Likewise.
19313 (fvar-tracking-assignments-toggle): Likewise.
19314 (fvar-tracking-uninit): Likewise.
19316 2017-01-07 Jakub Jelinek <jakub@redhat.com>
19318 PR translation/79018
19319 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
19322 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
19325 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
19326 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
19327 TARGET_LONG_JUMP_TABLE_OFFSETS.
19328 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
19329 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
19330 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
19331 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
19332 * config/m68k/m68k.md (tablejump expander): Likewise.
19333 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
19334 TARGET_LONG_JUMP_TABLE_OFFSETS.
19335 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
19336 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
19338 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
19339 David Holsgrove <david.holsgrove@xilinx.com>
19341 * common/config/microblaze/microblaze-common.c
19342 (TARGET_EXCEPT_UNWIND_INFO): Remove.
19343 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
19345 * config/microblaze/microblaze.c (microblaze_must_save_register)
19346 (microblaze_expand_epilogue, microblaze_return_addr): Handle
19348 (microblaze_eh_return): New function.
19349 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
19350 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
19351 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
19352 * config/microblaze/microblaze.md (eh_return): New pattern.
19354 2017-01-06 Jakub Jelinek <jakub@redhat.com>
19356 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
19357 GCC_DIAGNOSTIC_STRINGIFY): Define.
19359 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
19361 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19363 * config/arm/arm.md (<mcrr>): New.
19365 * config/arm/arm.c (arm_arch5te): New.
19366 (arm_option_override): Set arm_arch5te.
19367 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
19369 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
19370 (arm_mcrr_qualifiers): ... this. New.
19371 (MRRC_QUALIFIERS): Define to...
19372 (arm_mrrc_qualifiers): ... this. New.
19373 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
19375 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
19376 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
19377 (MRRCI, mrrc, MRRC): New.
19378 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
19379 VUNSPEC_MRRC2): New.
19381 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19383 * config/arm/arm.md (<mcr>): New.
19385 * config/arm/arm.c (arm_coproc_builtin_available): Add
19386 support for mcr, mrc, mcr2 and mrc2.
19387 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
19388 (arm_mcr_qualifiers): ... this. New.
19389 (MRC_QUALIFIERS): Define to ...
19390 (arm_mrc_qualifiers): ... this. New.
19391 (MCR_QUALIFIERS): Define to ...
19392 (arm_mcr_qualifiers): ... this. New.
19393 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
19395 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
19396 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
19397 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
19398 VUNSPEC_MRC2): New.
19400 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19402 * config/arm/arm.md (*ldc): New.
19406 * config/arm/arm.c (arm_coproc_builtin_available): Add
19407 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
19408 (arm_coproc_ldc_stc_legitimate_address): New.
19409 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
19410 'qualifier_const_pointer'.
19411 (LDC_QUALIFIERS): Define to...
19412 (arm_ldc_qualifiers): ... this. New.
19413 (STC_QUALIFIERS): Define to...
19414 (arm_stc_qualifiers): ... this. New.
19415 * config/arm/arm-protos.h
19416 (arm_coproc_ldc_stc_legitimate_address): New.
19417 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
19418 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
19419 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
19420 stc2, stcl, stc2l): New.
19421 * config/arm/constraints.md (Uz): New.
19422 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
19423 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
19424 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
19425 VUNSPEC_STC2L): New.
19427 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19429 * config/arm/arm.md (<cdp>): New.
19430 * config/arm/arm.c (neon_const_bounds): Rename this ...
19431 (arm_const_bounds): ... this.
19432 (arm_coproc_builtin_available): New.
19433 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
19434 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
19435 (CDP_QUALIFIERS): Define to...
19436 (arm_cdp_qualifiers): ... this. New.
19438 (arm_expand_builtin_args): Add case for 6 arguments.
19439 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
19440 (arm_const_bounds): ... this.
19441 (arm_coproc_builtin_available): New.
19442 * config/arm/arm_acle.h (__arm_cdp): New.
19444 * config/arm/arm_acle_builtins.def (cdp): New.
19446 * config/arm/iterators.md (CDPI,CDP,cdp): New.
19447 * config/arm/neon.md: Rename all 'neon_const_bounds' to
19448 'arm_const_bounds'.
19449 * config/arm/types.md (coproc): New.
19450 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
19451 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
19452 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
19453 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
19455 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19457 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
19458 (UBINOP_QUALIFIERS): New.
19460 (acle_builtin_data): New. Change comment.
19461 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
19462 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
19463 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
19464 arm_acle_builtins.def.
19465 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
19466 (arm_init_acle_builtins): New.
19467 (CRC32_BUILTIN): Remove.
19468 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
19469 crc32cb, crc32ch and crc32cw.
19470 (arm_init_crc32_builtins): Remove.
19471 (arm_init_builtins): Use arm_init_acle_builtins rather
19472 than arm_init_crc32_builtins.
19473 (arm_expand_acle_builtin): New.
19474 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
19475 * config/arm/arm_acle_builtins.def: New.
19477 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19479 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
19480 (arm_builtin_datum): ... this.
19481 (arm_init_neon_builtin): Rename to ...
19482 (arm_init_builtin): ... this. Add a new parameters PREFIX
19483 and USE_SIG_IN_NAME.
19484 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
19485 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
19486 'arm_builtin_datum'.
19487 (arm_init_vfp_builtins): Likewise.
19488 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
19489 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
19490 (arm_expand_neon_args): Rename to ...
19491 (arm_expand_builtin_args): ... this. Rename builtin_arg
19492 enum values and differentiate between ARG_BUILTIN_MEMORY
19493 and ARG_BUILTIN_NEON_MEMORY.
19494 (arm_expand_neon_builtin_1): Rename to ...
19495 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
19496 values, arm_expand_builtin_args and add bool parameter NEON.
19497 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
19498 (arm_expand_vfp_builtin): Likewise.
19499 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
19501 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19503 PR middle-end/77484
19504 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
19505 * predict.c (tree_estimate_probability_bb): Reverse direction of
19506 polymorphic call predictor.
19508 2017-01-06 David Malcolm <dmalcolm@redhat.com>
19510 * passes.c (execute_one_pass): Split out pass-skipping logic into...
19511 (determine_pass_name_match): ...this new function and...
19512 (should_skip_pass_p): ...this new function.
19514 2017-01-06 Nathan Sidwell <nathan@acm.org>
19516 * ipa-visibility.c (function_and_variable_visibility): Reformat
19517 comments and long lines. Remove extrneous if.
19518 * symtab.c (symtab_node::make_decl_local): Fix code format.
19519 (symtab_node::set_section_for_node): Fix comment typo.
19521 2017-01-06 Martin Liska <mliska@suse.cz>
19524 * lra-constraints.c: Rename invariant to lra_invariant.
19525 * predict.c (set_even_probabilities): Initialize e to NULL.
19527 2017-01-05 Martin Sebor <msebor@redhat.com>
19529 PR tree-optimization/78910
19530 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
19531 (format_integer): Correct off-by-one error in the handling
19532 of precision with negative numbers in signed conversions..
19534 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
19536 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
19538 2017-01-05 Jakub Jelinek <jakub@redhat.com>
19540 PR tree-optimization/71016
19541 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
19542 factor_out_conditional_conversion. Formatting fix.
19543 (factor_out_conditional_conversion): Add cond_stmt argument.
19544 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
19545 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
19548 2017-01-05 David Malcolm <dmalcolm@redhat.com>
19550 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
19551 read-rtl-function.o, and selftest-rtl.o.
19552 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
19553 (selftest::aarch64_test_loading_full_dump): New function.
19554 (selftest::aarch64_run_selftests): New function.
19555 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
19556 selftest::aarch64_run_selftests.
19557 * config/i386/i386.c
19558 (selftest::ix86_test_loading_dump_fragment_1): New function.
19559 (selftest::ix86_test_loading_call_insn): New function.
19560 (selftest::ix86_test_loading_full_dump): New function.
19561 (selftest::ix86_test_loading_unspec): New function.
19562 (selftest::ix86_run_selftests): Call the new functions.
19563 * emit-rtl.c (maybe_set_max_label_num): New function.
19564 * emit-rtl.h (maybe_set_max_label_num): New decl.
19565 * function.c (instantiate_decls): Guard call to
19566 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
19567 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
19569 * gensupport.c (gen_reader::gen_reader): Pass "false"
19570 for new "compact" param of rtx_reader.
19571 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
19572 rather than an empty string for NULL strings.
19573 * read-md.c: Potentially include config.h rather than bconfig.h.
19574 Wrap include of errors.h with #ifdef GENERATOR_FILE.
19575 (have_error): New global, copied from errors.c.
19576 (md_reader::read_name): Rename to...
19577 (md_reader::read_name_1): ...this, adding "out_loc" param,
19578 and converting "missing name or number" to returning false, rather
19580 (md_reader::read_name): Reimplement in terms of read_name_1.
19581 (md_reader::read_name_or_nil): New function.
19582 (md_reader::read_string): Handle "(nil)" by returning NULL.
19583 (md_reader::md_reader): Add new param "compact".
19584 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
19585 (md_reader::read_file): New method.
19586 * read-md.h (md_reader::md_reader): Add new param "compact".
19587 (md_reader::read_file): New method.
19588 (md_reader::is_compact): New accessor.
19589 (md_reader::read_name): Convert return type from void to file_location.
19590 (md_reader::read_name_or_nil): New decl.
19591 (md_reader::read_name_1): New decl.
19592 (md_reader::m_compact): New field.
19593 (noop_reader::noop_reader): Pass "false" for new "compact" param
19595 (rtx_reader::rtx_reader): Add new "compact" param.
19596 (rtx_reader::read_rtx_operand): Make virtual and convert return
19597 type from void to rtx.
19598 (rtx_reader::read_until): New decl.
19599 (rtx_reader::handle_any_trailing_information): New virtual function.
19600 (rtx_reader::postprocess): New virtual function.
19601 (rtx_reader::finalize_string): New virtual function.
19602 (rtx_reader::m_in_call_function_usage): New field.
19603 (rtx_reader::m_reuse_rtx_by_id): New field.
19604 * read-rtl-function.c: New file.
19605 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
19606 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
19607 (selftest::verify_three_block_rtl_cfg): New decl.
19608 * read-rtl-function.h: New file.
19609 * read-rtl.c: Potentially include config.h rather than bconfig.h.
19610 For host, include function.h, memmodel.h, and emit-rtl.h.
19611 (one_time_initialization): New function.
19612 (struct compact_insn_name): New struct.
19613 (compact_insn_names): New array.
19614 (find_code): Handle insn codes in compact dumps.
19615 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
19616 (bind_subst_iter_and_attr): Likewise.
19617 (add_condition_to_string): Likewise.
19618 (add_condition_to_rtx): Likewise.
19619 (apply_attribute_uses): Likewise.
19620 (add_current_iterators): Likewise.
19621 (apply_iterators): Likewise.
19622 (initialize_iterators): Guard usage of apply_subst_iterator with
19623 #ifdef GENERATOR_FILE.
19624 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
19625 (md_reader::read_mapping): Likewise.
19626 (add_define_attr_for_define_subst): Likewise.
19627 (add_define_subst_attr): Likewise.
19628 (read_subst_mapping): Likewise.
19629 (check_code_iterator): Likewise.
19630 (rtx_reader::read_rtx): Likewise. Move one-time initialization
19632 (one_time_initialization): New function.
19633 (rtx_reader::read_until): New method.
19634 (read_flags): New function.
19635 (parse_reg_note_name): New function.
19636 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
19637 Handle reuse_rtx ids.
19638 Wrap iterator lookup within #ifdef GENERATOR_FILE.
19639 Add parsing support for RTL dumps, mirroring the special-cases in
19640 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
19641 values, and calling handle_any_trailing_information.
19642 (rtx_reader::read_rtx_operand): Convert return type from void
19643 to rtx, returning return_rtx. Handle case 'e'. Call
19644 finalize_string on XSTR and XTMPL fields.
19645 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
19646 "(nil)" values were omitted. Call the postprocess vfunc on the
19648 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
19649 class ctor. Initialize m_in_call_function_usage. Call
19650 one_time_initialization.
19651 * rtl-tests.c (selftest::test_uncond_jump): Call
19652 set_new_first_and_last_insn.
19653 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
19654 * selftest-rtl.c: New file.
19655 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
19656 (selftest::get_insn_by_uid): New decl.
19657 * selftest-run-tests.c (selftest::run_tests): Call
19658 read_rtl_function_c_tests.
19659 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
19660 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
19663 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
19665 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
19666 operands in a special way. Assert that pos+len <= mode precision.
19668 2017-01-05 Jakub Jelinek <jakub@redhat.com>
19670 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
19671 3 argument Alias with unlimited for the negative form.
19672 (fno-vect-cost-model): Removed.
19674 2017-01-05 Martin Liska <mliska@suse.cz>
19676 * hsa-gen.c (gen_hsa_divmod): New function.
19677 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
19679 2017-01-05 Martin Liska <mliska@suse.cz>
19682 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
19685 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19687 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
19688 small constant length operands.
19690 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19692 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
19693 between loop iterations.
19695 2017-01-05 Martin Liska <mliska@suse.cz>
19698 * gimplify.c (gimplify_decl_expr): Compare to
19699 asan_poisoned_variables instread of checking flags.
19700 (gimplify_target_expr): Likewise.
19701 (gimplify_expr): Likewise.
19702 (gimplify_function_tree): Conditionally initialize
19703 asan_poisoned_variables.
19705 2017-01-04 Jeff Law <law@redhat.com>
19707 PR tree-optimizatin/78812
19708 * rtl.h (contains_mem_rtx_p): Prototype.
19709 * ifcvt.c (containts_mem_rtx_p): Move from here to...
19710 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
19711 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
19712 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
19713 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
19715 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19717 * input.c (assert_char_at_range): Default-initialize actual_range.
19719 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19721 * df-scan.c (df_ref_create_structure): Make regno unsigned,
19722 to match the caller.
19724 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19726 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
19727 insns after final jump in test to emit dummy move.
19729 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19731 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
19732 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
19734 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19736 * multiple_target.c (create_dispatcher_calls): Init e_next.
19737 * tree-ssa-loop-split.c (split_loop): Init border.
19738 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
19741 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
19746 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
19747 (altivec_register_operand): Do not return true if the operand
19748 contains a SUBREG mixing SImode and SFmode.
19749 (vsx_register_operand): Likewise.
19750 (vsx_reg_sfsubreg_ok): New predicate.
19751 (vfloat_operand): Do not return true if the operand contains a
19752 SUBREG mixing SImode and SFmode.
19753 (vint_operand): Likewise.
19754 (vlogical_operand): Likewise.
19755 (gpc_reg_operand): Likewise.
19756 (int_reg_operand): Likewise.
19757 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
19758 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
19759 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
19761 (rs6000_emit_move_si_sf_subreg): New helper function.
19762 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
19763 fixup SUBREGs involving SImode and SFmode.
19764 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
19765 numbers for the new peephole2 optimization.
19766 (peephole2 for SFmode unions): New peephole2 to optimize cases in
19767 the GLIBC math library that do AND/IOR/XOR operations on single
19768 precision floating point.
19769 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
19770 target macros to say whether we need to avoid SUBREGs mixing
19772 (TARGET_ALLOW_SF_SUBREG): Likewise.
19773 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
19774 (UNSPEC_SI_FROM_SF): Likewise.
19775 (iorxor): Change spacing.
19776 (and_ior_xor): New iterator for AND, IOR, and XOR.
19777 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
19778 (movdi_from_sf_zero_ext): Likewise.
19779 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
19780 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
19781 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
19782 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
19783 (fms<mode>4): Likewise.
19784 (fnma<mode>4): Likewise.
19785 (fnms<mode>4): Likewise.
19786 (nfma<mode>4): Likewise.
19787 (nfms<mode>4): Likewise.
19789 2017-01-04 Marek Polacek <polacek@redhat.com>
19792 * doc/invoke.texi: Document -Wpointer-compare.
19794 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19796 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
19799 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
19800 descriptions for -gdwarf-5 and emit them as uleb128 instead of
19803 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
19806 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
19807 documentation of the powerpc_popcntb_ok attribute.
19808 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19809 code to issue warning messages if a requested CPU configuration is
19810 not supported by the binary (assembler and loader) toolchain.
19811 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
19812 made to define a built-in function that has been disabled.
19813 (paired_init_builtins): Add assertion to prevent ICE if attempt is
19814 made to define a built-in function that has been disabled.
19815 (altivec_init_builtins): Add comment explaining why definition
19816 of the DST built-in functions is not preceded by an assertion
19817 check. Add assertions to prevent ICE if attempts are made to
19818 define an altivec predicate or an abs* built-in function that has
19820 (htm_init_builtins): Add comment explaining why definition of the
19821 htm built-in functions is not preceded by an assertion check.
19823 2017-01-04 Jeff Law <law@redhat.com>
19825 PR tree-optimizatin/67955
19826 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
19827 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
19828 the points-to solution does not include pt_null. Use DECL_PT_UID
19831 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
19833 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
19834 Use gen_int_mode instead of gen_lopwart for const_int operands.
19836 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19838 PR tree-optimization/71563
19839 * match.pd: Simplify X << Y into X if Y is known to be 0 or
19840 out of range value - has low bits known to be zero.
19842 2017-01-04 Alan Modra <amodra@gmail.com>
19844 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
19845 * configure: Regenerate.
19846 * config.in: Regenerate.
19848 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19851 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
19852 a substring of the message, but strcmp with the whole message. Ifdef
19853 ENABLE_NLS, translate the message first using dgettext.
19855 2017-01-03 Jeff Law <law@redhat.com>
19857 PR tree-optimizatin/78856
19858 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
19859 (mark_threaded_blocks): Remove code to truncate thread paths that
19860 cross multiple loop headers. Instead invalidate the cached loop
19861 iteration information and handle case of a thread path walking
19862 into an irreducible region.
19864 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
19867 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
19868 assertions. Add support for doing the signbit if the IEEE 128-bit
19869 floating point value is in a GPR.
19870 * config/rs6000/rs6000.md (Fsignbit): Delete.
19871 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
19872 Update the length attribute if the value is in a GPR.
19873 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
19874 the sign or zero extension instruction, since the value is always 0/1.
19875 (signbit<mode>2_dm2): Delete using <Fsignbit>.
19878 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
19879 extracting SImode to a GPR register so that we can generate a
19880 store, limit the vector to be in a traditional Altivec register
19881 for the vextuwrx instruction.
19883 2017-01-03 Ian Lance Taylor <iant@google.com>
19885 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
19887 2017-01-03 Martin Sebor <msebor@redhat.com>
19889 PR tree-optimization/78696
19890 * gimple-ssa-sprintf.c (format_floating): Correct handling of
19891 precision. Use MPFR for %f for greater fidelity. Correct handling
19893 (pass_sprintf_length::compute_format_length): Set width and precision
19894 specified by asrerisk to void_node for vararg functions.
19895 (try_substitute_return_value): Adjust dump output.
19897 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
19899 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
19901 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
19903 * doc/invoke.texi (SPARC options): Document -mlra as the default.
19904 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
19905 -mlra/-mno-lra was passed to the compiler.
19907 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
19909 PR rtl-optimization/65618
19910 * emit-rtl.c (try_split): Move initialization of "before" and
19911 "after" to just before the call to emit_insn_after_setloc.
19913 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
19915 * doc/md.texi (Standard Names): Remove reference to Java frontend.
19917 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
19919 * dwarf2out.c (gen_enumeration_type_die): When
19920 -gno-strict-dwarf, add a DW_AT_encoding attribute.
19922 2017-01-03 Jakub Jelinek <jakub@redhat.com>
19924 PR tree-optimization/78965
19925 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
19926 Change first argument from const call_info & to call_info &. For %n
19927 set info.nowrite to false.
19929 PR middle-end/78901
19930 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
19931 possibly throwing calls.
19933 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
19934 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
19935 and fns handling, rather than in a separate case SSA_NAME.
19937 2017-01-02 Jeff Law <law@redhat.com>
19939 * config/darwin-driver.c (darwin_driver_init): Const-correctness
19940 fixes for first_period and second_period variables.
19942 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
19945 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
19946 (*insvqi_1): New insn pattern.
19947 (*insvqi_1_mem_rex64): Ditto.
19948 (*insvqi_2): Ditto.
19949 (*insvqi_3): Rename from *insvqi.
19951 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
19953 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
19955 * doc/cfg.texi (Edges): Remove reference to Java.
19956 (Maintaining the CFG): Ditto.
19958 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19960 PR middle-end/77674
19961 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
19962 transparent aliases.
19964 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19966 PR middle-end/77484
19967 * predict.def (PRED_CALL): Update hitrate.
19968 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
19969 * predict.c (tree_estimate_probability_bb): Split CALL predictor
19970 into direct/indirect/polymorphic variants.
19972 2017-01-01 Jakub Jelinek <jakub@redhat.com>
19974 Update copyright years.
19976 * gcc.c (process_command): Update copyright notice dates.
19977 * gcov-dump.c (print_version): Ditto.
19978 * gcov.c (print_version): Ditto.
19979 * gcov-tool.c (print_version): Ditto.
19980 * gengtype.c (create_file): Ditto.
19981 * doc/cpp.texi: Bump @copying's copyright year.
19982 * doc/cppinternals.texi: Ditto.
19983 * doc/gcc.texi: Ditto.
19984 * doc/gccint.texi: Ditto.
19985 * doc/gcov.texi: Ditto.
19986 * doc/install.texi: Ditto.
19987 * doc/invoke.texi: Ditto.
19989 Copyright (C) 2017 Free Software Foundation, Inc.
19991 Copying and distribution of this file, with or without modification,
19992 are permitted in any medium without royalty provided the copyright
19993 notice and this notice are preserved.